-
Journal of Economic & Administrative Sciences Vol. 23, No.
2, December 2007 (44-70)
44
Determinants of Corporate Dividend Policy in Jordan:
An Application of the Tobit Model
Husam-Aldin Nizar Al-Malkawi
Department of Finance and Banking Faculty of Administrative and
Financial Sciences
Al-Ahliyya Amman University, Jordan Abstract This paper examines
the determinants of corporate dividend policy in Jordan. The study
uses a firm-level panel data set of all publicly traded firms on
the Amman Stock Exchange between 1989 and 2000. The study develops
eight research hypotheses, which are used to represent the main
theories of corporate dividends. A general-to-specific modeling
approach is used to choose between the competing hypotheses. The
study examines the determinants of the amount of dividends using
Tobit specifications. The results suggest that the proportion of
stocks held by insiders and state ownership significantly affect
the amount of dividends paid. Size, age, and profitability of the
firm seem to be determinant factors of corporate dividend policy in
Jordan. The findings provide strong support for the agency costs
hypothesis and are broadly consistent with the pecking order
hypothesis. The results provide no support for the signaling
hypothesis. 1. Introduction The topic of dividend policy is one of
the most enduring issues in modern corporate finance. This has led
to the emergence of a number of competing theoretical explanations
for dividend policy. No consensus has emerged about the rival
theoretical approaches to dividend policy despite several decades
of research. A range of firm and market characteristics have been
proposed as potentially important in determining dividend policy.
The attempt to test these competing models and refine them has in
turn spawned a vast empirical literature. The empirical work on
dividend policy has,
Tel. (+962) 799 666 527. E-mail address:
[email protected]. I am thankful and indebted to Michael
Rafferty, Stephane Mahuteau, and Roger Ham of the University of
Western Sydney who supervised the dissertation from which much of
this paper is derived.
-
Journal of Economic & Administrative Sciences December
2007
45
however, generally been focused on developed stock markets such
as the UK and US. The examination of dividend policy in emerging
stock markets has, until recently, been much more limited. Yet the
sorts of firm and market characteristics that may influence
dividend policy may in fact be more likely to be present in
developing markets in an exaggerated fashion than in developed
markets. This provided a central motivation for the present study.
This study seeks to add to that literature by providing a detailed
analysis of dividend policy in Jordan, an emerging market that has
been particularly poorly analyzed to date. By and large, emerging
stock markets have several similar characteristics so, to some
extent, corporate dividend behavior in Jordan may share some
important similarities with other emerging equity markets.
Consequently, the findings of such a detailed country case study
could form the basis of future comparative research into other
emerging markets. Such findings may also provide the basis for
reflection on empirical research in developed markets. The paper
uses a firm-level panel data set of all publicly traded firms on
the Amman Stock Exchange (ASE) between 1989 and 2000. The study
develops eight research hypotheses, which are used to represent the
main theories of corporate dividends. A general-to-specific
modeling approach is used to choose between the competing
hypotheses. The study examines the determinants of the amount of
dividends using Tobit specifications. The factors that affect
dividend policy in developed stock markets seem to apply for this
emerging market, but often in different ways and on a different
scale. The results suggest that the proportion of stocks held by
insiders and state ownership significantly affect the amount of
dividends paid. Size, age, and profitability of the firm seem to be
determinant factors of corporate dividend policy in Jordan. These
factors are found to positively affect the level of dividends. The
findings provide strong support for the agency costs hypothesis and
are broadly consistent with the pecking order hypothesis. The
results provide no support for the signaling hypothesis. The rest
of the paper is organized as follows. Section 2 gives a short
discussion of dividend policy theories. Section 3 formulates the
hypotheses. Section 4 describes the data. Section 5 explains the
methodology. Section 6 reports the results. The final section
summarizes and concludes the paper.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
46
2. Theoretical Consideration and Prior Research1 Financial and
business historians have shown that dividend policy has been bound
up with the historical development of the corporation. In its
modern form, however, dividend policy theory is closely tied to the
work of Miller and Modigliani (1961, hereafter M&M) and their
dividend policy irrelevance thesis. M&M demonstrate that under
certain assumptions including rational investors and a perfect
capital market, the market value of a firm is independent of its
dividend policy. In actual market practices however, it has been
found that dividend policy does seem to matter, and relaxing one or
more of M&Ms perfect capital market assumptions has often
formed the basis for the emergence of rival theories of dividend
policy. The bird-in-hand theory (a pre-Miller-Modigliani theory)
asserts that in a world of uncertainty and information asymmetry
dividends are valued differently to retained earnings (capital
gains). Because of uncertainty of future cash flow, investors will
often tend to prefer dividends to retained earnings. As a result, a
higher payout ratio will reduce the required rate of return (cost
of capital), and hence increase the value of the firm (see, for
example Gordon, 1959). This argument has been widely criticized and
has not received strong empirical support. The tax-preference
theory posits that low dividend payout ratios lower the required
rate of return and increase the market valuation of a firms stocks.
