Dividend Smoothness, Determinants and Impact of Dividend
Announcements on share prices: Empirical Evidence from Jordan
Bashar Khaled Abu Khalaf
Submitted for the degree of Doctor of Philosophy
Heriot-Watt University
School of Management and Languages
October 2012
The copyright in this thesis is owned by the author. Any quotation from the
thesis or use of any of the information contained in it must acknowledge
this thesis as the source of the quotation or information.
i
ABSTRACT
Dividends are the portion of the company’s earnings paid to the shareholders. Many
researchers have used different estimation techniques to estimate the smoothness of
dividends, determinants of dividends and the impact of dividend announcements on
share prices. They used these estimation techniques in order to understand better the
reason that companies smooth their dividends, be able to know what determines the
dividend decisions and determine if the market reaction to the dividend announcements
can be predicted. However, these investigations have been unsuccessful in finding an
agreement on the above three issues. Furthermore, the concentration of such empirical
investigations has been on the developed markets; rather few studies have been
conducted on the developing markets such as Jordan.
This study has tried to fill three main gaps in the previous literature by empirically
investigating the smoothness of dividends, the determinants of the two dividend
decisions and the impact of dividend announcements on share prices. In addition,
different empirical methods have been used in the three empirical chapters: firstly, fixed
and random effects techniques are used to check the smoothness of dividends; secondly,
the results of three estimation techniques (OLS, Tobit and Heckman’s simultaneous
technique) are compared to estimate the determinants of the two dividend decisions;
finally, the standard OLS event study methodology is compared with a hybrid method
(OLS/EGARCH) to check the impact of the dividend announcements on share prices.
The results of the first empirical chapter confirmed that the non-financial Jordanian
companies smooth their dividends in a moderate rate. In addition, our results contradict
the signalling theory; we find that large companies smooth their dividend faster than
small ones. Furthermore, in line with the agency cost theory; low leveraged firms
smooth their dividends faster than high leveraged firms. Also, our results confirmed
that highly profitable companies smooth their dividend more and this comes in line with
the signalling theory. The second empirical chapter concluded that Hechman’s model is
the better technique to estimate the determinants of the two dividend decisions. The
main determinants are profitability, growth, leverage and size.
The last empirical
chapter showed that the Hybrid method provided a higher level of significance and
ii
concluded that dividend has no impact on share prices on the dividend announcements
day. However, the results suggested that there was a significant information leakage
prior to the announcements. In addition, the market reacts significantly one day after
the dividend announcements and the third day.
iii
DEDICATION
This thesis is dedicated
…to my parents
…to my sisters, wife and daughter
For their endless support, love and encouragement
iv
ACKNOWLEDGEMENT
The accomplishment of this dissertation has been one of the most significant academic
challenges I have ever had to face. Without the support, patience and guidance of the
following people, this study would have not been possible.

Dr. Julian Fennema, who undertook to act as my first supervisor despite his
many other academic and professional commitments, his wisdom, knowledge
and commitment to the highest standards inspired and motivated me.
Dr.
Fennema offered invaluable supervision. Thanks for pushing me right through
the years to finish my research.

Dr. Robert Mochrie, who accepted to act as my second supervisor, his
encouragement, knowledge, patience and insightful discussions helped a lot all
the way through my PhD.

Dr. David Brown, former senior lecturer in Heriot Watt University, for the
insights he has shared even after starting his new job. In addition, I would like to
thank Dr. Mohamed Sherif and Dr. Ibrahim Boulis for their constant support.

Professor Ghassan Omet, for his continual support and his absolute confidence
in me.

The staff of the Student Support and Accommodation Office especially, the
director, Christine Johnston for the encouragement and help that I will never
forget.

I sincerely acknowledge the financial support from the University of Jordan.

I am indebted, for life, to my family for their sacrifice, love and sharing the load
of PhD study.

Farah Abu Khalaf, my daughter, who was born before this dissertation was
completed and who spent many days with relatives to allow me to focus. I am
deeply sorry for the time we spent apart.

My special thanks go to my office mate for many years, Eleni Chatzivgeri. She
was a true friend since we began to share an office in 2008. She has been a great
source of practical information, as well as being happy to be the first to hear my
outrage or glee at the day's current events.

I also extend my thanks to all my friends for their continual support and
encouragement. Special thanks go to Dr. Ziad Zurigat, Hesham Abdel Haleem
and Ismail AbuShaikha.
v
DECLARATION STATEMENT
ACADEMIC REGISTRY
Research Thesis Submission
Name:
School/PGI:
Version: (i.e. First,
Resubmission,
Final)
Degree Sought
(Award and
Subject area)
Declaration
In accordance with the appropriate regulations I hereby submit my thesis and I declare that:
1)
the thesis embodies the results of my own work and has been composed by myself
2)
where appropriate, I have made acknowledgement of the work of others and have made reference
to work carried out in collaboration with other persons
3)
the thesis is the correct version of the thesis for submission and is the same version as any
electronic versions submitted*.
4)
my thesis for the award referred to, deposited in the Heriot-Watt University Library, should be
made available for loan or photocopying and be available via the Institutional Repository, subject
to such conditions as the Librarian may require
5)
I understand that as a student of the University I am required to abide by the Regulations of the
University and to conform to its discipline.
*
Please note that it is the responsibility of the candidate to ensure that the correct version of the
thesis is submitted.
Signature of
Candidate:
Date:
Submission
Submitted By (name in capitals):
Signature of Individual Submitting:
Date Submitted:
For Completion in the Student Service Centre (SSC)
Received in the SSC by (name in
capitals):
Method of Submission
(Handed in to SSC; posted through
internal/external mail):
E-thesis Submitted (mandatory for
final theses)
Signature:
Date:
vi
List of Contents
Abstract
ii
Dedication
iv
Acknowledgements
v
Declaration Statement
vi
List of Contents
vii
Tables
ix
Chapter 1 – Introduction And Overview
1.1 Introduction
1.2 The Country
1.2.1 The government
1.2.2 Amman Securities Exchange
1.2.3 Growth
1.3 Motivation And Contribution of the Study
1.4 Objectives And Significance of the Study
1.5 Research Questions of the Study
1.6 Thesis Structure
1
1
2
2
3
5
6
9
10
11
Chapter 2 – Literature Review
2.1 Introduction
2.2 Dividend Policy Theories
2.2.1 Dividend Irrelevance Theory
2.2.2 Signalling Theory
2.2.3 Agency-Cost Hypothesis
2.2.4 Bird-In-Hand Theory
2.2.5 Tax Effect Hypothesis
2.3 Empirical Evidence
2.3.1 Empirical evidence of the smoothness of dividends
2.3.2 Empirical evidence of the determinants of dividends
2.3.3 Empirical evidence about the impact of
dividend announcement on share prices
2.4 Conclusion
13
13
14
14
16
19
23
24
27
27
31
35
39
Chapter 3 – Empirical Investigation of the Smoothness of Dividend
41
3.1 Introduction
41
3.2 Theoretical Background
43
3.3 Partial Adjustment Models
46
3.3.1 Symmetric Adjustment model
46
3.3.2 Asymmetric Adjustment model
48
3.4 Estimation Results
53
3.4.1 Descriptive Statistics
53
3.4.2 Estimation Results of model 3.1 (The symmetric Adjustment model)
54
3.4.3 Estimation Results of model 3.2 (The Asymmetric Adjustment model)
57
3.4.4 Estimation Results of model 3.3 (The Asymmetric partial model including the
interaction terms)
58
vii
3.4.4.1 The impact of large/small size
61
3.4.4.2 The impact of high/low leverage
63
3.4.4.3 The impact of high/low profitability
65
3.5 Conclusion
66
Chapter 4 –Empirical Investigation of the Determinants of the Two Dividend
Decisions
68
4.1 Introduction
68
4.2 Determinants of the two Dividend Decisions
69
4.2.1 Determinants of Dividend Decision- theoretical background and definition 69
4.2.1.1 Profitability
70
4.2.1.2 Size
71
4.2.1.3 Leverage
72
4.2.1.4 Ownership Structure
74
4.2.1.5 Growth Prospects
75
4.3 Model Specification
77
4.4 Statistical Analysis
83
4.4.1 Descriptive Statistics
84
4.4.2 Correlation Matrix
86
4.4.3 Comparison of results (OLS, Tobit, Heckman)
88
4.5 Conclusion
97
Chapter 5 – Empirical Investigation of the Impact of Dividend Announcements on
Share Prices
99
5.1 Introduction
99
5.2 Theoretical background
100
5.3 Model Specification
104
5.3.1 Standard Event Study
107
5.3.1.1 The event day
108
5.3.1.2 The event window
109
5.3.1.3 The estimation period
110
5.3.2 Hybrid Method (OLS/EGARCH)
113
5.3.3 Parametric and Non-parametric Tests
116
5.3.3.1 Parametric Test
116
5.3.3.2 Non-parametric Test
117
5.4 Statistical Analysis
117
5.4.1 Descriptive Statistics
118
5.4.2 Results from unit root tests
120
5.4.3 Discussion of results
120
5.5 Conclusion
130
Chapter 6 – Conclusion
133
6.1 Introduction
133
6.2 Key Findings and Implications
133
6.3 Limitations of the Study
135
6.4 Recommendations for future research
137
viii
Tables
Table 1.1: Relative Size of Arab Stock Markets
Table 3.1: The impact of different corporate theories on the smoothness of dividends
Table 3.2: Descriptive statistics
Table 3.3: Estimation results of the symmetric adjustment model
Table 3.4: Selected studies from literature
Table 3.5: Estimation Results of the asymmetric adjustment model
Table 3.6: Estimation Results of the asymmetric adjustment model with firm
characteristics interaction (profitability, leverage and size)
Table 4.1: The proxy used for the different variables, definitions and the expected
impact on the dividend decisions
Table 4.2: Descriptive Statistics: Dependent and Independent Variables
Table 4.3: Average Cross-Sectional Correlation Coefficients of variables
Table 4.4: Estimation results
Table 5.1: Implications of different Corporate Finance Theories in predicting the impact
of dividend announcements on share prices
Table 5.2: Summary of the Expected Average Abnormal Return
Table 5.3: Selected studies from literature
Table 5.4: Descriptive statistics
Table 5.5: Summary of the Expected Abnormal return, Actual Average Abnormal return
and the findings
Table 5.6: Results of the standards event study using the OLS market model and the
hybrid method using OLS/EGARCH
Table 5.7: Reason for opening and implications of different holding periods
Table 5.8: Summary of the Cumulative Abnormal Returns (CAR) for selected holding
periods around the announcement day
ix
1. Introduction And Overview
1.1 Introduction
Although the issue of 'corporate dividend policy' has for a long period of time been an
area of concern in the related financial literature and has been one of the major concerns
by financial scholars since the early 1960s, it remains one of the most puzzling and
interesting topics of research. Corporate dividend policy has attracted many research
papers both in theory and practice, which would indicate in fact how this issue was one
of the toughest challenges which corporate finance researchers were facing. This issue
was described in fact as a “puzzle with pieces that just do not fit together” (Black,
1976) and “one of the ten important unsolved issues in corporate finance” (Brealey
et.al., 2006). The different theories that emerged to answer the different questions of
dividend policy have resulted in a large number of empirical and theoretical research
papers but no consensus has been achieved (Allen & Michaely, 1995) i.e. three schools
of thought have been developed on whether dividend policies have an influence on
firms' values. The first school was led by Miller & Modigliani (1961) who set out the
'irrelevance theory', denoting that a firm’s value does not get affected by dividends. The
other two schools proclaim generally that dividends do affect the value of the firm either
positively or negatively.
In addition, the bulk of research that has empirically investigated the dividend policy
has been conducted in the context of developed countries, in particular, the UK and the
USA. Relatively speaking, there are few empirical investigations that have examined
dividend policy in developing countries like Jordan; where the capital market is not big
compared to developed markets. This is along with the legal and statutory differences,
which have made it hard to take a broad view of the outcomes obtained from any
empirical studies in the context of any of the developing markets, enlarging the need to
address this dividend policy issue. Lately, many of these developing countries have
tried to liberalise their capital markets, which would give their firms’ managers more
flexibility in selecting their companies’ capital structure and, therefore, affect their
dividend policies.
Jordanian companies, in this respect, have followed a model of financial liberalization
in order to give confidence to the foreign investment and make their capital market part
1
of the world’s market. As a result, new laws and regulations-such as the Transparency
Act 1998, the Companies Act 1998, and the Security Act 1997-were actually set up to
make this market operate in accordance with certain international standards.
In
addition, this had a positive impact on the performance and the efficiency of the
Jordanian market.
The ongoing significance of dividend payments today is an
interesting and, also, important phenomenon. In an attempt to unveil this puzzle, this
study examines this controversial issue in the Jordanian context and specifically in the
industrial and services sectors, which have not been examined in prior Jordanian
academic studies.
1.2 The Country
Jordan presents a large variety of landscapes. Its size is 89,342 kilometres squared. It is
strategically positioned at the intersection between Europe, Asia and Africa.
The
geographic centre of Jordan including the mountains and hills of the badia expands
eastwards into Saudi Arabia and Iraq. The Jordan River flows on a course of the fertile
rift valley, from the Western borders of Jordan into the Dead Sea. It is known to be the
lowest point on earth where the bulk of water slanders at 400 meter below sea level.
The access to the red sea is via the port of Aqaba in Jordan. A large number of
civilization have attempted to organize the strategic geography of Jordan; the
Mesopotamian, Sumerian and other empires to the South; besides, parts of Jordan have
been under the rule of Greek, Roman and Persian classical civilizations. Since the 7th
century, Arab and Islamic dynasties have ruled Jordan. From the 1920's, Jordan was
located under the British mandate in what was called Transjordan; in 1946 this name
expired and now the country is called the Hashemite kingdom of Jordan and governed
as a legitimate kingdom.
1.2.1 The Government
King Abdullah II is the head of the military and the most powerful political person in
the Hashemite Kingdom of Jordan which is a constitutional monarchy with a
representative government. His royal highness selects the councils of ministers that are
held responsible to a two- house parliament. 55 members of the upper house are
2
selected by the king himself, while 110 deputies of the lower house are selected by
popular vote. The Jordanian citizen’s rights and duties guarantee the right of free
worship, culture, private property, opinion and the press. The element of the Jordanian
political landscape has established responsibility to liberalization and an agreement of
greater democracy since the year 1989. Jordan has thus been on a stable path of
increasing democracy. The involvement of the Jordanian citizen in the community life
of the country has been of great importance to the stability of the nation and will be of
assistance in the country’s future. The political stability as well as the social climate
that has been present in Jordan continues to thrive under his majesty King Abdullah II.
King Husain Ben Talal’s legacy of reform has been conducted by his majesty, with the
view to commit the Kingdom to the aims of the capital market development as well as
privatization in addition to economic liberalization; furthermore, modernization of the
law has been important.
1.2.2 Amman Securities Exchange
The 1930s witnessed the first phase of establishing public shareholding corporations in
Jordan. Since the early 1930s, the public has had the chance to subscribe for and deal in
shares of these companies, while the early 1960s saw the beginning of the first issue of
bonds in Jordan (Amman Stock Exchange, 2002).
In the past, capital market exchanges were not constituted as organizational systems
since individual brokerage officers were handling purchase and sale transactions. It was
evident in the 1960s that there was a need for the establishment of a well-organized
market using the appropriate financial technology to govern and control these
transactions and Amman Financial Market (AFM) was established in 1976 as a result of
this need. The AFM was opened to the public on January, 1, 1978.
The AFM undertook two major functions; a stock exchange and a government securities
and exchange commission. It was a governmental tool to govern and manage the
Jordanian securities market. The main objective of the AFM was to mobilize savings
by promoting and encouraging investments to stimulate the national economy, to
regulate the issue of securities and to control the securities transactions in Jordan. The
AFM was charged with promoting and developing the securities market, to regulate and
3
manage the operations and activities of member firms dealing with securities as
underwriters or brokers.
In addition, the AFM undertook the responsibility of
regulating market trading.
A new securities law was passed in 1997, when the roles of the operator and regulator
were separated. According to this law, three financial entities were established;
1- The Jordan Securities Commission (JSC), which is a government agency
affiliated to the Prime Minister’s office and is the control and regulatory body
for the capital market in Jordan.
2- The Amman Stock Exchange (ASE) is the formal market for securities
transactions in Jordan.
This entity assumes a legal personality and is
independent financially and in management terms. The entity was set up as a
private sector non-profit making organization and it involves all financial
brokers. The transactions in this entity began in March of 1999 and it launched
its e-transactions in the 26th of March, 2000. These e-transactions replaced the
traditional forms of face-to-face transactions and the e-systems have organized
trading procedures and transmit deals to the securities depository centre for
settlement, thus promoting safety procedures and the efficiency of transactions.
3- The Securities Depository Centre (SDC). This centre regulates, monitors and
controls all deposits and transfers of securities between different parties traded
on the stock exchange.
financial brokers.
It also regulates settlement of stock prices among
It is a non-profit organization and is run by a board of
managers from the private sector. The United States of America is now helping
the centre to establish a high level clearing and settlement, depository and share
registration system. The private sector, non-profit entity of the ASE has a legal
and financial independence and runs the market. Although it is a non-profit
entity, the SDC monitors and controls settlements and maintenance of securities
ownership documents.
4
1.2.3 Growth
Since formation, the Amman Stock Market (ASM) has witnessed impressive growth.
For instance, and by the end of 2011, ASM listed companies totalled 227. Table 1.1
shows the different number of firms listed in the Arab stock markets. It also shows
figures on the market capitalization for each market in relation to other markets in the
region. The figures show that the largest number of listed companies is found in the
Jordanian market (ASE). As for the ASM it can be noted that it is the largest Arab
stock market with respect to the total number of companies listed (227). In addition, the
ASM capitalization is larger than the Bahraini, Palestinian, Tunisian, Omani and
Lebanese stock markets. Table 1.1 shows that the Saudi market has the highest market
capitalization of 38.3 % among all the capitalizations found in the Arab stock markets.
It is worth mentioning that the Egyptian Exchange Market size has decreased from 8.6
% of total capitalization in 2010 to about 5.4 % of total capitalization in 2011 which is a
result in the decrease of the index of this market.
Market
2010
2011
2010
2011
No. of firms
Market Capitalization
Abu Dhabi Securities Market
64
67
7.2
7.3
Amman Stock Exchange
247
277
3.1
3.3
Bahrain Stock Exchange
49
49
2.0
1.9
Saudi Stock Market
146
150
35.9
38.3
Kuwait Stock Exchange
214
216
12.6
11.4
Casablanca Stock Exchange
75
76
7.0
6.8
Tunis Stock Exchange
56
57
1.2
1.1
Dubai Financial Market
65
62
5.5
5.6
Palestine Stock Exchange
40
46
0.3
0.3
Muscat Securities Market
119
130
2.2
2.2
Qatar Exchange
43
42
12.6
14.5
Beirut Stock Exchange
24
25
1.8
1.9
Egyptian Exchanges
212
214
8.6
5.4
100
100
Total
Table 1.1: Relative Size of Arab Stock Markets (Arab Monetary Fund, 2011)
5
Furthermore, the assets of the listed Jordanian companies in the market have risen from
9 billion JD in 2000 to more than 28 billion JD in 2010, and these figures are a
significant sign of the ASM’s contribution in the Jordanian national economy. Finally,
it is worth noting that while figures in 2000 indicate that the total net income of all
listed companies in the ASM was only 250 million JD, this figure was more than 700
million JD by the end of 2010. The 2011 report prepared by the Arab Monetary Fund
revealed that the foreign investments in Jordan have increased dramatically throughout
the years. It is clear that these increases have had positive effects on the total cash
dividends distributed by the Jordanian companies listed in the ASM for the shareholders
(Arab Monetary Fund, 2011).
1.3 Motivation and Contribution of the Study
In this study there are three motivations which constitute the contribution of this
empirical study;
Firstly, Lintner (1956) stated that companies are mainly concerned with having a stable
dividend policy and companies will only increase dividends when the managers are
feeling confident that the company’s earnings can maintain the increase in the future.
However, dividends may be used by stockholders as a means of paying for their own
expenses; thus, the in-stability of dividends may raise concerns and cause them to view
shares as risky. In some cases, companies decrease dividends with a view to raising
internal funds available for future investments, which might be perceived as an incorrect
signal by investors (Woolridge & Ghosh, 1985). Thus, the share price of the company
might fall due to the fact that the reduction in dividends might be perceived as bad
future earnings of the company. This implies that if companies want to maximise their
stock price, they have to find the right balance between retaining their funds and paying
dividends. Based on Lintners’ partial adjustment model (Lintner, 1956) the existence of
the adjustment cost may prevent the companies’ ability to return back to their targets
instantly. Consequently, the partial adjustment mechanism permits the given firms’
observed dividend payment ratio to be not always equal to its target level. This means
that companies change their dividend, or adjust to their target, if the costs of getting
closer to the target dividend payout ratio are lower than the costs of staying away from
6
the target. The reason is rational, implying that dividend payments are not instantly
adjusted to their target level until the benefit of moving towards that target at least
outweighs its costs.
Despite the fact that much research has been done using the partial adjustment model
for the investigation of the smoothness of each company’s dividend policy, no evidence
is provided on the effect of the various company characteristics that may affect the
smoothness of dividends. That is to say, no research has been done using Symmetric
and Asymmetric partial adjustment models for the investigation of the impact of
different company characteristics such as size, profitability and leverage on the
smoothness of dividends. Therefore, the present thesis adds to the literature by firstly
using the symmetric and asymmetric partial adjustment model and secondly by
investigating the effect of different firm characteristics on the asymmetric partial
adjustment model. The previous studies assumed a symmetric adjustment toward the
target dividend payout ratio, and as mentioned above no study focused on the effects of
firm characteristics on the dividends’ smoothness. More specifically, the present thesis
investigates whether large/small sized companies, high/low leveraged companies and
high/low profitable companies have an asymmetric adjustment behaviour towards a
target dividend payout ratio.
Secondly, many papers have tried to explore the factors that affect the dividend decision
but no theory has identified the different variables that have an effect on dividends. The
empirical results so far are mixed. According to Brealey et.al. (2006), the mixed results
may be attributed to the fact that dividend decisions are usually associated with other
financing decisions; thus, it is difficult to understand the whole picture. Here lies the
motivation of chapter 4 of the present study. Chapter 4 of the present thesis identifies
one more possible reason for the mixed results presented within the literature. Most of
the empirical studies do not take into consideration that the dividend decision process
includes two stages: the payment decision (i.e., to pay or not to pay) and the amount
decision (i.e., how much to pay). Chapter four uses three estimation techniques (OLS,
Tobit and Heckman’s simultaneous model) with the view to identifying the best model
in determining the different factors that affect the two dividend decisions of the
Jordanian companies.
The simultaneous estimation method of Heckman (1979) is
applied in the present study due to the fact that it takes into account the possibility that a
7
variety of different determinants may affect, at the same time, each stage of the process
of the dividend decision.
In addition, Heckman’s simultaneous model has the
advantage of eliminating any sample selection bias that could arise from the fact that
many companies do not pay dividends.
Thirdly, many theories have been developed in an effort to explain the possible effects
of the dividend announcement on companies’ share prices, since the main target of all
companies is to keep their dividends smooth and increase their share price. Therefore,
several papers have tried to explore the effect of dividend announcements on share price
suing various estimation techniques (Nirmala et.al., 2011). Despite the fact that many
researchers have dealt with the effect of dividend announcements on share prices, the
results are mixed and our understanding is limited (Ali & Chowdhury, 2010).
The present thesis adds to the literature by identifying two reasons for the mixed results
obtained within the literature. The first reason for the mixed results can be attributed to
the fact that most of the studies that tried to investigate the effect of a dividend
announcement on share price have used the standard OLS market model. This model
assumes that the variance in the disturbance term is constant but in real life, share price
data usually suffer from heteroskedasticity in disturbances (Brown & Warner, 1985).
The debate in the previous empirical papers focused on the identification of the best
normal return model for the prediction of the variance and the mean of the return series
(MacKinlay, 1997). The present study is the first study that applies the OLS/EGARCH
hybrid method and compares it to the standard OLS method in order to examine the
effect of dividend announcements on share prices in the Jordanian market, using
companies from the industrial and services sectors. That is to say, the present thesis
adds to the previous empirical research by allowing EGARCH in a hybrid method to
treat the heteroskedasticity in the disturbance term.
8
1.4 Objectives and Significance of the Study
The topic of dividend policy is essential in determining the value of corporations and
has a clear effect on their performance. The value of a firm is affected by the amount of
dividend paid; hence, investors are in turns expected to inquire about them. Dividends
are used as an indication to assist the investors in their decision to invest in any firm,
just as a dividend policy influences and is influenced by the financing and investing
decisions.
However, and based on the literature recorded to date in this respect,
research on dividend policy and dividend decisions were found to be scarce in
developing markets.
The Jordanian Market is regarded as one of those emerging markets, which actually
highlights an important aspect of this study. The present study derives its significance
from the fact that it tackled an issue of great importance to Jordanian investors, which is
mainly 'dividends'; furthermore, it focuses attention onto the most prominent variables
underpinning these investors' decision. Moreover, the results of the tests generated
should be beneficial and of use in guiding the investors into developing an 'expectations
framework' on the dividends to be disbursed in the future by concentrating on the
variables that are related to a company’s dividend decision. This study should act as a
benchmark for the industrial and services companies of Jordan as well as their investors,
in a way which gives more credence and value to dividends.
The objectives of this thesis are primarily based on ascertaining the main determinants
of dividend decisions by Jordanian industrial and services corporations, as well as on
linking the findings to what is proposed in the financial literature. The study also aims
to investigate whether or not the theories advanced on dividend decisions can be
extended to the Jordanian market. In addition, the impact of dividend announcements
on share prices and the impact of firm characteristics on the smoothness of dividends
have been two other objectives that help in examining the behaviour of the industrial
and services sectors in Jordan. Actually, many arguments can be advanced to indicate
how important this present study is, with its research questions. The fact that dividend
policies are expected to affect companies' investments through their impact on these
companies' capital structures and consequently costs of capital, it is useful to examine
the listed Jordanian firms’ dividend policy. This type of analysis enables us to compare
the behaviour of the Jordanian companies in this thesis with their international
9
counterparts and give some recommendations that should help non-financial Jordanian
companies in answering at their dividend decisions. Finally, as will be explained later
on, the significance of this study is based on the fact that a company's dividend policy
will be examined using some econometric methods, which have not been used before in
the Jordanian context.
The research study documented will also be used to propose a set of recommendations
for all those corporations which mainly belong to the industrial and services sectors
regarding their dividend policies, building on the resultant outcomes of the certain tests.
It will also guide investors in their investment portfolio selections. In brief then, the
main objectives are to:

Determine whether the different firm characteristics (profitability, leverage and
size) have an impact on the smoothness of dividends for Jordanian non-financial
companies.

Understand the determinants of dividends in Jordan, and in particular, whether
the two dividend decisions (the decision to pay or not and how much to pay)
have the same determinants.

Understand the impact of dividend announcements on share prices in Jordan.
1.5 Research Questions of the Study
A set of research questions which the present study will pose in regard with the
dividend policy issue will be addressed in this thesis. Such questions as shown below
will be answered throughout the various forthcoming chapters via the examinations and
investigations conducted:

What features characterize the smoothness of dividends behaviour for listed nonfinancial Jordanian companies?

What are the dividend policy determinants, and in particular, do the company's
two dividend decisions have the same determinants?
10

What is the relationship between the announcement of a dividend and share
prices?

How can the dividend policy of the Jordanian companies be compared or
contrasted with the internationally published evidence?
1.6 Thesis Structure
The present study is divided into six chapters; four of them (i.e. Chapters 2-5) should be
considered to represent its main bulk, whereas Chapters 1 and 6 will be an introduction
and a conclusion respectively. The following is a brief description of what each chapter
of these generally contains:

Chapter 1: Introduction includes six sections and addresses in general
background details about Jordan as a country and presents the structure of the
study itself in terms of the motivation; significance; contribution and structure;
along with an introduction.

Chapter 2: Literature Review includes four sections and addresses selected
empirical evidence for the smoothness, determinants and the impact of dividend
announcement on share prices.
In addition, a review of different dividend
theories including the dividend irrelevance theory, signalling theory, agencycost hypothesis, bird-in-hand theory and lastly, the tax effect hypothesis; along
with an introduction and conclusion.

Chapter 3: Empirical Investigation of the Smoothness of Dividends includes five
sections and addresses the theoretical background of the dividend smoothness.
Then, an explanation of both model specification (symmetric and asymmetric
models); and estimation results (including the estimation results for the impact
of firms’ characteristics on the smoothness of dividends such as profitability,
size and leverage); along with an introduction and conclusion.

Chapter 4: Empirical Investigation of the Determinants of the two Dividend
Decisions includes five sections and addresses in general corporate dividend
policy determinants in terms of theoretical scope including profitability, size,
leverage, growth and ownership structure; model specification; statistical
11
analysis including descriptive statistics and a correlation matrix; along with an
introduction and conclusion.

Chapter 5: Investigation of the Impact of the Announcement of Dividends on
Share Prices includes five sections; it addresses the theoretical background;
model specification including the standard OLS event study methodology and
the hybrid method (OLS/EGARCH); estimation results; along with an
introduction and conclusion.