Because of the relative tax disadvantage of dividends compared to
capital gains investors require a higher before-tax risk adjusted
return on stocks with higher dividend yields (Brennan, 1970).
Several studies including Litzenberger and Ramaswamy (1979),
Poterba and Summers, (1984), and Barclay (1987) have presented
empirical evidence in support of the tax effect argument. Others,
including Black and Scholes (1974), Miller and Scholes (1982), and
Morgan and Thomas (1998) have opposed such findings or provided
different explanations. Another closely related theory is the
clientele effects hypothesis. According to this argument, investors
may be attracted to the types of stocks that match their
consumption/savings preferences. That is, if dividend income is
taxed at a higher rate than capital gains, investors (or
clienteles) in high tax brackets may prefer non-dividend or
low-dividend paying stocks, and vice versa. Also, the presence of
transaction costs may create certain clienteles. There are numerous
empirical studies on the clientele effects hypothesis but the
findings are mixed. Taking different paths, Pettit (1977), Scholz
(1992), and Dhaliwal, Erickson and Trezevant (1999) presented
1 See Lease et al. (2000) for a comprehensive review of the
literature.
-
Journal of Economic & Administrative Sciences December
2007
47
evidence consistent with the existence of clientele effects
hypothesis. Studies finding weak or contrary evidence include
Lewellen et al. (1978), Richardson, Sefcik and Thomason (1986),
Abrutyn and Turner (1990), among others. Despite the tax penalty on
dividends relative to capital gains, firms may pay dividends to
signal their future prospects. This explanation is known as the
information content of dividends or signaling hypothesis. The
intuition underlying this argument is based on the information
asymmetry between managers (insiders) and outside investors, where
managers have private information about the current performance and
future fortunes of the firm that is not available to outsiders.
Here, managers are thought to have the incentive to communicate
this information to the market. According to signaling models
(Bhattacharya, 1979, John and Williams, 1985, and Miller and Rock,
1985) dividends contain this private information and therefore can
be used as a signaling device to influence share price. An
announcement of dividend increase is taken as good news and
accordingly the share price reacts favorably, and vice versa. Only
good-quality firms can send signals to the market through dividends
and poor-quality firms cannot mimic these because of the
dissipative signaling costs (for example, transaction costs of
external financing, or tax penalties on dividends, or distortion of
investment decisions). Support for the signaling hypothesis can be
found, for example, in Pettit (1972), Michaely, Thaler and Womack
(1995), Nissim and Ziv (2001), and Bali (2003). Other researchers,
however, found limited support or rejected the hypothesis (see
Watts, 1973, Gonedes, 1978, and Conroy, Eades and Harris, 2000,
among others). Note that the aforesaid studies followed different
approaches. The information asymmetry between managers and
shareholders, along with the separation of ownership and control,
formed the base for another explanation for why dividend policy may
matter; that is, the agency costs thesis. This argument is based on
the assumption that managers may conduct actions in accordance with
their own self-interest which may not always be beneficial for
shareholders. For example, they may spend lavishly on perquisites
or overinvest to enlarge the size of their firms beyond the optimal
size since executives compensation is often related to firm size
(see Jensen, 1986). The agency costs thesis predicts that dividend
payments can reduce the problems associated with information
asymmetry. Dividends may also serve as a mechanism to reduce cash
flow under management control, and thus help to mitigate the agency
problems. Reducing funds under management discretion may result in
forcing them into the capital markets more frequently, thus putting
them under the scrutiny of capital suppliers
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
48
(Rozeff, 1982, and Easterbrook, 1984). Many researchers have
offered empirical support for agency explanations for why firms pay
dividends including Rozeff (1982), Lloyd, Jahera and Page (1985),
Jensen, Solberg and Zorn (1992), and Holder, Langrehr and Hexter
(1998), among others. Others including Denis, Denis and Sarin
(1994), Yoon and Starks (1995), and Lie (2000) provided little
support or reject the hypothesis. Theories discussed above
presented differing explanations for the determinants of corporate
dividend policy, and provide a partial solution to the dividend
puzzle (the debate between these explanations remains unresolved).