Chapter 6: Conclusion includes four sections and outlines the general findings of
the present study and the future research recommendations.
12
2. Literature Review
2.1 Introduction
Dividend policy is still a controversial issue in corporate finance and represents one of
the most fiercely debated topics in finance research. Decades have been spent on work
about the issues of dividend policy, or the 'dividend puzzle' as Black (1976) call it, so
the puzzle could be better understood. The basis upon which a payout decision is made
has been investigated in both emerging and developed markets, with a major role to be
played by the various differences in regulatory aspects among countries in dealing with
both payout ratios and the payout method to be used. A clear guideline for an optimal
dividend policy is yet to be realized, even with the many different studies on this
present subject; which leaves two puzzling questions about why firms decide to pay
dividends and why investors pay such close attention to them.
The dividend decision is highly important to corporations today since it determines the
level of funds which will be returned to investors and how much should be retained for
reinvestment purposes.
However, two opposing viewpoints have appeared in
connection with dividends; one side adheres to a 'perfect world set-up' (Miller &
Modigliani, 1961) and the other believes in a 'real world standpoint'. This failure to find
a common ground led many to acknowledge dividends as one of the most significant
unsettled problems in corporate finance. The rational of a firm’s dividend policy has
been developed by several theories.
In a world devoid of both taxes and transaction costs where all stakeholders are notified
of the distribution of firms' uncertain future cash flow, Modigliani & Miller (1958,
1961) argued according to their 'irrelevance theory' that there would not be any
difference in the value of companies according to the way cash flow is given to their
shareholders; in the form of dividends or capital gains. However, while we accept that
the afore-said irrelevance conclusion helps grasp the fundamentals of the dividend
policy, we must not forget the numerous theories which show how the dividend policy
can have an impact on a given firm’s value if any of Miller & Modigliani's assumptions
are relaxed.
This irrelevance theory will no longer be appropriate due to market
imperfections, where taxes, agency problems and information asymmetry exist. A
contrary stance was adopted by many others (e.g. Lintner, 1956) in accepting that
13
capital markets are not just perfect and, thus, dividends do matter. Although Fama &
French (2001) found that in the US, dividend is disappearing but many researchers still
beleive that both managers and investors appear today to prefer the payment of
dividends, and companies are more concerned with distributing a big portion of their
earnings as dividends while also attempting to keep their dividend status stable (Juma’h
& Pacheco, 2008).
This chapter attempts to outline the basic theories that explain dividend policies as well
as to provide an overview of the primary empirical studies on corporate dividend policy.
The chapter is organized into three main sections, which are as follows: Dividend
Policy Theories (Dividend Irrelevance Theory, Signalling Theory, Agency-Cost
Hypothesis, Bird-in-Hand Theory and lastly Tax Effect Hypothesis); then the empirical
evidence of Dividend Policy is outlined and discussed (smoothness, determinants and
the impact of dividend announcements on share value).
2.2 Dividend Policy Theories
A wide array of theories exists on the issue of dividends, which are not only restricted to
the five theories outlined below. However, the dividend irrelevance theory and four
alternative hypotheses do cover a great deal of the theoretical work which has thus far
been undertaken.
These four theories are the information content of dividends
(signalling) hypothesis, the agency-cost hypothesis, and the bird-in-the-hand hypothesis
(where the increasing dividend payments maximize the value of a firm) along with the
tax preference argument (where high dividend payouts reduce a firm's value).
2.2.1 Dividend Irrelevance Theory
Questions have been raised by Miller and Modigliani's (1961) seminal paper on
dividend policies over the issue of whether a firm’s value can be increased by higher
dividends. This idea was mainly based on the so-called bird-in-the-hand argument
which will be discussed in more detail shortly. For instance, Graham & Dodd (1934
cited by Frankfurter et.al., 2002, p.202) argue that “the sole purpose for the existence of
a corporation is to pay dividends and firms that pay higher dividends must sell their
14
shares at higher prices”. However, Miller and Modigliani verified through their study
that a dividend policy under certain assumptions about perfect capital markets would
not be relevant. They argued that “… given a firm’s investment policy, the dividend
payout policy it chooses to follow will affect neither the current price of its shares nor
the total returns to shareholders” (p.414). Such a dividend policy in a smooth running
capital market will have no impact on the given company's value.
In other words, under perfect market conditions, a firm’s value is not determined by its
dividends but by its investments. This will indicate to a firm’s shareholders that the
value of every share is based on the future cash flows, as it is not significant whether the
cash flows will be left in the company as cash reserves or will be distributed as
dividends. The value of such cash flows will be received by shareholders or in a form
of dividends or through an increase in the share price. This irrelevance theory relies on
the following idealistic assumptions: (1) Taxes on dividends and capital gains do not
differ, (2) when buying and selling securities there are no costs related to transactions
and flotations, (3) information (mostly symmetrical and costless) is free and equally
available to all those participating in the market, (4) the interests of managers do not
conflict with the interests of security holders (i.e. no agency problems), and (5) all those
participating in the market are price takers (Miller & Modigliani, 1961).
However, while the afore-said dividend-irrelevance argument seems convincing, it does
not provide any explanation or reason why companies themselves, the public,
investment analysts and anyone else harbouring any interest in the financial market are
so concerned with dividend announcements. The answer to this puzzle certainly lies in
the real world market conditions, which deviate from the above-stated idealistic perfect
market assumptions. For this reason, a dividend policy is generally perceived as an
expansion of the capital structure decision.
A company from a purely theoretical
perspective should distribute dividends when its investment opportunities are not
expected to generate such a high return as the required rate of return on equity.
Shareholders are provided via dividend payments with a direct source of value (i.e. of
immediate cash), rather than a higher equity value on a per share basis through
reinvestment of free cash flows into the given company.
15
Subsequent studies (e.g. Black & Scholes, 1974; Bernstein, 1996; Miller & Scholes,
1978, 1982; Miller, 1986; Hess, 1981) mostly supported the dividend irrelevance
hypothesis. Black & Scholes (1974) for instance studied the association between a
dividend yield and stock returns for identifying the impact of a dividend policy on stock
prices. Neither the high-yield nor the low-yield payout policy of a firm was found to
affect stock prices. They found empirical support for Miller & Modigliani’s dividend
irrelevance argument.
Besides, sufficient conditions were advanced, by Miller &
Scholes (1978, 1982), Hess (1981) Miller (1986) and Bernstein (1996), as to why
taxable investors would be neutral to dividends regardless of the certain tax
differentials.
However, mixed results have been documented about the dividend irrelevance theory of
Miller & Modigliani in emerging markets. In this regard, Ben Naceur & Goaied (2002)
studied 28 Tunisian companies listed on the Stock Exchange from 1990 to 1997. Their
evidence was found to support the irrelevance propositions of Miller & Modigliani of
capital structure and dividend. By contrast, Omet & Abu-Ruman’s (2003) evidence
proved not to be consistent with the dividend irrelevance argument. The CFOs of 47
Jordanian manufacturing listed companies were surveyed in order to ascertain their
opinions on dividend policies. They found out that most of the CFOs in question were
in agreement with the view that a dividend policy affects share prices, which would
certainly suggest that a dividend policy plays a major role in Jordan.
2.2.2 Signalling Theory
A firm will generally, just as signalling theory asserts, ensure that an increase in its
dividends will only occur when such an increase is certain to be associated at a later-on
date with a higher level of future cash flows. Investors in such a firm employing this
present value theory will take careful note of the firm’s credibility to measure the
likelihood that the future cash flow will indeed correspond to the present increase in the
dividend. A good level of credibility can be maintained by striving to avoid any
unexpected movements in either direction on its dividend payments.
With the presence of asymmetric information being commonplace, especially in
financial markets, too many economists are of the view that all those managers, who are
16
confident about the future prospects of their company, will announce an increase in the
dividends as a signal to the shareholders. Hence, companies that are publicly-held and
where the board of directors tries to manage shareholders’ expectation based on a
regular increase on the dividends, are more likely to take up a dividend policy under the
signalling hypothesis. In other words, dividends present a signal to the public equity
markets of management’s expectations of the company being capable of making future
cash flows.
Such public companies, practically, are usually unenthusiastic about
reducing dividend payments given the undesirable consequences for their share price
and likely anger of investors at shareholders meetings. The economics of such an
analysis involving information asymmetry lies at the root of studies which were mainly
conducted in both the 1960's and 1970's. More recent reviews (Heil & Robertson, 1991;
Moore, 1992; Kirmani & Rao, 2000; Riley, 2001) have taken quite a different outlook
from these early papers, which could be regarded as the starting point of this theoretical
approach.
The suggestion that capital markets take into consideration dividend announcements as
information for evaluating share price was actually put forward by Pettit (1972).
Subsequent theoretical models (e.g. Bhattacharya, 1979; John & Williams, 1985; and
Miller & Rock, 1985) suggest that any changes in a dividend policy carry news about
future cash flows. Specifically, good news is represented by an increase in dividends
while bad news is represented by a decrease. The resulting implication is that a change
in a dividend payment means that profitability will change in the same direction at some
future date. The reaction of the market to dividend announcements was examined by
Asquith & Mullins (1983) who used a sample of 168 firms that initiated dividends. The
magnitude of the initial dividends was found to be statistically and positively related to
the abnormal returns on the announcement day. Their results actually indicate that the
level of the dividend changes can also be of significance for share returns; Asquith &
Mullins’s (1986) finding also reinforced the view that initiating a dividend could also be
a signal offering more support to the information content of dividends hypothesis.
However, many empirical papers have contradicted the previous argument. And despite
the fact that Modigliani & Miller's pioneering work has helped academic researchers in
grasping the fundamentals of dividend policy; it has allowed them to challenge the longstanding irrelevance story by examining the topic of information conveyance innate to
17
the dividend policy changes. In this regard, Healy & Palepu (1988) have approached
the dividend policy under the realism of an imperfect capital market assumption, where
information asymmetry definitely exists.
They were actually amongst those who
supported the view that dividends disbursements signal future earnings improvements.
In other words, investors interpret dividend initiation and omission announcements as
managers' forecasts of a company’s future change in earnings.
Furthermore, Fama & Babiak (1968) and Watts (1973) documented results that are in
line with the findings reported by Lintner (1956) and they indicated that firms which
initiate (omit) dividends announcements report an increase (decrease) significantly in
their total earnings for at least one year after the current year of the given change in
dividend-policy.
Healy & Palepu's (1988) empirical research found supporting
evidence that earnings' increases (decreases) following the given firm’s respective
initiations (omissions) extend for at least one year afterwards.
Empirical studies that support the theory of an increase/decrease in dividend as a signal
of information about current and future cash flows an also examined abnormal stock
returns following dividend announcements.
This line of research focusing on the
information that can be provided by a dividend change was initially started by Denis
et.al. (1994), where their findings corroborated the signalling theory of cash-flow and
clientele assumptions on the way that a stock price reacts to an announcement of a
change in dividend, but did not provide much support for the hypothesis of
overinvestment. Indeed, they went much further in their research by providing detailed
evidence on the signalling theory of cash flow and the overinvestment hypothesis by
empirically testing for analysts' revisions of previous earnings forecasts in the
immediate aftermath of a dividend announcement. The value of a firm’s equity and
excess returns are directly linked, according to Denis et.al. (1994), to dividend
announcements.
Further, if information about future cash flows is conveyed by
dividends, analysts should revise their forecasts following dividend announcements;
which is a course of action empirically shown to accumulate benefits (Denis et.al.,
1994; Carroll, 1995).
Taking Jordan as an emerging market, El-Kouri & Al-mwalla (1997) investigated the
Jordanian capital market and the impact of any increase or decrease of dividend on
18
share prices. The announcements of a change in dividends were found to have an
almost insignificant impact on the prices of stocks. This result was later contradicted by
Omet & Abu-Ruman (2003); they interviewed 47 Jordanian chief financial officers who
confirmed the usage of dividends as a way to signal information to the market and this
would reduce the information asymmetry. Thus, their results are in line with the
signalling hypothesis.
The informational content of dividends was linked in another respect (Amihud & Li,
2002) to the phenomenon of increased institutional holdings; the authors point out that
institutional investors enjoy a better degree of information, while reducing the
informational asymmetry in the market. Empirical studies have shown generally that a
market reacts positively to an announcement of a dividend increase, while it reacts
negatively to a dividend decrease. It is of note as well, according to many scholars (e.g.
Ang, 1987; Allen & Michaely, 1995; Lease et.al., 2000), that the level of the market
reaction to a dividend decrease is much higher than the level at which a market reacts to
a dividend increase.
2.2.3 Agency-Cost Hypothesis
The agency problem is a separate stream of research with obvious implications for
dividend payout decisions. This stream has been well documented in the related finance
literature. The agency problem mainly occurs as a result of the separation of ownership
from management. It arises due to the information asymmetry which, to a large extent,
is due to a given firm's environment, resulting in moral hazard that is continually
affecting a firm's managers, when they carry out hidden actions. The recognition of this
potential agency conflict relating to the separation of control and corporate ownership is
not new, and can thus be traced back over more than three centuries, owing to the
differences between managerial and shareholder priorities. These problems were most
common in the British East Indies Company where any efforts to control managers were
found to be ineffective due to the inadequacies and costs associated with shareholderrelated monitoring (Kindleberger, 1984).
One of Miller & Modigliani’s assumptions regarding the concept of ‘perfect capital
markets’ is that the interests of managers and shareholders do not interfere. Though,
19
this assumption has proved extremely idealistic relative to the real world, where the
goals of a firm’s owners vary detectably from those of a firm’s managers. Under such
conditions, managers are often deemed deficient agents of shareholders (i.e. principals).
This happens because the interests of the managers are not essentially the same as those
of the shareholders, leading them to conduct themselves in a way which may prove
costly to shareholders, such as spending excessive premiums on their own needs or
overinvesting in projects regardless of whether their activities are profitable or not.
Shareholders, thus, bear (agency) costs in connection with monitoring managers’
behaviour where these agency costs are implicit costs arising from the potential conflict
of interest among shareholders and corporate managers. Distributing dividends could
be used as a tool to bring the interests of shareholders and managers into line and
relieve the agency problems between them, through decreasing the optional funds which
managers mainly hold (Rozeff, 1982; Easterbrook, 1984; Jensen, 1986; and Alli et.al.,
1993).
As a way of understanding corporate dividend policy, the issue of agency costs has been
broadly tackled in empirical studies. Rozeff (1982) used a large sample of US firms
and argued that the percentage of stock held by insiders was found to be negatively
related to the payout ratio, while the number of shareholders was found to be positively
related to the dividend payout ratio. Furthermore, Dempsey & Laber (1992) and Lloyd
et.al. (1985) updated Rozeff’s work investigating data over an extended period of 19811987, with highly supportive findings. In an attempt to explore the factors which
explain cross-sectional differences in insider ownership, dividend policy and debt,
Jensen et.al. (1992) followed a three-stage least squares method, using a sample of 565
firms in 1982 and 632 firms in 1987; their results were generally in line with Rozeff’s
(1982) findings and consequently with the agency-cost hypothesis.
Holder et.al. (1998) who investigated data for 477 US firms over a ten years period
(1980-1990), also reported that insider ownership was significantly and negatively
related to dividend payouts and that the number of shareholders and payouts are
positively related. Similarly, Saxena (1999) examined information for 333 companies
listed on the NYSE, out of which 235 were unregulated and 98 were regulated, over the
period of 1981-1990; they confirmed the previous findings. Furthermore, more than
4000 companies from 33 countries around the world were examined by La Porta et.al.
20
(2000) who included several emerging markets in their research and provided empirical
evidence for the agency-cost hypothesis. These researches were mostly in line with this
present hypothesis and provided support for the view that the dividend policy of a firm
can be determined by agency.
Firms were found additionally to restrict the ability of managers to overinvest by
committing to distribute higher dividends, and thus reducing the available free cash
flow for overinvestment (Jensen, 1986). The agency costs for overinvesting firms are
reduced by this increase which in turn leads to an increase in investment return over
time (Koch & Shenoy, 1999). In this scenario, a dividend policy which follows the
distribution of high dividends would act as a control mechanism since a rise in the
dividend payment will render an instant reduction in the potential cash available for
overinvestment. As a consequence, shareholders in their efforts to control the behaviour
of managers concerning investment can demand high dividends and leave them with
limited resources with which to abuse a firm’s financial resources. The possibilities that
managers might have are thus limited by this reduction in resources to accept negative
net present value (NPV) projects on their own behalf; hence the value of shareholders’
wealth is protected. However, it was found empirically that when a high dividend
policy is followed, managers’ incentives are reduced and their ability to adopt better
investment opportunities with a positive NPV value is tightened. Despite this inherent
drawback, the suggestion that high dividend payments control the agency cost has been
broadly adopted by many economists.
Another line of empirical research presented by Maury & Pajuste (2002) shows, in line
with the aforementioned findings, that the basic incentive for the agency model of
dividends is correct in that cash flows which may be diverted by corporate managers in
order to be used at the managers’ own discretion for personal use or in pursuing
unprofitable projects are paid out to investors. Such models according to the above-said
study are subdivided into two points of view: (i) one set of theories considers the
payment of dividends as a result of the agency conflict arising between the managers
and shareholders (see La Porta et.al., 2000; Faccio et.al., 2001) and (ii) the other stream
views dividend policies as substitutes for governance problems.
21
Corporate governance addresses the mechanisms that can be used by a firm’s
stakeholders in order to exercise control in order to protect their interests (John &
Senbet, 1998). It is then the relationship between that firm and the stakeholders, which
determines the given firm’s strategic direction and performance (Luo, 2005), and is
based on two major conditions: (i) conflict of interests or agency problem and (ii)
transaction costs (Hart, 1995). It is the matter of many definitions-including the ones
above-to raise one central issue, i.e. the fact that there is a conflict of interest between
outsiders and insiders, where the problems of corporate governance cannot be
completely resolved by contracts due to uncertainty, information asymmetry and
contracting costs.
Some mechanisms are then needed to solve corporate governance problems and control
the conflicts which arise. That is, to say that the mechanisms will depend on whether
the governance system is market-based (as in US and UK) or control-based (as in some
emerging countries). The market-based model is based on transparent information
disclosure, independent corporate boards, active take-over markets, widespread share
ownership and other factors. However, the control-based system gives emphasis to a
concentrated share ownership structure, reliance on family finance and a banking
system, insider corporate boards and limited disclosure (Du & Dai, 2005; Arslan &
Karan, 2006 and Claessens et.al., 2006).
Mechanisms which are available to resolve any potential problems will mostly depend
on whether such problems exist between (i) managers and owners or (ii) minor and
major shareholders. The resolution of problems associated with the first of these relies
on certain internal mechanisms just as those of ownership structure, board of directors,
executive compensation and financial disclosure. By contrast, resolution of problems
associated with the latter basically rely on external mechanisms such as those of the
protection of minority shareholders, the external take-over market, product market
competition and legal infrastructure (Gugler & Yurtoglu, 2003). Furthermore, corporate
governance is considered to be good when it is designed according to Jesover &
Kirkpatrick (2005: p.127) to “bridge the gap between the interests of those who run a
company and the shareholders, increasing investor confidence and lowering the cost of
capital for the company, and also to ensure that a company honors its legal
22
commitments, and forms value-creating relations with stakeholders including employees
and creditors”.
At the end of our illustration of the present hypothesis, and from an agency perspective,
a regular stream of dividends restricts managerial control over additional funds and
reduces the scale of resources under their control, which will then limit any potential
wastage and thereby mitigate the threat of an agency crisis. If the firm has not paid
dividends, they would open the door for their managers to adopt policies which gain at
the stockholders' expense (DeAngelo et.al., 2004). However, no evidence emerged
about this issue (Brav et.al., 2005) in that the payment of dividends is used to selfimpose discipline. There seems to be a good reason, if looking at it from an agency
perspective, why institutional holdings and dividend payouts are negatively correlated.
Naturally, institutions can engage in active monitoring (Gillan & Starks, 2000) whereby
the possibility of agency conflicts' is negated; then the disciplinary role of dividends
becomes at such a stage valueless.
2.2.4 Bird-in-Hand Theory
What the bird-in-hand theory mainly asserts is that dividends are relevant. The central
force of this theory is that any increase in dividend payments could simultaneously be
connected with an increase in the value of a firm. That is to say, the ‘bird-in-the hand’
of cash dividends is preferred by investors contrary to the uncertain capital gains which
may arise. As uncertainty about upcoming cash flows is reduced by a higher present
dividend, the cost of capital will be reduced by a high payout ratio and thus the share
value of the firm will be increased.
In this respect, three possible hypotheses were examined by Gordon (1959) for the
reasons behind an investor’s decision to buy a specific stock; to get both earnings and
dividends, to get only dividends, and lastly to get only earnings. Retained earnings
were found to have a lower impact on share price than dividends. Likewise, Gordon
(1963) asserted that the cost of equity as well as the required rate of return on equity can
be decreased by higher dividend payouts; this was proved correct by Gordon (1963) as
well by many others (e.g. Fisher, 1961; Lintner, 1962 and Walter, 1963).
23
Lastly, this theory was mainly called by Miller & Modigliani (1961) the bird-in-hand
fallacy, in that the risk of a firm’s operating cash flows is the determinant of the
company’s risk and not the method by which it payout its earnings. However, 54.9% of
investors were found according to managers of NASDAQ firms to have a preference for
cash dividends today relative to uncertain future price appreciations (Baker et.al., 2002),
which would not then provide any reinforcement for the bird-in-the-hand justification
about why firms pay dividends.
2.2.5 Tax Effect Hypothesis
Another violation of Miller & Modigliani’s (1961) assumptions concerns tax; it was
hypothesized by M&M (1961) that any potential tax effect was excluded by the
assumptions of an ideal capital market where taxes on dividends and capital gains are
identical; which makes the policy of dividends relevant. It was assumed that capital
gains and dividends were treated in the same for tax purposes. However, taxes exist in
reality and can have significant effects on the way dividends are distributed and the
value of a company. In general, capital gains and dividends are often treated in a
different way as regards taxes, and, due to the fact that it is after-tax returns that are
important for most investors, different taxes might influence their desire for dividends.
The supply of dividends may also be affected by taxes, with managers reacting to this
tax preference when they are looking to maximize shareholders’ wealth (i.e. the firm’s
value) by increasing the earnings retention ratio.
Alzahrani (2009) found that tax affects managers’ dividend payment decision; tax
systems vary noticeably from one country to another. In the USA, where companies
used to operate under a classical tax system, dividends got taxed twice. Corporate
income is taxed firstly under the corporate income tax and secondly under the individual
profits tax when paid as a dividend or recognized as a capital gain once the sale of a
firm’s shares occurs. Thus, the resulting implication is that dividends are taxed much
higher than the capital gains. Avner & Roni (2000) argued in this respect that the tax
benefit of capital gains over a dividend income is a significant feature of a firm’s
dividend policy which would have an influence on corporate considerations as well as
on the demand of investors for dividends.
This may impose additional costs on
24
investors, who are paid a smaller amount of their earnings in the form of dividends than
they would desire, which would perhaps discourage companies from paying dividends.
A firm's idea is to protect and increase its shareholder's value through retaining cash for
any future investment opportunities in order to generate more capital gains instead of
less valuable dividend payments. However in the UK, with a different tax system,
companies run their business under an imputation tax system, where the dividend is
taxed partially to the individuals; this would mean that shareholders receive a partial tax
reduction because of the tax which has been paid by corporations. The tax-system
difference may partially explain why British companies pay a considerable amount of
dividends, whereas their US counterparts used to pay low or no dividends (Egger et.al.,
(2012). The dividend policy followed by a company must take into consideration the
tax condition of its investors, i.e. different investors’ income-tax profiles imply a
different level of demand for corporate dividend payments. Personal incomes in the UK
for instance are taxed at different rates, ranging from 10% to 40% according to the level
of the individual’s income stream. Evidently, investors who have an income level with
a 10% tax burden will prefer a higher dividend payment than those who have an income
level that is 22% tax burden. No matter how considerate they are, it is fairly hard for
companies to provide an optimal dividend policy to satisfy every single clientele of its
shareholders.
According to the tax hypothesis, companies should fix their dividend-payment profile
(whether high or low), and then leave the investors to decide for themselves which firms
to invest in. In other words, when the payout ratio is fixed, shareholders can decide
which company is the best investment option for them. For example, investors who pay
a low rate of income tax will go for the sort of companies which give a high dividend
payment. For this reason, it is very important for companies in order to keep the
various shareholders satisfied to have a constant dividend policy. Otherwise, an amount
of confusion will decrease the amount of shareholders, the share price, and the
credibility of that given company, which will then decrease the shareholder’s value. In
addition, it is another way of explaining why managers are concerned with avoiding any
radical changes to their firms’ present dividend policy in either direction.
25
Having developed an after-tax version of the capital asset pricing model to estimate the
relationship between dividend yield and tax risk-adjusted returns, Brennan (1970)
maintained that a positive and linear relationship should exist between dividend yield,
stock pre-tax returns and market risk. The results of this study suggested that a higher
pre-tax risk adjusted return would be combined with higher dividend yield stocks which
would compensate investors for the disadvantages of tax. Actually many empirical
studies have attempted to estimate Brennan’s model and explain the relationship
between stock returns and dividend yields. Brennan’s model has been tested by Black
& Scholes (1974) who found that there is an insignificant effect of tax, and hence they
concluded that high or low-dividend yield shares do not have an impact on the share’s
returns either before or after taxes. On the other hand, Litzenberger & Ramaswamy
(1979) contradicted the results found by Black & Scholes by criticizing the
methodology that they used, i.e. the definition used for the dividend yield.1 Litzenberger
& Ramaswamy modified the model of Brennan (1970) by defining a monthly dividend
yield and they classified the shares into different yield classes; (i) a positive dividendyield class and (ii) a zero dividend-yield class, showing ultimately that the dividend
yield coefficient was highly significant and positive (see also Blume, 1980; Keim,
1985).
Later on, Miller & Scholes (1982) challenged Litzenberger & Ramaswamy’s
conclusion, criticizing the latter’s short-term (i.e. monthly) definition of a dividend
yield. Miller & Scholes basically argued that the estimation implemented to test the
effect of the differential tax treatment of capital gains and dividends on the returns of
stocks and the positive yield-return relation are inappropriate due to the fact that there is
an information bias. Hess (1981) found similar results during the period from 19261980 by examining the relationship between dividend yield and monthly stock returns.
In addition, Poterba & Summers (1984) proved that the hypothesis of tax-effect is
strongly supported in their analysis of the dividend yield-return relationship across
different tax regimes in the UK.
1
The definition used for dividend yield by Black & Scholes is a long term definition of dividend yield.
(The previous year’s dividends divided by the year end share price)
26
2.3 Empirical Evidence
The following three sections will discuss the empirical evidence about the smoothness,
determinants and the impact of dividends on share prices. More specifically, section
2.3.1 discusses the empirical evidence of the smoothness of dividends. Section 2.3.2
highlights the empirical evidence about the determinants of dividends. Then section
2.3.3 goes through the empirical evidence about the impact of dividends on share prices.
2.3.1 Empirical evidences of the smoothness of dividends
Based on a series of interviews made in the mid-1950s with 28 managers working in
different companies, Lintner identified fifteen variables concerning these companies’
dividend policies, the determinants of their policies and stability of their dividends.
Actually Lintner stated that the stability of dividends is the main concern of companies
and he argued that the first decision which managers have to make is if they need to
change dividends. And if they decide that a change is required then they need to decide
how large it should be. The findings of his research were two-fold: (i) firms had long
term target dividend payout ratio and dividend policy was adjusted according to the
target; and (ii) managers were not willing to change their dividends unless the higher
level could be maintained. Lintner went on to develop a model –the partial adjustment
model- which was to be consistent with his findings and to explain dividend payments
well; his model suggested that a dividend depends partly on a firm’s current earnings
and partly on the previous year’s dividend.
According to Brittain (1964, 1966) this model provided a remarkable clarification of
dividend and also confirmed Lintner’s findings that firms follow stable dividend
policies. Later on, Fama & Babiak (1968) examined several different models in order to
describe dividend behavior. The sample they used was 392 companies, while their
investigation period was 19 years (1946-1964). Among the several models they tested,
the two which were found to provide the best fit were Lintner’s partial adjustment
model and a similar model which included different measures as explanatory variables,
such as a lagged earnings term or excluded the constant term. They concluded that the
American firms do smooth their dividends.
27
The same conclusion was also reached by McDonald et.al. (1975) who used crosssectional data for 75 French firms over six years (1962-1968) and concluded that
Lintner’s partial adjustment model can explain dividend decisions. In 1985, Baker
et.al., in their effort to test Lintner’s conclusions, conducted a survey of 562 companies
listed on the New York Stock Exchange (NYSE). They found out that the same factors
affected dividend stability. In another study of 80 NYSE firms, DeAngelo & DeAngelo
(1990) demonstrated that corporations are more inclined to cut dividends rather than to
omit them, in order to preserve their corporate image, and enhance their bargaining
position with organized labour.
Leithner & Zimmermann (1993) enlarged the scope of dividend policy research outside
the USA to cover different statutory frameworks such as the UK, France, West
Germany, and Switzerland. They concluded that companies pursue smooth dividend
policies. In India as well, companies were found to prioritise stability in distributing
dividends and the study’s findings were in line with Lintner’s partial adjustment model
(Pandey & Bhat, 2004).
Dewenter & Warther (1998) conducted a comparison between two different markets
(US and Japan) over the period 1982-1993 by also using a partial adjustment model.
The US firms were found to be less willing to follow a volatile approach in their
dividend policy in contrast with their Japanese counterparts and were more willing to
smooth dividends than in the period investigated by Fama and Babiak during 19461964. In addition, Baker & Powell (1999) found that managers' perspectives regarding
the setting of dividend payments were in line at that time with those views reported by
managers interviewed by Lintner (1956).
They interviewed 603 Chief Financial
Officers in companies listed on the NYSE and concluded that 90% of participants
believed that dividends do affect firm value and profitability is one of the essential
figures that managers keep in mind when deciding on the level of dividend to payout.
Contrary to the great deal of evidence supporting the smoothness of dividends, Adaoglu
(2000) investigated data for the firms listed on the Istanbul Stock Exchange (ISE)
during 1985-1997 the commercial and industrial firms were found to follow unstable
dividend policies especially after 1995 when regulations in Turkey allowed the
companies to choose whether or not to pay a dividend; before 1995 it was mandatory to
28
pay at least 50% of earnings as a dividend. The result of Adaoglu (2000) confirmed the
argument raised by Glen et.al. (1995) that more empirical studies need to be conducted
in emerging markets in order to understand better the dividend behaviour followed by
their companies.
In a comparative study for eight emerging markets2 and the US over a ten year period,
Aivazian et.al. (2001) also found that less stable dividend policies were followed by the
emerging market firms in comparison to the US market companies. They argued that
the main reason for the different dividend behaviour between the emerging market and
the US was the different financial systems which existed. Similarly, Pandey (2001)
established the validity of the partial adjustment model in the emerging Malaysian
market during 1993-2000. He investigated the smoothness of dividends for 248 firms
and found that less stable dividend policies were followed by the Malaysian companies
compared to their developed country counterparts.
In a survey that included 34 firms from Portugal, Benzinho (2004) investigated the
dividend behaviour of the Portuguese companies over thirteen years (1990-2002) using
panel data regression.
He concluded that managers in Portugal do smooth their
dividends and that the profitability of the firm is an essential element in this decision.
He argued that his results are in line with Lintner’s partial adjustment model findings
since both variables – earnings and lagged dividends- were found to be significant. In
addition, during the ten years (1984-1993), Goergen et.al. (2005) investigated 221
German commercial and industrial firms using the partial adjustment model. They
concluded that the German firms follow a smooth dividend policy but they do not adjust
their dividends to their target ratio at the same rate as American firms which were
reported in Lintner (1956). In addition, Aivazian et.al. (2006) conducted a study on
data for American firms listed on the NYSE and concluded that companies which raised
debt in the public bond market did count the smoothness of their dividends as part of the
optimal choices which they made compared to firms which used private bank debt.
They argued that companies which raise debt in the public bond market have a superior
motivation to smooth their dividends in order to reduce agency problems and the
2
The eight emerging markets investigated are Malaysia, India, Korea, Jordan, Zimbabwe, Thailand,
Pakistan, Turkey and the United States.
29
information asymmetry between the managers of firms and the creditors. Consequently,
investors would be more willing to hold their debt.
Ben Naceur et.al. (2007) investigated the dividend behavior of firms in the Tunisian
market during the seven years of 1996-2002. Their sample included 48 listed companies
on the TSE (Tunisian Stock Exchange).
They found that Tunisian companies do
smooth their dividends and the profitability variable was a significant variable in
affecting the dividend behavior.
This suggests that highly profitable Tunisian
companies smooth their dividend to a greater extent than the low profitably firms.
Furthermore, the dividend behavior of Omani firms has been empirically investigated
during the period of 1989 to 2004 by Al-Yahyaee et.al. (2010). They found that the
Omani companies do smooth their dividends although this contradicted their
expectations.
They argued that most Omani firms pay their total profits out in
dividends; consequently, this leads to no funds for the retained earnings.3 Thus, most of
the companies experience high leverage where funds are raised by external debts
(banks).
Based on the previous literature about dividend smoothing, it is clear that no empirical
study has been done to check the impact of firm characteristics on the smoothness of
dividends in Jordan. In addition, most of the researchers have investigated the partial
adjustment model assuming a symmetric adjustment when the company is above or
below their target dividend payout ratio. Also, this symmetrical adjustment does not
differentiate between those cases where the company is above or below the target
dividend payout ratio. Thus, this is the first gap in literature that the current study will
try to investigate using both a symmetrical and an asymmetrical partial adjustment
model to check if there are any differences between both models. Then, the firm
characteristics will be introduced in an interaction term to investigate the impact of size,
leverage and profitability on the smoothness of dividends in Jordan.
3
It is worth to mention that since the Omani companies pay 100% of their profits as dividends then the
researchers should argue that the Omani companies do smooth their earnings instead of dividends.
30
2.3.2 Empirical evidence of the determinants of dividends
Many empirical studies have investigated the determinants of dividends in different
markets using different samples and different estimation techniques. The irrelevance
argument raised by M & M (Miller and Modigliani) in 1961 suggests that dividends
have no impact on the value of firms. However, many researchers have challenged this
argument and shown that dividends do affect the value of a company; the arguments of
these subsequent researches is based on the fact that if the M & M irrelevance
hypothesis is true then why do firms pay dividends and what are the reasons that make
the shareholders and investors want the dividends. Some of those researchers, such as
Shefrin & Statman (1985) used mental accounting and the prospect theory to explain the
reasons that make investors interested in dividends. In addition, in 1997 Shefrin &
Statman argued that the first step in solving the puzzle of dividends is to identify the
factors that affect the manager’s decisions when deciding their firm’s dividend policy.
Then, Fama & French (1998) showed that the dividend is an important element in
affecting the value of a firm. They argued that since the price of a share is the present
value of its future dividends from the intrinsic model then as a consequence dividends
affect share price and influence the value of the firm.
Ho (2003) has investigated the determinants of dividends in a comparative study
between Australia and Japan during the period of 1992-2001. The sample used in the
analysis included 332 companies from the Japanese and the Australian capital markets.
He concluded that the Japanese companies pay a lower dividend payout ratio than the
Australian companies. He referred to three main theories in explaining the results