3. Research Hypotheses and Selection of Proxy Variables Section 2
showed the main theoretical arguments for dividend policy along
with the empirical evidence. This in turn is now used as a guide to
formulate research hypotheses given the key characteristics of
Jordanian capital market. This section also presents the selection
of proxy variables that are often suggested in the literature and
their expected relationship to dividend policy as measured by
dividend yield. Hypothesis 1: Dividends serve as a bonding
mechanism to reduce agency problems The agency hypothesis of
dividends predicts that dividend payments can be used as a
mechanism to alleviate agency problems. To test this hypothesis two
proxy variables are used. The first proxy for agency costs is the
natural logarithm of number of shareholders (STOCK), which is used
to measure ownership dispersion (see, for instance, Rozeff, 1982,
and Deshmukh, 2003). The hypothesized relation between dividend
payouts and STOCK is expected to be positive. That is, in order to
alleviate the agency costs associated with an increasing number of
shareholders firms should pay more dividends, other things being
equal. The second proxy for agency costs is the percentage of a
firms common stock held by insiders (INSD) (see, for example,
Rozeff, 1982, Jensen et al., 1992, and Holder et al., 1998). It has
been argued that agency costs may be reduced if insiders (managers,
directors, and other executive officers) increase their ownership
in the firm, because this can help to align the interests of both
managers and shareholders (Jensen and Meckling, 1976). Therefore,
the higher the proportion of managers in firm ownership, the less
is the need for using dividends as a device to mitigate agency
costs. Hence, INSD variable is expected to bear a negative relation
to dividend payouts.
-
Journal of Economic & Administrative Sciences December
2007
49
Hypothesis 2: Ownership and control structures affect corporate
dividend policy Although the relation between dividend payouts and
a firms ownership structure can be examined within the agency
theory framework, in the case of Jordan it is important to address
the relationship separately because of the variety of owners of
Jordanian firms (for instance, families, institutions, government,
and foreign investors). To test the link between dividend policy
and ownership structure, a set of four dummy variables were
included to describe the ownership structure of the firm: family
(FAML), state (STATE), institution (INST), and multiple (MULT). To
identify the ultimate owner of the firm we employed the 10-percent
threshold level of ownership, which is the criterion used by the
ASE throughout the study period. We propose that dividend payouts
are negatively related to FAML and positively to STATE, INST and
MULT, ceteris paribus. Hypothesis 3: Dividends as a signaling
device The information content of dividends (signaling) hypothesis
predicts that dividends can be used to signal firms future
prospects and only good-quality firms can use such a device. The
hypothesis can be examined by identifying the relationship between
information asymmetry and dividend payouts. A potential proxy for
the degree of information asymmetry is the trading volume of a
firms shares. In general, investors tend to invest in securities
that are better known in the market, that is, with less information
asymmetries. Therefore, other things being equal, the higher the
information asymmetry of a security the lower its trading volume.
This analysis uses annual share turnover (TURN) as a proxy measure
for information asymmetry (see, for example, Bartov and Bodnar,
1996, and Leuz and Verrecchia, 2000). Using this proxy for
information asymmetry, the signaling hypothesis predicts an inverse
relationship between share turnover and dividend payouts.