achieved; these are the signalling, agency and transaction cost theories. He concluded
that dividend is significantly and positively affected by the size of the firms in Australia
while in Japan dividend is significantly affected by risk and liquidity. Ho (2003) found
that the more liquid the Japanese firms were the more dividends they would be able to
pay but the more risk the company faced then the lower the dividend disbursed.
Additionally, Abor & Amidu (2006) empirically investigated the determinants of
dividends for twenty companies listed on the Ghana Stock Exchange (GSE). They
concluded that in Ghana, dividend policy is determined by cash flow, growth,
profitability and investment opportunities. They argued that the higher the profitability
of the company the more the dividend paid because highly profitable companies
31
experience better cash flow than their low profitability counterparts and this means
better liquidity. On the other hand, they found that the higher the market to book ratio
the lower the dividend since the company would be in the growth stage of its life cycle
with many investment opportunities; so the managers would reduce or cut dividends to
make the growth of the company possible.
Ahmed & Javid (2009) confirmed the results achieved by Abor & Amidu (2006) but in
a different market- Pakistan.
Their sample included 320 companies listed in the
Pakistani market – the Karachi Stock Exchange (KSE) -over the six years (2001-2006).
They concluded that there is a positive relation between dividend policy and
profitability but a negative relation with the size of the firm. Their results supported the
signalling theory since more profitable firms pay high dividends to signal the managers’
confidence about the future of the company’s profitability and liquidity. In addition,
small size companies pay more dividends to signal more information about their firms
and this reduces the information asymmetry between the managers and investors.
Furthermore, the determinants of dividends have been investigated in a comparative
study done by Denis & Osobov (2008) using data from 6 countries (Germany, UK,
USA, Japan, France and Canada) that covered the period of 1994 till 2002. They found
that dividend policy is affected by many variables that have been documented in
literature such as size, growth and profitability. In Tunisia, Ben Naceur et.al. (2007),
investigated the determinants of dividends for 48 companies listed on the Tunisian
Stock Exchange (TSE) and found that the determinants of dividends for the Tunisian
firms were in line with the published factors for companies from the developed
countries. More specifically, they concluded that profitable companies have more cash
flow than unprofitable firms and this allows them to pay more dividends. Besides firms
with rapid growth pay more dividends to signal confidence in the future of the company
and this make investor’s think positively about buying the shares of this company.
Reddy (2002), used two main theories -the signalling theory and the trade off theory- to
explain the determinants of dividends in India during the period 1990-2001.
He
investigated data for companies listed on Bombay Stock Exchange (BSE) and compared
the results to the American companies listed on the NYSE. He concluded that large
size, profitable and rapidly growing companies are more likely to pay dividends than
32
small size, unprofitable and slow growth firms.4 In addition, the determinants of the
dividend policy of 245 companies listed on the Gulf Cooperation Council (GCC)
country stock exchange during the period of 1999 to 2003 have been investigated by AlKuwari (2009). He concluded that companies listed in the GCC countries use dividends
to reduce the agency problem and improve their reputations in the market. His results
suggested that profitability, size of the firms, leverage and government ownership are
the main determinants of dividend policy. He found that leverage affects dividend
policy negatively but profitability, size and government ownership had a positive
relationship with the amount of dividend paid.
Moreoevr, Nazir, et.al. (2010) empirically investigated the determinants of dividend
policy in the Pakistani market using both the fixed and random effects techniques.
Their sample included 73 companies listed on the Karachi Stock Exchange (KSE)
during the five years of 2003-2008. They concluded that the main determinants of
dividend policy in Pakistan are the same as those reported by Ahmed & Javid (2009) for
the same market. They found that the signalling theory holds in the Pakistani market
since profitability affects dividend policy positively but size affects dividends
negatively.
In a study that investigated 266 manufacturing and services companies in the USA in
2007, Gill et.al. (2010) concluded that the sample firms’ dividend policies were
determined by leverage, tax, growth and profitability. They found that the dividend
policies of the manufacturing companies were affected by the tax, profitability and
growth but the dividend policies for the services companies were affected by the same
variables excluding the tax variable. Furthermore, Agyei & Marfo-Yiadom (2011) used
the random and fixed effect technique to examine the determinants of dividend policy in
Ghanaian banks during 1999-2003. They found that the variables that have a significant
impact on the dividend policy of banks in Ghana are profitability, leverage, growth and
age. They concluded that the age of the companies and the growth variables affect the
dividend policy negatively but leverage and profitability affect the dividend policy
He argued that rapidly growing companies are more likely to pay dividends; to signal the manager’s
confidence in the future cash flows of the company. This would increase the confidence of investors in
investing in the shares of the company.
4
33
positively. They argued that such results are in line with the agency theory; no support
was found for the free cash flow theory since the cash variable was insignificant.
The Saudi market has been investigated in a recent empirical study by Al-Ajmi & Abo
Hussain (2011). They used panel data for 54 listed companies on the Saudi Stock
Market (SSM) and covered the period of 1990-2006. They concluded that Al-Zakat is a
major and significant determinant of dividend policy in Saudi Arabia. In addition, the
life cycle of the firm, cash flow and profitability are the other significant determinants
of dividend policy. They also examined if the Saudi firms follow a stable dividend
policy and confirmed that the lagged dividend payment was a significant determinant of
the current dividend payment.
In a more recent empirical study that examined the dividend policy of 44 listed
companies on the Abu Dhabi Stock Exchange of the United Arab of Emirates (UAE)
market during the five years of 2005-2009 a comprehensive analysis was undertaken.
Mehta (2012) investigated the determinants of dividend policy in the UAE for different
sectors such as energy, real estate, telecommunications, construction, industrial and
health care. He included different variables to investigate the determinants of the
dividend decision such as the risk, profitability, leverage, size and liquidity of the
company. He used OLS regression to estimate the relationship and found that size and
profitability were the most important determinants of dividend policy in UAE.
Based on the previous selected literature about the determinants of dividend policy, it is
apparent that no empirical study has differentiated between the two dividend decisions
(the probability to pay and the amount of payment decisions) that are included in the
dividend process. In other words, most of the researchers that have investigated the
determinants of the dividend decision have studied either the probability to pay a
dividend decision or the amount of payment decision -how much to pay. Thus, this is
the gap in literature that the current study will try to investigate by comparing three
estimation techniques (OLS, Tobit and Heckman’s simultaneous model) to check if
there is any difference in the determinants of the two dividend decisions.
34
2.3.3 Empirical evidence about the impact of dividend announcement on share prices
The effect of dividend announcements on share prices has been well documented within
the literature. One of the first studies investigating the relationship between share prices
and dividend changes was the one of Pettit (1972) who used as a sample of 1,000
dividend announcements of 625 companies listed on the New York Stock Exchange.
The author used data from two periods; monthly from January 1964- June 1968 and
daily from 1967-1969.
The second time period was used for the analysis of the
observed changes on abnormal returns during the announcement of dividends, while the
dividend announcements were categorized into seven groups. The results indicated that
there was a significant effect on the dividend announcement day as well as the day
following the dividend announcement.
Asquith & Mullins (1983) also investigated the effect of dividend announcements on
share prices.
More specifically, they investigated the announcement of dividend
initiation on the share prices; their argument is based on the idea that the effect of a
dividend announcement on share prices can be examined by studying abnormal returns
around the announcement of the dividend initiation.
The dataset included 168
companies and the time period included data from 1963 to 1980. Their sample focused
on companies that announced the payment of dividends for the first time as well as the
companies who started paying dividends again after stopping their dividend payments
for at least ten years.
The results indicated a statistically significant and positive
relationship between the share price and the dividend initiation announcement,
suggesting that the dividend announcements provide useful information to investors.
The research of Asquith & Mullins (1983) was expanded on by Healy & Palepu (1988)
who included in their dataset announcements of dividend omissions. Their sample was
separated into two categories; one for announcements of dividend omissions and one for
the announcements of dividend initiations.
The dividend omission announcements
included 172 companies that either decided not to pay dividends or companies that
decided not to pay dividends after 10 years (1986-1996) of continuous dividend
payments.
On the other hand, the category of dividend initiation announcements
included 131 companies that either decided to pay dividends after 10 years of their last
dividend payments or companies that decided to pay dividends for the first time. The
35
results were in line with Asquith & Mullins’s (1983) findings revealing a positive and
significant relationship between dividend announcements and share prices.
The effect of dividend announcements on share prices was also investigated by Aharony
& Swary (1980) who used as their sample 3500 dividend announcements over a time
period of 15 years. Their dataset included not only dividend cut announcements but
also news of neutral and positive changes in dividends. As a result their sample was
separated into three categories; the first category included dividend decreasing
announcements, the second neutral announcements while the third one included
dividend increasing announcements. The event period included 51 days around the date
of the dividend announcement while the market model was used for the estimation of
the cumulative abnormal returns. The results revealed that the companies that had
announced a decrease in their dividend payment experienced negative abnormal returns,
while those companies that had announced an increase in their dividends experienced
positive abnormal returns.
The results were line with the information content of
dividend hypothesis.
The relationship between share prices and dividend announcements was also examined
by Impson & Karafiath (1992). They aimed to examine the different responses of
shares prices to changes in the dividend payout ratio. The used as a sample 116
announcements of dividends from companies listed on the New York Stock Exchange
over a 16 year period. More specifically, their sample included companies that had
announced at least a 20% change in their dividend payment. They hypothesized that
abnormal returns and payout ratios would be negatively correlated, while abnormal
return and dividend announcements would be positively correlated; there were more
negative abnormal returns than positive abnormal returns due to the percentages of
positive and negative payout ratio changes respectively. The sample was separated into
two categories: negative and positive payout changes.
The results showed that
irrespective of the changes in the dividend payout ratio there was no significant effect of
the dividend announcement on the companies’ share price. On the contrary, the effect
of a negative announcement of dividends (a drop in the payout ratio) was found to be
statistically significant and negatively related to the abnormal return.
36
The relationship between dividend announcements and share price changes was also
investigated by Impson (1997) who did not find any significant relationship between the
share price and the announcement of a positive dividend change. In a later study
therefore, he included only negative announcements of dividends in his sample. His
sample included share prices from both regulated and unregulated entities. The total
number of the negative dividend announcements used were 600 over 20 years from
companies listed on the New York Stock Exchange. The sample was divided into two
groups: public companies and unregulated companies.
For the group of public
companies, there were 48 dividend decreasing announcements and 18 omission
announcements, while the unregulated firms there were 556 announcements about a
decrease in dividends and 84 announcements about dividend omissions. Two different
regressions were run in order to compare the results for the two groups. The results
revealed a negative relationship between omissions or announcements of dividend cut
and abnormal returns in both public and unregulated companies. Yet, the effect was
weaker for unregulated companies than public companies.
Lonie et.al. (1996) used 620 dividend announcements from UK companies for a 6
month period to identify the effect of the announcements on the share prices of UK
companies.
They concluded that there is a significant impact of the dividend
announcement on the abnormal returns for the announcement day and one day prior to
the announcement.
In addition, the findings showed that those companies that
announced an increase in their dividend payment or decided not to change their
dividend payment experienced positive abnormal returns, while all those companies that
announced a decrease in their dividends experienced negative abnormal returns.
Likewise, McCluskey et.al. (2007) used 674 dividend announcements for 50 Irish firms
over a time period of 16 years (1987-2001). The event window used was 41 days
around the day of the announcement.
The results showed that Irish companies
experienced positive abnormal return on the dividend announcement day. Specifically,
the results were partly in line with Lonie et.al. (1996) revealing that companies that
announced an increase in their dividends or kept their dividends at the same level
experienced positive abnormal return, while the relationship between the dividenddecreasing announcements and abnormal returns was found to be insignificant for the
Irish companies.
37
Travlos et.al. (2001) also examined the relationship of dividend announcements and
share prices in the Cypriot market, using the market model. The time period under
investigation was from 1985 until 1995 and they managed to collect the data for 31
companies that announced 181 dividend announcements. The results revealed that the
Cypriot companies experienced positive abnormal returns on the day of the dividend
news but they were found to be insignificant. The reason for the insignificance may be
due to the fact that during the period under investigation, Cyprus did not have a formal
Stock Exchange. Similarly, the study of Ali & Chowdhury (2010) also revealed an
insignificant relationship between the share price changes and dividend announcements
for Bangladeshi companies. They applied the standard OLS event study methodology
for twenty five listed banks in DSE.
Once more, the insignificant relationship
uncovered was attributed to the fact that the Dhaka Stock Exchange (DSE) was
inefficient during the period they investigated (Jan 2008- Sep 2008).
An insignificant relationship between dividend announcements and share prices was
also found by Khan (2011) who investigated the Pakistani market and used the market
model.
The time period included 5 years (2005-2009).
dividend announcements used was 639.
The total number of the
Khan (2011) also interviewed 23 firm
executives as a means of investigating the above mentioned effect. The results revealed
an insignificant relationship between share price changes and dividend announcements.
The insignificance was attributed to the fact that managers focus more on the
company’s liquidity and earnings than the dividend payment.
In addition, Zuguang & Ahmed (2010) used 218 dividend announcements of Chinese
companies listed on the Shanghai Stock Exchange in order to investigate whether the
dividend announcements have an impact on share prices covering the period from
January 2005 till December 2009. The companies used belong to the 180 index which
includes companies that have the biggest size and greatest liquidity in the Chinese
market. After the initiation of the “China reform and opening up policy”, the securities
markets in China have developed in line with other economic sectors in a very rapid
fashion. They used three models to capture the effect of dividend announcements
(market model, market adjusted return model and mean adjusted returns model) and
they found that the dividend announcement and share prices were significantly and
positively related. The results also showed that the days before the announcement day
38
did not affect significantly the share price of the Chinese companies irrespective of the
model used. They concluded therefore, that there was no leakage of information in the
Shanghai Stock Exchange and that this market was effective supervision.
In a comparative study by Karim (2010) between the New York Stock Exchange
(NYSE) and the London Stock Exchange (LSE), he investigated the impact of 189
dividend announcements of LSE shares and 213 dividend announcements of NYSE
equities on share prices. He used the standard OLS event study as a methodology
covering the recession years between 2006 till 2008.
The results showed an
insignificant relationship between dividend announcements and share prices in the
NYSE but a significant one in the LSE.
Irum et.al. (2012) investigated four sectors of Pakistan using data from 2005 to 2010.
More specifically, they examined companies listed on the Karachi Stock Exchange
using an event window of 21 days (10 days before and 10 days after the dividend
announcement day). The data included 18 dividend announcements from the oil and
gas sector, 17 dividend announcements from fertilizer sector, 27 from oil and gas
marketing sector and 11 from the refineries sector, while the time period included the
years 2005 to 2010. Normally, the announcement of dividend increases is perceived as
a good signal regarding the future of a company which in turns increases the share price
of the company. Despite the fact that the semi-strong form of the EMH suggests that
share prices are affected by public information, the results of Irum et.al. (2012) showed
no effect of dividend announcement on share prices.
Therefore, Karachi Stock
Exchange was not thought to be an efficient market based on their results.
2.4 Conclusion
Having long puzzled the finance profession, the literature suggests that share value
should not be affected by dividend policies in an ideal and efficient capital market
(Miller & Modigliani, 1961). However, several theories have observed that an increase
in the dividend payment results in an increase in share prices - in contrast to M&M
(1961); research suggests that dividends are of great significance. While the financial
literature contains a vast range of papers both from a theory and practice perspective on
39
the behaviour of a corporate dividend policy, the issue of dividend policy will continue
to be an issue of interest for researches. This could be the reason so many empirical
studies have attempted to test different aspects of dividend policy, including the
smoothness and determinants of dividend policy and the effect of dividend
announcements on share prices.
This chapter has reviewed such issues as the smoothness, determinants of a dividend
decisions, impact of dividend announcements on share prices and the theoretical
background of dividend policy. The review has highlighted the essential points and the
gaps in previous studies. Firstly, no previous investigations have estimated the effect of
the company’s characteristics on the smoothness of their dividend. In other words, no
empirical investigations have tested the effects of size, profitability and leverage on the
smoothness of dividend policy. Secondly, the empirical studies that have investigated
the determinants of dividend policy have neglected to consider that the dividend process
includes two decisions (the payment decision -to pay or not- and the amount decision how much to pay).
Finally, the relationship between dividend announcements and share prices has been
discussed in most of the papers as an important and helpful relation to understand from
both the managers’ and the shareholders’ perspectives. More specifically, what should
be concentrated on is the effect of the increase and decrease of dividends on the market
share price since this will aid the investors in their investing decision and the managers
in knowing what can drive price movements.
In addition, after investigating if
companies in the Jordanian market smooth their dividends or not and clarifying the
determinants that affect the firms’ dividend decisions, a conclusion can be arrived at if
the relationship between dividend announcements and market share prices can be
explored. In addition, the greatest amount of empirical research has generally focused
on investigating the factors affecting dividend policy in the developed countries, while
little research has been done for the developing countries such as Jordan. The present
study is an effort to provide evidence for the Jordanian market, investigating the
dividend policy theories in a Jordanian context.
40
3. Empirical Investigation of the Smoothness of Dividend
3.1 Introduction
Lintner (1956) argued that companies are mainly concerned with having a stable
dividend policy and will only increase dividends when the managers feel confident that
the company’s earnings are going to maintain the increase in the future. In addition,
many stockholders use dividends to pay their own expenses and this makes them
unhappy if the dividend payments are not stable. Some companies might send incorrect
signals to market participants when they reduce dividends to increase the funds
available for investment opportunities; this could result in investors pushing down the
price of the stock because they have misinterpreted the reduction of dividends as a sign
of lower future earnings for the company. This suggests that companies have to find the
right balance between their internal funds and paying dividends, which would maximise
the stock price (Gitman, 2007). Smoothness of the dividend indicates that there will not
be any immediate response in the dividend to a proportional shift of profits.
Smoothness means that dividends regularly adapt to the long term level of profits for
the company. Leary & Michaely (2011) stated that till now, no clear answer has been
found for the reason behind the smoothing of dividends; but they argued that one
possible reason for a stable dividend is to reduce any doubt in the minds of investors
about the future cash flows.
A large body of research both in theory and practice has been conducted, as seen in
Chapter Two, in regard to dividend policy, with several theories having appeared and
arguments tested.
Most noticeable is Miller & Modigliani's (1961) irrelevance
hypothesis that the dividend policy of a firm does not affect its value. Though, several
market “defects” (such as information asymmetry, agency problem and transaction
costs) exist, which provide some basis for developing various dividend policy theories,
such as the signalling and agency cost theories (Barclay, 1987 and Poterba & Summers,
1984). Some theories have emerged which recognise that a dividend policy is relevant
as a result of information asymmetries, which are generally categorised as signalling
theory. This theory states that information can be conveyed by firms to the market
through the payment of dividends. Others such as Bhattacharya (1979) and Bali (2003)
have argued that other market imperfections such as agency costs have an impact on a
company’s dividend policy; agency theory recognices that the costs associated with
41
conflicts between shareholders and managers can be reduced by dividends. Debate
among these theoretical models is still happening today.
The contribution of this chapter to the literature comes in different ways. This chapter
is the first to empirically investigate the impact of the different firm characteristics (size,
leverage and profitability) on the smoothness of dividends for Jordanian non-financial
companies (industrial and services) using both Symmetric and Asymmetric partial
adjustment models.
Consequently, this empirical chapter will extend the previous
literature by firstly empirically investigating the asymmetric partial adjustment model
whereas previous studies have assumed a symmetric adjustment of dividends toward
target dividend payout ratio. Secondly, this chapter investigates what possible firm
characteristics affect the asymmetric partial adjustment model; testing the hypothesis of
whether large/small size, high/low leveraged and high/low profitable companies have
different asymmetric adjustment behaviours towards the target dividend payout ratio.
In addition, most of the research carried out so far has focused on developed countries,
whereas very little attention has been paid to emerging countries, where the capital
market is not big, developed, or suffers from a lot of regulations and insufficient
supervision (Mody, 2004). This is in addition to the legal and statutory differences that
exist between countries, which make it hard to take a broad aggregate view of the
outcomes obtained from empirical studies conducted in developing markets. Lately,
many developing countries have tried to liberalise their markets, which would allow
their firms’ managers more flexibility in selecting their companies’ capital structure
and, therefore, affect their dividend policies. The Jordanian economy, in this respect,
has followed a model of financial liberalization in order to give confidence to foreign
investors and make their capital market more integrated into the world’s market. As a
result, new laws and regulations-such as the Transparency Act 1998, the Companies Act
1998, and Security Act 1997-were actually enacted to make the stock market operate in
accordance with certain international standards; a considerable level of performance and
efficiency improvements have been witnessed recently in the Jordanian market (Amman
Financial Market, 2005). Finally, this empirical investigation will compare the results
obtained to the published empirical investigations from other developed markets.
42
The following section provides an explanation of the theoretical background and the
empirical models used to check the impact of firms' characteristics (i.e. size, profit, and
leverage) on the smoothness of dividends using symmetric and asymmetric partial
adjustment models. In particular, Section 3.2 explains the smoothness of dividends
based on different theories; the agency cost and the signalling theories are mainly used.
Then, Section 3.3 describes the specification of the symmetric and asymmetric models.
Section 3.4 shows the empirical results based on the different firm characteristics (such
as size, leverage and profitability). Finally, the chapter is concluded in Section 3.5.
3.2 Theoretical Background
An entity’s dividend policy is argued to be irrelevant according to Miller & Modigliani
(1961).
This 'irrelevancy proposition' involves the assumption that the firm's value
relies only on how profitable its investments are (i.e. income flow from its assets); the
way this income is divided between dividends and retained earnings is irrelevant. This
irrelevancy proposition assumes the existence of a perfect market where there is no
information asymmetry (i.e. both managers and investors have access to the same
information regarding any investment opportunities) and no transaction or bankruptcy
costs (i.e. issuing debt is absolutely safe).
This means that both investors and
corporations can borrow at the same rate. Conversely, dividends become relevant with
the presence of market imperfections. In the real world, where markets are not perfect,
managers and investors do not have access to the same information regarding the
company’s investment opportunities. This information asymmetry means that value is
not always reflected in share prices, which makes managers willing to share their own
knowledge in order to achieve a comparable market value equivalent to the real worth
of the company (Michaely et.al., 1995).
In this context, dividends can be used by managers as a signalling mechanism to convey
their insider information to the market (Miller & Rock, 1985 and John & Williams,
1985). This dividend signalling theory ensures that the market views dividends as a
signal of a management’s point of view regarding the firm’s future profitability, and
share prices respond accordingly. Figures of earnings and dividends may be two of the
most significant available signals (Aharony & Swary, 1980), and dividends can be used
43
as a simple, concise signal of the management’s interpretation of the company’s recent
performance and future forecasts (Asquith & Mullins, 1986).
Empirical research
would tend to support this view since share price is positively related to an increase and
negatively related to a decrease (or omission) of dividends (Pettit, 1972, and Bali,
2003).
An explanation for this signalling view of dividends is provided by the partial
adjustment framework; it puts forward a reason as to why a company would try to
smooth its dividends.
In ground-breaking research, Lintner (1956) characterises
corporate dividend behaviour as a partial adjustment model, by which dividends are
observed to be sticky and companies pay a great deal of attention to the stability of their
dividends.
Thus, managers try to smooth dividends over the years and they do not
increase their dividends unless they feel confident about the future earnings of the
company (Kumar & Lee, 2001). A dividend increase can be perceived as a signal of a
substantial increase in the distribution of earnings, and consequently an increase of the
given firm's value (Aivazian et.al., 2003).
More specifically, Lintner (1956) formed a quantitative model to capture smoothing in
dividends via a partial adjustment towards a target payout ratio.
He argued that
managers are not willing to cut dividends as they believe that if they do so, the
reputation of their company will be affected in a negative way (i.e. a negative signal)
which in turn affects the firm’s value.
Many empirical investigations have argued that because markets are not perfect and
there are agency problems as well as information asymmetry, firms’ investment
decisions and financial decisions are related and dependent on each other. Gordon
(1962) and Adedeji (1998) stated that leverage, investment and dividend decisions are
connected and the major factor that affects the previous decisions is the profitability of
the company. Based on the signalling theory we would expect profitable companies to
smooth their dividends faster than their less profitable counterparts.
In addition, Baskin (1989) concluded that the asymmetry of information restricts the
company’s ability to raise funds externally which suggests that this will affect the
dividend decision. As well, Lintner (1956) argued that the main reason why companies
44
smooth their dividends is the need to reduce the information asymmetry. Diamond &
Verrecchia (1991) stated that generally the regulators and the market pay more attention
to large companies given that they are more mature and more likely to disclose full
information about the performance of the company than small size companies;
information asymmetry is less for large firms. Based on the signalling theory we expect
small size firms to adjust their dividend faster than large firms to reduce the
information asymmetry and signal their confidence in the financial position of the
company.
Another explanation of why companies smooth their dividends has been provided by
agency theory. Jensen & Meckling (1976) contradicted the M & M irrelevance theory
by arguing that there is a conflict of interest between the managers and shareholders;
thus, the managers may not act exclusively on behalf of shareholders. In addition,
Jensen (1986) argued that the free cash flow gives the managers the incentive to invest
in less profitable projects which increases the cost of the agency problem. Frankurter &
Wood (2002) confirmed Jensen’s (1986) predictions that using dividends or increasing
the leverage of the company would reduce the free cash flow available to managers and
consequently reduce agency costs. Dividend and debt can be used as a combined
strategy according to Jensen (1986) where they help in aligning the different interests
between managers and shareholders and mitigate the agency problem.
Therefore,
companies with severe agency problems are expected to adjust their dividend. Based on
the agency theory, we can expect that high leveraged firms to adjust their dividends
toward the target slower than their low leveraged counterparts. Table 3.1 summarises
the expected impact of the different firm characteristics on the smoothness of dividends.
Different Corporate Theories
Firm characteristic
Impact on smoothing
Signalling
Profitability
positive
Size
negative
Leverage
negative
Agency theory
Table 3.1: The impact of different corporate theories on the smoothness of dividends
45
3.3 Partial Adjustment Models
The partial adjustment mechanism recognises that a firms’ observed dividend payment
ratio will not always be equal to its target level.
This means that companies change
their dividend and adjust to their target if the costs of getting closer to the target
dividend payout ratio are lower than the costs of staying away from the target. The
reason is rational, implying that dividend payments are not instantly adjusted to their
target level until the benefit of moving towards that target at least outweighs its costs.
Typically the benefit is the positive market reactions and consequently the increase in
the market share price, while the cost would be the transaction costs.
But the
assumption is that the adjustment benefits as well as the costs of reducing and
increasing dividends are symmetrical, when adjusting towards the target dividend
payout ratio; this view does not differentiate between firms which are below and above
their target dividend payout ratios.
Consequently, there are two types of partial
adjustment model; symmetric and asymmetric.
The following subsections will outline the symmetric and asymmetric partial adjustment
models in order to empirically investigate if the non-financial Jordanian companies in
the industrial and services sectors have a target dividend payout ratio and if they move
steadily to their target dividend payout ratio. The models used in this chapter were
tested and analysed using different econometric techniques, such as the pooled and
panel data techniques. The panel data can be estimated by the use of a fixed or random
effects technique and helps detect the effects of the firm- and time-specific
heterogeneities.5
3.3.1 Symmetric Adjustment model
The main assumption of the target dividend payout ratio is that companies try to find the
optimal balance between the benefit, which is mainly the increase in the share price
when the shareholders react positively to the payment of dividends, and the cost, which
typically would be the transaction cost.
Furthermore, Leary & Michaely (2011)
suggested that in the absence of transaction costs, a company’s observed dividend
5
It is worth noting that the most important advantage of using the panel data approach is that it usually
gives a large number of observations, increases the degrees of freedom and hence, improving the
efficiency of the econometric estimates.
46
payout ratio should be its long-run target or optimal ratio. According to Lintner (1956)
and Fama & Babiak (1968) the existence of the adjustment cost may prevent companies
from returning back to their targets instantly.
However, less developed financial
markets face financial constraints, creating financial obstacles which make partial
adjustments toward the target possible.
The following model (3.1) will be tested as the symmetric adjustment model to achieve
the first objective of this chapter; empirically investigating the difference between the
symmetric and asymmetric dividend payout ratio models. To investigate the symmetric
adjustment model, we need to construct the dividend deviation variable Ddev i ,t to
determine if the actual lagged dividend payment deviates from the target dividend
payout ratio. As mentioned before, the Ddev i ,t can be calculated by subtracting the
lagged actual dividend payment from the target dividend payout ratio of the current year
D *i ,t  Div i ,t 1 .6 The previous literature has suggested two main proxies for the target
dividend payout ratio, either the mean of the industry during the period under
investigation or the median of the dividend payout ratio for each firm during the period
under study (Leary & Michaely, 2008). In this empirical investigation we used the
median of the dividend payout ratio7 and for the robustness of the results we checked
our findings with the industry mean.
Based on the preceding discussion, the symmetric adjustment model can be presents as:
Div i ,t   0   1 Ddev i ,t   i ,t
(3.1)
Where ΔDivi,t is the change in the dividend payment ( Div i ,t  Div i ,t 1 ) and the ɛi,t is the
error term. The adjustment coefficient is  1 which captures the adjustment in dividend
changes to the target dividend payout ratio.
We hypothesise that the adjustment
coefficient in model (3.1) lies between 0  1  1 which suggests a partial adjustment
towards the target; thus, companies do not react immediately to adjust their dividend
6
The target dividend payout ratio is calculated as the multiplication of target ratio
ri by the current profits
Pi ,t as follows D *i ,t  ri Pi ,t
7
I chose to use the median rather than the mean, because the mean has a disadvantage of getting affected
by the outliers.
47
payout ratio to the target payout ratio.
Indeed, if 1  1 , the actual changes in
dividends correspond with the desired changes. Conversely, if 1  0 , no changes in
dividends toward the desired level are undertaken since the actual change at time t is the
same as the one observed in the previous time period. Therefore, the  1 coefficient on
the dividend deviation variable Ddev i ,t is significant, companies would appear to have a
target dividend payout ratio and they adjust their dividends toward the target ratio over
time.
Model (3.1) assumes a symmetrical  1 for dividends below and above the target
dividend payout ratio which suggests that there is no difference in the costs when
moving above or below the target dividend payout ratio. To clear up any differences
between these two scenarios (below and above the target dividend payout ratio), we
develop the asymmetric adjustment model in the next section.
3.3.2 Asymmetric Adjustment model
When the actual dividend payout ratio is below or above the target dividend payout
ratio then this suggests that the companies may experienced different adjustment rates.
According to Lintner (1956), managers usually take decisions that maximize the wealth
of the companies and consequently their shareholders. Thus, managers try to choose the
dividend payout ratio that would help them to achieve the target of value maximization.
Fama & French (2002) confirmed Lintners’ argument and they emphasized that firms
that seek to maximize their value determine their target dividend payout ratio by
comparing the benefits and the costs of the last dollar of dividends. Thus, the main aim
of the managers of such companies is to keep their actual dividend payout ratio at the
target level, which can be achieved by smoothing the actual payment toward the target
level.
In addition, Lambrecht & Myers (2010) pointed out that since the adjustment costs are
present then this restricts the ability of the companies to go back immediately to their
target dividend payout ratio. In other words, this suggests that if a company’s dividend
is above the target dividend payout ratio then this company needs to adjust its dividends
downward, accordingly if the company’s dividend is below the target dividend payout
48
ratio then this company needs to adjust its dividends upward. Leary’s & Michaely’s
(2011) argument is in line with Lintners’ (1956) view that companies smooth the
dividend payment toward the target dividend payout ratio and managers are unwilling to
cut dividends which suggests an asymmetric adjustment to the target dividend payout.
By definition the variable Ddevi,t (Di,t *– Divi,t-1) indicates that when Ddevi,t>0 then the
company’s dividend payout ratio is below the target dividend payout ratio and when
Ddevi,t<0 then the company’s dividend payout ratio is above the target dividend payout
ratio.
Therefore, to empirically investigate if the Jordanian companies have different
adjustment rates when the company is above or below the target dividend payout ratio,
we construct model (3.2) by dividing the values of Ddevi,t into two variables as follows:
Ddevibelow
 Ddevi ,t if D *i ,t  Div i ,t 1  0 and zero otherwise
,t
Ddeviabove
 Ddevi ,t if D *i ,t  Div i ,t 1  0 and zero otherwise
,t
When we substitute the two constructed variables Ddev iabove
and Ddev ibelow
instead of
,t
,t
Ddev i ,t in model 3.1, then the model would be as follows:
Div i ,t   0  1 Ddevibelow
  2 Ddeviabove
  i ,t
,t
,t
3.2
Where ΔDivi,t and ɛi,t as explained in model 3.1, Ddevi,tbelow and Ddevi,tabove are the
dividend payment below and above the target dividend payout ratio respectively.
The above model (3.2) is the asymmetric adjustment model that is used to check if the
rate of adjustment differs when the dividend is above or below the target dividend
payout ratio. In addition, model (3.2) allows the null hypothesis that δ1=0, δ2=0 to be
checked and the joint test of δ1= δ 2 to be investigated; the first hypothesis takes into
consideration that the two coefficients have to be greater than zero to converge and not
equal (δ 1>0, δ 2>0). Thus, the above mentioned adjustment coefficients (δ 1, δ 2) capture
the magnitude of response of dividend payout ratio when it is below and above the
target respectively.
Consequently, we can argue that if the adjustment costs of
increasing dividends are lower than decreasing dividends, (then δ 1> δ 2), then the speed
49
of adjustment for dividends above the target would be slower than for dividends below
the target.
According to Lintner (1956), one of the basic determinants of dividend policy is the
company’s profits.
He hypothesised that usually companies have the tendency to
smooth their dividends based on the achieved profits (cash flow).
For instance,
managers would increase their dividends if they can maintain the same dividend payout
in the future. This view is also supported by Jensen et.al. (1992) and Fama & French
(2002) who found a positive relation between dividends and profitability.
Based on signalling theory, managers usually use dividends to signal inside information
to the market (Miller & Rock, 1985 and John & Williams, 1985).
The dividend-
signalling theory argues that the market evaluates dividends as a signal of a
management’s point of view regarding the firm’s future profitability; and share prices
respond accordingly.
Earnings and dividend information may be two of the most
significant measures available for signalling future prospects (Aharony & Swary, 1980);
dividends can be used as a simple, concise signal of management’s interpretation of a
company’s recent performance and oncoming forecasts (Asquith & Mullins, 1986).
Therefore, we can expect the companies with high levels of profitability smooth (adjust)
their dividend more rapidly than companies that report low levels of profitability.
Therefore, to empirically investigate whether the adjustment above or below the target
dividends varies when companies experience high/low profits, two new variables
(Hprof for high profitability and Lprof for low profitability) have been constructed as
follows:

Hprof = companies with high profitability (interaction between a dummy
variable which equals one for the firms that have a profitability (ROA) greater
than the median, and equals zero otherwise (i.e. zero to the firms that have a
return on assets lower than the median) multiplied by the profitability variable).

Lprof = companies with low profitability (interaction between a dummy variable
that equals one for the firms that have a profitability (ROA) lower than the
50
median, and equals zero otherwise (i.e. zero to the firms that have a return on
assets higher than the median) multiplied by the profitability variable).8
In addition, leverage has been mostly asserted to be an essential element in clarifying a
firm’s dividend policy. Jensen et.al. (1992) supported the argument that leverage and
dividend policy are negatively correlated, suggesting that debt and dividends serve as a
substitute mechanism for mitigating the agency cost of free cash flow by reducing the
amount of funds under management control. Hence, companies with high debt levels
will not need to pay dividends as low debt companies in order to reduce agency costs.
For high leverage firms, paying the fixed charges-including interest payments and
principle amounts-serve to strip management of their free cash flows, and this generally
supports the agency cost theory of dividend policy. Consequently, we can expect that
highly leveraged companies adjust their dividends at a slower rate than low leveraged
companies.
Thus, to empirically investigate whether the adjustment above or below the target
dividends ratio varies when companies experience high/low leverage, two new variables
(Llev for low leverage and Hlev for high leverage) have been constructed as follows:

Hlev = companies with high leverage (interaction between a dummy variable
that equals one for the firms that have a total debt ratio greater than the median,
and equals zero otherwise (i.e. zero for the firms that have a total debt to total
assets ratio lower than the median) multiplied by the total debt ratio variable).

Llev = companies with low leverage (interaction between a dummy variable that
equals one for the firms that have a total debt to total assets ratio lower than the
median, and equals zero otherwise (i.e. zero for the firms that have a total debt
ratio greater than the median) multiplied by the total debt ratio variable).9
Furthermore, the signalling theory suggests that in the presence of information
asymmetry between investors and managers in small firms, the best way to send
8
To check for robustness of results the return on equity has been used as another proxy of profitability to
construct the high/low probitability variables.
9
To check for robustness of results the total debt to equity has been used as another proxy of leverage to
construct the high/low leverage variables.
51
information about the value of the company is dividends. That is, more dividends
should be paid by small companies with high levels of information asymmetry in order
to signal the financial situation of the company (Deshmukh, 2003). The empirical
investigation of Ramachandran & Packkirisamy (2010) found that large size companies
are more likely to pay dividends than their small size counterparts; they relate this
finding to the maturity (age) argument that large firms are more mature than small firms
and pay dividends to keep their good image and retain the confidence of shareholders.
However, in a study done by Ghosh & Woolridge (1988) they concluded that dividends
help small firms to send information to the market which reduces any information
asymmetry which may be present. Therefore, we can expect small companies to adjust
their dividends more than large companies. Accordingly, to empirically investigate
whether the adjustment above or below the target dividends ratio varies for small/large
companies, two new variables (Lsize for large size and Ssize for small size) have been
constructed as follows:

Lsize = large size companies (interaction between a dummy variable that equals
one for the firms that have a natural logarithm of total assets greater than the
median, and equals zero otherwise (i.e. zero for the firms that have a natural
logarithm of total assets lower than the median) multiplied by the natural
logarithm of total assets variable).

Ssize = small size companies (interaction between a dummy variable that equals
one for the firms that have a natural logarithm of total assets lower than the
median, and equals zero otherwise (i.e. zero for the firms that have a natural
logarithm of total assets greater than the median) multiplied by the natural
logarithm of total assets variable).10
10
To check for robustness of results the natural logarithm of market capitalisation has been used as
another proxy of size to construct the large/small size variables.
52
Based on the previous constructed variables and when included in model (3.2), the new
model (3.3) would be as follows:
Divi ,t   0  1Lprof i ,t   2 Hprof i ,t  3 Hlevi ,t   4 Llevi ,t  5 Lsizei ,t   6 Ssizei ,t
  7 Ddevibelow
 8 Ddeviabove
 9 Lprof i ,t * Ddevibelow
 10 Hprof i ,t * Ddevibelow
,t
,t
,t
,t
 11Lprof i ,t * Ddeviabove
 12 Hprof i ,t * Ddeviabove
 13Hlevi ,t * Ddevibelow
 14 Llevi ,t * Ddevibelow
,t
,t
,t
,t
 15 Hlevi ,t * Ddeviabove
 16 Llevi ,t * Ddeviabove
 17 Lsizei ,t * Ddevibelow
 18Ssizei ,t * Ddevibelow
,t
,t
,t
,t
3.3
 19 Lsizei ,t * Ddeviabove
  20Ssizei ,t * Ddeviabove
  i ,t
,t
,t
Where Lprofi,t, Hprofi,t, Hlevi,t, Llevi,t, Lsizei,t, and Ssizei,t are low profitability, high
profitability, high leverage, low leverage, large size and small size respectively as
explained before. Ddevi,tbelow and Ddevi,tabove are already explained in model 3.2.
Ddevi,tabove(Lprofi,t+ Hprofi,t+ Hlevi,t+ Llevi,t+ Lsizei,t+ Ssizei,t) is the interaction
between the dividend deviation above the target with the different firm characteristics.
Ddevi,tbelow(Lprofi,t+ Hprofi,t+ Hlevi,t+ Llevi,t+ Lsizei,t+ Ssizei,t) is the interaction
between the dividend deviation below the target with the different firm characteristics.
In model 3.3 where the coefficients (β9, β10, β11, β12, β13, β14, β15, β16, β17, β18, β19, and
β20) are significant and positive then dividend smoothing exists and these coefficients
should not be equal for an asymmetric adjustment to exist. For example, if the
coefficient (β9) is significant and positive then low profitable companies smooth their
dividends when the dividend payment is below the target payout ratio.
3.4 Estimation Results
The following four subsections discuss and analyse the results of the three previously
developed models.
Section 3.4.1 shows the descriptive statistics of the company
characteristics (high/Low profitability, High/Low leverage and Large/Small size) as
shown in table 3.2. Section 3.4.2 discusses the results of the symmetric adjustment
model while section 3.4.3 explains the results of the asymmetric adjustment model.
Section 3.4.4 analyses the results of model 3.3 which investigates the impact of the
different firm characteristics on the smoothness of dividends.
3.4.1 Descriptive Statistics
The data used for the analysis has been obtained from firms' annual reports which were
found on the ASE’s website. To empirically investigate the smoothing behaviour for
53
the Jordanian non-financial companies during the period 1997-2006, this study has
collected data for companies that pay dividend for at least 5 years of the study period.
This method helps in minimising the likelihood of any spurious results. Therefore, the
total number in the final sample is 65 listed Jordanian firms, out of which 37 were
industrial and 28 were service companies under the condition that all firms have to be
listed over the period under study. The sample excluded the companies that did not pay
dividends for more than five years throughout the period under investigation.
In
addition, companies that have been newly listed on the ASE after 1997 or companies
that have been merged during the study have not been included in the final sample. In
addition, all companies’ financial years had to run from the first of January till the end
of December for the ten years. Therefore, if a company started during the period or had
no past information then it was excluded from the sample. Hausman’s test was used to
choose the best estimation method for the specific dataset and the method which proved
to have estimated the results in the best way will be presented.
The subsequent
subsections will explain and discuss the two different models. The following table
shows the summary statistics of time-series cross-sectional means, minimum values,
maximum values, and standard deviations for variables included in the given model for
the sample firms across the 10-year time period.
Hprof
Lprof
Hlev
Llev
Lsize
Ssize
Mean
0.111
-0.029
0.548
0.184
17.384
15.420
Std. Dev
0.119
0.088
0.101
0.098
1.039
0.666
Min.
0.037
-1.098
0.307
0.001
16.301
11.533
Max.
1.500
0.036
0.976
0.306
20.047
16.290
Table 3.2: Descriptive Statistics
Note: Hprof is the high profitability. Lprof is the Low profitability. Hlev is the high
leverage ratio as. Llev is the low leverage ratio. LSize is the large size of the firms.
SSize is the small size of the firms.
3.4.2 Estimation Results for model 3.1 (The symmetric adjustment model)
Table 3.3 below shows the results of the estimation of the partial adjustment model
(model 3.1) for the industrial and services companies in the Jordanian market. The best
specification of the results presented below is based on the random effect model due to
the insignificance of the Hausman Test and the significant Lagrange Multiplier (LM),
54
which suggests that the hypothesis of the existence of non firm-specific effects is
rejected. The model is significant and the null hypothesis is rejected since all slope
coefficients are not jointly zero at the 1% level.
Dependent variable Div i ,t
Intercept
0.019**
(0.048)
Ddevi,t
0.514*
(0.000)
R2
0.20
p-value (F-stat)
(0.000)
LM Test
49.57
(0.000)
Hausman Test
2.02
(0.889)
Table 3.3: Estimation results of the
symmetric adjustment model
Note: ΔDivi,t is the change in dividend payment and
Ddevi,t is the dividend deviation from the target
dividend payout ratio. Figures in parentheses are the
probabilities of significance based on the standard
errors which are corrected for heteroskedasticity.
The symbols *, ** and ***denote statistical
significance at 1%, 5% and 10% levels respectively.
The results above revealed that the value of the constant term α0 is statistically
significant, positive and its value equal to +0.019. This finding implies that a given
firm resists reducing dividends so not to affect the reputation of the company negatively
and this comes in line with the signalling theory. This result comes in line with
Lintners’ (1956) suggestion that “the constant ... will be generally positive to reflect the
greater reluctance to reduce than to raise dividends”.
In addition, the adjustment
coefficient value of the dividend deviation variable Ddev i ,t is also positive, significant
and its value equal to 0.514. This implies that the Jordanian firms smooth dividends
toward the target. Table 3.4 shows some selected studies to compare the adjustment
rate with different empirical studies.
55
Author
Market
Adjustment Rate
Period of study
Aivazian et.al. (2006)
USA
0.24
1981-1999
Goergean et.al. (2005)
Germany
0.25
1984-1993
Pandey & Bhat (2004)
India
0.71
1989-1997
Aivazian et.al. (2003)
Korea
0.50
1980-1990
Malaysia
0.65
Thailand
0.65
Zimbabwe
0.39
Adaoglu (2000)
Turkey
1.00
1985-1997
Short et.al. (2002)
UK
0.38
1988-1992
Table 3.4: Selected studies from the literature
Based on the above table, the adjustment rate towards the target for the Jordanian firms
(0.514) is higher when compared to developed markets such as the US, Germany and
the UK where adjustment rates of 0.24, 0.25 and 0.38 are reported respectively which
suggests that the Jordanian companies adjust their dividend faster (on average over 2
years). On the other hand, the Jordanian companies are slower when compared to
companies in developing markets such as India, Thailand and Turkey with adjustment
rates of 0.71, 0.65 and 1 respectively.
The main reason for the difference in the adjustment rate between Jordan, the developed
and developing markets can be found in the arguments of Al-Malkawi (2007). He
argued that since 1990, Jordan has followed a policy of financial liberalization; this has
allowed Jordanian companies to raise their required funds through easy access to the
capital market. Consequently, this had increased the possibility of smoothing dividends
toward the target since the dividend decision is not constrained by funding
requirements.
Taking into consideration, the fact that the results reported for the
Turkish companies suggest that there is no smoothing towards the target and companies
do not follow stable dividend policy.
The finding of Adaoglu (2000) means that
56
Turkish firms during the period under study experienced no transaction costs or any
other adjustment costs which facilitated a complete adjustment toward the target
dividend payout ratio. An interesting conclusion that can be drawn is that the Jordanian
firms do move toward the target dividend payout ratio in a moderate fashion, however.
3.4.3 Estimation Results of model 3.2 (The Asymmetric partial adjustment model)
In this section, the asymmetric partial adjustment model has been examined to
investigate if the adjustment rate varies when dividend payment is below or above the
target. The best specification of the results presented below is based on the random
effect model due to the insignificance of the Hausman Test and the significance of the
Lagrange Multiplier (LM), which suggests that the hypothesis of the existence of no
firm-specific effects is rejected. The dependent variable is the change in dividend
Div i ,t . The explanatory variables are the dividend payment below the target dividend
payout ratio Ddev ibelow
and the dividend payment above the target dividend payout ratio
,t
. The model is significant and the null hypothesis is rejected since all slope
Ddev iabove
,t
coefficients are not jointly zero at the 1% level.
Dependent variable Div i ,t
Intercept
0.021
(0.005)
Ddevi,t below
0.651*
(0.000)
Ddevi,t above
0.327*
(0.000)
R2
0.24
p-value (F-stat)
(0.000)
N
585
LM Test
6.98
(0.008)
Hausman Test
1.03
(0.597)
Table 3.5: Estimation results of the asymmetric
adjustment model
57
Note: ΔDivi,t is the change in dividend payment. Ddevi,tbelow is the
dividend payment below the target dividend payout ratio. Ddevi,tabove
is the dividend payment above the target dividend payout ratio.
Figures in parentheses are the probabilities of significance based on the
standard errors which are corrected for heteroskedasticity. The
symbols *, ** and ***denote statistical significance at 1%, 5% and
10% levels respectively.
Table 3.5 presents the estimated adjustment coefficients (δ
1
and δ 2) and shows
significant positive adjustment coefficients for both dividend deviations above and
below the target; they are both significantly different. 11 This result suggests that the
dividend adjustment in Jordan is asymmetric rather than symmetric; the adjustment
coefficient for the dividend below the target (0.651) is higher than the adjustment
coefficient of the dividend above the target (0.327). The higher adjustment rate for
below the target than the adjustment rate for above the target suggests that the Jordanian
companies are more interested in increasing rather than cutting dividends. This result
implies that the benefit of increasing dividends (agency cost) is higher than that of
decreasing them. The behaviour involving a reluctance to reduce dividends is in line
with the signalling theory and confirms that the Jordanian companies try to signal good
news to the market and make the shareholders feel more confident about the future cash
flow of the company. In addition, this result confirms Lintner’s (1956) finding for
American firms, Pandey’s & Bhat’s (2004) result for Indian firms and Aivazian’s et.al.
(2003) finding for different emerging markets. Although these studies have examined
dividends during different time periods as shown in table 3.3 they still confirm that
managers think of dividend as a signalling tool.
3.4.4 Estimation Results of model 3.3 (The Asymmetric partial adjustment model
including the interaction terms)
The asymmetric partial adjustment model has been examined while taking account of
the different firm characteristics (size, profitability and leverage) to check if the
adjustment rate varies when high/low leveraged firms, large/small firms and high/low
profitable companies make dividend payments below or above the target.12 The best
The null hypothesis that the two coefficients of the dividend deviation for below and above target δ 1=δ 2
is rejected since the wald test is significant at the 1% level (12.56 and p-value of 0.000)
12
The null hypothesis that the coefficients of the interaction between the firm characteristics and the
dividend deviation for below and above target is rejected since the wald test is significant at the 1% level.
11
58
specification of the results presented below is based on the random effect model due to
the insignificance of the Hausman Test and the significance of the Lagrange Multiplier
(LM), which suggests that the hypothesis of the existence of no firm-specific effects is
rejected.
The results reported in table 3.6 suggest that the dividend adjustment in the Jordanian
firms during the period (1997-2006) is not only asymmetric for below/above the target
but also asymmetric for the upward/downward dividend adjustments when companies
experience high/low profits, high/low leverage and even for large/small size companies.
The positive coefficients (β7, β8, β9, β10, β11, β12, β13, β14, β15, β16, β17, β18, β19 and β20) are
significant and these coefficients are not equal which suggests the existence of
asymmetrical adjustments in the Jordanian companies. In addition, the coefficients (β2,
β3 β4, β5 and β6) are significant which suggests that the small/large size, high/low
leverage as well as highly profitable Jordanian companies are unwilling to cut
dividends; the only exception is the insignificance of the variable Lprofi,t which implies
that the Jordanian companies are willing to cut their dividends when they experience
low profitability.
59
Dependent variable Div i ,t
Independent Variables
Beta
Coefficient
Intercept
0
0.051***
probability
(0.065)
2
3
0.698**
(0.029)
0.195**
(0.055)
0.254*
(0.000)
Lsizei,t
4
5
0.650*
(0.000)
Ssizei,t
6
0.540*
(0.000)
Ddevi,tbelow
7
0.759*
(0.000)
Ddevi,tabove
8
0.258*
(0.000)
Lprofi,t*Ddevi,tbelow
9
0.348*
(0.000)
Hprofi,t*Ddevi,t below
10
0.579*
(0.001)
Lprofi,t*Ddevi,tabove
11
0.185*
(0.000)
Hprofi,t*Ddevi,tabove
12
13
0.390**
(0.055)
0.385**
(0.030)
0.853*
(0.000)
Hlevi,t*Ddevi,t above
14
15
0.353***
(0.074)
Llevi,t* Ddevi,t above
16
0.435***
(0.062)
Lsizei,t*Ddevi,t below
17
0.550*
(0.000)
Ssizei,t*Ddevi,t below
18
0.350*
(0.000)
Lsizei,t* Ddevi,t above
19
0.560**
(0.025)
Hprofi,t
Hlevi,t
Llevi,t
Hlevi,t*Ddevi,t below
Llevi,t*Ddevi,t below
(0.014)
 20
Ssizei,t* Ddevi,t above
0.240*
R2
0.29
p-value (F-stat)
(0.000)
(0.001)
LM Test
10.37
Hausman Test
12.64
(0.449)
Table 3.6: Estimation results of the asymmetric adjustment model with firm characteristics
interaction (profitability, leverage and size)
Note: ΔDivi,t is the change in dividend payment. Ddevi,tbelow and Ddevi,tabove are the dividend payment below/ above the
target dividend payout ratio respectively. Hprofi,t*Ddevi,t below and Lprofi,t*Ddevi,t below: are the interaction term between
the high profitability and low profitability with the dividend deviation below the target respectively. Hlevi,t*Ddevi,t below
and Hlevi,t*Ddevi,t above is the interaction term between the high leveraged companies with the dividend deviation below
and above the target respectively. Llevi,t*Ddevi,t below and Llevi,t*Ddevi,t above: are the interaction terms between the low
leveraged firms with the dividend deviation below and above the target respectively. Lsizei,t*Ddevi,t below and
Lsizei,t*Ddevi,t above: are the interaction terms between the large size companies with the dividend deviation below and
above target respectively. Ssizei,t*Ddevi,t below and Ssizei,t*Ddevi,t above: are the interaction terms between the small size
companies with the dividend deviation below and above target respectively. Lprofi,t, Hprofi,t, Hlevi,t, Llevi,t, Lsizei,t, and
Ssizei,t are low profitability, high profitability, high leverage, low leverage, large size and small size respectively. Figures
in parentheses are the probabilities of significance based on the standard errors which are corrected for heteroskedasticity.
Results are robust; we run regression with the same variables but constructing them with other proxies and the results did
not change. The symbols *, ** and ***denote statistical significance at 1%, 5% and 10% levels respectively.
60
The following subsections will explain the impact of firm characteristics (size, leverage
and profitability) on dividend smoothness.
3.4.4.1 The impact of large/small size
As I explained when developing the model in the previous section (section 3.3.2),
signalling theory suggests that dividends are the best way to send information about the
value of the company in the presence of information asymmetry in small firms (Eddy &
Seifert, 1988).
Specifically, small companies with a great deal of information
asymmetry should smooth their dividends more than large firms in order to signal the
strong financial situation of the company (Deshmukh, 2003). Thus, the assumption is
that dividends help small firms to send information to the market more than large firms
in order to move smoothly towards the target dividend payout ratio. No empirical
investigation has been done to the best of this author's knowledge having tackled the
topic of signalling and smoothness of dividends in Jordan. This Chapter hypothesizes
that small firms adjust their dividends toward the target dividend payout ratio in the
Jordanian market faster than the large companies.
The results above show that different smoothing behaviour is followed by small/large
Jordanian firms.
Lsize*Ddevi,t
below
The estimated coefficients on Lsize, Ssize, and its interactions
, Ssize*Ddevi,t
below
and Ssize*Ddevi,t
significant at 1 % level and Lsize*Ddevi,t
above
above
are found to be statistically
at 5% level. The positive estimated
coefficients ( 17  18 , 19 and 20 ) suggest that large Jordanian firms smooth their
dividends and adjust to their dividend payout ratio faster than smaller firms. This result
confirms the argument of Al-Najjar & Hussainey (2009) that small firms in Jordan may
experience more transaction costs than large firms since they may have to raise funds to
pay their dividends through issuing debt. Therefore, one could conclude that large firms
adjust their dividends faster than small firms. The results also show that the size
variable for large/small firms is found to be statistically significant and positive at the
1% level, this suggests that the size of the firms has an impact on the smoothness of
dividends. In Jordan, large firms can raise funds easier by having greater access to the
market (assuming that they are listed) and they are more mature than small size firms
and can therefore borrow more easily.
In addition, the regulators and the market
61
concentrate more on large firms than small ones because they disclose more (Al-Najjar
& Hussainey, 2009).
This finding suggests that the size characteristic of the firm has a considerable effect on
the smoothness of the dividend policy of Jordanian firms.
In addition, this result
supports prior research from Deshmukh (2003); although he investigated a sample from
a different market (73 companies in the American market during the period 1990-1997),
he found that the possibility of large companies smoothing their dividends is higher than
small firms. Thus, his findings suggest that large American firms move towards the
target dividend payout ratio faster than small ones. In another study using American
data, Fama & French (2001) investigated dividends for the period during 1978-199 and
concluded that larger and more profitable companies are more likely to smooth their
dividends compared to small size companies since large firms can afford to pay more
dividends through their profits and in case they need funds then they have easier access
to the market.
In addition, the results in this thesis confirm the predictions advanced by the theoretical
model developed by Redding (1997), who suggested that the probability of large and
liquid companies paying dividends is higher than the possibility for small firms because
paying more dividends would lower the available liquidity to managers and this would
reduce the agency problem. Moreover, our results support the findings of Eriotis (2005)
who empirically investigated the Greek market during the period 1996-2001 and stated
that smoothing dividends and the adjustment toward a target dividend is dependent on
the size of firms; larger firms adjsut dividends faster than small firms.
On the other hand, the results of this chapter contradict signalling theory which assumes
that companies with high levels of information asymmetry need to pay more dividends
to convey the information about the company in comparison with the firms which
experience lower levels of information asymmetry. The findings of this thesis contrast
with the results of Ahmed & Javad (2009) where the authors empirically investigated
the Karachi Stock Exchange (KSE) in Pakistan during the years (2001-2006); they
concluded that large companies do not pay dividends which are as stable as those paid
by smaller firms; this is related to the investment behaviour of Pakistani firms; large
62
companies tend to invest more in their assets instead of paying dividends to grow more
in the market and be more powerful than other companies.
3.4.4.2 The impact of high/ low Leverage
In prior literature, leverage has been documented as an essential element in explaining a
firm’s dividend policy. However, mixed results have been presented in the related
literature about whether leverage has a negative/positive influence on dividend policy.
As discussed in section 3.2, the agency cost argument raised by Jensen et.al. (1992)
supports the idea that leverage negatively affects the payment of dividend. He argued
that dividends and debts are two alternatives to solve the agency cost problem; reducing
the amount of funds under management control. Thus, there is a low possibility that
high leveraged company paying more dividend than a low leveraged firm; high
leveraged companies need to service their loans by paying the fixed charges-including
interest payments and principle amounts.
This generally supports the agency cost
theory of dividend policy. Therefore, it was proposed that high leverage firms adjust
their dividends toward the target dividend payout ratio slower than low leveraged ones
in the Amman Stock Exchange.
Table 3.5 provides the estimation results of model 3.3 for the Jordanian companies
under investigation. The findings in this table support the hypothesis that different
dividend smoothing behaviours are followed by the high/low leveraged Jordanian firms.
The estimated coefficients on Hlev, Llev, and their interactions Hlev*Ddevi,tbelow,
Llev*Ddevi,tbelow are found to be statistically significant at the 1 % level. The estimated
coefficient for the interaction variable Llev*Ddevi,t
statistically
significant
at
the
10%
level.
above
The
and Hlev*Ddevi,t
estimated
above
is
coefficients
(13  14 , 15 and16 ) suggest that high leveraged firms smooth their dividends and
adjust to their dividend payout ratio at a slower rate than low leveraged firms.
Therefore, one could conclude that the dividend adjustment for the low leveraged firms
is faster than for the high leveraged ones; the transaction cost of moving towards the
target dividend payout ratio for the high leveraged companies maybe higher than the
low leveraged ones. The results also show that the leverage variable for high/low
leveraged firms is found to be statistically significant and positive at the 1% level. This
63
finding suggests that the leverage characteristic of the firm has a considerable effect on
the smoothness of the Jordanian firm’s dividend policy. Our results are in line with the
predictions of agency cost theory; companies in Jordan use dividends to cut down on
the funds which are available in hands of managers so as not to invest in unprofitable
projects in order to reduce the agency cost. Thus, our results suggest that low leveraged
Jordanian companies adjust their dividends toward the target payout ratio faster than the
high leveraged ones. This result is in line with Al-Shubiri’s (2011) argument regarding
the high leveraged Jordanian firms; he found that high leveraged companies in Jordan
suffer from high transaction costs which lead to them having a weak financial position
since they are not able to pay higher dividends and stay away from raising more debt.
Faccio et.al. (2001) arrived at the same results; they reported a significant relation
between leverage and dividend payment suggesting that high leveraged firms pay a
lower rate of dividends. Thus, low leveraged companies get to their target dividend
faster than high leveraged ones. Faccio et.al. (2001) empirically investigated data for
different counties in Asia and Europe (Taiwan, Singapore, Hong Kong, Thailand, Japan,
Indonesia, Philippines, South Korea, Malaysia, Germany, France, Spain, United
Kingdom and Italy) during the period 1992 till 1996 and they came to the same
conclusion as the current thesis. They argued that high leveraged companies find
themselves out of cash to pay dividends and not in the position to raise more funds; thus
they reduce their rate of adjustment toward the target.
Furthermore, Gugler & Yurtoglu (2003) examined data for German companies during a
seven year period (1992-1998) and argued that high leverage firms smooth their
dividend at a lower rate than others in the market; to have enough cash flow in order to
meet their financial obligations. Gugler & Yurtoglu (2003) supported the arguments of
Baskin (1989) and Allen (1993); they argued that paying a dividend would increase the
company’s need to generate funds externally to finance new investment opportunities
because paying a dividend reduces the amount of funds available internally through
retained earnings.
On the other hand, our results contradict the predictions of signalling theory; when high
leverage companies move to their target payout faster than low leverage firms, those
companies try to signal that even though they need to service their debt they still feel
64
confident about the future cash flows of the company. Thus, this behaviour makes
shareholders feel more confident about the future of such a company. Utami & Inanga
(2011) contradict our result; they examined data from the Indonesian market during the
period of 1994-2007, and argued that, although their results contradicted many
empirical papers, high leverage companies in their sample smoothed their dividends
more (and approached their target payout ratio more quickly) than low leverage ones;
managers in Indonesian firms try to signal their favourable financial expectations of
future cash flows and their confidence to service their debts.
3.4.4.3 The impact of high/low profitability
As discussed in section 3.2, the dividend signalling theory predicts that managers
usually use dividends as a tool to signal their confidence in the firms future of profits
(cash flows) for a given market; consequently share prices react positively (Lintner,
1956). According to Juma’h & Pacheco (2008), one of the basic firm characteristics
that affect dividends is the company’s profitability. Their analysis was supported by
Fama & French (2002) who argued that profitability affect the payment of dividends
positively; companies with high profits tend to have better cash flow and are more
likely to move their dividends towards the target dividend payout ratio faster.
In
addition, Gill et.al. (2010) also confirmed that most of the empirical papers have agreed
on the rate of profits and dividends in signalling information available to shareholders
about the company future financials. Hence, it is proposed that companies with high
profits to move their dividends to the target payout ratio faster than companies with low
profits in the Amman Stock Exchange.
The results support the hypothesis that different dividend behaviours are followed by
the high and low profit Jordanian firms, where the estimated coefficients on Hprof, and
its interactions Hprof*Ddevi,tbelow, Lprof*Ddevi,tbelow and Lprof*Ddevi,tabove are found to
be statistically significant at the 1 % level and Hprof*Ddevi,tabove at the 5% level. The
positive estimated coefficients ( 9  10 , 11and12 ) suggest that high profitable firms
adjust their dividends and move to their dividend payout ratio faster than low profitable
firms; the transaction cost of moving towards the target dividend payout ratio for the
high profitable companies is lower than the low profitable ones. The results also show
65
that the variable of high profitability is found to be statistically significant and positive
at the 5% level, suggesting that profitability affects the dividend smoothness in Jordan.
These results are in line with signalling theory suggesting that the managers of highly
profitable firms tend to signal their greater confidence in the future cash flows and to
assure the shareholders that the company will keep smoothing dividends toward the
target payout ratio. This result also agrees with the findings of Al-Malkawi (2007)
where he concluded that highly profitable Jordanian companies pay higher dividends
than less profitable firms.
Thus, this suggests that highly profitable Jordanian
companies experience better cash flows and this allows them to move to the target
dividend payment ratio more quickly.
Pandey (2001) arrived at same result when he found that an essential determinant of the
smoothness of dividend policy in the Malaysian market was profitability. He argued
that companies with higher profits are more willing to keep smoothing their dividends;
he related this to signalling theory. In addition, Adaoglu (2000) argued that managers
in Turkey pay close attention to profitability measures when it comes to paying
dividends; more than any other characteristics of the firm.
Furthermore, Abor &
Amidu, (2006) and Franklin & Muthusamy (2010) have empirically investigated the
relationship between dividend payment and profitability in Ghana and India
respectively. Both studies concluded that profitability had a highly significant impact
on a firm’s ability to pay dividends. Thus, we accept the hypothesis that companies
with high profits smooth dividend more than companies with low profits.
3.5 Conclusion
This chapter of the study is an extension of the empirical work published on corporate
dividend policy. The chapter has investigated data for 65 companies (37 industrial and
28 services) listed on the Amman Stock Exchange (ASE) covering the period 19972006. Both pooled data and panel data are used in this chapter as techniques for the
estimation of results. Fixed or random effects techniques are usually used for the
estimation of panel data; in this study, the random effects technique was found to be the
best model.
66
This empirical chapter has examined the smoothness of dividends using symmetric and
asymmetric partial adjustment models. Thus, this chapter has extended the previous
literature by empirically investigating the asymmetric partial adjustment model;
previous studies have assumed a symmetric adjustment toward the target dividend
payout ratio. Jordanian companies tend not to cut their dividends which makes the
downward movement when above the target dividend payout ratio less pronounced.
Secondly, this chapter has investigated the possible firm characteristics which affect the
asymmetric partial adjustment model; the hypothesis of large/small size, high/low
leveraged and high/low profit companies making asymmetric adjustments towards their
target dividend payout ratios was studied.
The results suggest that Jordanian firms move toward a target payout ratio at a
reasonably moderate rate of adjustment. In addition, the process of moving towards the
target adjustment is asymmetrical rather than symmetrical; the Jordanian companies
have different adjustment rates when they are above or below the target. Furthermore, a
number of interesting results have been found in terms of the characteristics of
companies (size, profitability and leverage) and their impact on the speed of adjustment
toward the target dividend.
The impact of size on the smoothness of dividends
contradicts signalling theory, implying that small firms with high levels of information
asymmetry are slower in moving towards the target dividend payout ratio than large
firms. Further, according to agency cost theory, companies with high leverage smooth
their dividends at a lower rate towards the target dividend payout ratio than low
leveraged ones; which indicates that companies in Jordan use dividends to decrease
funds in hands of managers so that they cannot invest in unprofitable projects in order to
reduce agency cost. Moreover, the effect of profitability on the stability of paying
dividends has found to be positive significant and in line with the signalling theory.
The higher the company’s profit the better is the financial position of the firm; so it can
maintain a stable dividend payment.
Thus, it can be proposed that the partial
adjustment model for the Jordanian listed firms is asymmetric and is influenced by the
size, leverage and profitability.
67
4. Empirical Investigation Of The Determinants Of The Two Dividend
Decisions
4.1 Introduction
A theoretical framework was established in Chapter Two along with a discussion about
empirical evidence in Chapter Three regarding dividend smoothness.
The key,
influential and relevant dividend policy theory was advanced by Miller & Modigliani's
(1961) in their Nobel-prize-winning Irrelevance Theory.
Also, a set of opposing
theories were outlined in order to make clear why corporate dividend policy seems to be
of significance in practice. Some theories like the tax effect Hypothesis have emerged
to explain that dividend policy is important because of the presence of different tax
treatments. Others, like signalling theory and the agency theory argue that dividend
strategies followed by companies can be affected by market imperfections like (a)
information asymmetry, which mean that companies can signal information to the
market by making a dividend payment; and (b) agency cost, where dividends can
minimize the cost of controlling the conflict between managers and shareholders. The
debate among these theoretical models still continues.
This chapter will contribute to test the existing literature in different ways. This chapter
is the first empirical investigation to apply Heckman’s simultaneous model and
compares the finding to those from the Tobit and OLS estimation techniques to
determine the possible factors that affect the two dividend decisions (i.e. the decision on
whether to pay dividends or not and the decision about the amount of the dividend
payment). In addition, empirical studies have concentrated on the world's developed
capital markets where the researchers used various estimation techniques to determine
the factors that should affect the dividend decisions. This chapter also adds to the
literature by providing new evidence on the factors that can affect the dividend
decisions of companies in an emerging market (Jordan). This point is essential since the
distain which dividend decisions encounter in the developing markets requires us to
consider the possible variables that might have an effect on dividends in Jordan which is
one of the developing countries. Finally, this empirical investigation will compare the
results obtained about Jordan to the published empirical investigations of the developed
markets.
68
Investigating the determinants of dividend decisions in the Jordanian market is certainly
important since Jordan possesses some features that are closely related to those of
developed countries.
Such features would include an adequate level of corporate
governance and a developed bank-based financial system. The ASE is growing strong
in terms of market capitalization and trading volume, yet high level of ownership
concentration might reduce agency costs and information asymmetries; however, this
may lead to conflicts between large major owners and small minor shareholders
(Amman Financial Market, 2008). These characteristics are used to explain our results,
because they have several implications for the corporate dividend decisions in Jordan.
The reminder of this chapter is organised as follows. Section 4.2 discusses the different
possible determinants of dividend decisions, their related hypotheses and the proxies
used to represent such variables. The model specification is described in Section 4.3. A
discussion of the analysis and the results of the different estimation techniques are
presented in section 4.4. The last section (section 4.5) gives a conclusion to the chapter.
4.2 Determinants of the Two Dividend Decisions
This part of the chapter discusses and explains the factors that might affect the two
dividend decisions by formulating five testable hypotheses. In addition, every factor
will be connected to the relevant theories to test the chosen proxy.
4.2.1. Determinants of dividend decisions- theoretical background and definition
Many research papers have been written about the factors which might affect dividend
decisions. In previous research, there are several variables that have been used as a
proxy for dividends. The proxies used throughout this literature are the dividend payout
ratio, the dividend to total assets ratio, the dividend yield and the natural logarithm of
actual dividend payment.
The independent variables include; profitability, size,
leverage, ownership and growth (Al-Malkawi, 2007; Al-Kuwari, 2009 and Agyei &
Marfo-Yiadom, 2011 among others). The dividend payout ratio (calculated by dividend
over net income) is employed in this empirical research to investigate the factors which
69
can influence dividend decisions. The use of the dividend payout ratio rather than the
other proxies mentioned is reasonable because this ratio takes into consideration the
dividend paid out of the net income of the company. Such a consideration is important,
because the managers of different companies pay dividends out of the companies’
profits. In addition, as revealed in the previous empirical chapter and confirmed by
Lintner (1956), the dividend depends partially on the firm's current earnings and
partially on the lagged dividend. In addition, the use of the dividend yield as a measure
of dividend decisions is questionable due to the possibility of share pricing problems13
of this ratio (Aivazian et.al., 2003). To make sure that the results are robust, the natural
logarithm of actual dividends of the company will be used as another proxy for dividend
decisions (Redding, 1997).
4.2.1.1 Profitability
The basic idea behind the choice of this variable dates back to Miller & Modigliani's
(1961) argument and maintains that for a given business the firm value is determined by
just the earnings influence of the assets used and the investment policy implemented by
the company. The value of this firm can be altered by changing the combination
between dividends and retained earnings. Therefore, companies with good levels of
profitability are capable of achieving sizeable amounts of free cash flows and new
opportunities for firms to invest; then the higher dividend payment does not affect them.
In this respect and in line with the pecking-order theory, companies prefer firstly to use
the internal financing sources, followed by debt and external equity acquired by the
issuance of stocks respectively. If the company achieves more profit then it would have
more internal financing which will allow them able to pay larger dividends.
In addition, the trade off theory argues that companies try to find the balance between
the costs (i.e., flotation costs and financial distress) and the benefits (i.e., reduced free
cash flow problems) of dividends (Fama & French, 2002 and Barclay et.al., 1995). As
firms with more profitability incur lower costs of financial distress and superior free
13
The ASE used to have a weak order-driven market making system before the year 2000. The orders to
sell or buy were prioritized for execution in terms of time and price manually. In mid 2000 the French
government provided the ASE with a new Electronic Trading System (ETS). Consequently, this system
helped in better safety for investors and more transparency. The ETS matches the buy and sell orders in a
good fair order which helped to determine the security prices more accurately.
70
cash flows, other variables remain the same; then the dividend payment is likely to
continue for firms with high profitability. Also signalling theory predicts a positive
correlation between dividend and corporate income which has been well documented by
different researchers (Goergen et.al. (2005) and Faccio & Lang (2002) among others).
The main argument of signalling theory is that the managers of profitable companies
will pay more dividends to signal their confidence about the future cash flows and the
ability to maintain a healthy financial position in the future.
Based on the above
discussion a positive relationship is expected between profitability and the two dividend
decisions.
Several studies have used two proxies for profitability; return on assets ROA (Abor &
Amidu, 2006) or return on equity ROE (Gwilym et.al., 2004; Aivazian et.al., 2003).
We investigate company profitability using the return on assets (ROA) as a proxy. The
return on assets measures the efficiency of management in generating earnings from the
invested capital (assets). The higher the ROA, the better, since the firm is earning more
profit and consequently the better is the position of the firm to pay dividends. In
addition, return on assets has been chosen to facilitate comparability with other
empirical studies. To make sure of the robustness of results, the other proxy (ROE) is
also used since return on equity measures the efficiency of a firm at generating profits
from its shareholders' equity.
4.2.1.2 Size
Many empirical investigations have acknowledged size as an important and positively
related determinant of a firm’s dividend decisions. The related literature claims that
there is an inverse relation between the size of the firm and the possibility of bankruptcy
(Rajan & Zingales, 1995; Ferri & Jones, 1979; Titman & Wessels, 1988). Specifically,
large companies have easier access to the market and can borrow at a competitive rate14.
In addition, small firms may have more conflicts between their shareholders and
creditors than larger firms (Marsh, 1982 and Charitou & Vafeas, 1998).
14
Higgins (1972) asserted that larger firms should be more able to pay dividends than small firms since
they have easy access to capital markets. In addition, Holder et.al. (1998) related the transaction costs to
the size of the firm, in other words, they discussed that large companies may pay higher dividends to
investors than small companies since they have easy access to the market and can borrow funds at low
cost.
71
Furthermore, more diversification and stability of cash flows tend to characterise larger
firms as a result; they will be more willing to payout higher dividends. This proposition
is documented by the findings of several empirical investigations such as Adedeji
(1998), Charitou & Vafeas (1998), and Ooi (2001). Thus, large companies are more
likely to pay dividends as part of their profits since their cash flows or profits are more
stable.
In the case of small companies, which face more volatile cash flows and
earnings, paying high dividends might result in financial problems which might raise
the possibility of having to approach the capital market to raise external debt (an
expansive choice). Denis & Osobov (2008), Barclay et.al. (1995) and Fama & French
(2001) concluded that size affects the dividend in a positive significant way and they
related this to the competitive advantage of large firms compared to small ones. In view
of the fact that larger firms face lower issuing costs and higher agency costs a positive
relationship is expected between size and the two dividend decisions.
In previous studies, size has been measured by several proxies such as the natural
logarithm of total sales, the natural logarithm of total assets and the natural logarithm of
market capitalisation. This empirical chapter uses the natural logarithm of total assets
as a proxy for size since it is an obvious measure; the more assets the company has the
larger is the firm, while the use of the natural logarithm corrects for scale effects by
treating as equal the same percentage variation rather than the same numerical variation;
see for example (Deshmukh, 2003; Redding, 1997; Eddy & Seifert, 1988; Ghosh &
Woolridge, 1988; Rajan & Zingales 1995). Furthermore, the natural logarithm of total
assets has been applied as a size measure to facilitate comparability with other empirical
studies. The natural logarithm of sales is not a good proxy for size since it can be
affected by seasonality.
For robustness of results the natural logarithm of market
capitalisation will be used as a check on the findings for assets.
4.2.1.3 Leverage
Different explanations have been documented in the literature concerning the impact of
leverage on dividend decisions. Based on signalling theory, a positive association is
expected between leverage and dividend decisions since high leverage firms tend to
keep paying dividends even though they need to service their loans, by paying the
principle amount and the interest. Thus, the managers of companies try to signal
72
financial health and confidence to investors about the future of the company. In other
words, the managers keep paying a dividend despite the high leverage because they
believe that the high leverage will make them invest money in profitable projects and be
able to pay the loans, interest as well as the dividends.
On the other hand, many empirical investigations have concluded that dividend is
negatively affected by leverage (Al-Malkawi, 2007; Gugler & Yurtoglu, 2003 and
Faccio et.al., 2001). The investigations carried out by these researchers argue that
highly levered companies try their best to maintain the internal cash flow in order to
meet the firm’s financial obligations and protecting the creditors; instead of paying the
existing cash to their shareholders. Companies may be allowed to distribute extra
dividends from the proceeds from issuing debt (i.e. leverage). Though, such a proposed
increase in external debt would raise the cash flow variability and, therefore, an inverse
correlation is to be seen between leverage and dividend decisions (Aivazian et.al.,
2003).
Moreover, Jensen & Meckling's (1976) agency theory provides us with another
illustration to why leverage might be negatively connected with dividends. Such a
theory implies that dividends and leverage work substitute mechanisms for finding a
solution for the conflict of interest between shareholders and managers. In other words,
Jensen (1986) suggests that firms can use debt as an alternative to dividends to decrease
agency cost. Based on the argument of the agency theory we expect that the payment
and the amount of dividend decisions are negatively associated with financial leverage.
Two proxies have been used in literature to test the effect of leverage on dividend: the
total debt to total assets ratio and the total debt to equity ratio (Al-Malkawi, 2007 and
Mollah, 2001). This chapter employs the total debt to total assets ratio to investigate if
debt has an effect on dividend because this measure indicates the percentage of the
firms’ assets acquired via debt. In other words, the higher the total debt to total assets
ratio the higher is the risk associated in the operations of the company and the lower is
the future borrowing capacity which in turn lowers the firms’ financial flexibility. In
addition, it has been used in the present investigation in order to compare the findings
with results from other empirical studies. The total debt to equity ratio has been used to
check for the robustness of results since it indicates the percentage debt and equity that
73
the firm has used to finance the assets of the company. Thus a high percentage would
make the company suffer from high risk because of the additional interest expense.
4.2.1.4 Ownership Structure
The Dividend decision is one of the financial decisions where recent studies have
investigated the effect of ownership structure. Based on the previous empirical research
done, ownership structure is one of the factors that can affect the dividend decision
(Mancinelli & Ozkan, 2006; Qian et.al., 2007 and Datta et.al., 2005).
In today’s corporate environment, the conflict of interest between insiders (owners),
outsiders and executives of companies can arise because of the separation between
management and ownership.
Jensen & Meckling (1976) analysed the relationship
between managers and shareholders as a contract. Moreover, the manager may take a
decision which will be in their own interest but not in the interest of the firm. La Porta
et.al. (2000) explained that managers might take advantage of the company by using the
authority they have to benefit themselves by allocating firm assets to their own use
through perks; high salaries or the usage of company assets for themselves. Therefore,
the dividend and all other financial policies of companies may be affected by ownership
structure (Gugler & Yurtoglu, 2003; Schooley & Barney, 1994 and La Porta et.al.,
2000). Researchers have recommended that paying a dividend can be essential in
minimizing the conflict between outside and inside owners. Dividends paid by insiders
reduces the free cash in the company so insiders cannot misuse corporate funds
especially for their own benefit; and the investors would get the dividends they deserve
(La Porta et.al., 2000).
Some scholars argue that the controlling shareholder influences the dividend decisions
significantly. Cronqvist & Nilsson (2003) suggest that controlling financial institutions
and controlling corporations are run by professional managers who focus more on
maximizing firm value than the welfare of minority shareholder; controlling families are
assumed to be more prone to extract private benefits from control. Khan (2006) argues
that institutional ownership is associated with higher dividend levels due to lower
agency costs, and that firms controlled by families are associated with lower dividend
levels due to higher agency costs. For example, there is less need to pay dividends
74
when the managers of companies are the owners, simply because there should be no
conflict of interest. Based on the previous discussion, we expect low dividend payment
when the ownership of a company is family controlled simply because using dividends
is less valuable as a way to reduce information asymmetry or agency costs.
Moreover, the large institutional investors can reduce agency costs since they can
monitor the managers of the firm which is an important role to play. Many researchers
have confirmed that large institutional investors play an important role in reducing
agency costs such as Crutchley et.al. (1999) and McConnell & Servaes (1990). The
better financial position of institutional investors relative to small shareholders gives
them the chance to takeover inefficient companies, which increases the obligations on
managers of companies to work in an efficient way. For example, Shleifer & Vishny
(1986) noted that small investors demand that companies pay high dividends to
compensate them and attract the large institutional investors who will keep monitoring
the managers of such companies. In addition, many empirical investigations have
highlighted that large institutional investors prefer to invest in stocks which pay high
dividends (Han et.al., 1999 and Short et.al., 2002). Therefore, we expect a positive
relationship between dividend decisions and the large institutional shareholders.
Two proxies are used in this empirical chapter to investigate the impact of corporate
ownership structure on dividend decisions: the percentage of large institutional
shareholders and the percentage of family ownership (Al-Kuwari, 2009; and Gugler,
2003). We employ the 5 % threshold level of ownership which is the criteria applied by
the Amman Stock Exchange (ASE) throughout the period of this study.
4.2.1.5 Growth Prospects
An analysis of previous empirical investigations has documented that researchers have
explained the relationship between dividend and growth prospects in different ways.
One of the explanations that have been discussed is based on the pecking order theory
that companies who has good growth opportunities and large projects prefer to use the
internal funding sources to pay for investments. These companies tend not to pay or
decrease the dividend just to avoid a situation where they might have to have recourse
to costly external financing.
Alternatively, to stop managers from entering into
75
unprofitable investments, companies with fewer investment opportunities and slow
growth pay higher cash dividends. So, the dividend can play a motivation role, by
diminishing the agency costs of free cash flows and removing resources from the firm
(Al-Malkawi, 2007 and Jensen, 1986).
As a result, several studies have found that
dividends are lower in companies with high growth opportunities (these companies have
lower free cash flows), in comparison to companies with lower growth opportunities
(Jensen et.al., 1992; Rozeff, 1982 and Dempsey & Laber, 1992).
Other researchers have investigated the level of a firm’s investment opportunity set to
differentiate growth from non-growth firms (Gaver & Gaver, 1993; Murali & Welch,
1989; and Titman & Wessels, 1988). One agreed result of these studies is that debt in
growth companies is lower compared to non-growth companies is lower which
minimizes their need for costly external financing. This argument is consistent with
Myers (1984), who noted that investment policy can be a substitute for dividend
decisions; consequently, the free cash flow reduces the agency problem.
In a study
which has tested data from countries with high legal protection, La Porta et.al. (2000)
concluded that fast-growth companies paid lower dividends, as the shareholders were
legally protected; they wait to receive their dividends when the investment opportunities
are good. However, in countries where shareholders have low levels of legal protection,
companies increase the dividend payment; to build a strong reputation for a company,
even though it has good investment opportunities. Based on the previous discussion we
expect that the two dividend decisions are positively/negatively associated with growth
opportunities.
Several proxies have been used in many research studies for growth prospectus such as
the market to book ratio of equity, the change in total assets and the market to book
ratio of assets (Rozeff, 1982; Alli et.al., 1993; Jensen et.al, 1992; and Manos, 2002).
This empirical investigation uses the market to book ratio of equity as a proxy for
growth opportunities for two reasons: firstly, the logic behind this is simple; if a
company’s market value is greater than its book value of equity then shareholders
expect growth and secondly to facilitate comparability with other empirical papers. The
market to book ratio of assets has not been used because of the difficulty of getting the
76
market value of assets.15 To check for robustness of results the change of total assets
will be used as another proxy for growth.
Table 4.1 outlines the main variables 16 used, the definition of each variable and the
expected impact on the two dividend decisions based on the previous subsections.
Proxy
Definition
Expected sign
Prof (ROA)
ROA=return on assets (return/ total assets)
Positive
Size (Ln(TA))
Ln(TA)=Natural logarithm of total assets
Positive
Lev (TD/TA)
TD/TA=total debt to total assets
Negative
Growth (MB)
MB= market to book ratio of equity
Positive/Negative
Ownership Structure (LIS, FS)
LIS= Large Institutional Shareholders
Positive
FS=Family Shareholders
Negative
Table 4.1: The proxies used for the different variables, definitions and the expected
impact on the two dividend decisions.
4.3 Model Specification
Managers and investors treat dividends as an essential and crucial part of the investment
process; therefore, too many research papers have investigated this issue. Even though
dividends have been widely researched, the academic understanding of this topic has
remained limited (Brealey et.al., 2006). Unfortunately, till now no theory has managed
to clarify the different variables which affect dividends. Empirical investigations have
frequently arrived at conflicting results, which have failed to offer a clear explanation.
There might be some reasons behind the disappointing development of this area of our
understanding.
15
In the original paper done by Brainard & Tobin, (1968) then Tobin (1969) the Q ratio has been defined
as the market value of assets divided by their replacement cost. Researchers such as Lindeberg & Ross,
(1981) and Lewellen & Badrinath, (1997) criticized this ratio because of the difficulty in calculating the
market value of assets and their replacement cost.
16
This chapter used the main variables used throughout the literature of dividends. However, different
studies have included more variables such as Baker et.al. (2001) and this has been included as one of the
limitations and an area of future research to be undertaken.
77
For example, Brealey et.al. (2006) suggested that since the dividend decision is related
to all other financing decisions then it is not easy to understand the topic unless you
consider the whole picture. This empirical investigation starts with a discussion that
there is another reason behind getting the different results in most prior studies on
dividends. That is, a majority of the preceding studies do not differentiate or clarify that
there are two stages in the dividend decision process; the payment decision (i.e., to pay
or not to pay) and the amount decision (i.e., how much to pay).
This chapter applies three different estimation techniques: OLS, Tobit and Heckman’s
simultaneous model in an attempt to determine the best model in explaining the factors
affecting dividend decisions in Jordan. Heckman’s (1979) simultaneous estimation
method is applied since it takes into consideration both the likelihood of various
probable determinants affecting each stage of the dividend decision making process at
the same time. Also, the method can exclude potential sample selection bias 17 that
might arise from the fact that in many cases a substantial number of companies do not
pay any dividends.
Explanation of Models
Most empirical studies have investigated the determinants of dividends by employing a
linear regression, fixed effects and random effects model. Such models investigate the
amount of dividends either for the firms who pay dividend (Fama & French, 2002) or to
all the companies in the sample (Rozeff, 1982). Although, in real life scenarios, there
are cases where the sample used in a study may be limited by censoring. On the other
hand, preceding research into dividend that has investigated the factors that affect the
amount of the dividend payment does not take the censoring problem into consideration.
The censoring problem arises due to the fact that a substantial number of firms do not
pay dividend even when most firms do. In addition, a number of studies have often
used limited dependent variable regressions to investigate the decision of the payment
(i.e., probability of paying dividends) (Denis & Osobov, 2008; DeAngelo et.al., 2006
and Fama & French, 2001). Even though this is a useful methodology, it cannot test for
17
Sample selection bias is referred usually to the selection effect; selecting certain firms instead of
randomly selecting the companies in the sample. This can happen when applying the methodology of data
collection.
78
the possibility that the amount of dividends might be determined in a different way than
the payment itself, i.e. it does not differentiate between the two dividend decisions; the
possibility of paying dividends and the determinants of the amount of dividends.
In the case of Jordan in particular, some firms do not pay dividends. Thus, in my
sample, the dependent variable (i.e., the amount of dividend) is continuous only to the
right of zero, i.e. it can take on only non-negative values, with a substantial number of
the observations being exactly zero (i.e., censored). This implies as suggested by Kim
& Maddala, (1992) a censoring problem and requires the use of a Tobit model or
Heckman’s model which is normally suggested as a substitute (Kennedy, 2003 and
Wooldridge, 2001). In the case of a censored dependent variable, a useful analysis can
be performed using Tobit. However, the Tobit technique has the disadvantage of not
being able to account for the two-stage nature of dividend decisions.
The structural equation of the Tobit model can be written as:
𝑦𝑖∗ = 𝑥𝑖′ 𝛽 + 𝜀𝑖
𝜀𝑖 ~𝑁(0, 𝜎 2 )
𝑤𝑖𝑡ℎ
(4.1)
Where: 𝑦𝑖∗ is the dependent variable which is censored for the values less than or equal 0
and observed for values greater than 0. 𝑥𝑖′ is the vector of the explanatory variables
observed for all cases. 𝛽 is the vector of coefficients to be estimated and 𝜀𝑖 is the error
term which is assumed to be independently normally distributed, that is 𝜀𝑖 ~ 𝑁 (0, 𝜎 2 ).
The censored variable, observed over the entire range, is defined by the following
measurement equation:
𝑦𝑖∗
𝑦𝑖 = {
0
𝑖𝑓 𝑦∗𝑖 > 0
𝑖𝑓 𝑦∗𝑖 ≤ 0
(4.2)
When estimating the above structural equation using OLS while having a censored
sample, this gives inconsistent estimates; i.e. it overestimates the slope and
79
underestimates the intercept or vice versa (Woolbridge, 2001 and Gujarati, 2003).18 The
use of Tobit has been suggested by previous studies and when substituting equation
(4.1) into (4.2), this results in:
𝑥′𝑖 𝛽 + 𝜀𝑖
𝑦𝑖 = {
0
𝑖𝑓 𝑦∗𝑖 > 0
𝑖𝑓 𝑦∗𝑖 ≤ 0
(4.3)
Note that the 0’s in the dependent variable can be either a censored observation (i.e.
cannot be observed) or a “true” 0. So as explained above, in our data some of the
observations must be censored, otherwise 𝑦𝑖 would equal 𝑦𝑖∗ and the model would not
be a Tobit model, it would be a linear model and as discussed earlier the use of OLS
would result in inconsistent results.
Therefore, the use of the Tobit Maximum
Likelihood (ML) estimator is recommended19 (Greene, 2003).
On the other hand, Heckman’s simultaneous estimation has been much less frequently
used than other methods of analysis, in spite of its possible advantages in investigating
firm dividend decisions (e.g., OLS, Tobit, logistic, Generalized Methods of Moments
(GMM), etc.). After checking the previous literature, only a limited number of studies
have employed Heckman’s procedure to investigate a company’s dividend decisions
(Kim & Jang, 2010), but the possible advantages of the technique do not seem to be
fully explained by even the work done in these studies. Those empirical investigations
that use Heckman’s procedure mainly do so because of potential sample selection bias,
and they pay no attention to other vital advantages of the procedure. Heckman’s model
allows us to estimate the two stages of the dividend decision while understanding the
differences between those two stages in a simultaneous estimation. Alternatively, this
approach allows for the possibility that some potential firm-specific factors might play
18
The result of Profitability in table 4.4 shows how the results changed when the censoring problem has
been taken into account.
19
The Tobit maximum likelihood function is given by
𝑙𝑛ℒ = ∑ 𝑙𝑛 [𝜎 −1 𝜑 (
𝑦𝑖 >0
𝑦𝑖 − 𝑥𝑖′ 𝛽
𝑥𝑖′ 𝛽
)] + ∑ 𝑙𝑛 [1 − Φ (
)]
𝜎
𝜎
𝑦𝑖 =0
Where the first term is the likelihood function to be estimated when y greater than 0, and the second for
y=0, and Φand φ represent the cumulative density and probability distribution functions of the standard
normal distribution’ (Greene, 2003).
80
different roles in the two stages of the dividend decision making process: the payment
decision (i.e., whether or not to pay dividends) and how much to pay. Consequently,
this research is unlike any other earlier research in terms of the motivation that leads us
to apply Heckman’s procedure.
As already emphasized, this study adopts the
Heckman’s procedure not only to check sample selection bias but also to investigate the
possible differences between the payment decisions and the amount of dividends paid.
Heckman’s (1979) model
Let us consider a model with two latent variables
𝑑𝑖∗ = 𝑧𝑖′ 𝛾 + 𝑣𝑖
(4.4)
𝑦𝑖∗ = 𝑥𝑖′ 𝛽 + 𝜀𝑖
(4.5)
With the error terms 𝜀𝑖 and 𝜐𝑖 are independently and jointly normally distributed with
covariance ρσε , as follows:
(𝑣𝑖 , 𝜀𝑖 )~𝛮 (0, [
1
𝜌𝜎𝑒
𝜌𝜎𝑒
])
𝜎𝑒2
The two latent variables cannot be observed but an indicator 𝑑𝑖 can be measured when
the latent variable 𝑑𝑖∗ is positive. The value if the variable 𝑦𝑖 = 𝑦𝑖∗ is only observed if the
indicator is 1; as follows:
1
𝑖𝑓 𝑑𝑖∗ > 0
𝑑𝑖 = {
0
𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
∗
𝑦
𝑖𝑓 𝑑𝑖 = 1
𝑦𝑖 = { 𝑖
𝑛. 𝑎.
𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
(4.6)
(4.7)
In other words, the first stage in Heckman’s model estimates the variables using the
probit estimation regression that investigates the decision equation of paying dividends
𝑑𝑖∗ (i.e., the likelihood of paying dividends), it takes on a value of one if a firm pays
dividends and zero otherwise. The first step of the probit estimation in Heckman’s
model produces the inverse Mill’s ratio (IMR), λ, by calculating the ratio of the value of
the standard normal density function to the value of the standard normal cumulative
distribution function as follows:
𝜑(𝑧′ 𝛾)
𝜆 = Φ(𝑧′𝑖 𝛾)
𝑖
(4.8)
81
Heckman’s second stage model investigates the relationship between the amount of
dividends and the determinants of dividends. (i.e. the regression equation 𝑦𝑖∗ which
determines the value of 𝑦𝑖 and the IMR obtained from the probit analysis of the firststage). This stage of the model tests how the explanatory variables affect the amount of
dividends, while the potential of the sample selection bias would be corrected. A
significant coefficient for the IMR shows that sample selection bias was present in the
sample but is now corrected.20
The first stage of Heckman’s model, the Probit technique:
DDP = β0 + β1 Lev + β2 Growth + β3 Hprof + β4 Lprof + β5 Size + β6 LIS + β7 FS + υ
(4.9)
Where:

DDP = Dummy of Dividend payment; takes 1 if dividend payment > 0, 0 if
dividend payment equals 0.