Hypothesis 4: Firm growth and investment opportunities are
negatively associated with dividend payouts Firms with high growth
and investment opportunities will need the internally generated
funds to finance those investments, and thus tend to pay little or
no dividends. This prediction is consistent with the pecking order
hypothesis 2 proposed by Myers and Majluf (1984). Accordingly, we
expect
2 The pecking order hypothesis suggests that firms finance
investments first with the internal finance, and if external
financing is necessary, firms prefer to issue debt
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
50
the firms growth and investment opportunities, as measured by
market-to-book ratio (MBR), to be negatively related to dividends
payouts (see, for instance, Deshmukh, 2003, and Aivazian et al.,
2003). Moreover, mature companies are likely to be in their
low-growth phase with less investment opportunities (see Barclay et
al., 1995, and Grullon et al., 2002). These companies do not have
the incentives to build-up reserves as a result of low growth and
few capital expenditures, which enable them to follow a liberal
dividend policy. On the contrary, new or young companies need to
build-up reserves to face their rapid growth and financing
requirements. Hence, they retain most of their earnings and pay low
or no dividends. Therefore, the age of the firm (AGE) is used as a
second proxy for the firms growth opportunities. Several studies
have related firm growth to age (Farinas and Moreno, 2000, and
Huergo and Jaumandreu, 2004, among others). Other things held
constant, as a firm gets older its investment opportunities decline
leading to lower growth rates, consequently reducing the firms
funds requirements for capital expenditures. Therefore, dividend
payout should be positively related to the firms age. Yet, we do
not expect the impact of age to always be linear. Thus, we allow
the effect of age to be non-linear by including the age squared
(AGESQ). If the coefficient on AGESQ appears to be negative, then
our assumption of a quadratic relationship between age and
dividends is true. Hypothesis 5: The firm size is positively
associated with dividend payouts A large firm typically has better
access to capital markets and finds it easier to raise funds with
lower cost and fewer constraints compared to a small firm. This
suggests that the dependence on internal funding decreases as firm
size increases. Therefore, ceteris paribus, large firms are more
likely to afford paying higher dividends to shareholders. In this
study the firms market capitalization of common equity (MCAP) is
used as a measure for size (see, for example, Deshmukh, 2003).
Based on the above discussion and consistent with previous research
the size variable is expected to have a positive relationship with
dividend payouts. Hypothesis 6: The firm debt is negatively
associated with dividend payouts
before issuing equity to reduce the costs of information
asymmetry and other transactions costs. (Myers 1984, and Myers and
Majluf, 1984).
-
Journal of Economic & Administrative Sciences December
2007
51
When a firm acquires debt financing it commits itself to fixed
financial charges embodied in interest payments and the principal
amount, and failure to meet these obligations may lead the firm
into liquidation. The risk associated with high degrees of
financial leverage may therefore result in low dividend payments
because, ceteris paribus, firms need to maintain their internal
cash flow to pay their obligations rather than distributing the
cash to shareholders. Moreover, Rozeff (1982) points out that firms
with high financial leverage tend to have low payouts ratios to
reduce the transaction costs associated with external financing.
Therefore, other things being equal, an inverse relationship
between financial leverage ratio, defined as the ratio of total
short-term and long-term debt to total shareholders equity (DER),
and dividends is expected. Hypothesis 7: There is a positive
relationship between a firms profitability and dividend payouts The
decision to pay dividends starts with profits. Therefore, it is
logical to consider profitability as a threshold factor, and the
level of profitability as one of the most important factors that
may influence firms dividend decisions. In his classic study,
Lintner (1956) found that a firms net earnings are the critical
determinant of dividend changes. The pecking order hypothesis may
provide an explanation for the relationship between profitability
and dividends. That is, taking into account the costs of issuing
debt and equity financing, less profitable firms will not find it
optimal to pay dividends, ceteris paribus. On the other hand,
highly profitable firms are more able to pay dividends and to
generate internal funds (retained earnings) to finance investments.
Fama and French (2002) used the expected profitability of assets in
place for testing the pecking order hypothesis. In another study,
Fama and French (2001) interpreted their results of the positive
relationship between profitability and dividends as consistent with
the pecking order hypothesis. Based on the above discussion,
profitability is expected to be a key determinant of corporate
dividend policy in Jordan. To test this hypothesis, the after tax
earnings per share (EPS) is used as a measure of a firms
profitability. The hypothesized relationship between EPS and
dividends is positive. Hypothesis 8: The relative tax disadvantage
of dividends induces lower dividend payouts The tax-preference
theory proposed that companies should retain rather than distribute
their income because of the preferential tax treatment of
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
52
capital gains versus dividends. Prior to 1996, investors paid no
taxes on income received from Jordanian companies in both the form
of capital gains and dividends. However, in 1996 the Jordanian
Income Tax Authority imposed a 10 percent tax rate on dividends3,
resulting in making dividend payments costly for investors. Since
this study covers the period between 1989 and 2000, which includes
five years following the implementation of taxes on dividends, it
is important to examine if the resulting relative tax disadvantage
of dividends induces Jordanian firms to reduce their payout ratios.
In order to examine any change in dividend payouts following the
change in tax law we introduced a dummy variable (DTAX), which
divides the study into two periods, post-tax (1996-2000) and
pre-tax (1989-1995). Based on the tax-preference argument, the
association between DTAX and dividends is anticipated to be
negative. 4. The Data The data employed in this study is derived
from the annual publications of the ASE. Based on a 12-year period
(1989-2000) and 160 companies listed on the ASE, a panel dataset
was constructed4. This dataset consist of all companies listed on
the ASE covering four sectors: industrial, service, insurance, and
banks. These companies ranged from old to newly established ones,
and some companies were de-listed during the study period.
Therefore, the number of observations for each company is
different. In order to gain the maximum possible observations,
pooled cross-section and time-series data is used. Because the
number of observations for each company is not identical, this
results in an unbalanced panel. The analysis is based on annual
data, because the data set is annual, and the ultimate sample
consists of 759 firm-year observations. The present study includes
both dividend-paying as well as non-dividend-paying firms. The
exclusion of non-dividend-paying firms results in a well-known
selection bias problem (see for instance, Kim and Maddala, 1992,
and Deshmukh, 2003). 5. Methodology 5.1 Choosing between
Hypotheses: The General to Specific Method In Section 3 eight
hypotheses were developed and a set of related proxy variables were
identified. The selected variables constitute the general model to
be tested in order to determine the factors that may affect
dividend policy
3 This law was again amended in 2001 and implemented as of
January 2002 removing the 10 percent tax on dividends. 4 The list
of companies used in the analysis is available upon request.
-
Journal of Economic & Administrative Sciences December
2007
53
in the case of Jordan. To choose between the competing
hypotheses and to arrive at the best model that fits the data the
general-to-specific method is used. The general-to-specific method
is generally referred to as the London School of Economics approach
to econometric modeling (see Hendry, 1995). Based on this approach,
the analysis starts with a general unrestricted model, which
includes all the variables that were identified and supported by
theories of dividend policy. Next, nested procedures are followed
to arrive at the best-fitted model. Further, in testing the
competing models the likelihood-ratio (LR) test is carried out. The
statistic LR = -2[Log(LR)-Log(LUR)] follows a 2 ( )k , where k is
the number of restrictions and the null model is the restricted
model. This test enables us to see whether the additional
parameters in the unrestricted model significantly increase the
likelihood. In other words, the test is used to confirm whether or
not the unrestricted model is statistically different from the
restricted model. The general model to be estimated using the Tobit
specifications, for firm i in period t [mathematical signs indicate
the hypothesized impact on dividend policy as measured by dividend
yield (DYLD)] can be written as: DYLD = 0 + 1 STOCK - 2 INSD - 3
FAML + 4 STATE + 5 INST + 6 MULT + 7 AGE - 8 AGESQ 9 MBR
- 10 DER - 11 TURN - 12 DTAX + 13 MCAP + 14 EPS 15 NONFIN + ,
(1)
where the variables are defined in Table 1 below. The table also
provides a summary of the research hypotheses and identifies the
dependent variable used in the regressions.
Table 1 Summary of Research Hypotheses and Proxy Variables
H1: Agency costs
STOCK: natural log of number of common stockholders INSD:
percentage held by insiders
Positive Negative
H2: Ownership structure*
FAML: family dummy = 1 if firm is family owned, and 0 otherwise.
STATE: state dummy = 1 if firm is owned by the government or its
agencies, and 0 otherwise. INST: institution dummy = 1 if firm is
owned by an institution, and 0 otherwise. MULT: multiple owners
dummy = 1 if firm has more than one type of owners who control at
least 10% of the stock, and 0 otherwise.