Lev = Leverage is the ratio of total debt/total assets.

Growth (MB) = the ratio of the Market to book value.

Hprof = High profitable companies (interaction between dummy variable which
equals one for the firms that have a profitability (ROA) greater than the median,
and equals zero otherwise, multiplied by the return on assets variable).

Lprof = Low profitable companies (interaction between dummy variable which
equals one for the firms that have a profitability (ROA) lower than the median,
and equals zero otherwise, multiplied by the return on assets variable).21

Size = natural logarithm of total assets.
20
In model 9, the industry and time dummy variables are included in the estimation to capture both -the
time and industry-specific effects on dividend decisions.
21
The pecking order theory argued that the firms prefer to use the internal funds then debt and equity
respectively to finance their investments in order to reduce the transaction costs and the costs of
information asymmetry (Myers & Majluf, 1984). This hierarchy of finance may have an effect on the
dividend decision. Thus, low profitable firms will not find it optimal to pay dividends because of the
costs incorporated with issuing debt. However, high profitable firms are in better position to have better
internal funds. This argument has been in line with the descriptive statistics, i.e. the mean dividend
payout ratio for high profitable firms was 50% while 16.5% for lower profitable firms and the mean
leverage for high profitable firms was 25% while 35% for low profitable firms. Therefore, the
profitability affects the dividend amount but most importantly high profitable firms have better
probability in paying dividends than lower firms. Thus, the two measures of high profitability and low
profitability have been included in the first stage and the proxy return on assets (ROA) has been included
as a measure of profitability in the second stage when we applied the simultaneous estimation of
Heckman model.
82

FS = 5% of shares or more owned by the family shareholders.

LIS = 5% of shares or more owned by the large institutional shareholders.
The second stage regression of Heckman’s model:
𝐷𝑖𝑣𝑟𝑎𝑡𝑖𝑜 = 𝛽0 + 𝛽1 𝐿𝑒𝑣 + 𝛽2 𝐺𝑟𝑜𝑤𝑡ℎ + 𝛽3 𝑆𝑖𝑧𝑒 + 𝛽4 𝐿𝐼𝑆 + 𝛽5 𝐹𝑆 + 𝛽6 𝑃𝑟𝑜𝑓 + 𝛽7 𝜆 + 𝜀(4.10)
Where:

Divratio= Dividend payout ratio equals to dividend divided by net income.

Prof (ROA) = profitability is the ratio of net income divided by total assets.

Lambda (𝜆) = Inverse Mill Ratio (IMR) as explained in expression 8.

Lev, Growth, Size, Prof, LIS and FS as defined in expression 9.
After explaining the two stages of the model and explaining our variables for the
possible factors that might have an impact on corporate dividend decisions, a
simultaneous estimation of the above stated two stages model will be analysed by the
use of Heckman’s maximum likelihood function.22
4.4 Statistical Analysis
Based on the selection of variables from international evidence and the availability of
all data the analysis was conducted, all the relevant data were collected for a total of 85
listed non-financial Jordanian companies during the period 1997-2006. The first subsection begins with descriptive statistics for size, growth, leverage, 5% of stocks or
more owned by family shareholders, 5% of stocks or more owned by large institutional
shareholders and the profitability ratio, as shown in Table 4.2. This Table reveals
several noteworthy properties for all the variables such as the means, the minimum and
maximum values together with standard deviations, followed by a discussion of the
data. In the second sub-section, the Pearson Correlation Matrix between the variables
22
Heckman’s maximum likelihood function is given by 𝑙𝑛ℒ = ∑𝑑𝑖=0 𝑙𝑛[𝛷(−𝑧𝑖′ 𝛾)] +
∑𝑑𝑖=1 𝑙𝑛 [𝜎 −1 𝜙 (
(𝑦𝑖 −𝑥𝑖′ 𝛽)
𝜎
)] + ∑𝑑𝑖 =1 𝑙𝑛 [𝛷 (
𝑧𝑖′ 𝛾+𝜌𝜎 −1 (𝑦𝑖 −𝑥𝑖′ 𝛽)
)]
(1−𝜌2 )1/2
83
of the present study is presented for the sample period (see Table 4.3). This is followed
by a final sub-section, which mainly focuses on comparing the results of the three
techniques (OLS, Tobit and Heckman’s simultaneous model) demonstrated in Table
4.4.
4.4.1 Descriptive Statistics
The descriptive statistics describe the collected data for the pooled sample for the nonfinancial firms during the period of 1997-2006. The sample consists of 85 non-financial
companies including 53 industrial and 32 service firms under the condition that all firms
have to be listed over the period under study. The other condition that the data had to
satisfy is that, all companies’ financial years had to run from the first of January till the
end of December for the ten years. Therefore, if a company started during the period or
had no past information then it was excluded from the sample. The data was collected
from the annual reports issued by the firms or the publications of the Amman Stock
Exchange.23
Table 4.2 shows the summary statistics of time-series cross-sectional means, minimum
values, maximum values, and standard deviations for variables included in the given
model for the sample firms across the 10-year time period. ROA is the earnings divided
by total assets; size is the natural logarithm of total assets; LIS is the proportion of
shares owned by the largest institutional investors (i.e. those who own 5% or more of
the given firm's shares); FS is the proportion of shares owned by family investors; Lev
is total debt divided by total assets; and Growth (MB) is the market value of subscribed
shares divided by book value.
23
The total number of firms in ASE was 220 including 136 non-financial firms. Our sample represents
62.5% of the total number of non-financial firms. This implies that 51 companies have been eliminated
for one of the reasons mentioned above.
84
Prof
Size
Mean
0.037
Std. Dev
0.126
LIS
FS
Divratio Lev
MB
16.357 0.371
0.097
0.319
0.307
1.328
1.301
0.275
0.169
0.429
0.220
0.817
Min.
-1.098 11.533 0.000
0.000
-0.747
0.001
0.065
Max.
1.500
0.940
2.365
0.976
5.902
20.047 0.970
Table 4.2: Descriptive Statistics: Dependent and Independent Variables
Note: Divratio is dividend payout ratio, Prof is the return on assets ratio, LEV is the
leverage ratio, Size is the size of the firms, MB is the market to book ratio, LIS is the shares
owned by the large institutional shareholders and FS is the shares owned by the family
shareholders. See operational definitions of the variables.
Prof is evidently the most volatile of all variables, having a standard deviation to mean
ratio of (0.126) compared to (1.301), (0.275), (0.169), (0.429), (0.220) and (0.817) for
the size, shares owned by largest institutional shareholder, shares owned by family
shareholders, dividend payout ratio, leverage and growth ratios, respectively. For the
full period of 1997-2006, the dividend payout ratio averages 31.9 % per year. On
average, the industrial and service firms had a general tendency to pay out 31.9 % of net
income as dividends.
The lowest value for the dependent variable Divratio occurred at (-0.747). What we
could infer from this is that, in a particular year, a particular firm did pay out a dividend
although they suffered net loss. On the other hand, the maximum value tabulated for
Divratio was striking at (2.365), which means again that, in a certain year, a certain
company paid 237% of its net income in form of dividends. 24
The maximum and minimum values of profitability are 1.5, -1.098 respectively, which
shows a high variation in the profitability variable. The mean value is 0.037 is a low
average with a standard deviation of 0.126 which means that the companies in Jordan
were reporting poor levels of profitability. One of the explanations for this is that
Jordan as discussed in chapter one is in the centre of the Middle East and gets affected
easily by any political or unstable movements in the area. For example the Palestinian
and Gulf war’s adversely affected the Jordanian economy since most of the companies
faced a reduction in their exports which reduced their sales and consequently their
24
The firm that paid out dividends although they experienced losses during 2000 is Jordan Tanning
(JOTN). In addition, the striking value of 2.365 has been documented in 2006 for The Jordan Press
Foundation/ Al-Ra’i.
85
profits. The descriptive statistics of this study which are reported in Table 4.2 show that
the mean value of leverage is similar that reported in other developed countries. We
noticed that the mean value of leverage is 0.307 and the standard deviation is 0.220,
which means that nonfinancial companies tend to borrow (short term and long term
debt) on average of 30.7 % of their assets. The maximum and minimum values of
leverage are 0.001 and 0.976 respectively, which show that for one firm a striking
97.6% of the company’s assets have been financed through debt; this is normal in the
service sector in Jordan where firms start their businesses with a loan from the bank.
Our mean level of leverage is similar to that reported by German firms who have a
mean of 0.38 (Rajan & Zingales, 1995). In addition, British companies have a mean
leverage of 0.38 according to Adedeji (1998). Different empirical studies such as,
Flannery & Rangan (2006) and Hovakimian et.al. (2001) have reported a mean value of
leverage of 0.25 and 0.27 respectively, suggesting that American firms use retained
earnings or equity instead of debt. The market to book ratio which measures the growth
variable reported a mean of 1.328 with a standard deviation of 0.817.
4.4.2 Correlation Matrix
Multicollinearity is a possible issue between the independent variables in any data set.
This study applied a test to check if there is a correlation between any of the variables.
Simply if multicollinearity exists then the investigation of the relationship between the
variables would be not possible (Gujarati, 2003). To check for multicollinearity the
variance inflation factor (VIF) is usually used by researchers; this explores the degree
that variables explain each other. Generally, if the VIF results in a value of more than
10 then this suggests a harmful level of collinearity. After applying the VIF test our
results show that no multicollinearity exists since the range of VIF for all variables
between 1.03 and 1.32 with a mean of 1.19.
86
Divratio
Divratio
1.000
LEV
-0.0246
LEV
SIZE
MB
ROA
LIS
FS
1.000
VIF
1.32
(0.4746)
SIZE
MB
ROA
LIS
FS
0.2143
0.4324
1.000
1.27
(0.000)
(0.000)
0.1959
0.0195
0.1304
(0.000)
(0.5704)
(0.000)
0.2383
-0.1216
0.0402
0.2105
(0.000)
(0.000)
(0.042)
(0.000)
-0.0139
0.0778
0.0569
0.1696
-0.0137
(0.6847)
(0.0234)
(0.0975)
(0.000)
(0.6905)
-0.0721
-0.0368
-0.1391
0.0194
-0.0287
-0.3922
(0.0355)
(0.2842)
(0.000)
(0.5728)
(0.4036)
(0.000)
1.000
1.09
1.000
1.03
1.000
1.22
1.000
1.21
Table 4.3: Average Cross-Sectional Correlation Coefficients of Variables
Note: Divratio is dividend payout ratio, Prof is the return on assets ratio, LEV is the leverage ratio,
Size is the size of the firms, MB is the market to book ratio, LIS is the shares owned by the large
institutional shareholders and FS is the shares owned by the family shareholders. The value in the
upper case in each box refers to the correlation coefficient between any two variables, and the
value in the lower case in each box refers to the (Statistical Significance) denoted by the P-value
that the coefficient value is statistically significant at the 1% Level. See operational definitions of
the variables.
The correlation matrix can be used as a second way to test for mulitcollinearity. The
main problems faced when using this as a way to test for multicollinearity is that
researchers have not agreed on a level that can determine if multicollinearity exists or
not. On other hand, different researchers have interpreted correlation values of more
than 0.70 to suggest that muticollinearity may be present (Bryman & Cramer, 2001).
As it is obvious in the correlation matrix above, the correlation coefficients of the
independent variables have a low value which suggests then that the likelihood of a
multicollinearity problem is minimal. With the exception of the correlation between
size and leverage, i.e. correlation value of (0.432), no high correlation appeared between
any of the other variables.25
Table 4.3 above presents the Pearson Correlation Matrix, to depict the correlation
between the set of independent variables, and the dependent variable of the Dividend
payout ratio. Four relationships were found to be statistically significant at the 1%
25
Although fourteen significant correlations have been found but no high correlation appeard and the
highest was between the size and leverage.
87
level, one of which was between the independent and dependent variables: leverage,
size, growth and profitability.
One of the points that we can analyse using the
correlation matrix is that the size has a positive correlation with profitability, leverage
and growth. So we would expect that larger firms have higher profitability and they can
access the market easily to raise more debt and their potential growth opportunities are
greater than other firms.
4.4.3 Comparison of results (OLS, Tobit, Heckman)
To investigate the above relationships, this chapter use the OLS, Tobit and Heckman’s
models.
The following table reports the coefficients, standard errors and the
significance of the variables under investigation, such as growth prospects, ownership
structure, size, profitability and leverage.
Results of hypothesis testing will be
embedded in the analysis, based on the results of the three estimations. This chapter
uses data for a total of 85 listed Jordanian industrial and service firms during the period
of 1997-2006.26
The OLS regression presented in the first column of table 4.4 includes information
about the four independent variables which were found to significantly affect the two
dividend decisions. Thus, the ownership structure variables (two proxies were included;
family ownership and large institutional shareholder) were not reported because they
were found to be insignificant.
The coefficients of the leverage, market to book,
profitability and size were found to be significant and their signs are in line with the
hypotheses of this chapter. Due to the fact that the dependent variable includes a high
number of observations that take the value zero, the Ordinary Least Square estimation is
not ideal.
26
This time period has been selected because of the availability of data.
88
OLS
Tobit
Column 1
Heckman
Column 2
Column 3
Std
Coef
Err
Prob
Coef
Std Err
Prob
Coef
Std Err
Prob
lev
-0.358
0.102
0.001
-0.988
0.230
0.000
-0.195
0.101
0.055
mb
0.022
0.010
0.025
0.053
0.020
0.008
-0.299
0.164
0.069
prof
0.639
0.157
0.000
2.665
0.378
0.000
2.345
0.251
0.000
size
0.070
0.016
0.000
0.252
0.037
0.000
0.069
0.014
0.000
constant
-0.763
0.265
0.004
-4.265
0.604
0.000
2.188
0.268
0.000
lev
-0.755
0.279
0.007
mb
-0.850
0.380
0.025
Hprof
18.739
2.315
0.000
Lprof
2.140
0.410
0.000
size
0.271
0.041
0.000
constant
-4.848
0.721
0.000
0.316
0.019
ρ (rho)
-0.654
0.086
lambda
-0.206
0.038
σ (sigma)
LL
1.063
0.043
-799.94385
-518.3096
Table 4.4: Estimation results27
Note: Divratio is dividend payout ratio, prof is the return on assets ratio, lev is the leverage ratio, Size is the size
of the firms, mb is the market to book ratio, Hprof is high profitable firms and Lprof is low profitable firms. See
operational definitions of the variables.
Based on previous empirical studies, the existence of zero values for the dependent
variable means that our data is censored and needs to be estimated by the Tobit model.
The Tobit results are reported in column two of table 4.4. As the OLS regression
results, the ownership structure variable was found to be insignificant and the rest of the
independent variables were significant but their coefficients are higher. The Tobit
model makes an assumption that our model is censored based on the dividend payout
ratio which is the dependent variable but it does not take into consideration that this
censoring might be attributable to a correlation of the dividend payout ratio with another
variable. For instance, companies before setting the dividend payment must first decide
27
The ownership structure variables (LIS is the shares owned by the large institutional shareholders, FS is
the shares owned by the family shareholders) have been found to be insignificant in affecting the dividend
decisions; therefore their results are not reported in the three estimation techniques.
89
on whether they will distribute dividends or not.
Heckman’s model takes into
consideration this possible correlation of the dividend payout ratio with the probability
of paying dividends by excluding all those companies that decided not to distribute
dividends.
The results of the simultaneous estimation of Heckman model are reported in column 3
of table 4.4 and are consistent with the results of OLS and Tobit model. In addition, the
estimated coefficient of the inverse Mill is significantly different from zero which
provides evidence for sample selectivity. This result contradicts the findings of Kim &
Jang, (2010) who investigated the determinants of the two dividend decisions using the
Heckman two steps approach.
They reported insignificant inverse Mill ratio and
showed that their sample of American firms did not have a selection bias problem. The
correlation between the error terms in the selection and outcome equations presented by
the ρ (rho) is -0.656. The negative ρ (rho) can be interpreted as that any component of
the error that makes the selectivity equation more likely to happen this will affect the
output equation in a negative way. In addition, the standard error of the residual in the
dividend equation which is presented by the sigma (σ) is 0.315. One more interesting
result that the Heckman model provides is the test if ρ=0 by the Likelihood ratio test.
This test compares the joint likelihood of the regression model of the observed dividend
data and the independent probit model for the selection equation against the Heckman
model likelihood, and since χ2=16.86 with a probability of Chi2 =0.000 then this
justifies the Heckman selection equation with the Jordanian data.
Wooldridge (2006) argued that there are three ways to choose between models with
multiple restrictions (the Lagrange multiplier, Wald test and the Likelihood ratio). He
confirmed that the Lagrange multiplier is usually used for binary response models and
rarely to test exclusion restrictions. In addition, since the Wald test only requires the
estimation of the unrestricted model and usually the estimation of both the restricted and
unrestricted models are easy to estimate then the LR test becomes very attractive.
The Likelihood ratio test (LR) 28 between the Tobit model and Heckman’s model is
563.268, which is much higher than the critical value of χ2 distribution with eight
28
The LR test and the F test (which is in the linear regression) follow the same concept, i.e. the F test
measures the increase in the sum of squared residuals when variables are dropped from the model and the
90
degrees of freedom of 20.09 at the 99% level.
This means that Heckman model
provides a statistically significant increase in the likelihood when compared to Tobit
model. The following analysis compares our results with previous empirical studies.
Profitability
To measure profitability, Return on Assets (ROA) was used as a proxy. The results
show an agreement with most of the published empirical studies and confirm our
expectation. We found that profitability determines the dividend payment level and it is
statistically significant at the 1% level. This result confirms the findings of Heckman’s
Probit stage (the payment decision; whether to pay dividend or not) that highly
profitable firms increase the possibility of paying dividends more than low profitable
firms which confirms both the pecking order theory and the signalling theory. Thus,
Jordanian managers use the dividend to signal their confident expectations about the
future of their firms.
In addition, the result is in line with the earlier empirical results reported in the previous
chapter about the stability of dividend. In other words, this result supports the argument
that dividend is positively related to management’s expectations of future profits and
this is due to the fact that executives are not willing to cut dividends (Lintner, 1956;
DeAngelo et.al., 1992; & Baker et.al., 2001). This result is in line with the empirical
work done by Kim & Jang, (2010) who applied the two steps of Heckman’s model to
investigate the determinants of dividends in the American market and found a positive
significant relationship between profitability and the first step of the dividend decision
(i.e. the payment decision). But their findings contradict the result reported in the
second step of the Heckman model that the amount of dividend is not explained by
profitability.
LR test measures the difference between the log-likelihood functions between unrestricted and restricted
models. Thus, a decision can be made after comparing the result of the LR test and the critical value of χ2
since the maximum likelihood estimation maximises the log likelihood function because dropping
variables generally leads to a smaller log-likelihood.
The likelihood ratio test is calculated by:
L(m1 )
𝐿𝑅 = −2 ln (
) = 2(ll(m2 ) − ll(m1 ))
L(m2 )
91
This result has been confirmed by Fama & French (2001) in spite of using a different
estimation technique (logit regression) when explaining dividend data for the American
market. They argued that less profitable companies will not find it optimal to pay
dividends while more profitable firms are more able to generate retained earnings to
finance investments and pay dividends. They reported that profitability has a positive
and significant effect on dividends and they related this positive relationship to the
pecking order hypothesis.29 In addition, our result is in line with the empirical results
reported by Ben Naceur et.al. (2007), although they used two different estimation
techniques (Panel regressions and GMM); they also found evidence that profitability is
one of the main determinants that affects dividends positively in the Tunisian Stock
Exchange. Anand (2004) used a factor analysis framework to check if profitability
affect dividends in the Indian market and found consistent results with the current
findings. Moreover, Al-Kuwari (2009) used the Tobit estimation technique which is
one of the techniques used in this chapter and documented similar results when he
reported that profitability is positively related to dividends. The following researchers
have also reported that profitability is a major determinant of dividend policy (Pandey,
2001, and Abor & Amidu, 2006). Finally, with the aim of checking the robustness of
our results ROE was employed as another proxy for profitability. This alternative did
not yielded results that were different from the findings using ROA.
Company’s size
The co-efficient on size of the firm was found to be positive and statistically significant
at the 1% level. The natural logarithm of total assets was used as a proxy for the size of
firm.
This result agrees with the findings from the literature and supports our
expectations. The significant relationship between dividend and size is in line with the
predictions of transaction costs theory which suggests that larger firms pay high
dividends because they can access the market more easily than small firms and borrow
at a lower cost. This result also confirms the first stage of the results for Heckman’s
model; size affects the probability of paying dividends and it is also one of the
determinants of the amount of dividend in the Jordanian market.
29
The pecking order hypothesis suggests that the first selection for companies to finance investments is
the internal financing; then, companies prefer to issue debt, while the last resort would be the issue of
equity in order to reduce the transaction costs and the costs of information asymmetry (Mayers & Majluf,
1984).
92
This result is consistent with the results reported by Fama & French (2001) although
they used a different estimation technique (i.e. logit regression) and found that size was
an important determinant of dividend payouts. In addition, in the two studies by AlMalkawi (2007) & Al-Kuwari (2009) which used the Tobit estimation technique, they
found supporting evidence of a positive significant impact of size on dividends. They
argued that larger firms pay higher dividends in order to reduce their agency costs. This
result has also been confirmed by Kim & Jang (2010) who documented a positive
significant impact of size on the first step of Heckman’s model but an insignificant
relationship in the second step, i.e. size determines the possibility of paying dividends
but not the amount of dividends.
On the other hand, our result contradicts the findings of an insignificant size effect
reported in some of the countries tested by Aivazian et.al. (2003); they used a pooled
OLS estimation technique to investigate if size has an impact on dividends in different
emerging market countries and the USA. The estimation technique applied in their
empirical investigation did not take the censoring problem into consideration and this
may be the reason for the inconsistent results.
Although they found that size was
insignificant, it was positively related to dividends. Moreover, Ben Naceur et.al. (2007)
expected the sign of the relationship between size and dividends to be positive but they
found that the size of the firm significantly affects the dividend payment in a negative
fashion, using panel and GMM estimation techniques. These latter findings are more in
line with the result when the natural logarithm of market capitalisation is used as a
proxy for size in the current thesis; this alternative proxy did not yield any significant
result.
Leverage
The empirical result for leverage documents a significant negative relationship between
leverage and dividends. The debt ratio was used as the proxy for leverage; the result is
consistent with what is reported in a majority of the literature. In other words, leverage
does affect the two dividend decisions negatively; the payment and the amount
decisions. This suggests that if the leverage ratio of a company is increased then the
possibility of paying dividends is lower and if they still make a dividend payment this
payment will typically decrease. In other words, rather than paying dividends high
93
leveraged firms act as if they need to retain their internal funds in order to meet their
financial obligations that they need to satisfy.
Furthermore, and by recalling the argument of Jensen (1986) that debt can be a key
factor in minimizing the agency costs of free cash flow, one can see why leverage
affects the dividend decisions of companies. Thus, if companies obtain debt, a fixed
commitment will be made to creditors that will reduce the discretionary funds which
could be accessible by managers and will reduce the possibility of investing in negative
net present value investments. This implies that firms with high leverage are more
likely to pay low dividends. Although Jordanian firms are not generally highly geared,
the results obtained here provide support for Jensen’s argument, since the bond market
is still under development so the main place to get loans is banks; this implies that firms
are financially constrainted. Fazzari et.al. (1988) mentioned that companies that face
financial constraints retain most of their earnings. Because of this, in Jordan, debt may
be an important factor in determining dividends.
Using OLS as an estimation technique, Agrawal & Jayaraman (1994) and Aivazian
et.al. (2003) results are consistent with the findings of this chapter; they also found
empirical support for the significant negative relationship between leverage and
dividends. They argued that to control for the agency cost of free cash flow, leverage is
a substitute for dividends and firms with high leverage are typically committed to pay
creditors and this will reduce the cash available for managers. In addition, the result in
this chapter has been confirmed by Fama & French, (2002) and Al-Kuwari, (2009) who
used Logit and Tobit regressions respectively. They found that an increase in debt
raised the instability of cash flows and, therefore, firms with high debt to equity ratios
had a tendency to pay lower dividends; thus the level of dividend payments was
negatively affected by the level of leverage. In addition, Agyei & Marfo-Yiadom’s
(2011) findings are consistent with this result in that they reported a significant negative
relationship between dividend and leverage, where they estimated the determinants of
dividends by fixed and random effects panel regressions.
On the other hand, Ben Naceur et.al. (2007) found that leverage had an insignificant
impact on dividends in the Tunisian Market. They compared their results using two
estimation techniques (the panel and the GMM estimation techniques); with these they
94
found that leverage did not affect dividends arguing that their result contradicted the
common view that high leveraged firms were riskier than low leveraged firms and
needed to reduce dividend due to the fact that they had to reduce their reliance on
external costly financing. Finally, with the aim of checking the robustness of this result
in the current chapter the total debt to equity ratio was chosen as another proxy for
leverage and this did not yield a different result.
Ownership structure
This study has adopted two measures for ownership structure: the family and the large
institutional ownership. The results showed that these two measures had no impact on
both dividend decisions in all of the three estimation techniques. The insignificant
coefficients of the ownership structure variables indicate that these variables have no
influence on corporate dividend decisions in Jordan. The existence of family ownership
or large institutional shareholders did not seem to have an impact either on the
probability of paying dividends or the amount of dividends.
This finding agrees with the result of Al-Malkawi (2007) who used a similar estimation
technique to this study (Tobit) and concluded that ownership structure whether
represented by family ownership or large institutional shareholders had no effect on the
dividend decisions. In addition, our result is in line with the conclusion of Ben Naceur
et.al. (2007) who found no significant relationship between dividend and ownership
structure although they used a panel estimation method with random and fixed effects
and compared it to the GMM results; they concluded that both estimations gave the
same results.
This result contradicts my expectation that ownership structure would have an impact
on the two dividend decisions; consequently it contradicts the agency cost view that
family companies reduce dividends simply because the owners are also the managers so
there is no conflict of interest.30 And it contradicts the notion that institutional investors
are expected to monitor the managers of companies and affect their decisions positively,
so high dividends should be paid to reduce the cash available for managers. Our results
30
It is worth to mention that although the result of family ownership was insignificant but the coefficient
was negatively affecting decisions, the payment and amount of dividends.
95
contradict the findings of Rozeff (1982) who discovered that insider ownership structure
was negatively related to the dividend payment; the findings of the current chapter
about ownership structure also conflict with the conclusions of Mancinelli & Ozkan
(2006) and Qian et.al. (2007).
Growth prospects
This study used the ratio of the market value of equity to the book value of equity as the
proxy for the measurement of growth. The result confirms our expectation that there is
a negative and significant link between growth opportunities and the two dividend
decisions. This means that firms with a high market to book ratio would have lower
probability of paying dividends as well as paying lower dividends.
The negative relationship between growth and dividend decisions can be explained by
the pecking order theory where the Jordanian companies follow the pecking order of
financing. For example if the industrial or service companies have a preference to
finance their growth from internal sources of funds instead of the more expensive other
choices like debt and equity then the possibility of having enough cash to pay dividends
is reduced. Thus, the relationship between dividend and growth should be negative.
Our result is in line with Agyei & Marfo-Yiadom (2011) who reported a significant
negative relationship between dividend and growth, despite the fact that they employed
a different estimation technique (the panel fixed and random effect). In addition, our
result was confirmed but only in the first step of Heckman’s model when applied by
Kim & Jang (2010) who used the two step Heckman procedure, i.e. they found a
significant impact of growth opportunities on the probability of paying dividends, but an
insignificant relationship between growth and the amount of the dividend paid.
Similarly, Barclay et.al. (1995) and Fama & French (2001) asserted that growth
prospects were a determinant of dividends and affects dividends significantly although
they applied a different estimation technique (logit regression).
The negative significant coefficient for growth contradicts the explanation provided by
the agency model, which argues that debt is a substitute for dividends. This model
argues that a number of companies may use dividends as a tool to establish a reputation
for treating shareholders favourably in capital markets (La Porta et.al., 2000).
In
96
general, companies with high growth prospects will need more external funds in the
future and for this reason, will be strongly motivated to increase the reputation of a firm
and to treat shareholders better by paying more dividends. If this is the case in the
Jordanian non-financial companies, where firms with high growth pay dividends more
than low growth companies, this will lead to higher stock prices and being able to raise
more capital by issuing new shares. Our result contradicts the findings of Al-Kuwari
(2009) where he investigated the Gulf Co-operation Council country stock exchanges
(GCC) using the Tobit regression method where he found an insignificant relationship
between growth opportunities and dividend decisions. In addition, Ben Naceur et.al.
(2007) reported an insignificant relationship between growth and dividends which
contradicts the significant relationship that we found in this empirical chapter. This
inconsistency of results may be due to the fact that they used a different estimation
method (GMM). Finally, with the aim of checking the robustness of our results, the
change of the natural logarithm of total assets has been chosen as another proxy for
growth opportunities and these yielded similar significant results.
4.5 Conclusion
Most of the previous empirical investigations of dividends have ignored the fact that
there are two decisions included in the dividend process: the decision to pay dividends
(whether to pay or not) and the decision about the amount of dividend (how much to
pay). In addition, debates among theoretical models are still continuing, in my view the
major reason is the different estimation techniques, model specifications and data used
by various researchers. This chapter provides more empirical evidence on investigating
the determinants of dividend decisions in the Jordanian market, using firm-level data
from non-financial companies. Based on the sample period of 1997-2006, this chapter
has used three estimation techniques (OLS, Tobit and Heckman’s ML model). The
Heckman model was found to be the best estimation technique since the likelihood ratio
test showed that the Tobit model did not provide any statistically significant increase in
the likelihood over the Heckman model. The coefficients of the variables of the OLS
regression were biased since the dependent variable (dividend payout ratio) is censored
(a high number of the observations take the value of zero, which means that many
companies do not pay dividends).
97
The determinants of dividend decisions of Jordanian companies actually yielded some
interesting results. The most interesting result of the chapter is the insignificance of a
firm's ownership structure on corporate dividend decisions. In addition, the results
indicate that growth, profitability, size and leverage are significant factors in impacting
on the two dividend decisions, which supported the internationally available evidence. I
found that profitability and size are significantly and positively related to dividend
payments.
However the empirical investigation revealed a negative relationship
between dividend, growth opportunities and leverage, which is statistically significant at
the 1% level. In summary, the two dividend decisions in developed as well as in
developing countries such as Jordan, are somehow affected by the same factors.
Finally, one of the ways that can broaden this study is to empirically investigate the
impact of government ownership on the determinants of dividend decisions and maybe
compare the results with another emerging market. Another future research could
divide the sample used into different sectors to identify the different characteristics that
would affect one sector and not the other.
98
5. Empirical Investigation Of The Impact of Dividend Announcements
On Share Prices
5.1 Introduction
The first two empirical chapters concentrated on investigating the impact of different
firm characteristics on the smoothness of dividends and the determinants of the two
dividend decisions. This chapter investigates the impact of dividend announcements on
stock prices.
This is one of the topics addressed by academic researchers in the
substantive literature. It is one of the most studied topics in order to examine the
importance of dividends and their impact on share price since Modigliani & Miller
(1961). This theory, stipulates that the dividend has no impact on the market value of a
share. In contrast, Gordon (1959, 1963) presented an alternative theory; this theory
pointed to the existence of a positive relationship between dividends and stock prices.
Moreover, signalling theory states that due to market imperfections managers use
dividends to signal their expectations about the future of the firm and to reduce the
information asymmetry between managers and outsiders; where investors react to this
signal, dividend changes consequently affect share prices.
The contribution of this chapter to the extant literature occurs in a number of ways.
Firstly, this chapter is the first to apply an OLS/EGARCH hybrid method and compares
it to the standard OLS event study methodology to investigate the impact of dividend
announcements on share prices; with special reference to industrial and services
companies that are listed on the ASE. In other words, this chapter adds to previous
empirical investigations by allowing EGARCH in a hybrid method to treat any
heteroskedasticity in the disturbance term.
Also it uses a larger sample compared to
most studies that have applied the event study methodology. In addition, this chapter
fills a gap in this area for an emerging market (Jordan) since the concentration of the
empirical investigations in this field has been on developed markets. Investigating the
impact of dividend announcements on share prices is important since the strength of an
organization, in today's international business, is heavily depended upon the support of
its investors.
This fact emphasizes that if a business owner wants to register his
presence in the industry for long term, he must pay attention to the values which
investors place on the business. So, continuous support from investors is essential for
the growth of a business since they provide strong financial backing to organizations as
99
important stakeholders. The Dividend is a tool that helps the management in achieving
its primary objective of maximizing the market value of the company. Finally, the
results of our empirical investigation will be compared to the published results of
studies conducted on data from developed markets.
This chapter continues with section 5.2 which analyses the theoretical background of
the impact of dividend announcements on share prices. Then section 5.3 explains the
standard event study and the hybrid method. Finally, section 5.4 analyses the results
and section 5.5 concludes the chapter.
5.2 Theoretical Background
This section discusses the impact of dividend announcements on the sensitivity of share
prices and formulates the hypothesis which will be tested.
The impact of dividend announcement on share price
The maximization of share price is assumed to be the basic concern of all the
companies, and has led to the development of several theories that outline different
predictions about the impact of dividends on share prices. The Dividend Irrelevance
Theory was introduced in 1961 by Miller and Modigliani (M&M) and supports the view
that in the absence of costs related to transactions and taxes, where no one can affect the
price of a security, all individuals value a business on the basis of investments or profit
and not dividends. Therefore, paying dividends would not affect the firm’s value and
its share price (M&M, 1961). Consequently, what determine the firm’s value are just
the profits achieved and the risk of assets but not the different combinations of retained
earnings and dividends. Despite the fact that, the prefect market described by M&M
does not exist in reality, other theories have emerged which challenge the Dividend
Irrelevance Theory.
For example one of the imperfections of the market that has led to the development of
different theories in finance such as the signalling theory, focuses on the information
content of dividend hypothesis. Based on the signalling theory, dividend payments are
usually used to send information in an indirect way about the company so that the
100
managers show their confidence about future earnings and the stability of future cash
flows. Thus, the shareholders will react to the declaration of dividend changes since
they contain important information about the managers’ expectation of current and
future cash flows which would lead to changes in the share price. Consequently,
dividend increases/decreases signal information of increases/decreases in expected cash
flows which should lead to a raise/fall in the share price. In addition, according to this
theory, companies that do not change their dividend from what the market expected
should earn no significant abnormal returns. However, research has not been clear
regarding the effect of the information delivered by the declaration of dividends on
share price and whether investors prefer companies to retain profits instead of paying
dividends, especially in Jordan.
Other theories started building on the assumptions of signalling theory, such as the cash
flow hypothesis, also called cash flow signalling theory. Among those who developed
the cash flow hypothesis were John & Williams (1985), Bhattacharya (1979, 1980) and
Miller & Rock (1985), who theorized that the decision taken by managers to increase or
decrease dividends will affect the company’s cash flow.
Miller & Rock (1985)
investigated the effect of a dividend payment on the company’s cash flow. They
assumed that the information about the internal cash flow of the company is not fully
shared by all interested parties, but the information regarding the asset values and
investments of the company is available to everyone. They argued that the probability
of paying a dividend is highly correlated with the company’s cash flow other factors
remaining constant. In other words, when a company declares that it will pay higher
dividends than predicted in the market then it signals confidence among managers about
the future cash flows which will result in an increase in share price. Therefore, the cash
flow signalling theory assumes that an increase/decrease in dividend payment will result
in an increase/decrease in share prices, taking into consideration that the level of
dividends is related to the level of stability of earnings which impacts the share price.
On the other hand, the Free Cash Flow Hypothesis was introduced by Jensen (1986)
who argued that free cash flow is the cash remaining after accepting all positive net
present value projects.
He suggested that the movement of share prices on the
announcement of dividends can be explained by the different interests of managers and
shareholders about the dividend payment.
In other words, this theory expects an
101
upward movement in the share price when there is an increase in dividend, since it
relates the increase of dividends to higher free cash flow. Further, Jensen (1986) noted
that managers have incentives to enlarge the size of their companies, in order to increase
the resources available to them and their compensation because all these factors are
closely associated with the company’s size. Therefore, availability of cash in excess of
that needed for positive NPV investments can lead the manager towards the problem of
over investment i.e. investing in non-profitable projects.
The management might
extract some excess funds from the free cash flow for over investment. So a good
option would be to increase the dividend payment in order to decrease the free cash
flow under the control of manager.
Control of the free cash flow prevents the
investment in projects which have negative NPVs and also reduces the agency costs
between managers and shareholders.
Moreover, Jensen & Meckling (1976) introduced a new theory called the agency cost
theory which made similar assumptions. They argued that when managers are not the
owners of the company this will give them the right to use the company’s assets for
their own benefit. Furthermore, they considered that dividends could be used as a
technique to reduce the free cash flow. For example, when the company pays dividends
this will lead to a decrease in available funds with the company. Consequently, the
possibility of managers using the free cash for their own benefit decreases dramatically.
In this case, agency cost theory relates the increase in dividend positively to a firm’s
value and its share price. Furthermore, according to the findings of chapter three, it
appears firms in the Jordanian market follow a smooth dividend policy. Based on the
above arguments, table 5.1 outlines the implications of the different corporate theories
about the effect of dividends on share price.
In addition, based on the previous
discussion, table 5.2 shows the expected average abnormal returns.
102
Theory
Dividend
Expected Price Effect
announcement
Irrelevance M&M Theory
Cash Flow Signalling
Theory/ Information
Content Hypothesis
Increase
No effect
Decrease
No effect
Increase
Positive price effect
Decrease
Negative price effect
Jensen’s Free Cash flow Increase
Hypothesis
Agency Cost Theory
Positive price effect
Decrease
Negative price effect
Increase
Positive price effect
Decrease
Negative price effect
Table 5.1: Implications of different Corporate Finance Theories in predicting the impact of
dividend announcements on share price.
Dividend
Expected AAR
change
(t=0)
Dividend
Positive AAR>0
Increase
Dividend
Negative AAR<0
Decrease
Dividend
No change AAR=0
No Change
Table 5.2: Summary of the Expected
Average Abnormal Return
103
5.3 Model specification
Usually the main aim of all companies is to maximise shareholder wealth; it is argued
that this can be achieved by increasing their share price and keeping their dividends
smooth in order to reduce risk. For that reason different empirical investigations have
tried to explain the impact of dividend announcements on share prices using different
estimation techniques. Although investigating the effect of dividend announcements on
share prices is of interest to many empirical researchers, academic understanding of the
topic remains limited (Ali & Chowdhury, 2010; Bhatia, 2010). In addition, although
different empirical papers have agreed based on the different theories outline in the
previous section that share prices react positively to dividend increase and negatively to
dividend decrease announcements, they document mixed results. For example, on one
hand, Chen et.al. (2009) investigated the Chinese market during the period 2000 till
2004 and reported that dividend announcements have a positive impact on stock prices.
On the other hand, Ali & Chowdhury (2010) reported no impact of dividend
announcements on share prices in Bangladesh over the period January 2008 to
September 2008.
They suggested that the reason behind the fact that dividend
announcements have not affected the share prices is because of the impact of insider
trading in the commercial banks of Bangladesh.
This empirical investigation starts with a discussion that there are two other reasons
behind the different results which emerge from most of the studies on the impact of
dividend announcements on share prices. Firstly, a majority of the preceding studies
that have employed the event study methodology have used the standard OLS market
model which assumes constant variance in the disturbance term but, in real life,
heteroskedasticity in disturbances is generally present in share price data (Brown &
Warner, 1985). The debate in the previous empirical papers concentrated on which
normal return model is more appropriate to predict the variance and the mean of the
return series (Fama & French, 1993; MacKinlay, 1997). However, the choice of the
mean equation has been the major focus of the debate (i.e. mean return model, market
model, multifactor model 31 , etc.) and researchers have paid little attention to the
Mackinlay (1997) stated that “ the gains from employing multifactor models for event studies are
limited”, so applying Fama & French’s (1992) 3 factor model or Carhart’s (1997) 4 factor model will add
a small marginal explanatory power thus the multifactor models will achieve a little reduction in the
variance of abnormal returns.
31
104
variance equation. 32 However, researchers have recently started to suggest that an
alternative to the OLS model be used. For example, Wang et.al. (2002) applied the
GARCH model while Yamaguchi (2007) and Takeda & Tomozawa (2008) applied the
EGARCH in their event study methodology to overcome the heteroskedasticity 33
problem in the stock price data. However, one likely disadvantage of Yamaguchi
(2007) and Takeda & Tomozawa’s (2008) empirical studies is their use of annual data
because of the high possibility several events occur during the year – not just dividends.
In this chapter, we apply the EGARCH approach when investigating the effect of
dividend announcements on share prices; then we compare the results of the hybrid
method (OLS/EGARCH) to the standard OLS method. Secondly, the variations in the
samples, causes mixed results between the different empirical studies.
Empirical
investigations on large sample sizes estimate the parameters of interest in a more
accurate way. In this chapter we investigate data for 81 companies from the industrial
and service sectors in Jordan during the period from 2005-2010; other studies have used
smaller sample sizes as outlined in table 5.3.
32
MacKinlay (1997) argued that the market model is a possible development of the mean return model
simply because the variance of the abnormal is reduced and this in turn can increase the possibility to
check the effects of the event under investigation.
33 Gujarati (2003) stated that “The assumption of homoskedasticity is when the basic version of the least
squares model assumes that the expected value of all error terms, when squared, is the same at any given
point. However, data suffers from heteroskedasticity when the variances of the error terms are not equal,
in which the error terms may reasonably be expected to be larger for some points or ranges of the data
than for others. The standard warning is that in the presence of heteroskedasticity, the regression
coefficients for an ordinary least squares regression are still unbiased, but the standard errors and
confidence intervals estimated by conventional procedures will be too narrow, giving a false sense of
precision”.
105
Author
Market
Sample size
Period of study
Khan (2011)
Pakistan
23 companies
2005-2009
Ali & Chowdhury (2010)
Bangladesh
25 Banks
Jan.2008-Sept.2008
Asamoah (2010)
Ghana
3 companies
Jan.2003-Dec.2005
Akbar & Baig (2010)
Pakistan
79 companies
July 2004-June2007
Kaleem & Salahuddin (2006)
Pakistan
24 companies
2002-2003
Takeda & Tomozawa (2008)
Japan
30 companies
1998-2005
Uddin (2003)
Pakistan
137 companies
Sept.2002-Oct.2003
Table 5.3: Selected studies from literature
This chapter is different since it applies a hybrid method (combining EGARCH with
OLS) in an attempt to determine the impact of dividend announcements on share prices
in Jordan for 82 companies (31 service and 51 industrial companies). Closing daily
stock prices data has been used for the period 2 January 2005- 29 December 2010 to
investigate the impact of dividend announcements on stock prices. To the best of the
author’s knowledge, the period covered in this chapter is more up to date compared to
other existing empirical studies which help in understanding the reaction of share prices
to the dividend announcements in a more comprehensive way. The main aim of using
up to date data in this chapter is to be able to provide relevant and reliable findings. In
addition, 492 dividend announcements were collected where certain criteria had to be
fulfilled in order to include an observation in the sample such as; the company had to
have trading data during the period of the dividend announcement, the company that
announced dividends had to have a dividend announcement in the previous year and the
company had to be listed on the Amman Stock Exchange (ASE) either as an industrial
or a service company. The following sections will explain the standard event study
methodology and then the hybrid method (the combination of the OLS and EGARCH
approaches).
106
5.3.1 Standard Event study
The main purpose of applying the event study methodology in most of the papers within
the accounting and finance area is to investigate the effect of an event on a company’s
market value. In other words, the aim of an event study is to investigate if there are any
abnormal returns caused by an unanticipated event (e.g. dividend announcement). The
results of an event study can have implications in forming government regulations or
accounting policies (Chen et.al., 2006).
The main reason why the event study methodology is so popular in empirical
investigations is that it avoids investigating accounting-based measures (these measures
have been criticised because they often do not give a true indication of the company’s
performance, i.e. managers select different accounting procedures that may
“manipulate” the accounting profits) (Benston, 1982; Nobanee et.al., 2009). Stock
prices are less susceptible to manipulation by insiders. The event study methodology is
used in this chapter due to the fact that, through this technique, any abnormal returns
can be calculated by the differentiation of the actual and expected returns. Additionally,
abnormal returns are considered to be the best reflection for the announcements of
different events in the markets (Asamoah, 2010).
This chapter applies the event study methodology to empirically investigate the impact
of dividend announcements on share prices for two sectors (industrial and service firms)
in the ASE. The event study that has been used to investigate the effect of dividend
announcements on share prices follows MacKinlay’s (1997) argument that “Using
financial markets data, an event study measures the impact of a specific event on the
value of the firm”. In addition, Clarke et.al. (2001) argued that the philosophy of using
the event study comes from the Efficient Market Hypothesis, where all of the
information available for investors in the market should be reflected in the prices of the
securities in an efficient and unbiased manner. Therefore, the challenge would be to
calculate the abnormal return on the shares compared to the expected return when the
dividend is announced.
107
The following subsection explains the event study methodology. More specifically, it
specifies the event date and period as well as the estimation period.34 The following
figure (5.1) shows the event study.
Event Window
5.3.1.1 The event day
The determination of the exact date of the event (i.e. the dividend announcement) is of
major significance because selection of the correct event date can help yield a more
relevant and reliable test of the information content of the announcement. In addition,
most researchers have defined the event date as the date that the new information about
a certain event is released to all market participants through different channels such as
corporate releases or the financial press (Asamoah, 2010 and Akbar & Baig, 2010). The
general assembly meeting date will be used as the event day in investigating the impact
of dividend announcements on share prices in Amman Stock Exchange (ASE). This
date has been chosen as the event date because according to Jordanian company law
which was enacted in 1997, article 191 A and B, states that the general assembly
meeting day is the date that the company announces the right of the shareholders to
obtain their dividends. Additionally, article 64 states that the general assembly meeting
has to be held by the company during the first four months of the year (Ministry of
Industry and Trade, 1997). 35
34
As MacKinlay (1997) stated that the event study methodology starts by identifying the event day (i.e.
the announcement day of the event under investigation). Then specifying the event window (this step
estimates the company’s stock returns) and the estimation window which estimates the model parameters.
35
All of the firms in our sample over the period 2005 till 2010 have had their dividend announcements
during the month of April every year.
108
In the ASE the dividend process is almost like that of most of the other markets. The
procedure starts when the company’s board of directors voting to either pay a dividend
or not, how much dividends to payout of the profit (the dividend decisions that were
investigated in the previous empirical chapter) and specify a date on which they will
announce the decision to the public. Based on the article 145 in the company law, the
company has to announce the timing of the GAM in two local daily newspapers at least
14 days before the GAM; it must also announce it once either on radio or television
within a period of three days before the GAM. Usually, the company also sends letters
of invitation to the shareholders to attend the General Assembly Meeting (GAM).
During the GAM the board of directors announces the dividend per share they have
decided to recommend for payment and discusses other things like the auditor’s
report.36 Additionally, some companies decide within the GAM meeting the date that
the dividend will be paid but article 191 C obliges the companies to pay the determined
dividends within 45 days from the date of the GAM. Based on the above argument, we
specify the general assembly meeting as the event day because it is the essential day
where dividends are approved and shareholders get notified (Ministry of Industry and
Trade, 1997).
5.3.1.2 The event window
MacKinlay (1997) defined the event window as the period that we are interested in
when calculating the abnormal returns as a first step in the event study process.
Different empirical investigations have used different ranges of days as windows when
calculating the abnormal returns (the window has ranged from -1 to 1, to -30 to 30 in
studies such as Zuguang & Ahmed (2010) and Vieira & Raposo (2007)). Moreover,
Nobanee et.al. (2009) and Jong (2007) have argued that the selection of the length of
the event window is subjective; they agreed that researchers have to give some time for
the market to react for the new information.
36
As discussed in Armitage (1995), there is always the possibility of having the problem of
contaminating events in the same period, i.e. news other than the event under investigation which may
affect the share price. It is difficult to check all of the other news that would happen within the same day
of the GAM but in this empirical chapter we have excluded three companies that the announcement date
was within the first month of the year. The reason of this exclusion is that the earnings announcement at
the end of the year may have a possible impact on the share price.
109
This chapter applies an event window of (-5 to 5)37 which is used by many empirical
studies simply because we hope that this gives more reliable findings and restricts the
share price reaction to the event under investigation, i.e. the dividend announcements.
Additionally, we can always argue that the larger the event window the more likely it is
that the share prices react to different news available in the market. Therefore, choosing
this event window (the eleven-day period) helps in investigating the share price reaction
to the dividend announcements while reducing the possibility that the share price will be
affected by other news. To check for robustness of results different event windows will
be employed such as +1 to -1 and+3 to -3.
5.3.1.3 The estimation period
The event study methodology selects an estimation period as a number of days prior to
the event period due to the assumption that the expected returns during the estimation
period are independent of the release of the new information. The estimation period is
employed to estimate the parameters of the market model. This step of the event study
is used to predict the normal return. Reinganum (1981) and Morse (1984) stated that
using daily stock returns is more powerful than monthly data when applying event study
methodology since using daily data is important to investigate the daily reactions of
share prices on an event day; it also increases the statistical power of the significance
tests. This conclusion has been supported by Brown & Warner (1985) and Zhang et.al.
(2010). Thus, based on the preceding reasoning, this chapter applies the daily stock
returns for the Jordanian companies (industrial and service sectors) during the period
January 2005 till December 2010 to facilitate comparability with other published
empirical studies. The estimation period for this empirical chapter will be from day -21
to -51 prior the announcement day so the chapter will use 30 trading daily observations.
The main reason for choosing 30 days for the estimation period is that all the companies
in our sample had their dividend announcements during the month April in each year.
37
Generally before the dividend announcement day rumors start to spread about the dividend. Thus,
there is always a possibility that the price would react to such rumors before the event day. Because of
this the event window has started before 5 days of the dividend announcements. It is vital to have the
event window to start before the event date simply because this can give us a good idea if the Jordanian
market (ASE) suffers from information leakages before the dividend is formally announced in the
General Assembly Meeting. This in turn; helps in understanding the effectiveness of regulation and
supervision.
110
In addition, at the end of every year the companies announce their earnings. Thus to
avoid any misleading estimation in the parameters of the market model we left a gap of
15 days between the event window and the estimation period and at least a month
between the end of the year and the estimation period. 38
The following equations will be used in applying the standard event study methodology:
The daily percentage return39 will be calculated as the following equation:
𝑅𝑡 =
𝑃𝑡 −𝑃𝑡−1
𝑃𝑡−1
∗ 100
(5.1)
Where:
𝑅𝑡 is the return on day t
𝑃𝑡 is the price of the share at closing on day t
𝑃𝑡−1 is the price of the share at closing on day t-1
Most of the previous empirical investigations that used the event study applied the
market model40 which was proposed by Fama (1976) to calculate the expected return.
The market model has been used to calculate the expected return because it adjusts for
risk and market factors. In addition, the market index has been used as a proxy for the
return on market (Dodd & Warner, 1983). The market model is a linear regression
model and it relates the return of firm i at time t 𝑅𝑖𝑡 to the return of the market at time t
𝑅𝑚𝑡 as follows:
38
The announcement of earnings at the end of the year might affect the share prices, simply because the
investors might react to the announcement if it is bad or good. Thus, the volatility of share prices might
increase and to avoid any effect on the estimation period we ignored the first month after the earnings
announcement.
39
The return equation has been used following Uddin (2003) and Kadioglu (2008).
40
Jong (2007) stated that previous empirical investigations had to choose between two models other than
the market model to be able to calculate the abnormal returns (Market adjusted returns and mean adjusted
returns). This chapter will use the two other models to check for robustness of results although the
market adjusted returns method gives an assumption that the beta of every share is equal to one and the
mean adjusted returns method ignores the wide movements in share prices.
111
𝜀𝑖𝑡 ~𝑁(0, 𝜎𝜀2𝑖𝑡 )
𝑅𝑖𝑡 = 𝛼𝑖 + 𝛽𝑖 𝑅𝑚𝑡 + 𝜀𝑖𝑡 ,
(5.2)
Where:
𝑅𝑖𝑡 is the stock return for firm i on day t
𝑅𝑚𝑡 is the market return on day t
𝜀𝑖𝑡 is the error term, normally distributed with mean equal 0 and variance equal 𝜎𝜀2𝑖𝑡
The above equation will be estimated using ordinary least squares to get the estimated
parameters 𝛼𝜄 and 𝛽𝜄 during the estimation window. Then the expected returns over the
event window (from t-5 to t+5) will be calculated for every dividend announcement.
After getting the expected return then the abnormal returns for each day within the event
window can be found by subtracting the fitted expected return from the actual return as
follows:
AR it = εit = R it − αi − βi R mt
(5.3)
Where:
𝐴𝑅𝑖𝑡 is the abnormal return on stock i at time t
𝑅𝑖𝑡 is the actual return on stock i on day t
𝛼𝑖 , 𝛽𝑖 and 𝑅𝑚𝑡 as explained in equation 2
Then, the daily abnormal returns will be calculated during the eleven days of the event
window (from t-1 to t+1) for each dividend announcement. Using equation 5.4, the
abnormal returns are added within the event window to get the cumulative abnormal
return (CAR) and this can be calculated for each dividend announcement over the event
window for the eleven days:
𝑡
2
𝐶𝐴𝑅 (𝑡1 , 𝑡2 ) = ∑𝑡=𝑡
𝐴𝑅𝑖𝑡
1
(5.4)
112
5.3.2 Hybrid Method (OLS/EGARCH)
The hybrid method is proposed as another alternative for this event study methodology
with the advantage that it takes into consideration what the standard event method
ignored the autoregressive conditional heteroskedasticity.
The average cumulative
abnormal returns variance is an essential component in calculating the test statistic;
thus, we cannot compromise the accuracy of the variance forecast. The potential bias
caused by ignoring any autoregressive heteroskedasticity present has rarely been dealt
within the literature of event study methodology. On the other hand, in the finance
literature volatility clustering has been extensively investigated. Engle (1982) proposed
modeling volatility by a class of stochastic process models like the autoregressive
conditional heteroskedasticity (ARCH)41 family of models. The ARCH-LM test is used
to identify any ARCH effects by testing the residuals from the preliminary Ordinary
Least Squares (OLS), which regresses the squared residuals on a constant and q lagged
values of the squared residuals. Bolerslev (1986) generalized the ARCH process to
form
a
new
model
called
the
Generalized
Autoregressive
Conditional
Heteroskedasticity (GARCH) by allowing past conditional variances in the current
conditional variance equation. The typical GARCH (1, 1)42 model is as follows:
𝑅𝑖𝑡 = 𝛼𝑖0 + 𝛽𝑖0 𝑅𝑚𝑡 + 𝜀𝑖𝑡 , 𝜀𝑖𝑡 = √ℎ𝑖𝑡 𝜈𝑖𝑡 , 𝜈𝑖𝑡 ~𝛮(0,1) 𝑎𝑛𝑑
(5.5)
2
2
ℎ𝑖𝑡 = 𝜔𝑖 + 𝛼𝑖 𝜀𝑖,𝑡−1
+ 𝛽𝑖 ℎ𝑖,𝑡−1 = 𝜔𝑖 + 𝛼𝑖 ℎ𝑖,𝑡−1 𝜈𝑖,𝑡−1
+ 𝛽𝑖 ℎ𝑖,𝑡−1
(5.6)
Where:
𝜀𝑖𝑡 : is the error term and equal to the product of its standard deviation and the
Gaussian white noise 𝜈𝑖𝑡 with zero mean and unit variance.
ℎ𝑖𝑡 : is the variance of the residuals of the return equation.
𝛼𝑖 : is the weight assigned to the long run mean variance
𝛽𝑖 : is the error between actual return and predicted return
𝜔𝑖 /(1 − 𝛼𝑖 − 𝛽𝑖 ): is the estimated conditional variance in the past time period.
43
41
ARCH processes are defined as mean zero, serially uncorrelated processes with non-constant variances
conditional on the past variances.
42
The (1,1) after the GARCH term is the standard notation where the first number refers to the number of
autoregressive lags of ℎ𝑖𝑡 or ARCH terms and the second number refers to the number of moving average
2
lags of 𝜀𝑖,𝑡
43
In order to keep the long run variance and conditional variance nonnegative, 𝛼𝑖 >0, 𝛽𝑖 >0 and 𝜔𝑖 >0 are
required.
113
One advance of the standard GARCH is the Exponential GARCH (EGARCH). It
recognizes that volatility is likely to rise in reaction to a dividend decrease and to fall in
reaction to dividend increase (Zhang et.al., 2010). This is due to the fact that the market
deals with the rise in dividend as good news about the future of the company while the
decrease of dividend will be perceived as bad news and investors will be uncertain
about the future of the company; this uncertainty will lead to perceptions that this
company as a risky investment. In addition, Oskooe & Shamsavari (2011) have argued
that volatility reacts more to bad news (negative shocks like dividend decrease) than to
good news (positive shocks like dividend increase) due to the fact that there is a
negative correlation between the future volatility and current stock returns.
This
reaction of the stock returns to the new available news is identified in the literature as
asymmetric volatility (Goudarzi & Ramanarayanan, 2011). In other words, volatility in
stock markets tends to be lower in a rising market than in a falling one. According to
Wu (2001); ‘Asymmetric volatility is mainly observed when a certain stock market
crashes consequently the market experience high decrease in the share prices which
leads to high market volatility’.
Even though the magnitude of volatility clustering is captured in the GARCH model, it
is incapable of taking account of the sign of the rise and fall of volatility. In addition,
the nonnegative limitation on the weights of the GARCH model ignores the possibility
that the variance process can go up and down in an oscillatory manner. In other words,
modeling the stock return behavior using a GARCH model may not be appropriate
because it imposes symmetry on the conditional variance structure.
Then, Nelson
(1991) introduced the EGARCH model to overcome the previous mentioned limitations.
The following equations show the simple EGARCH (1, 1):
𝑅𝑖𝑡 = 𝛼𝑖0 + 𝛽𝑖0 𝑅𝑚𝑡 + 𝜀𝑖𝑡 , 𝜀𝑖𝑡 = √ℎ𝑖𝑡 𝜈𝑖𝑡 , 𝜈𝑖𝑡 ~𝛮(0,1) 𝑎𝑛𝑑
log(ℎ𝑖𝑡 ) = 𝜔𝑖 + 𝛽𝑖1 log(ℎ𝑖𝑡−1 ) + 𝛾𝑖1
𝜀𝑖𝑡−1
√ℎ𝑖𝑡−1
+ 𝛼𝑖1
|𝜀𝑖𝑡−1 |
√ℎ𝑖𝑡−1
− √2⁄𝜋
(5.7)
(5.8)44
44
Logarithms on both sides of the variance equation (8) have been imposed to ensure that the conditional
variances are positive.
114
Where:
𝜔𝑖, 𝛼𝑖1, 𝛽𝑖1and 𝛾𝑖1 are parameters for conditional variance estimation.
𝛽𝑖1indicates the impact of the last period measures on the conditional variance.
𝛼𝑖1 is a coefficient which measure the effect of previous period in the information set
and explain the past standardized residuals influence on the current volatility.
𝛾𝑖1signifies the asymmetry effect in the variance.
Comparing the EGARCH with the GARCH model one can identify three unique
advantages. Firstly, the use of the logarithm in the EGARCH model ensures that the
conditional variance is positive, consequently, the use of non-negative constraints in the
estimation of the GARCH model is not necessary. Secondly, the parameter 𝛾𝑖1enters
equation 8 with a negative sign (𝜀𝑖𝑡 <0 news) which produces more volatility than good
news. The last advantage is that the degree of persistence in this model is measured
only by the 𝛾𝑖1 parameter. (if 𝛾𝑖1<1 then the model is not integrated).
Therefore, the EGARCH fits the aims of this chapter; because of its flexibility in
estimating the parameters and the ability to treat information asymmetry. To apply the
hybrid method the following steps will be undertaken:

Based on the dividend announcements events we construct the stock return
sample. For each event we conduct a preliminary OLS regression using the data
from the estimation period.
Then we apply the ARCH-LM test 45 on the
residuals of the OLS regression which identifies the events with significant
autoregressive conditional heteroskedasticity in the disturbance. Subsequently,
we separate the events into two groups: events with ARCH effects and events
without ARCH effects.
45
Hong & Shehadeh (1999) argued that the ARCH-LM test is the most popular test to check if the null
hypothesis of no ARCH effect is true. They continued that the ARCH-LM test is easy to apply (easy to
calculate) and is asymptotically locally powerful if the true alternative is ARCH (q). In addition, Granger
& Terasvirta (1993) mentioned that the other alternatives (the portmanteau tests of Box & Pierce (1970)
and Ljung & Box (1978)) are asymptotically equivalent to Engle’s (1982) LM test.
115

For the events with significant ARCH effects we use the EGARCH to estimate
the abnormal returns and the variance of the abnormal returns. For the events
without ARCH effects we use the standard OLS market model estimates.46

The results achieved from the EGARCH and the OLS will be combined together
for all the events to get the cumulative abnormal returns and the variance of the
cumulative abnormal returns.

Then we can test the null hypothesis that there is no reaction to dividend
announcements.
5.3.3 Parametric and Non-parametric Tests
In order to test the hypothesis mentioned in section 5.2, we will apply the two main
types of test that have been used in literature; parametric and non-parametric tests.
Bartholdy et.al. (2007) argued that non-parametric tests do better than parametric tests
due to the fact that parametric tests assume that the abnormal returns of the companies
should be normally distributed. Usually empirical event studies apply the t-test as a
parametric test and Corrado’s rank test as a non-parametric test. MacKinlay (1997)
stated that for robustness of results the two types of results should be used in
conjunction with each other; this was illustrated by Campbell & Wasley (1993) where
they found that non-parametric rank test provided more reliable inferences than
parametric tests. The following subsections explain briefly the parametric and nonparametric tests.
5.3.3.1 Parametric test
Parametric tests are applied when the assumption is that the returns are normally
distributed. Lyon et.al. (1999) stated that “The central limit theorem guarantees that
the distribution of the mean abnormal return measure converges to normality as the
number of firms is independent and identically distributed drawings from a finite
variance distribution”. Therefore, the assumption given is that with our relatively large
sample we expect the t-statistic to be well specified. The aim of applying the t-test is to
check if the abnormal returns of the companies in the days before and after the event
day differ significantly from zero. Thus, the null hypothesis that the share prices do not
46
Goudarzi & Ramanarayanan (2011) stated that when the ARCH-LM test is insignificant then there is
no need to use the EGARCH model because this will lead to a misspecified model.
116
react to the dividend announcements can be tested by calculating the test statistic as
MacKinlay (1997) explained as follows:
𝐶𝐴𝑅(𝑡1 ,𝑡2 )
𝜃=
1
~𝛮(0,1)
(5.9)
𝑣𝑎𝑟(𝐶𝐴𝑅(𝑡1 ,𝑡2 ))2
5.3.3.2 Non-parametric tests
Mackinlay (1997) argued that there are two common types of non-parametric tests that
are used in the event study methodologies; the Wilcoxon sign test and Corrado’s rank
test. In this study we apply the rank test for two main reasons. Firstly, the sign test has
a weakness of not being well specified with daily data which are used in this empirical
investigation because of the possibility that the data suffer from skewness. Secondly,
Maynes & Rumsey (1993) concluded that since the distribution of ranks is uniform then
the rank test has been found to be a superior test static to test the significance of
abnormal returns. For example, Seiler (2000) justified using the rank test because he
found that the rank test was well specified and independent of the degree of skewness.
The rank test is simple to apply and the assumption is that the abnormal returns are not
symmetrically distributed around the mean. The event period and the estimation period
errors are ranked for every share then the average rank is subtracted from the rank of the
event day error. Thus, the test statistic can be calculated by dividing the average
difference by the standard deviation of the average difference over the event period and
estimation period (Armitage, 1995)
5.4 Statistical Analysis
Before proceeding with the results of the analysis as well as the hypothesis testing, it is
necessary to analyse the descriptive statistics of the data. Thus, section 5.4.1 begins
with the descriptive statistics for the variables under investigation; the daily stock prices
and the market index returns. Section 5.4.2 briefly explains the results of the unit root
test. Finally, section 5.4.3 analyses and compares the results of the standard event study
methodology with the hybrid OLS/EGARCH method.
117
5.4.1 Descriptive statistics
The closing daily stock price data were collected for the period 2 January 2005- 29
December 2010 from the Amman stock exchange (ASE) and the Jordan Security
Commission (SEC). This chapter also used other sources like various stock market
publications (for example, annual reports and the company’s guide) to either find a
missing observation or to double check the observations collected for accuracy. 47 To
investigate the impact of the event - dividend announcements on share prices, we
managed to collect 492 dividend announcements for Jordanian industrial and service
companies from the Trade of Ministry- Companies Control Department. (The Trade of
Ministry-Companies Control Department, 2011) The data included in the investigation
of the announcements effect on share prices is 82 companies (31 service and 51
industrial companies) which resulted in a sample of 20172 observations.48
stock return
market index return
Mean
0.06
0.0547
Standard Deviation
2.311
1.3786
Skewness
-0.05
-0.0362
Kurtosis
5.209
1.5051
Minimum
-25.327
-4.6373
Maximum
23.809
4.7798
Table 5.4: Descriptive statistics
Note: the descriptive statistics describe the stock returns for the whole
sample collected during the 6 years period.
As we can see from table 5.4, the mean for the market return sample is slightly lower
than the stock return. In addition, the standard deviation of the market index return is
considerably lower than that for the average stock return. This is expected since the
return for an individual stock fluctuates more than the market index return; investing in
one company is riskier than investing in the whole market. The more diversified the
portfolio of investment, the less risk is incurred. The market index is a portfolio of the
47
The General Assembly Meeting dates for twenty companies from the industrial sector in 2010 have
been double checked from their annual reports or Amman Stock Exchange website.
48
The total number of firms in ASE was 220 including 136 non-financial firms. Our sample represents
60% of the total number of non-financial firms. This implies that 54 companies have been eliminated
because the data is not available throughout the years for those companies.
118
companies listed in the market; such diversification tends to decrease risk and
consequently decrease return. Moreover, the wide range between the minimum and
maximum in the typical stock return (-25.327 % to 23.809 %) and the market return (4.6373 % to 4.7798 %) indicates that the Jordanian market faced a high degree of
variability of changes in prices over the period 2005-2010. This result is in line with
findings from other emerging markets; Oskooe & Shamsavari (2011) reported a wide
range in the returns of the Iranian Stock Market – they reported (-5.4330% to 5.2581%).
Likewise, with respect to the mean, the high standard deviation of the Jordanian market
returns (1.3786) which is higher than the Iranian Market Returns (0.516) points to the
fact that the Jordanian market is risky in nature with high levels of volatility.
Furthermore, the negative Skewness 49 indicates that the Jordanian stock return
distribution is skewed to the left of its mean and has a long left hand tail which means
that large negative returns tend to occur more often than large positive ones in the
Jordanian market. These result co-insides with the empirical findings of Al-Fayoumi
et.al. (2009); they reported that the returns in the Jordanian market are negatively
skewed. To explain it in a different way, large negative movements in stock prices are
not usually matched by equally large positive movements. One of the reasons that
could cause this negative movement in the Jordanian market is Herding behavior.
Demirer et.al. (2007) defined Herding as “a behavior pattern that is correlated or cointegrated across individual investors, firms in an industry or even country stock
markets”.
Therefore, herding arises when investors choose to copy the observed
decisions of others or movements in the markets instead of following their own beliefs
or their own analysis of market/information.
Demirer et.al. (2007) empirically
investigated if herding behavior exists in different markets including Jordan. They
concluded that Jordanian investors imitate the behavior of other investors in the market
which could lead to a dangerous level of speculation.
Hence, the distribution of
Jordanian stock return series is not-symmetric (the stock returns distribution is
asymmetric to the left with a few extreme values). In addition, the Kurtosis50 for the
stock returns is (5.209), which indicates that the Kurtosis in the Jordanian stock returns
series is bigger than the normal value of 3. Consequently, based on the above argument
49
50
Skewness is a measure of the asymmetry of the probability distribution curve.
Kurtosis describes the shape of the distribution curve.
119
it can be concluded that the distribution of stock returns departs from normal
distribution51.
The following subsection shows the empirical results of stationary tests because the first
step before dealing with the return series is to check the unit root tests of the return
series. In other words, the following section discusses if the series have been found to
be stationary or not.
5.4.2 Results from unit root tests
The first step to be applied is to check if the return series follow a random walk. One of
the suggestions that the series are random is that the series never returns to it mean
value. This empirical investigation applies the two commonly used tests to check for a
unit root; the Augmented Dickey-Fuller (ADF) and Phillips Perron (PP) test. We apply
these two tests to examine the distribution properties of the return series. The result of
the ADF 52 and PP 53 tests reject the result of a unit root at the 1% significant level.
Thus, the series are stationary under both tests the PP and ADF tests.
5.4.3 Discussion of results
To investigate the impact of dividend announcements on share prices in the Jordanian
market (ASE) during the period 2005-2010, this empirical investigation applied and
compared two estimation methods -the standard OLS event study methodology and the
hybrid method OLS/EGARCH. A summary of the results is reported in table 5.5 while
the detailed results are reported in table 5.6 which includes the average abnormal returns
during the eleven days of the event window (t-5 till t+5). The abnormal return for every
firm has been calculated using the market model as explained in section 5.3.1.3.
The table also shows the results of the parametric test (t test) and the nonparametric test
(Corrado’s rank test). In addition, table 5.7 shows the implications of choosing the
51
The Normal distribution has a zero (0) Skewness and Kurtosis equal to three (3). However, it is worth
noting that Dyckman et.al. (1984) argued that any “non-normality of daily abnormal returns has little
effect on event study tests.” (as cited in Strong, 1992, p. 542). In addition, Brown & Warner (1985, p.
25) stated that “the non-normality of daily returns has no obvious impact on event study methodologies”.
52
The ADF test resulted in -4.708 rejecting the unit root test at 1% significant level.
53
The PP test resulted in -26.128 confirming the result found in the ADF test.
120
different holding periods. Moreover, table 5.8 shows the cumulative abnormal return
for four different holding periods around the dividend announcement day for the whole
sample.
Dividend change
Dividend
Expected AAR
Actual AAR
(t=0)
(t=0)
Positive AAR>0
Negative AAR<0
Findings
-
(Not significant)
Increase
A significant negative AAR
before three days of the
announcement day
-
No positive market reaction
on
the
dividend
announcement day
-
A significant positive AAR
after three days of
the
announcement.
Dividend
Negative AAR<0
Negative AAR<0
-
(Not significant)
Decrease
A significant negative AAR
before three days of the
announcement day.
-
No
significant
market
negative
reaction
on
the
dividend announcement day
-
A significant negative AAR
during
the
three
days
following the announcement.
Dividend
Constant
No change AAR=0
Negative AAR<0
-
(Not significant)
No significant AAR before
the announcement day.
-
A significant positive AAR
after three days of
the
announcement (t+3) and a
significant
negative
after
days
five
of
AAR
the
announcement (t+5).
Table 5.5: Summary of the Expected Average Abnormal Return, Actual Average Abnormal
Return and the findings.
121
Days
All Events
Dividend Increase
Dividend Decrease
Constant Dividend
Sample N=492
Sample N=136
Sample N=152
Sample N=204
OLS
AAR
-5
-0.108
-4
-0.077
-3
###
OLS/EGARCH
OLS
t-test
AAR
t-test
AAR
-1.008
-0.113
-1.079
###
OLS/EGARCH
OLS
OLS/EGARCH
OLS
t-test
AAR
t-test
AAR
-0.06
0.274
-0.09
-0.545
-0.728
-0.12
-0.897
-0.14
OLS/EGARCH
t-test
AAR
t-test
AAR
t-test
AAR
t-test
-0.12
0.144
-0.17
-0.941
-0.09
-0.606
-0.11
-0.479
-1.116
-0.22
0.869
- 0.21
0.128
-0.15
-0.875
-0.12
-0.145
-0.635
-0.083
-0.163##
-1.421***
-0.167##
-1.527**
-0.25#
-2.189*
-0.26#
-2.247*
-0.33##
-1.895**
-0.35##
-1.857**
-0.21
-1.094
-0.27
-1.011
-2
0.029
0.252
0.015
0.205
0.05
0.628
0.07
0.048
-0.03
-0.186
- 0.11
0.62
-0.01
-0.113
-0.08
-0.489
-1
-0.067
-0.643
-0.068
-0.674
-0.08
-0.372
-0.14
-1.185
-0.05
-0.321
-0.18
-1.069
-0.06
-0.414
-0.16
-1.073
0
-0.101
-0.689
-0.11
-0.761
-0.34
-1.153
-0.35
-1.041
-0.18
-0.886
-0.20
-1.192
-0.15
0.505
- 0.12
0.407
1
0.198###
-1.560**
0.204##
-1.617**
0.08
0.328
0.11
-0.549
-0.47#
-2.260*
-0.53#
-2.486*
-0.20
-1.072
-0.15
-0.796
2
-0.118
-0.921
-0.109
1.012
0.19
0.689
0.19
0.223
-0.49#
-2.744*
-0.54#
-2.738*
0.28
-1.014
0.35
-1.028
3
0.273###
-2.209***
0.295##
2.392**
0.39##
1.558***
0.47##
1.638**
-0.54#
-2.885*
-0.58#
-3.542*
0.42#
-2.393*
0.55#
-3.176*
4
0.003
0.021
0.005
0.037
0.09
0.337
0.15
0.791
0.25
1.041
0.21
1.111
-0.25
-0.380
-0.32
-0.506
5
-0.041
###
-0.329
-0.037
###
0.301
-0.07
-0.266
-0.12
-0.624
-0.02
0.098
-0.06
-0.475
-0.06
##
-1.473
**
-0.09
##
-1.887**
Table 5.6: Results of the standard event study using the OLS market model and the hybrid method using OLS/EGARCH
Note: AAR is the Average Abnormal Return. The results are robust simply because we found almost the same results when the mean adjusted model and market adjusted model have been
applied. The symbols *, ** and ***denote statistical significance at 1%, 5% and 10% levels respectively for t-test. The symbols #, ## and ### denote statistical significance at 1%, 5% and 10%
levels respectively for Corrado’s rank test. EGARCH stands for Exponential Generalised Autoregressive Conditional Heteroscedasticity. When the current dividend is greater than the previous
year dividend then that dividend announcement has been included in the dividend increase sample. In addition, if the current dividend is less than the previous year dividend then that dividend
announcement has been included in the dividend decrease sample and if the current dividend equals the previous year dividend then this dividend announcement has been included in the
constant dividend sample.
122
Holding periods
Reason
-5 till -1
The
-3 till -1
+1 till +3
+1 till +5
Implication
periods
before
the Efficiency in regulation
announcement to test information and
leakage.
effectiveness
of
supervision.
The periods after the announcement Information
content
of
to check the effect of dividend dividend and duration of
announcement on share price and price adjustment.
market efficiency.
Table 5.7: Reason for opening and implications of different holding
periods.
Note: this table shows four different holding periods in the first column. The second column
gives the reason for checking these periods. Then the last column states the implication of the
different holding periods.
OLS
OLS/EGARCH
Test
CAR%
Test statistic
Holding periods
CAR %
statistic
(-5 till -1)
-38.6
-0.239
-41
-0.917
(-3 till -1)
-20.1
-0.012
-22
-0.035
(+1 till +3)
35.3
2.190*
39
2.88*
(+1 till +5)
31.5
1.954**
35.8
1.99**
Table 5.8: Summary of the Cumulative Abnormal Returns (CAR) for selected
holding periods around the announcement day.
Note: The symbols * and ** denote statistical significance at 1% and 5% levels respectively when
applying the t-test.
The main interesting point is that no matter the methodology used the significant
average abnormal returns for both methods are the same but the level of significance is
higher when the hybrid method is applied. The difference in the level of significance
between the two methods is due to the different estimates of the ARCH events and this
confirms and supports the use of the hybrid method (OLS/EGARCH) over the standard
OLS event study method. Applying the ARCH-LM test revealed that 91 events out of
492 (18.5%) suffered from ARCH effects.
123
In addition, on most of the days, in the three different samples, a negative average
abnormal return54 in the event window even when the abnormal returns were estimated
by the two different methods. The reason behind this might be that the whole world
was facing an economic depression from 2006 till now and the period under
investigation in this chapter is from 2005 till 2010. Also, Omet (2011) reported that the
liquidity of ASE (Amman Stock Exchange) has been relatively high during the years
2007 till 2009. He found that the stock market crash in 2009 was the major reason for
liquidity to decline. This result has been supported by pointing out the collapse in the
stock price of most of the listed firms; the weighted price index of the Jordanian stock
market has fallen from 7519.3 points in 2007 to 5520.1 points by the end of 2009.
Furthermore, the result indicate that the announcement day (t=0) abnormal returns, in
the three different samples are insignificant. This suggests that either investors in the
Jordanian market do not react quickly enough to the news of dividends or the signalling
theory does not hold in the Jordanian market. The insignificant impact of dividend
announcements on the share price supports the dividend irrelevance theory proposed by
M & M in 1961 which states that dividends have no impact on share price. In addition,
Omet (2011) argued that the market-making mechanism in the ASE is order-driven;
subsequently the orders placed by investors are prioritized according to time and price.
After submitting successive buy and sell orders, traders provide liquidity 55 for other
participants who demand immediacy by placing counter market orders. However, in
other stock markets such as the United States of America (USA), they have market
makers (liquidity providers) who stabilize stock prices by trading on their own accounts.
Thus, the non-availability of such liquidity providers in the Jordanian market may affect
the share price.
A more detailed analysis of the results reported in table 5.6 shows some interesting
points. On the dividend announcement day (t=0) the average abnormal returns for the
whole sample is -10.1 % and -11 % for the OLS and the hybrid method respectively; for
54
This confirms the negative skweness of the return data mentioned in the descriptive statistics which
means that large negative returns tend to occur more often than large positive ones in the Jordanian
market.
55
It is worth noting that operationally efficient (Liquid) stock markets allow traders to get their orders
executed as quickly (immediacy) and as cheaply as possible. In other word, liquidity refers to the ease
with which buyers and sellers of securities promptly get their orders executed with minimal impact on the
price.
124
the dividend increase sample the figures are -34% and -35% for the OLS and the hybrid
method respectively; for the dividend decrease sample the numbers are -18% and -20%
for the OLS and the hybrid method respectively; and for the constant dividend sample
the values are -15% and -12% for the OLS and the hybrid method respectively.
However, all returns in different samples are found to be insignificant according to both
the parametric and nonparametric tests (the t test and the rank test). This result is in line
with the findings of Travlos et.al. (2001) who investigated the impact of dividend
announcements on share prices in Cyprus. They investigated the period of 1985-1995
using the market model and found insignificant but positive abnormal returns. They
argued that this result is due to the lack of a formal stock exchange according to the
Cyprus Chamber of Commerce (CCCI).
Transactions are dealt with through
brokers/dealers of a decentralized network in the over the counter market. In other
words, during the period under their study there were no official market makers in
Cyprus and this may be the reason why our results are in line with the findings for the
Cyprus market.
In addition, by using the same standard event methodology and using the market model,
Ali & Chowdhury (2010) concluded that dividend announcements do not affect share
prices in Bangladesh when they investigated the banking industry.
They also
investigated a different period from Jan.2008-Sept.2008 and stated that during this
period the Dhaka Stock Exchange (DSE) was being manipulated and speculation was
rife. In other words, the employees from the exchange, brokers and insiders were the
speculators in the market; they sell and buy to gain in the short term which leads to
inefficient dividend information. Al-Shiab (2006) also argued that the Jordanian market
went through a period of speculation where the public started to lose trust with trading;
as a result, the government started to work on improving regulations to move towards
international standards.
Khan (2011) also used the market model in the standard event methodology and found
that dividend announcements have no impact on share prices in Pakistan. He used
mixed methods (interviews and standard OLS event study) to investigate the effect of
dividend announcements on share prices during the period 2005-2009.
After
investigating 639 events and interviewing 23 company executives he found similar
results; that dividends have no impact on share prices in Pakistan. He argued that the
125
main reason for the insignificant effect of dividend announcements on share price was
the fact that managers concentrate more on the earnings as well as the liquidity of the
firm and not the dividend payment. So managers in Pakistan prefer to increase the
earnings and the liquidity of the firm and hope that investors will react to this news
positively.
On the other hand, Aamir & Shah’s (2011) results contradict our findings and even
contradict Khan (2011) who investigated the same market; they found a positive
significant impact of dividend announcements on share prices in Pakistan and they used
the standard event methodology with the market model. The contradiction between
Aamir & Shah (2011) and our results is based on the different sample and the different
market; mainly they concentrated on large well established companies in the Pakistani
market while we chose 82 companies listed in the service and industrial sectors in the
Jordanian market.56 Thus, the managers of these companies appeared to use dividends
to signal a profitable future about the company where shareholders get more confidence
in the management and price shares to give a return on their investments.
Moreover, Zuguang & Ahmed (2010) investigated the impact of dividend
announcements on share prices in China; their sample included 218 dividend
announcements from stocks listed in the Shanghai Stock Exchange (SSE) 180 index.
The 180 SSE index includes the main shares listed by size, liquidity and sector to reveal
the overall performance of SSE. After the initiation of the “China reform and opening
up policy” the securities markets in China have developed in line with other economic
sectors in a very rapid fashion. They investigated the response of the Chinese market to
dividends by comparing the market model returns, mean adjusted returns and market
adjusted returns and they contradicted our results by reporting a positive significant
impact of the dividend announcements on share prices. We would argue that this
contradiction is based on the sample that has been chosen from the main 180 shares in
the Chinese market where the companies are characterized by large size and high levels
of liquidity in the market.
56
The main difference between my study and their research, they selected just large well established
companies but we chose 82 companies from two sectors (industrial and service) with no preference of
size.
126
The insignificance of the market reaction on the dividend announcement day might be
because there is a leakage of information in the Jordanian market a few days before the
announcement day. Thus, the market did not get surprised at the news of the dividend
announcement. For instance, the significant results on day (t-3) and day (t-4) before the
GAM with average abnormal returns of -16.3% and -7.7% respectively based on the
OLS method and -16.7% and -8.3% respectively based on the hybrid method suggest
that the market reacts early before dividend is announced.
The dividend increase
sample reported a significant average abnormal return in day (t-3) of -25% and -26%
based on the OLS and the hybrid method respectively and also the dividend decrease
sample has reported a significant average abnormal return on day (t-3) of -33% and 35% based on the OLS and the hybrid methods respectively.
But there was no
significant average abnormal return in the days before the announcement day when the
company announces no change in dividends. This result suggests that the Jordanian
stock market is inefficient in terms of regulation and ineffectiveness of supervision in
general when the companies decide to change their dividend payment compared to the
previous year - to increase or decrease the dividend. Nobanee et.al. (2009) argued that
there is a leakage of information in the Jordanian market; investors get news about
dividend payments before the announcement day. This confirms that the supervision in
the Jordanian market may be inefficient and the authorities have to apply regulations
better.
This result is in line with Kaleem & Salahuddin (2006) who investigated data for
Pakistani companies and found significant abnormal returns before the announcement
of dividends.
They argued that the Pakistani market suffered from ineffective
supervision during the period under investigation.
Although they investigated a
different market, a different period of time and a different sample they concluded that
there was information leakage in the LSE.
In addition, Uddin (2003) empirically
examined companies in the Dhaka Stock Exchange (DSE) and also found significant
abnormal returns a few days before the announcement date. They related this to the
information leakage where regulation was not effective; they argued that the DSE seems
to support the dividend irrelevance theory.
On the other hand, Akbar & Baig (2010) who investigated the reaction of 79 firm’s
share prices to dividend announcements in the Karachi Stock Exchange (KSE) during
127
the period 2004-2007 resulting in 129 cash dividend announcements found insignificant
average abnormal returns before the day of the announcement. This result contradicts
the result achieved even in Kaleem & Salahuddin’s (2006) empirical investigation on
the second market of Pakistan. This result can be interpreted as the KSE has an
efficient regulation system and effective supervision compared to LSE the reason might
be that KSE is the first market in Pakistan and it is recognized to be the most liquid and
biggest market in Pakistan.57 Moreover, our results contradict the findings of Zuguang
& Ahmed (2010) who reported insignificant pre-event window abnormal returns in the
three models that they used (market model, mean adjusted returns and market adjusted
returns). They argued that the there was no information leakage in the SSE and that the
supervision of the exchange was effective.
So the contradiction is based on the
effectiveness of supervision since Al-Shiab (2006) mentioned that the Jordanian
government is working on improving the efficiency of the market.
Furthermore, based on the whole sample, our results support the argument that investors
in the Jordanian market have a good deal of confidence in the performance of shares
after the dividend announcement event and reject the null hypothesis that the average
abnormal returns are close to zero. The largest average abnormal returns of 27.3% and
29.5%, in the whole sample, based on the OLS and the hybrid method respectively is
recorded for day t+3 implying that investors responded favourably to the dividend
signal three days after the disclosure of the information. Thus, significant average
abnormal returns can be earned by investors in the first day (t+1) and the third day (t+3)
after the dividend announcement day (t=0); there abnormal returns were significant at
5% and 10% level for day one (t=+1) according to t-test and the rank test respectively
and at the 10% level for day three according to the two tests (t test and rank test) based
on the standard OLS method. But both days where reported significant abnormal
returns at the 5% level when the hybrid method is applied. In more detail, the dividend
increase sample reported a significant positive average abnormal returns of 39% and
47% based on the OLS and the hybrid method respectively on day (t+3). The dividend
decrease sample reported a significant decrease of average abnormal returns at the 1%
confidence level during the three days (t+1 till t+3) suggesting that when companies
It is worth noting that based on Business Week – it is a weekly business magazine published by
Bloomberg L.P. and it’s headquarter is in New York - KSE was declared in 2002 as the “Best Performing
Stock Market of the World”.
57
128
announce a decrease in dividend payments then investors react relatively quickly after
the announcement day. The constant dividend sample recorded a significant positive
average abnormal return at the 1% confidence level in day (t+3) suggesting that when
companies announce constant dividend payment then investors take least three days to
react to the announcements.
Our result for the whole sample is in line with the findings of Chen et.al. (2009) and
Zuguang & Ahmed (2010); both empirical studies employed the market model of the
standard event methodology in the Chinese market and reported a positive significant
reaction to share prices after the dividend announcement. Both studies argued that the
Chinese market reacted significantly to the new information available in the market (the
dividend announcement) in a quick fashion; Zuguang & Ahmad (2010) reported a
significant average abnormal return one day after the announcement while Chen et.al.
(2009) found that there was a significant average abnormal return during the first three
days after the announcement. Although both studies relate to the Chinese market, the
significant results were not the same possibly because they investigated different time
periods.58
However, this result contradicts Karim (2010) who investigated the American market
and found that there was no reaction in the market after the dividend announcement. He
investigated 189 dividend announcements and their impact on the share prices listed in
the NASDAQ100. The empirical investigation of Karim (2010) was conducted around
the same time period as that in this empirical chapter but he has applied just the
standard OLS event methodology. He found insignificant average abnormal returns
after the dividend announcements which meant that investors in New York ignored the
information content of dividend if there was any.
Furthermore, Bhatia (2010) concluded that there was no reaction after dividend
announcements in India. Bhatia (2010) concentrated on popular companies in the
Indian market so it was not a randomly selected sample from the National Stock
Exchange of India (NSE). Bhatia (2010) related the reason behind his findings of no
significant reaction of the average abnormal returns after the dividend announcements
58
Chen et.al. (2007) investigated the period 2000-2004 while Zuguang & Ahmad (2010) empirically
investigated the period (2005-2009).
129
to the dividend policy followed by most of the popular firms in India. They pay a
constant dividend every year based on the face value of the share which is not the case
in the Jordanian market. In Jordan every company decides whether to pay or not based
on the performance of the company and the policy followed by the board of directors.
As we can see in table 5.8, based on the whole sample, the cumulative abnormal returns
are highest in the period of (+1 till +3) the first day after the announcement day till the
third day with a cumulative average return of 35.3% and 39% based on the OLS and the
hybrid method respectively. This result has been found to be significant at the level of
1%. In addition, the holding period after the dividend announcement for the five days
(+1 till +5) has been found to be significant at the 5 % level with a cumulative abnormal
return of 31.5% and 35.8 % based on the OLS and the hybrid method respectively.
Both periods suggest that shareholders can earn sizeable returns if they hold their shares
for the five days after the announcement. The significance of those two holding periods
confirms that the Jordanian market incorporates the information content of the dividend
announcements and reacts to the new information available in the market by adjusting
the share prices within three days. On the other hand, the cumulative abnormal return
for the two holding periods before the dividend announcement (-5 till -1) and (-3 till -1)
were insignificant according to the two parametric tests with a negative cumulative
abnormal return. This result suggests that if shareholders hold the shares just till the
dividend announcement then this will lead to loss in their wealth.
5.5 Conclusion
The impact of dividends on a company’s share price is a very important research subject
not only for the literature but also for the supervision and regulation of different capital
markets. Even though the impact of dividend on share prices and company value is still
a puzzle, the legal system in different capital markets accepts the fact that a dividend
announcement is price sensitive information.
On the other hand, the Dividend
Irrelevance Theory was introduced in 1961 by Miller and Modigliani (M&M) and
suggests that in the absence of costs relating to transactions and taxes, no firm can affect
the price of a security by selecting a particular payout policy.
Therefore, paying
dividends would not affect the firm’s value and its share price. Afterwards, different
130
academics have tried to explain the impact of dividends on share prices using signalling
theory, free cash flow theory and agency cost theory.
This chapter has examined the announcement of cash dividends on share prices in the
Jordanian capital market. It has investigated whether cash dividend announcements
result in an abnormal return around the announcement day in the Amman Stock
Exchange. The abnormal returns are calculated using the market model, over an event
window between t-5 and t+5 days.
The study used data for 492 events from 82
companies (51 industrial and 31 services) during the period 2005-2010. This empirical
chapter applied the OLS/EGARCH hybrid method and compared it to the standard OLS
event study methodology to investigate the impact of dividend announcements on share
prices.
This chapter adds to the previous empirical investigations by allowing
EGARCH in a hybrid method to treat the heteroskedasticity in the disturbance term.
The main findings of this empirical chapter are; first, no matter the methodology used
the significant average abnormal returns for both methods (OLS or hybrid method) are
the same but the level of significance is higher when the hybrid method has been
employed in the Jordanian market.
Second, the results are insignificant on the
announcement day (t=0) suggesting that dividends do not affect the share price; it
supports the dividend irrelevance theory proposed by M & M in 1961 which states that
dividends have no impact on share price and suggests that either the investors in the
Jordanian market do not react quickly enough to the news of the dividends or the
signalling theory does not hold in the Jordanian market. Third, the adjustment of prices
to new information starts on the first day after the announcement day and it continues
for at least 3 days. The most significant price adjustment takes place in the first three
days. Fourth, there is a significant relationship between cash dividends and abnormal
returns prior to the announcement day which implies that there is information leakage
prior to the information becoming publicly available.
Thus, the regulation and
supervision of the capital markets in Jordan can be said to be ineffective and inefficient
in preventing insider trading in relation to cash dividends.
The negative relationship between cash dividends and abnormal share returns on the
Amman Stock Exchange can be explained by the fact that the whole world is facing an
economic depression from 2006 till now and the period under investigation in this
131
chapter is from 2005 till 2010. Also, Omet (2011) reported that the liquidity of ASE
(Amman Stock Exchange) is relatively poor during the years 2007 till 2009. Finally,
one of the ways that can broaden this empirical study is to investigate the impact of
stock dividends on the share price and maybe compare it to another emerging market.
Another way is to divide the sample into different sectors to identify the different
characteristics that would affect one sector but not the other.
132
6. Conclusion
6.1 Introduction
The main emphasis of this study has been on the behaviour of the dividend policies of
listed non-financial Jordanian firms on the Amman Stock Exchange. Three objectives
had to be accomplished: firstly an investigation of the smoothness of dividend, secondly
an investigation of the factors affecting the two dividend decisions, and finally an
investigation of the impact of dividend announcements on the prices of shares of
Jordanian listed firms.
The dataset was extracted from two different sources, the ASE database (Amman Stock
Exchange) and the annual reports of the companies (the yearly shareholding company’s
guide). The basic aim of this chapter is to present the main conclusions of the different
empirical chapters on dividend policy.
The limitations of the study will also be
discussed. The structure of this chapter is given as follows: a summary of the key
findings of this study will be discussed in section 6.2; this includes a review of the three
empirical
chapters.
Section
6.3
highlights
the
limitations
and
possible
recommendations for any further research.
6.2 Key Findings And Implications
Chapter three generally examined the empirical evidence on whether or not Jordanian
firms that are listed on the ASE over the periods 1997-2006 smooth their dividend
policy. In this chapter, various investigations from models were developed in order to
examine the smoothness of dividend and the effect of the different characteristics of
Jordanian firms on the smoothness of dividend was explored. Results revealed that the
companies in the Amman stock market followed a smooth dividend policy.
Characteristics such as size, leverage and profitability do affect the Jordanian market’s
smoothness of dividend. Jordanian managers tend to smooth and regulate their long
term dividend policies. On the other hand, signalling theory did not support the effect
of the size on dividend smoothness, since it was shown that small firms that have a high
level of information asymmetry do not smooth their dividend faster than large size firms
in order to increase their corporate values and reduce any information asymmetry.
133
Leverage was found to have an influence on the smoothness of dividend in the Amman
Stock Exchange listed companies.
The high leveraged companies smooth their
dividends slower than the low leveraged ones since the high leveraged companies have
to service the debt and make sure not to face financial problems in the coming years.
Finally, it is concluded herein in terms of corporate profitability that there is a
significant impact between a firm’s profitability and its dividend policy in the Jordanian
market. High profitable firms smooth their dividend faster than low profitable ones;
because high profitable firms are in a better cash flow than low profitable ones. It can
be indicated that firms in general smooth their dividend payment which involves
moving toward the target dividend payout ratio and following the signalling theory
(signal that managers are confident about the firm’s successful future income).
The second empirical chapter of this study is an investigation on the determinants of the
two dividend decisions by applying Heckman’s (1979) simultaneous two-step model
which investigates the dividend decisions and explicitly distinguishes the difference
between these two decisions. Results showed that the Jordanian firms indicated that
growth, size, profitability and leverage were significant factors in impacting those
dividend decisions, which would not contradict the internationally available evidence.
It was found that profitability and size are positively connected to dividend decisions,
while empirical results show a negative correlation between the dividend decisions and
leverage. An interesting result was the insignificance of a firm’s ownership structure in
impacting corporate dividend decisions.
In summary, dividend policy in both
developed and emerging countries are affected by identical factors.
The final chapter shows the empirical investigation for the impact of dividend
announcements on share prices. It adopted the standard event study methodology and
compared it to the hybrid method (OLS/EGARCH). In specific, this chapter allowed
the hybrid method to treat the heteroskedasticity in the disturbance term. The findings
show that both methods used concluded the same significant average abnormal returns
but the hybrid method was superior in the level of significance.
In addition, the
announcement day resulted in an insignificant impact on the share prices in the
Jordanian market which came in line with the M & M irrelevance theory. On the other
hand, the results showed that the Jordanian market experience information leakage and
134
this was supported by the significant average abnormal return before the announcement
day.
6.3. Limitations Of The Study
In addition to the useful insights provided by this research there are some drawbacks
that could reduce the extent to which these findings could be generalized. One of the
limitations of this study could be the period under investigation (1997-2006). Within
this timeline Amman Stock Exchange has seen an increase in share prices and an
increase in the market index and this has been followed by the world financial crisis and
the drop in the market index through the years of 2005-2010; due to lack of liquidity. As
a result, during the ten years of 1997-2006, the Jordanian firms of the non financial
sector generally prefer to go to the Stock Market to gain funds. Consequently the
outcomes of this study are restricted within its period. Although the research focuses in
Jordan, its results could be useful while attempting to generalize the rival models of
dividend policy.
Another limitation can be the selection of just the main determinants (variables) selected
in the literature that affected dividend policy. However, other comprehensive studies
have included several other variables to check the impact on dividend policy (Baker
et.al., 2001). The main reason of not including all other variables is that the main
contribution of the second empirical chapter – the determinants of the two dividend
decisions – was on the methodology applied not the variables that affect the dividend
policy.
Compared to Jordan, other developing markets have a high number of structural
features that are presented in the Jordanian capital market. Hence, the outcomes of this
study can facilitate a comparative investigation in Jordan as well as other developing
countries. The dataset chosen for the present study could be perceived as a limitation,
since it was obtained from the annual reports of the firms and publications of the
Amman Stock Exchange. All efforts were made to assure the security of the data which
still though remains a problem.
135
On the other hand, this is not a difficulty faced only by the developing capital markets;
it may be faced by the developed ones as well. It does not worth anything if the data
failed to accomplish the basic regression assumptions’ and caused biased estimates of
the coefficient and standard error resulting in invalid p values of the t test and the f test.
Thus, the present study has assured that the assumptions of the regressions have been
met by the well collected data.
The statistical software used in this study is STATA which was used in analysing the
data collected. Using STATA helped a lot in the analyses, since it can be counted one
of the best software’s available nowadays for statistical purposes. In our study, we
investigated several methods of testing data and others showing how to meet regression
assumptions. For instance, we test for the existence of any particular observation that it
is significantly different from all other observations; these are known as unusual
influential data (outliers) and make a huge difference in the results of a regression
analysis.
Furthermore, we test for homoscedasticity, linearity, multicollinareaty and normality of
residuals.
In addition, to make sure that the data is accurate, a sample has been
compared to four different sources of data involving the firm’s yearly reports, CD’S
(which include the balance sheet and income statement of each company listed in ASE),
the Amman stock exchanges publications, as well as the website of the ASE.
The
comparison of these different sources shows that there is no difference in the data.
The different proxy variables that have been used in the present study could be
considered as a third limitation. Despite the fact that the proxies used for the variables
were extracted from the theories and the empirical investigations done for dividend
policy, they may not perfectly stand for the theoretical propositions, since they remain
proxies. Additionally, it is far from easy to determine the proxies that are unrelated to
one another. Moreover, all of the empirical investigations done on dividend policy
faced a general problem which is the issue of determining the proxies for different
variables.
136
6.4 Recommendations For Further Research
The present thesis has implemented new models and utilized a justified testing
methodology for 85 industrial and service firms in Jordan from 1997-2006. The study
reached interesting results after estimating the models, which helped in understanding
and clarifying more the corporate finance of non-financial companies in Jordan.
Furthermore, it has given the basis for estimating the degree of the explanatory power of
different models that can clarify corporate finance theory. The empirical investigations
done in this thesis provided some insight which might constitute the basis for future
research in Jordan. The findings could be used for a comparison with other empirical
investigations that can be done in other developing markets.
Consequently, more
research can be applied to widen the empirical investigations to other emerging markets.
The motivation for this kind of studies of other emerging markets comes from the
contradicting outcomes and the drawbacks of those currently existing studies. The basic
problem of the existing literature is that most of the empirical investigations have been