Negative Positive Positive Positive
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
54
H3: Signaling TURN: share turnover (proxy for information
asymmetry)
Negative
H4: Investment opportunities
MBR: market-to-book ratio AGE: age of the firm AGESQ: the square
of AGE
Negative Positive Negative
H5: Size MCAP: natural log of market capitalization Positive H6:
Financial leverage
DER: debt-to-equity ratio Negative
H7: Profitability EPS: the after-tax earnings per share Positive
H8: Taxes DTAX: tax dummy = 1 for the years 1996-2000,
and 0 otherwise.Negative
Control variable NONFIN: is a dummy variable to control for
industry effects = 1 if a firm belongs to non-financial sector, and
zero otherwise
Negative/ Positive
Dependent variable
DYLD: dividend yield (dividend-to-price ratio)
* The 10% threshold level of ownership is used to identify the
ultimate owner of the firm. 5.2. Tobit Estimation In considering
the dividend decision firms have only two options, either to pay or
to not pay dividends. In Jordan, many companies do not pay
dividends at all, and even those who pay dividends do not pay them
continuously5. This gives the dependent variable (dividends) a
special feature in that it takes two outcomes. It is either equal
to zero or positive. Dividends can never be negative. Therefore,
OLS is not an appropriate method to analyze the payment of
dividends, because of the nature of the dependent variable. Indeed
there is what one calls a mass point in 0 because the dividends
paid by firms can only be positive or nil. The appropriate
technique in this case is to apply Tobit estimation. Kim and
Maddala (1992) explicitly supported this claim (see also Anderson,
1986, and Huang, 2001). The evaluation of the determinants of the
amount of dividends is carried out using the general specification
of the censored data estimation, namely the Tobit model. Indeed,
the observed dependent variable, the amount of dividend paid by
each firm, may either be zero or positive. The data are then
censored in the lower tail of the distribution.
5 Of the 1511 cash dividend observations in our sample, 853
observations (56.45 percent) are zero dividends.
-
Journal of Economic & Administrative Sciences December
2007
55
The estimation involves the following general structure: the
latent underlying regression for the amount of dividend paid by the
firm is defined as (see Verbeek, 2000):
iti'it*it xy ++= , (2) with the observed dependent variable
being such that: yit = 0, if 0y*it , = y*it , if 0y
*it > .
The random effects i and the error term it are assumed to be 2
(0, )iid N and 2 (0, )iid N , respectively, and independent of
xi1,
xiT. To estimate the maximum likelihood estimation (MLE) method
is used6. The software package STATA (version 8) is used to perform
the estimation. The main emphasis in the analysis of the results
from MLE will be on the significance of each estimated coefficient
as well as on the overall significance of the model as judged by
the Chi-square ( 2 ) statistics derived from the Wald test
statistic. The Wald test statistic follows a 2 distribution with
degrees of freedom equal to the number of coefficient restrictions.
6. Results Table 2 presents summary statistics of all variables
used in the analysis. The table reports the mean, standard
deviation, minimum, maximum, coefficient of variation (CV)7, and
the number of observations for each of the dependent and
independent variables. The CV indicates that there is a significant
variation among the explanatory variables. In order to ensure that
our results are robust, several diagnostic tests were performed. In
attempting to detect multicollinearity, we computed the variance
inflation factors (VIF) for the independent variables. The
estimated VIF values were small (much less than 10, the rule of
thump) with an average of 1.64 indicating an absence of
multicollinearity between the variables. The correlation matrix
also confirmed the absence of multicollinearity among the
explanatory variables used in the regressions (see Appendix 1).