done on small samples, which is considered to be a sample selection problem. A
straight forward further research could be done in dividend policy using samples from
such countries that could be representative and comprehensive. Regarding the case of
Jordan, further research could include the investigation of the role of the Amman Stock
Exchange in the development of the Jordanian economy. Corporate financial decisions
as well as market efficiency could also be investigated. Moreover, further research may
broaden the examination of this thesis by including financial firms in the sample. This
could result into more effective outcomes and therefore, provide a greater proof of
evidence for any policy implications mentioned above. If other researchers investigated
the financial sector then this would help in knowing the differences between financial
and non-financial firms.
Regarding the first empirical chapter; dividing the sample into different sectors and
checking the different firm characteristics on the smoothness of dividends; would help
in understanding better the main factors that would increase or decrease the adjustment
toward the target dividend payout ratio. In addition, the second empirical chapter – the
determinants of the two dividend decisions- may be improved by including all of the
variables mentioned in the literature to conclude with a comprehensive picture of the
determinants of the two dividend decisions. Moreover, the future research of the last
137
empirical chapter - the impact of dividend announcements on share prices- would be to
split the data by year to check if the results would be different simply because this
would help in identifying the impact of the financial crisis on the market.
138
REFERENCES
Aamir, M. and Shah, S.Z.A. (2011). ‘Dividend Announcements and the Abnormal
Stock returns for the event firm and its rivals’, Australian Journal of Business and
Management Research, Vol. 1, No. 8, pp. 72-76.
Abor, J. and Amidu, M (2006). ‘Determinants of the Dividend payout ratio in Ghana’,
The Journal of Risk Finance, Vol. 7, No. 2, pp. 136-145.
Adaoglu, C. (2000). ‘Instability in the Dividend Policy of the Istanbul Stock Exchange
(ISE) Corporations: Evidence from an Emerging Market’, Emerging Market Review,
Vol. 1, pp. 252-270.
Adedeji, A. (1998). ‘Does the Pecking Order Hypothesis Explain the Dividend Payout
Ratios of Firms in the UK?’, Journal of Business Finance & Accounting, Vol. 25, pp.
1127-1155.
Agrawal, A. and Jayaraman, N. (1994). ‘The Dividend Policies of All-equity Firms: A
Direct Test of the Free Cash Flow Theory’, Managerial and Decision Economics, Vol.
15, pp. 139-148.
Agyei, S.K. and Marfo-Yiadom, E. (2011). ‘Determinants of Dividend Policy in
Chana’, International Research Journal of Finance and Economics, Vol. 61, pp. 99108.
Aharony, J. and Swary, I. (1980). ‘Quarterly Dividend and Earnings Announcements
and Stockholders' Returns: An Empirical Analysis’, Journal of Finance, Vol. 35, pp. 112.
Ahmed, H. and Javad, A. (2009). ‘Dynamics and Determinants of Dividend Policy in
Pakistan’, International Research Journal of Finance and Economics, Vol.28, pp. 56–
61.
Aivazian, V., Booth, L. and Cleary, S. (2001). ‘Dividend Policy in Developing
Countries’, Paper Presented at the Eighth Annual Conference, Multinational Finance
Conference, Italy, 1- 42.
Aivazian, V., Booth, L. and Cleary, S. (2003). ‘Dividend Policy and the Organisation of
Capital Markets’, Journal of Multinational Financial Management, Vol. 13, pp. 101121.
Aivazian, V., Booth, L. and Cleary. S. (2006). ‘Dividend smoothing and debt ratings’,
Journal of Financial and Quantitative Analysis, Vol. 41, No. 2, pp. 439-453.
139
Akbar, M. and Baig, H.H. (2010). ‘Reaction of Stock Prices to Dividend
Announcements and Market Efficiency in Pakistan’, The Lahore Journal of Economics,
pp. 103-125.
Al-Ajmi J. and Abo Hussain H., (2011). ‘Corporate dividends decisions: evidence from
Saudi Arabia’, Journal of Risk Finance, Vol. 12, No.1, pp.41-56.
Al-Fayoumi, N.A, Khamees, B.A. and Al-Thuneibat, A.A. (2009). ‘Inrofmation
Transmission among Stock Return Indexes: Evidence from the Jordanian Stock
Market’, International Research Journal of Finance and Economics, Vol. 24, pp. 194208.
Ali, M.B. and Chowdhury, T.A. (2010). ‘Effect of Dividend on Stock Price in Emerging
Stock Market: A Study on the Listed Private Commercial Banks in DSE’, International
Journal of Economics and Finance, Vol. 2, No. 4, pp. 52-64.
Al-Kuwari, D. (2009). ‘Determinants of the Dividend Policy in Emerging Stock
Exchanges: The Case of GCC Countries’, Global Economy & Finance Journal, Vol. 2,
No. 2, pp. 38-63.
Allen, D.E. (1993). ‘The Pecking Order Hypothesis: Australian Evidence’, Applied
Financial Economics, Vol. 3, pp. 101-112.
Allen, F., Bernardo, A.E. and Welch, I. (2000). ‘A Theory of Dividends Based on Tax
Clienteles’, Journal of Finance, Vol. 55, pp. 2499-2536.
Allen, F. and Michaely, R. (1995). ‘Dividend Policy’, in R. A. Jarrow, V. Maksimovic,
and W. T. Ziemba, eds.: Handbooks in Operations Research and Management
Sciences: Finance (Elsevier, New York).
Alli, K.L., Khan, A.Q. and Ramirez, G.G. (1993). ‘Determinants of Corporate Dividend
Policy: A Factorial Analysis’, The Financial Review, Vol. 28, pp. 523-547.
Al-Malkawi, N.H. (2007). ‘Determinants of Corporate Dividend Policy in Jordan: An
application of the Tobit Model’, Journal of Economic & Administrative Sciences, Vol.
23, No. 2, pp. 44-70.
Al-Najjar, B. and Hussainey, K. (2009). ‘What Drives Firms’ Capital Structure and
Dividend Policy?’, Working Paper. Middlesex University.
Al-Shiab, M. (2006). ‘The Predictability of the Amman Stock Exchange using the
Univariate Autoregressive Integrated Moving Average (ARIMA) Model’, Journal of
Economic and Administrative Sciences, Vol. 22, No. 2, pp. 17-35.
Al-Shubiri, F.N. (2011). ‘Capital Structure and Market Power: Evidence from Jordanian
Banks, Managing Global Transitions, vol. 9, No. 3, pp. 289-310.
140
Al-Yahyaee, K.H., Pham, T. and Walter, T. (2010). ‘Dividend stability in a unique
environment’, Managerial Finance, Vol. 36, No. 10, pp. 903-916.
Alzahrani, M. and Lasfer, M. (2009). ‘The Impact of Taxation on Dividends:A crossCountry Analysis’. Available at SSRN:http://ssrn.com/abstract=1343826.
Amihud, Y. and Li, K. (2002). ‘The declining information content of dividend
announcements and the effect of institutional holdings’, Working paper, New York
University.
Amihud, Y. and Murgia, M. (1997). ‘Dividends, Taxes, and Signalling: Evidence from
Germany’, Journal of Finance, Vol. 52, pp. 397-408.
Amman Financial Market (1997), The Twentieth Annual Report. Research and Studies
Department, Amman, Jordan.
Amman Financial Market (1998), The Twenty First Annual Report. Research and
Studies Department, Amman, Jordan.
Amman Financial Market (1999), The First Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2000), The Second Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2001), The Third Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2002), The Fourth Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2003), The Fifth Annual Report. Research and Studies
Department, Amman, Jordan.
Amman Financial Market (2004), The Sixth Annual Report. Research and Studies
Department, Amman, Jordan.
Amman Financial Market (2005), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2006), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2007), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2008), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
141
Amman Financial Market (2009), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
Amman Financial Market (2010), The Seventh Annual Report. Research and External
Relations Department, Amman, Jordan.
Anand, M. (2004). ‘Factors influencing dividend policy decisions of corporate India’,
The ICFAI Journal of Applied Finance, Vol. 10, pp. 5-16.
Andres, C., Betzer, A., Goergen, M. and Renneboog, L. (2008). ‘The dividend policy of
German firms: A panel data analysis of partial adjustment models’, Tilburg University,
Discussion Paper, No. 2008-67.
Ang, J.S. (1987). ‘Do Dividends Matter? A Review of Corporate Dividend Theories and
Evidence’, Salomon Brothers Center for the Study of Financial Institutions and the
Graduate Schools of Business Administration of New York University, New York.
Arab Monetary Fund (2011), ‘Economic Statistics Bulletin of Arab Countries’ available
on the internet at http://www.amf.org.ae/statistics.
Armitage, S. (1995). ‘Event study methods and evidence on their performance’, Journal
of Economic Surveys, Vol. 9, No. 1, pp. 25-52.
Arslan, O. and Karan, M.B. (2006). ‘Ownership and Control as Determinants of
Corporate Debt Maturity: A Panel Study of an Emerging Market’, Corporate
Governance: An International Review, Vol. 4, pp. 312-324.
Asamoah, G.N. (2010). ‘The Impact of Dividend Announcement on Share Price
Behaviour in Ghana’, Journal of Business & Economics Research, Vol. 8, No. 4, pp.
47-58.
Asquith, P. and Mullins, D.W.Jr. (1983). ‘The impact of Initiating Dividend Payments
on Shareholders' Wealth’, Journal of Business, Vol. 56, pp. 77-96.
Asquith, P. and Mullins, D.W.Jr. (1986). ‘Signalling with, Dividends, Stock
Repurchases, and Equity Issues’, Financial Management, Vol. 15, pp. 27-44.
Avner, K. and Roni M. (2000), ‘Dividends and Taxes: A Reexamination’, Financial
Management, Vol. 29.
Baker, H.K., Farrelly, G.E. and Edelman, R.B. (1985). ‘A Survey of Management
Views on Dividend Policy’, Financial Management, Vol. 14, pp. 78-84.
Baker, H.K, Powell, G.E. and Veit, E.T. (2002). ‘Revisiting Managerial Perspectives on
Dividend Policy’, Journal of Economics and Finance, Vol. 26, pp. 267-283.
Baker, H.K, Theodore, E. and Powell, G.E. (2001). ‘Factors Influencing the Dividend
Policy Decisions on NASDAQ Firms’, Financial Review, Vol. 38, pp. 19-38.
142
Baker, H.K., and Powell, G.E. (1999). ‘How Corporate Managers View Dividend
Policy’, Quarterly Journal of Business and Economics, Vol. 38, pp.17-35.
Bali, R. (2003). ‘An Empirical Analysis of Stock Returns Around Dividend Changes’,
Applied Economics, Vol. 35, pp. 51-61.
Barclay, M.J. (1987). ‘Dividends, Taxes, and Common Stock Prices: The Ex-Dividend
Day Behavior of Common Stock Prices Before the Income Tax’, Journal of Financial
Economics, Vol. 19, pp. 31-44.
Barclay, M.J., Smith, C.W. and Watts, R.L. (1995). ‘The Determinants of Corporate
Leverage and Dividend Policies’, Journal of Applied Corporate Finance, Vol. 7, pp. 419.
Bartholdy, J., Olson D., and Peare, P. (2007). ‘Conducting event studies on a small
stock exchange’, The European Journal of Finance, Vol. 13, pp. 227-252.
Baskin, J. (1989). ‘An Empirical-Investigation of the Pecking Order Hypothesis’,
Financial Management, Vol. 18, pp. 26-35.
Ben Naceur, S. and Goaied, M. (2002). ‘The Relationship Between Dividend Policy,
Financial Structure and Firm Value’, Applied Financial Economics, Vol. 12, No. 12, pp.
843-849.
Ben Naceur, S., Goaied, M. and Belanes, A. (2007). ‘On the determinants and dynamics
of dividend policy’, International Review of Finance, Vol. 6, No. 1-2, pp. 1-23.
Benston, G.J. (1982). ‘Accounting numbers and economic values’, Antitrust Bulletin,
Vol. 27, pp. 161-215.
Benzinho, J. (2004). ‘The dividend policy of the Portuguese corporations. Evidence
from Euronext Lisbon’, available at SSRN: http://ssrn.com/abstract=609461.
Bernatzi, S., Michaely, R. and Thaler, R.H. (1997). ‘Do Changes in Dividends Signal
the Future or the Past?’, Journal of Finance, Vol. 52, pp. 1007-1034.
Berk, J. and DeMarzo, P. (2009). ‘Corporate Finance: The Core’, Pearson Education,
Prentice Hall
Bernstein, P.L. (1996). ‘Dividends: The Puzzle’, Journal of Applied Corporate Finance,
Vol. 9, pp. 16-22.
Bhatia, P. (2010). ‘A Study of Dividend Announcements on Stock Returns of Popularly
Traded Companies in India’, Sri Krishna International Research and Educational
Consortium, pp. 178- 189.
Bhattacharya, S. (1979). ‘Imperfect Information, Dividend Policy, and "the Bird in the
Hand" Fallacy’, Bell Journal of Economics, Vol. 10, pp. 259-270.
143
Bhattacharya, S. (1980) ‘Nondissipative Signalling Structures and Dividend Policy’,
Quarterly Journal of Economics, Vol. 95, pp. 1-24.
Black, F. (1976). ‘The Dividend Puzzle’, Journal of Portfolio Management, Vol. 2, pp.
5-8.
Black, F. and Scholes, M.S. (1974). ‘The Effects of Dividend Yield and Dividend
Policy on Common Stock Prices and Returns’, Journal of Financial Economics, Vol. 1,
pp. 1-22.
Blume, M.E. (1980). ‘Stock Returns and Dividend Yields: Some More Evidence’,
Review of Economics and Statistics, Vol. 62, pp. 567-577.
Bodla, B.S., Pal, K. and Sura, J.S. (2007). ‘Examining Application of Lintners Dividend
Model in Indian Banking Industry’, The IUP Journal of Bank Management.
Bolerslev, T. (1986). ‘Generalized Autoregressive Conditional Heteroskedasticity’,
Journal of Econometrics, Vol. 31, pp. 307-27
Box, G.E.P. and Pierce, D.A. (1970). ‘Distribution of Residual Autocorrelations in
Autoregressive Integrated Moving Average Time Series Models’, Journal of American
Statistical Association, Vol. 65, pp. 1509-1526.
Bradley, M., Capozza, D.R. and Seguin, P.J. (1998). ‘Dividend Policy and Cash-Flow
Uncertainty’, Real Estate Economics, Vol. 26, pp. 555-580.
Brainard, W. and Tobin, J. (1968). ‘Pitfalls in Financial Model Building’, American
Economic Review, Vol. 58, pp. 99-122.
Brav, A., Graham, J., Harvey, C. and Michaely, R.. (2005). ‘Payout policy in the 21st
century’, Journal of Financial Economics, Vol 77:, pp. 483-527.
Brealey, R.A., Myers, S.C. and Allen, F. (2006). ‘Corporate Finance’, 8th edition,
McGraw-Hill Irwin.
Brennan, M.J. (1970). ‘Taxes, Market Valuation and Corporate Financial Policy’,
National Tax Journal, Vol. 23, pp. 417-427.
Brittain, J.A. (1964). ‘The Tax Structure and Corporate Dividend Policy’, American
Economic Review, Vol. 54, pp. 272-287.
Brittain, J.A. (1966). ‘Corporate Dividend Policy’, The Brooking Institution,
Washington, DC.
Brown, S.J. and Warner, J.B. (1985). ‘Using daily stock returns: The Case of Event
Studies’, Journal of Financial Economics, Vol. 14, pp. 3-31.
Bryman, A. and Cramer, D. (2001). Quantitative Data Analysis with SPSS Release 10
for Windows, London: Routledge.
144
Campbell, C.J. and Wasley, C.E. (1993). ‘Measuring security performance using daily
NASDAQ returns’, Journal of Financial Economics, Vol. 33, pp. 73-92.
Carhart, M.M. (1997). ‘On Persistence in Mutual Fund Performance’, Journal of
Finance, Vol. 52, No. 1, pp. 57–82.
Carroll, T.J. (1995). ‘The Information Content of Quarterly Dividend Changes’, Journal
of Accounting, Auditing and Finance, Vol. 10, pp. 293-320.
Carvalhal-da-Silva, A. and Leal, R. (2003) ‘Corporate Governance, Market Valuation
and Dividend Policy in Brazil’, Working Paper No. 390, (COPPEAD Graduate
Business School, Fedral University of Rio de Janeiro, Brazil).
Chang, R.P. and Rhee, S.G. (1990). ‘The Impact of Personal Taxes on Corporate
Dividend Policy and Capital Structure Decisions’, Financial Management, Vol. 19, pp.
21-31.
Charitou A. and Vafeas, N. (1998). ‘The Association Between Operating Cash Flows
and Dividend Changes: An Empirical Investigation’, Journal of Business Finance and
Accounting, Vol. 25, pp. 225-249.
Chen, D.H., Blenman, L.P., Bin, F.S., and Chen, J. (2006), ‘The effects of Open Market
Reforms on the Behavior of China’s Stock Prices’, International Research Journal of
Finance and Economics, Vol. 5.
Chen, D.H., Huang, H.H.L. and Cheng, T. (2009). ‘The announcement effect of cash
dividend changes on share prices: An empirical analysis of China’, The Chinese
Economy, Vol. 42, No. 1, pp. 62-85.
Claessens, S., Klingebiel, D. and Schmukler, S.L. (2006). ‘Stock market development
and internationalization: Do economic fundamentals spur both similarly?’, Journal of
Epirical Finance, Vol. 13, pp. 316-350.
Clarke, J., Tomas J., and Gershon M., (2001) “The Efficient Market Hypothesis.” In
Expert Financial Planning: Investment Strategies from Industry Leaders. Arffa, R. C..
New York, NY. Wiley & Sons.
Cronqvist, H. and Nilsson, M. (2003). ‘Agency Costs of Controlling Minority
Shareholders’, Journal of Financial and Quantitative Analysis, Vol. 38, pp. 695-719.
Crutchley, C.E., Jensen, M.R.H., Jahera, J.S. Jr. and. Raymond, J.E. (1999). ‘Agency
problems and the simultaneity of financial decision making: the role of institutional
ownership, International Review of Financial Analysis, Vol. 8, pp. 177-197.
145
Crutchley, C. and Hansen, R. (1989). ‘A Test of the Agency Theory of Managerial
Ownership, Corporate Leverage and Corporate Dividends’, Financial Management,
Vol. 18, pp. 36-46.
Datta, S., Iskandar-Datta, M. and Raman, K. (2005). ‘Managerial Stock Ownership and
the Maturity Structure of Corporate Debt’, An article submitted to The Journal of
Finance, Manuscript 1325.
DeAngelo, H. and DeAngelo, L. (1990). ‘Dividend Policy and Financial Distress: An
Empirical Investigation of Troubled NYSE Firms’, Journal of Finance, Vol. 45, pp.
1415-1431.
DeAngelo, H., DeAngelo, H. and Skinner, D.J. (1992). ‘Dividends and losses’,
Journal of Finance, Vol. 47, pp. 1837-1863.
DeAngelo, H., DeAngelo, L. and Skinner, D.J. (1996). ‘Reversal of Fortune: Dividend
Signalling and the Disappearance of Sustained Earnings Growth’, Journal of
Financial Economics, Vol. 40, pp. 341-371.
DeAngelo, H., DeAngelo, L. and Stulz, R.M. (2004). ‘Dividend Policy, Agency Costs,
and Earned Equity’, Financial Economics Working Paper, No.10.
DeAngelo, H., DeAngelo, L. and Stulz, R.M. (2006). ‘Dividend policy and the
earned/contributed capital mix: A test of the lifecycle theory’, Journal of Financial
Economics, Vol. 81, pp. 227–254.
Demirer R, Gubo D. and Kutan, A.M. (2007). ‘An analysis of cross-country herd
behavior in stock markets: A regional perspective’, Journal of International Financial
Markets, Institutions and Money, Vol. 3, pp. 123-142.
Dempsey, S.J. and Laber, G. (1992). ‘Effects of Agency and Transaction Costs on
Dividend Payout Ratios: Further Evidence of the Agency-Transaction Cost
Hypothesis’, Journal of Financial Research, Vol. 15, pp. 317-321.
Demsetz H. (1983). ‘The structure of ownership and the theory of the firm’, Journal of
Law and Economics, Vol. 26, pp. 375-390.
Demsetz H. and Lehn K. (1985). ‘The structure of corporate ownership: Causes and
consequences’, Journal of Political Economy, Vol. 93, No. 6, pp. 1155-1177.
Denis, D.J., Denis, D.K. and Sarin, A. (1994). ‘The Information Content of Dividend
Changes: Cash Flow Signalling, Overinvestment, and Dividend Clienteles’, Journal of
Financial and Quantitative Analysis, Vol. 29, pp. 567-587.
146
Denis, D.J. and Osobov, I. (2008). ‘Why do firms pay dividends? International
evidence on the determinants of dividend policy’, Journal of Financial Economics,
Vol. 89, pp. 62–82.
Deshmukh, S. (2003). ‘Dividend Initiations and Asymmetric Information: A Hazard
Model’, Financial Review, Vol. 38, pp. 351-368.
Dewenter, K.L. and Warther, V.A. (1998). ‘Dividends, Asymmetric Information, and
Agency Conflicts: Evidence from a Comparison of the Dividend Policies of Japanese
and U.S. Firms’, Journal of Finance, Vol. 53, pp. 879-904.
Diamond, J.J. (1967). ‘Earnings Distribution and the Valuation of Shares: Some
Recent Evidence’, Journal of Financial and Quantitative Analysis, Vol. 2, pp. 15-30.
Diamond, D.W. and Verrecchia. R.E. (1991). ‘Disclosure, liquidity, and the cost of
capital’, The Journal of Finance, Vol. 46, pp. 1325-1360.
Dodd, P. and Warner, J.B. (1983). ‘On Corporate Governance: A Study of Proxy
Contests’, Journal of Financial Economics, Vol. 11, No. 1, pp. 401-438.
Du, J. and Dai, Y. (2005). ‘Ultimate Corporate Ownership Structures and Capital
Structures: Evidence from East Asian Economies’, Corporate Governance: An
International Review, Vol. 13, No. 1, pp. 60-71.
Dyckman, T., Philbrick, D. and Stephan, J. (1984), ‘A Comparison of Event Study
Methodologies Using Daily Stock Returns: A Simulation Approach’, Journal of
Accounting Research, Vol. 22, pp. 1-30.
Easterbrook, F.H. (1984). ‘Two Agency Costs Explanations of Dividends’, American
Economic Review, Vol. 74, pp. 650-659.
Eddy, A. and Seifert, B. (1988). ‘Firm Size and Dividend Announcements’, Journal of
Financial Research, Vol. 14, pp. 295-302.
El-Kouri, R. and Almwalla, M. (1997). ‘The Effect of Dividend Changes on Security
Prices: The case of Jordanian Companies’, Abhath Al-Yarmouk, Vol. 13, pp. 87-94.
Engle, R. (1982). ‘Autoregressive Conditional Heteroskedasticity with Estimates of
the Variance of United Kingdom Inflation’, Econometrica, Vol. 50, pp. 987-1007.
Eriotis, N. (2005). ‘The Effect Of Distributed Earnings And Size Of The Firm To Its
Dividend Policy: Some Greek Data’, International Business & Economics Journal,
Vol. 4, No. 1, pp. 45–51.
Eriotis, N. and Vasiliou, D. (2003). ‘Dividend Policy: An empirical Analysis of the
Greek Market’, International Business and Economics research Journal, Vol. 3, No.
3, pp. 49-57.
147
Faccio, M., Lang, L. and Young, L. (2001). ‘Dividend and Expropriation’, American
Economic Review, Vol. 91, pp. 54-78.
Faccio, M. and Lang, L. (2002). ‘The Ultimate Ownership of Western European
Corporations’, Journal of Financial Economics, Vol. 65, pp. 365-395.
Fairchild, R.J. (2003). ‘Dividend Smoothing and Optimal Re-Investment’, Managerial
Finance, Vol. 29, pp. 35-48.
Fama, E.F. (1974). ‘The Empirical Relationships Between the Dividend and
Investment Decisions of Firms’, American Economic Review, Vol. 64, pp. 304-318.
Fama, E.F. (1976) Foundations of finance. New York: Basic Book.
Fama, E.F. and Babiak, H. (1968). ‘Dividend Policy: An Empirical Analysis’, Journal
of American Statistical Association, Vol. 63, pp. 1132-1161.
Fama, E. and French K. (1992). ‘The cross-section of expected stock returns’, Journal
of Finance, Vol. 47, pp. 427-465.
Fama, E. and French K. (1993). ‘Common risk factors in the returns on stocks and
bonds’, Journal of Finance, Vol. 33, pp. 3-56.
Fama, E. and French K. (1998). ‘Taxes, financing decisions, and firm value’, Journal
of Finance, Vol. 53, pp. 819-829.
Fama, E.F. and French, K.R. (2001). ‘Disappearing Dividends: Changing Firm
Characteristics or Lower Propensity to Pay?’, Journal of Financial Economics, Vol.
60, pp. 3-43.
Fama, E.F. and French, K.R. (2002). ‘Testing Trade-Off and Pecking Order
Predictions About Dividends and Debt’, The Review of Financial Studies, Vol. 15, pp.
1-33.
Fazzari, S.M., Hubbard, R.G. and Petersen, B.C. (1988). ‘Financing Constraints and
Corporate Investment’, Brooking Papers on Economic Activity, Vol. 1, pp. 141-195.
Ferri, M.G. and Jones, W.H. (1979). ‘Determinants of financial structure: A new
methodological approach’, The Journal of Finance, Vol. 34, pp. 631-644.
Ferris, S., Sen, N. and Yui, H. (2006). ‘God save the queen and her dividends:
Corporate payouts in the U.K’, Journal of Business, Vol. 79, pp. 1149–1173.
Fisher G.R. (1961). ‘Some Factors Influencing Share Prices’, Economic Journal, Vol.
71, pp. 121-141.
Flannery, M.J. and Rangan, K.P. (2006). ‘Partial Adjustment toward Target Capital’,
Journal of Financial Economics, Vol. 41, pp. 41-73.
148
Fluck, Z. (1999). ‘The Dynamics of Management-Shareholder Conflict’, Review of
Financial Studies, Vol. 12, pp. 379-404.
Frankfurter, G.M., Kosedag, A., Schmidt, H. and Topalov, M. (2002). ‘The Perception
of Dividends By Management’, Journal of Psychology and Financial Markets, Vol. 3,
pp. 202-217.
Frankfuter G.M.. and Wood B.G. (2002). ‘Dividend policies and their empirical test’,
International Review of Financial Analysis, Vol. 11, pp. 111-138.
Franklin, S.J. and Muthusamy, K. (2010). ‘Leverage, Growth and Profitability as
Determinants of Dividend Payout Ratio- Evidence from Indian Paper Industry’, Asian
Journal of Business Management Studies, Vol. 1, No. 1, pp. 26-30.
Friend, I. and Puckett, M. (1964). ‘Dividend and Stock Prices’, American Economic
Review, Vol. 54, pp. 656-682.
Gaver, J.J. and Gaver, K.M. (1993). ‘Additional Evidence on the Association Between
the Investment Opportunity Set and Corporate Financing, Dividend, and
Compensation Policies’, Journal of Accounting and Economics, Vol. 16, pp. 125-160.
Ghosh, C and Woolridge, J.R. (1988). ‘An Analysis of Shareholder Reaction to
Dividend Cuts and Omissions’, Journal of Financial Research, Vol. 14, pp. 281-294.
Gill, A., Biger, N. and Tibrewala, R. (2010). ‘Determinants of Dividend Payout
Ratios: Evidence from United States’, The Open Business Journal
Gillan, S.L., and Starks L.T., (2000), Corporate governance proposals and shareholder
activism: The role of institutional investors, Journal of Financial Economics 57, 275305.
Gitman, L. (2007). ‘Principles of Managerial Finance’, 12th edition, Addison Wesley.
Glen, J.D., Karmokolias, Y., Miller, R.R. and Shah, S. (1995). ‘Dividend Policy and
Behavior in Emerging Markets’, Discussion Paper No. 26, (International Financial
Corporation).
Goergen, M.., Renneboog, L. and Correia da Silva. L. (2005). ‘When do German firms
change their dividends?’, Journal of Corporate Finance, Vol. 11, No. 1, pp. 375-399.
Gombola, Mi.J. and Liu, F.Y.L. (1993). ‘Considering Dividend Stability in the
Relation Between Dividend Yieldsand Stock Returns’, Financial Review, Vol. 16, pp.
139-150.
Gordon, M.J. (1959). ‘Dividends, Earnings, and Stock Prices’, Review of Economics
and Statistics, Vol. 41, pp. 99-105.
149
Gordon, M.J. (1962). ‘The Investment, Financing and Valuation of the Corporation,
Irwin Series in Economics.
Gordon, M.J. (1963). ‘Optimal Investment and Financing Policy’, Journal of Finance,
Vol. 18, pp. 264-272.
Goudarzi, H. and Ramanarayanan, C.S. (2011),’Modeling Asymmetric Volatility in
The Indian Stock Market’, International Journal of Business and Management, Vol. 6,
No. 3.
Granger, C.W.J. and Teräsvirta, T. (1993). ‘Modelling Non-Linear Econometric
Relationships’, Oxford: Oxford University Press.
Greene, W.H. (2003). ‘Econometric analysis’, Prentice Hall.
Gugler, K. (2003). ‘Corporate Governance, Dividend Payout Policy, and the
Interrelation Between Dividends, R&D, and Capital Investment’, Journal of Banking
and Finance, Vol. 27, pp. 1297-1321.
Gugler, K. and Yurtoglu, B.B. (2003). ‘Corporate Governance and Dividend Pau-Out
Policy in Germany’, European Economic Review, Vol. 47, pp. 731-758.
Gujarati, D.N. (2003). Basic Econometrics, 4th Edition, Irwin/McGraw-Hill: New
York.
Gwilym, O.A., Morgan, G. and Thomas, S. (2000). ‘Dividend Stability, Dividend
Yield and Stock Return: UK Evidence’, Journal of Business and Accounting, Vol. 27,
pp. 261-281.
Gwilym, O.A., Seaton, J. and Thomas, S. (2004). ‘Dividends Aren’t Disappearing:
Evidence from the UK’, Working Paper, University of Southampton
Han, K.C., Lee, S.H. and Suk, D.Y. (1999). ‘Institutional Shareholders and
Dividends’, Journal of financial and Strategic Decisions, Vol. 12, pp. 53-62.
Hart, O. (1995). ‘Firms, Contracts and Financial Structure’, Oxford University Press.
Healy, P.M. and Palepu, K.G. (1988). ‘Earnings Information Conveyed by Dividend
Initiations and Omissions’, Journal of Financial Economics, Vol. 21, pp. 149-176.
Heckman, J.J. (1979). ‘Sample Selection Bias as a Specification Error’, Econometrica,
Vol. 47, No. 1, pp. 153-161.
Heil, O. and Robertson, T. (1991). ‘Towards a theory of Competitive Market
Signalling: A research agenda’, Strategic Management Journal, Vol. 12, No. 6, pp.
403-418.
Hess, P.J. (1981). ‘The Dividend Debate: 20 Years of Discussion, In "The Revolution
in Corporate Finance"’, 1992 (Blackwell Publishers, Cambridge, Massachusetts).
150
Higgins, R.C. (1972). ‘The corporate dividend-saving decision’, Journal of Financial
and Quantitative Analysis, Vol. 7, No. 2, pp. 1527-1541.
Ho, H. (2003). ‘Dividend Policies in Australia and Japan’, International Advances in
Economic Research, Vol. 9, pp. 91-100.
Holder, M.E., Langrehr, F.W. and Hexter, J.L. (1998). ‘Dividend Policy Determinants:
An Investigation of the Influences of Stakeholder Theory’, Financial Management,
Vol. 27, pp. 73-82.
Hong, Y. and Shehadeh, R.D. (1999). ‘A New Test for ARCH Effects and Its FiniteSample Performance’, Journal of Business and Economic Statistics, Vol. 17, No. 1,
pp. 91-108.
Hovakimian, A., Opler, T. and Titman, S. (2001). ‘The Debt-Equity Choice’, Journal
of Financial and Quantitative Analysis, Vol. 36, pp. 1-24.
Impson, C.M. and Karafiath, I. (1992), A Note on the Stock Market Reaction to
Dividend Announcements. The Financial Review, Vol. 27, pp. 259–271.
Impson, C. M. (1997), Market Reaction to Dividend Decrease Announcements: Public
Utilitie vs Unregulated Industrial Firms, The Journal of Financial Research, Vol 20,
pp. 407-422.
Irum, M., Rafique, M. and Hassan, A. (2012) Effect of Dividend Announcement on
Share Prices of Petroleum Industry of Pakistan, Journal of Basic and Applied
Scientific Research, Vol. 2, No. 7, pp. 6503-6511.
Jensen, G.R., Solberg, D.P. and Zorn, T.S. (1992). ‘Simultaneous Determination of
Insider Ownership, Debt, and Dividend Policies’, Journal of Financial and
Quantitative Analysis, Vol. 27, pp. 274-263.
Jensen, M.C. (1986). ‘Agency Costs of Free Cash Flow, Corporate Finance, and
Takeovers’, American Economic Review, Vol. 76, pp. 323-329.
Jensen, M.C. and Meckling, W.H. (1976). ‘Theory of the Firm: Managerial Behavior,
Agency Costs and Ownership Structure’, Journal of Financial Economics, Vol. 3, pp.
305-360.
Jesover F. and Kirkpatrick, G. (2005). ‘The Revised OECD Principles of Corporate
Governance and their Relevance to Non-OECD Countries’, Corporate Governance,
Vol. 13, No. 2, pp. 127-136.
John, K. and Senbet, L. (1998). ‘Corporate Governance and Board Effectiveness,
Journal of Banking and Finance, Vol. 22, pp. 371-403.
151
John, K. and Williams, J. (1985). ‘Dividends, Dillution, and Taxes: A Signalling
Equilibrim’, Journal of Finance, Vol. 40, pp. 1053-1070.
Jong, F.D. (2007). ‘Event Study Methodology’. Tilburgh University.
Juma'h, A.H. and Pacheco, C.J. (2008). ‘The Financial Factors Influencing Cash
Dividend Policy: A Sample of U.S. Manufacturing Companies’, International Metro
Business Journal , Vol. 4, No. 2, pp. 23-43.
Kalay, A. and Michaely, R. (2000). ‘Dividends and Taxes: A Re-Examination’,
Financial Management, Vol. 29, pp. 55-75.
Kaleem, A. and Salahuddin, C. (2006). ‘Impact of Dividend Announcement on
Common Stock Prices at Lahore Stock Exchange (Pakistan)’, South Asian Journal of
Management, Vol. 13, No.2.
Karim, M. (2010). ‘Announcement Effect of Dividend on the Stock Price of Enlisted
Companies in Developed Countries: A Comparative Study between London Stock
Exchange & New York Stock Exchange’, available on the internet at
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1624363.
Keim, D.B. (1985). ‘Dividend Yields and Stock Returns: Implications of Abnormal
January Returns’, Journal of Financial Economics, Vol. 14, pp. 473-489.
Kennedy, P. (2003). ‘A Guide to Econometrics’, Fifth Edition, MIT Press.
Khan, N.U. (2011). ‘Dividend policy and the stock market reaction to dividend
announcements in Pakistan’, PhD Thesis, Dundee University
Khan, T. (2006), ‘Company dividends and ownership structure’, Economic Journal
Vol.116, pp. 172- 189.
Kim, B.S. and Maddala, G.S. (1992). ‘Estimation and Specification Analysis of
Models of Dividend Behavior Based on Censored Panel Data’, Empirical Economics,
Vol. 17, pp. 111-124.
Kim, J. and Jang, S. (2010). ‘Dividend behavior of lodging firms: Heckman's two-step
approach’, International Journal of Hospitality and Management, Vol. 29, No. 3, pp.
413-420.
Kindleberger, C.P. (1984). ‘A Financial History of Western Europe', London: George
Allen & Unwin.
Kirmani, A. and Rao, A.R. (2000). ‘No Pain, No Gain: A Critical Review of the
Literature on Signalling Unobservable Product Quality’, Journal of Marketing, Vol.
64, No. 2, pp. 66- 79.
152
Koch, P.D. and Shenoy, C. (1999). ‘The Information Content of Dividend and Capital
Structure Policies’, Financial Management, Vol. 28, pp. 16-35.
Kumar, P.K. and Lee, B.S. (2001). ‘Discrete Dividend Policy with Permanent
Earnings’, Financial Management, Vol. 30, pp. 55-76.
La Porta, R., Lopez-De-Silanes, F., Shleifer, A. and Vishny, R.W. (2000). ‘Agency
Problems and Dividend Policies Around the World’, Journal of Finance, Vol. 55, pp.
1-33.
Lambrecht, B.M. and Myers, S.C. (2010). ‘A Lintner model of dividends and
managerial rents’, MIT Sloan School of Management working paper.
Lang, L.H.P., and Litzenberger, R.H. (1989). ‘Dividend Announcements: Cash Flow
Signalling vs. Free Cash Flow Hypothesis’, Journal of Financial Economics, Vol. 24,
pp. 181-191.
Lasfer, M. (1996). ‘Taxes and Dividends: The UK evidence’, Journal of Banking and
Finance, Vol. 20, pp. 455-472.
Leary, M.T. and Michaely, R. (2008). ‘Why Firms Smooth their Dividends: Empirical
Evidence’, Johnson School Research Paper Series No. 11-08, available on the internet
at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1274291.
Leary, M.T. and Michaely, R. (2011). ‘Determinants of Dividend Smoothing:
Empirical Evidence’, Review of Financial Studies, Vol. 24, No. 10, pp. 3197-3249.
Lease, R.C., John, K., Kalay, A., Loewenstein, U. and Sarig, O.H. (2000). ‘Dividend
Policy: Its Impact on Firm Value’ (Harvard Business School Press, Boston,
Massachusttes).
Leithner, S. and Zimmermann, H. (1993). ‘Market value and aggregate dividends: A
reappraisal of recent tests, and evidence from European markets’, Swiss Journal of
Economics and Statistics, Vol. 129, pp. 99-122.
Lewellen, W.G. and Badrinath, S.G. (1997). ‘On the measurement of tobin's Q’,
Journal of Financial Economics, Vol. 44, pp. 77-122.
Lindeberg, E.B. and Ross, S.A. (1981). ‘Tobin’s q Ratio and Industrial Organization’,
Journal of Business, Vol. 54, No. 1, pp. 1-32.
Lintner, J. (1956). ‘Distribution of Incomes of Corporations Among Dividends,
Retained Earnings, and Taxes’, American Economic Review, Vol. 46, pp. 97-113.
Lintner, J. (1962). ‘Dividends, Earnings, Leverage, Stock Prices and Supply of Capital
to Corporations’, The Review of Economics and Statistics, Vol. 64, pp. 243-269.
153
Litzenberger, R.H. and Ramaswamy, K. (1979). ‘The Effect of Personal Taxes and
Dividends on Capital Asset Prices’, Journal of Financial Economics, Vol. 7, pp. 163195.
Ljung, G.M. and Box, G.E.P. (1978). ‘On a measure of lack of fit in time series
models’, Biometrika, Vol. 65, pp. 297-303.
Lloyd, W.P., Jahera, J.S. and Page, D.E. (1985). ‘Agency Costs and Dividend Payout
Ratios’, Quarterly Journal of Business and Economics, Vol. 24, pp. 19-29.
Lonie, A. A., Abeyratna G., Power D. M. and Sinclair C. D.. (1996). ‘The stock
market reaction to dividend announcements: A UK study of complex market signals’.
Journal of Economic Studies, Vol. 23(1): pp. 32–52.
Loewenstein, U. and Kato, K. (1995). ‘The Ex-Dividend Day Behavior of Stock
Prices: the Case of Japan’, The Review of Financial Studies, Vol. 8 No. 3, 817-847.
Lyon, J., Barber, B. and Tsai, C. (1999). ‘Improved methods of tests of long-horizon
abnormal stock returns’, Journal of Finance, Vol. 54, pp. 165-201.
Luo Y. (2005). ‘Corporate Governance and Accountability in Multinational
Enterprises: Concepts and Agenda’, Journal of International Management, Vol. 11,
pp. 1-18.
MacKinlay, A. (1997). ‘Event studies in economics and finance’, Journal of Economic
Literature, Vol. 35, pp. 13-39.
Mancinelli, L. and Ozkan. A. (2006). ‘Ownership structure and dividend policy:
Evidence from Italian firms’, The European Journal of Finance, Vol. 12, No. 3, pp.
265-282.
Manos, R. (2002). ‘Dividend Policy and Agency Theory: Evidence on Indian Firms’,
Working Paper No. 41, (Institute for Development Policy and Management,
University of Manchester).
Marsh, P. (1982). ‘The choice between equity and debt: An empirical study’, The
Journal of Finance, Vol. 37, pp. 121-144.
Maury, C.B. and Pajuste, A. (2002). ‘Controlling Shareholders, Agency Problems, and
Dividend Policy in Finland’, The Finnish Journal of Business Economics
(Liiketaloudellinen Aikakauskirja), Vol. 51, pp. 15-45.
Maynes, E. and Rumsey, J. (1993). ‘Conducting event studies with thinly traded
stocks’, Journal of Banking and Finance, Vol. 17, pp. 145–157.
154
McCluskey, T., Burton, B. and Power, D. (2007). ‘Evidence on Irish
Financial
Directors‟ Views about Dividends, Quantitative Research in Accounting &
Management, Vol. 4, No. 2, pp. 115-132
McConnell J.J. and Servaes, H. (1990). ‘Additional evidence on equity ownership and
corporate value’, Journal of Financial Economics,Vol. 27, pp. 595-612.
McDonald, J.G., Jacquillat, B. and Nussenbaum, M. (1975). ‘Dividend, Investment
and Financing Decisions: Empirical Evidence on French Firms’, Journal of Financial
and Quantitative Analysis, Vol. 10, pp. 741-755.
Mehta, A. (2012). ‘An Empirical Analysis of Determinants of Dividend Policy –
Evidence from the UAE Companies’, Global Review of Accounting and Finance, Vol.
3, No. 1, pp. 18-31.
Michaely, R., Thaler, R.H. and Womack, K.L. (1995). ‘Price Reactions to Dividend
Initiations and Omissions: Overreaction or Drift?’, Journal of Finance, Vol. 50, pp.
573-608.
Miller, M.H. (1986). ‘Behavioral Rationality in Finance: The Case of Dividends’,
Journal of Business, Vol. 59, pp. 451-468.
Miller, M.H. and Modigliani, F. (1961). ‘Dividend Policy, Growth, and the Valuation
of Shares’, Journal of Business, Vol. 34, pp. 411-433.
Miller, M.H. and Rock, K. (1985). ‘Dividend Policy Under Asymmetric Information’,
Journal of Finance, Vol. 40, pp. 1031-1051.
Miller, M.H. and Scholes, M.S. (1978). ‘Dividends and Taxes’, Journal of Financial
Economics, Vol. 6, pp. 333-264.
Miller, M.H. and Scholes, M.S. (1982). ‘Dividend and Taxes: Some Empirical
Evidence’, Journal of Political Economy, Vol. 90, pp. 1118-1141.
Ministry of Industry and Trade (1997),’The Company Law’, The Hashemite Kingdom
of Jordan. No 22, the official Gazette No. 4202.
Modigliani, F. and Miller M.H. (1958). ‘The Cost of Capital, Corporation Finance and
the Theory of Investment: Reply’, American Economic Review, Vol. 49, No. 4, pp.
655-669.
Modigliani, F. and Miller M.H. (1958). ‘The Cost of Capital, Corporation Finance and
the Theory of Investment’, American Economic Review, Vol. 48, pp. 261-297.
Moh'd, M.A., Perry, L.G. and Rimbey, J.N. (1995). ‘An Investigation of Dynamic
Relationship Between Agency Theory and Dividend Policy’, The Financial Review,
Vol. 30, pp. 367-385.
155
Mollah, A. (2001). ‘Dividend Policy and Behaviour, and Security Price Reaction to
the Announcement of Dividends in an Emerging Market: A Study of Companies
Listed on the Dhaka Stock Exchange’, Business School, Leeds, UK, University of
Leeds.
Moore, M. (1992). ‘Signals and Choices in a Competitive Interaction: The Role of
Moves and Messages’, Management Science, Vol. 38, pp. 483-500.
Morck R., Shleifer A. and Vishny R.W. (1988). ‘Management Ownership and Market
Valuation’, Journal of Financial Economics, Vol. 20, pp. 293-315.
Morse, D. (1984). ‘An Econometric Analysis of the Choice of Daily Versus Monthly
Returns in Test of Information Content’, Journal of Accounting Research, Vol. 22,
No. 2, pp. 605 - 623.
Murali, R. and Welch, J.B. (1989). ‘Agents, owners, control and performance’,
Journal of Business Finance and Accounting, Vol. 16, No. 3, pp. 385-398.
Myers, S.C. (1984). ‘The Capital Structure Puzzle’, Journal of Finance, Vol. 39, pp.
575-592.
Myers, S.C. and Majluf N.S. (1984). ‘Corporate Financing and investment decisions
when firms have information that investors do not have’, Journal of Financial
Economics, Vol. 13, pp. 187- 221.
Nazir, M. S., Nawaz, M. M., Anwar, W., & Ahmed, F., (2010). “Determinants of
Stock Price Volatility in Karachi Stock Exchange: The Mediating Role of Corporate
Dividend Policy”, International Research Journal of Finance and Economics, Vol. 55.
Nelson, D.B. (1991). ‘Conditional heteroskedasticity in asset returns: A new
approach’, Econometrica, Vol. 59, pp. 347-370.
Nirmala, P.S., Sanju, P.S., Ramachandran, M. (2011) ‘Determinants of Share Prices in
India’, Journal of Emerging Trends in Economics and Management Sciences, Vol. 2:
pp. 124-130.
Nobanee, H., Haddad, A.E., AlShattarat, W.K. and AlShattarat, H.K. (2009). ‘An
Analysis of the Information Content of Dividend Change Payments at Amman Stock
Exchange’, Middle Eastern Finance and Economics, Vol. 5, pp. 62-78.
Omet, G. (2011). ‘The Global Financial Crisis and its impact on Operational
Efficiency: The Case of the Jordanian Stock Market’, 5th Meeting of the MENA-OECD
Working Group on SME Policy, Entrepreneurship and Human capital Development,
Casablanca, Marocco.
156
Omet, G. and Abu-Ruman, G. (2003). ‘Dividend Policy in the Jordanian Capital
Market: Empirical Findings and Survey Results’, Second International Finance
Conference, March 13-15 (Hammamet, Tunisia).
Ofer, A.R. and Thakor, A.V. (1987). ‘A Theory of Stock Price Responses to
Alternative Corporate Cash Disbursement Methods: Stock Repurchases and
Dividends’, Journal of Finance, Vol. 42, pp. 365-394.
Omet, G. and Khasawneh, J. (2002). ‘The Stability of Dividend Policy and the
Importance of Dividends and Earnings in Share Valuation: A Study of Jordanian NonFinancial Companies’, Abhath Al-Yarmouk, Vol. 18, pp. 153-165.
Omet, G. and Mashharawe, F. (2003). ‘The Capital Structure Choice in Tax
Contrasting Environments: Evidence from the Jordanian, Kuwaiti, Omani and Saudi
Corporate Sectors’, The Economic Research Forum 10
th
Annual Conference,
December (Marrakesh, Morocco).
Omran, M. and Pointon, J. (2004). ‘Dividend Policy, Trading Characteristics and
Share Prices: Empirical Evidence from Egyptian Firms’, International journal of
Theoretical and Applied Finance, Vol. 7, pp. 121-133.
Ooi, J.T.L. (2001). ‘Dividend Payout Characteristics of U.K. Property Companies’,
Journal of Real Estate Portfolio Management, Vol. 7, No. 2, pp. 133-142.
Oskooe, S.A.P. and Shamsavari, A. (2011). ‘Asymmetric Effects in Emerging Stock
Markets- The Case of Iran Stock Market’, International Journal of Economics and
Finance, Vol. 3, No. 6, pp. 16-24.
Pandey, I.M. (2001). ‘Corporate Dividend Policy and Behaviour: The Malaysian
Experience’, Working Paper No. 2001-11-01, (Indian Institute of Management
Ahmedabad).
Pandey, I.M. and Bhat, R. (2004). ‘Dividend behaviour of Indian companies under
monetary policy restrictions’, Working Paper, Indian Institute of Management.
Pettit, R.R. (1972). ‘Dividend Announcements, Security Performance, and Capital
Market Efficiency’, Journal of Finance, Vol. 27, pp. 993-1007.
Poterba, J.M. and Summers, L.H. (1984). ‘New Evidence That Taxes Affect the
Valuation of Dividends’, Journal of Finance, Vol. 39, pp. 1397-1415.
Qian Y.M., Tian Y. and Wirjanto T.S. (2007). ‘An Empirical Investigation into the
Capital-Structure Determinants of Publicly Listed Companies: A Static Analysis’,
University of Waterloo.
157
Rajan, R.G., and Zingales, L. (1995). ‘What Do We Know About Capital Structure?
Some Evidence From International Data’, Journal of Finance, Vol. 50, pp. 1421-1460.
Ramachandran, A. and Packkirisamy, V. (2010). ‘The Impact of Firm Size on
Dividend Behaviour: A Study With Reference to Corporate Firms across Industries in
India’, Managing Global Transitions, Vol. 8, No. 1, pp. 49-78.
Redding, L.S. (1997). ‘Firm Size and Dividend Payouts’, Journal of Financial
Intermediation, Vol. 6, pp. 224-248.
Reddy, S.Y. (2002). ‘Dividend Policy of Indian Corporate Firms: An Analysis of
Trends and Determinants’ NSE Research Initiative, Serial No.19,
Reinganum, M. (1981). ‘A Misspecification of Capital Asset Pricing: Empirical
Anomalies Based on Earnings Yields and Market Values’, Journal of Financial
Economics, Vol. 9, pp. 19-46.
Riley, J. (2001). ‘Silver Signals: Twenty-Five Years of Screening and Signalling’,
Journal of Economic Literature, Vol. 39, pp. 432-78
Rozeff, M.S. (1982). ‘Growth, Beta and Agency Costs as Determinants of Dividend
payout Ratios’, The Journal of Financial Research, Vol. 5, pp. 249-259.
Saxena, A.K. (1999). ‘Determinants of Dividend Payout Policy: Regulated Versus
Unregulated Firms’, Working Paper, (State University of West Georgia).
Schooley, D.K., and Barney, L.D.Jr. (1994). ‘Using Dividend Policy and Managerial
Ownership to Reduce Agency Costs’, Journal of Financial Research, Vol. 17, pp.
363-373.
Seiler, M.J. (2000). ‘The Efficacy of Event-Study Methodologies: Measuring EREIT
Abnormal Performance Under Conditions of Induced Variance’, Journal of Financial
and Strategic Decisions, Vol. 13, No. 1, pp. 101-112.
Sembenelli, A. (1993). ‘Signalling, Financial Hierarchy and Agency Theory as
Explanations for Dividend Behaviour: Evidence from Italian Firm Data’, Managerial
and Decision Economics, Vol. 14, pp. 37-45.
Shefrin, H. and Statman, M. (1985). ‘The disposition to sell winners too early and ride
losers too long: theory and evidence’, Journal of Finance, Vol. 40, pp. 77-90.
Shefrin, H. and Statman, M. (1997). ‘Comparing expectations about stock returns to
realised returns’, Working paper, Leavey school of Business, Santa Clara University.
Shefrin, H. and Statman, M. (1997). ‘Behavioral Portfolio Theory’, Working paper,
Leavey school of Business, Santa Clara University.
158
Shleifer, A., and Vishny, R.W. (1997). ‘A Survey of Corporate Governance’, Journal
of Finance, Vol. 52, pp. 737-783.
Shleifer, A., and Vishny, R.W. (1986). ‘Large Shareholders and Corporate Control’,
Journal of Political Economy, Vol. 94, pp. 464-488.
Short, H., Zhang, H. and Keasey, K. (2002). ‘The Link Between Dividend policy and
Institutional Ownership’, Journal of Corporate Finance, Vol. 8, pp. 105-122.
Smith, C. and Wessels, R. (1992). ‘The Investment Opportunity Set and Corporate
Financing, Dividend and Compensation Policies’, Journal of Financial Economics,
Vol. 32, pp. 263-292.
Stacescu, B. (2006). ‘Dividend Policy in Switzerland’, Financial Markets Portfolio
Manangement, Vol. 20, pp. 153-183.
Strong, N. (1992). ‘Modelling abnormal returns: A review article’, Journal of Business
Finance and Accounting, Vol. 19, No. 4, pp. 533-553.
Takeda, F. and Tomozawa, T. (2008). ‘A change in market responses to the
environmental management ranking in Japan’, Ecological Economics, Vol. 67, pp.
465-472.
Titman, S. and Wessels, R. (1988). ‘The Determinants of Capital Structure Choice’,
Journal of Finance, Vol. 43, No. 1, pp. 1-19.
Tobin, J. (1969). ‘A General Equilibrium Approach to Monetary Theory’, Journal of
Money, Credit, and Banking, Vol. 1, No. 1, pp. 15—29.
Travlos, N., Trigeorgis, L. and Vafeas, N. (2001). ‘Shareholder wealth effects of
dividend policy changes in an emerging stock market: The case of Cyprus’,
Multinational Finance Journal, Vol. 5, No. 2, pp. 87–112.
Trojanowski, G. (2004). ‘Ownership Structure and Payout Policy in the UK’,
European Financial Management Association Annual Meeting, June 30-July 3 (Basel,
Switzerland).
Uddin, M.H. (2003). ‘Effect of Dividend Announcement on Shareholders‟ Value:
Evidence from Dhaka Stock Exchange’, Global Corporate Governance Forum
(Sponsored by World Bank), 13-16 December, Hyderabad, (INDIA).
Utami, S.R. and Inanga, E.L. (2011). ‘Agency Costs of Free Cash Flow, Dividend
Policy, and Leverage of Firms in Indonesia’, European Journal of Economics,
Finance and Administrative Sciences, Vol. 33, pp. 7-24.
159
Vieira, E.S. and Raposo, C.C. (2007). ‘Signalling with Dividends? New Evidence
from Europe’, available on the internet at
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1004523.
Walter, J.E. (1963). ‘Dividend Policy: Its Influence on the Value of the Enterprise’,
Journal of Finance, Vol. 18, pp. 280-291.
Wang, K., Erickson, J. and Gau, G.W. (1993). ‘Dividend Policies and Dividend
Announcement Effects for Real Estate Investment Trusts’, Journal of the American
Real Estate and Urban Economics Association, Vol. 21, No. 2, pp. 185-201.
Wang, Z., Salin, V., Hooker, N.H. and Leatham, D. (2002). ‘Stock market reaction to
food recalls: A GARCH application’, Applied Economics Letters, Vol. 9, pp. 979-987.
Watts, R. (1973) ‘The Information Content of Dividends’, Journal of Business, Vol.
46, pp. 191-211.
Wie, G., Zhang, W. and Xiao, J.Z. (2003). ‘Dividend Policy and Ownership Structure
in China’, SRNN Working Paper Series No. 463924, [online], (www.ssrn.com).
Woolridge, J.R. (1983). ‘Dividend Changes and Security Prices’, Journal of Finance,
Vol. 38, pp. 1607-1615.
Wooldridge, J.M. (2001). ‘Econometric Analysis of Cross Section and Panel Data’,
Cambridge, Mass.: MIT Press.
Wooldridge, J.M. (2006). ‘Cluster-Sample Methods in Applied Econometrics: An
extended analysis, Mimeo.
Wu, G. (2001), ‘The determinants of asymmetric volatility’. Review of Financial
Studies, Vol. 14, pp. 837-859.
Yamaguchi, K. (2007). ‘Reexamination of stock price reaction to environmental
performance: A GARCH application’, Ecological Economics, Vol. 56, pp. 23-35.
Yoon, P.S. and Starks, L.T. (1995). ‘Signalling, Investment Opportunities, and
Dividend Announcements’, Review of Financial Studies, Vol. 8, pp. 995-1018.
Zhang, R., Simons, K.L. and Stern, D.I. (2010). ‘News Media as a Channel of
Environmental Information Disclosure: Evidence from an EGARCH Approach’,
Rensselaer Working Papers in Economics 1001.
Zuguang, H. and Ahmed, M.U. (2010). ‘Dividend Announcement Effect on Stock
Return: An Event Study on Shanghai Stock Exchange’, WRI Global Congress on
Intelligent Systems, GCIS.
160