6 See StataCorp (2003) and Verbeek (2000) for the likelihood
function. 7 The coefficient of variation CV is defined as the
standard deviation over the mean.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
56
Table 2 Descriptive Statistics for the Dependent and Independent
Variables
Variable Mean Std. Dev. Min Max CV Obs. DYLD 0.027 0.035 0 0.400
1.282 1316 STOCK 3.068 0.746 1.114 5.407 0.243 885 INSD 0.289 0.246
0 0.945 0.851 936 FAML 0.303 0.460 0 1 1.517 1181 STATE 0.210 0.408
0 1 1.941 1181 INST 0.539 0.499 0 1 0.926 1181 MULT 0.368 0.483 0 1
1.311 1182 AGE 15.738 13.533 0 70 0.860 1598 AGESQ 430.717 703.841
0 4900 1.6341 1598 MBR 1.290 1.025 -0.020 12.670 0.795 1382 DER
2.143 6.939 -177.242 69.270 3.238 1511 TURN 0.299 0.671 0 7.993
2.243 1491 DTAX 0.500 0.500 0 1 1.000 1592 MCAP 15.731 1.403 11.156
21.356 0.089 1320 EPS 0.282 1.774 -8.388 28.299 6.285 1514 NONFIN
0.738 0.440 0 1 0.597 1920 Note: Variables are defined in Table
1.
The likelihood ratio (LR) test reported at the bottom of table
of the results (presented below) provides a formal test for the
pooled (Tobit) estimator against the random effects panel
estimator. For all estimated regressions, the results of the LR
test indicate that the panel-level variance component is important
and, therefore, the pooled estimation is different from the panel
estimation. The reported coefficient of Rho ( ), which is the
panel-level variance component, provides a similar test. In
attempting to test for heteroskedasticity, a two-stage test
procedure is employed. An important test for heteroskedasticity is
the Lagrange Multiplier test devised independently by Breusch and
Pagan (1979) and Godfrey (1978). The results of the test show that
our models do not suffer from a heteroskedasticity problem. For
example, in the general model (Model 1 of Table 3) the observed
Chi-square value of 2.581 is not significant at the 5 percent level
(the 5 percent critical value from a Chi-square distribution with 1
degree of freedom is 3.841). Therefore the null hypothesis of
homoskedasticity (or no heteroskedasticity) is not rejected. Table
3 gives the results of the maximum likelihood estimation (MLE) of
the random effects Tobit model. The table shows three models in
which
-
Journal of Economic & Administrative Sciences December
2007
57
dividend policy is measured by the dividend yield (DYLD). The
second column of each model reports the estimated marginal effect,
which is the probability that the dependent variable is
uncensored8. That is, the probability of a non-dividend-paying firm
to pay dividends conditional on a change in each of the explanatory
variables. The Wald test statistics reject the null hypothesis that
the parameters in the regression equations are jointly equal to
zero (models 1-3). The general model (Model 1) includes fifteen
variables and encompasses all of the models, with 759 firm-year
observations. Of the fifteen variables used in the model, twelve
have the hypothesized signs with the exception of STOCK and MBR.
Seven variables are statistically different from zero. To control
for industry effects, a dummy variable (NONFIN) is included taking
a value of one if a firm belongs to industrial or services sectors,
and zero otherwise (for firms in the insurance or banking sectors).
In the process of moving from the general to the specific model,
six variables (STOCK, FAML, INST, MULT, MBR, and TURN) were dropped
from Model 1 generating Model 2. The likelihood ratio test (LRT) is
performed to test Model 2 (restricted) against Model 1
(unrestricted) and see whether this process statistically provides
additional explanatory power to the model. In this case, the LRT
statistic is LR= 2 [270.47 - 272.33] = 3.72, and the critical value
from a 2 distribution, with 6 degrees of freedom, is 12.59 at the 5
percent level of significance. Since the computed value is less
than the critical value, the null hypothesis is not rejected. That
is, the null model (Model 2), which is the restricted model, cannot
be rejected. Therefore, Model 1 does not provide a statistically
significant increase in likelihood over Model 2, which supports our
exclusion of the aforesaid variables. These procedures are repeated
until we arrived at the best model that fits our data, that is
Model 3. The variables included in Model 3 possess the anticipated
sign and are statistically significant. From models 2 and 3, the
coefficient of NONFIN was not statistically different from zero,
suggesting that industry effects may not be important. At the
aggregate level, the regressions (models 1 3) are highly
significant at the 1 percent level or better. The regression
results of Table 3 show that the variable STOCK has a negative sign
but is not significantly different from zero9, indicating that
ownership dispersion does not appear to influence dividend policy
in Jordan.
8 For computing the marginal effects see, for example, Greene
(1999). 9 Using US data, Alli et al. (1993) obtained a similar
result.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
58
Table 3 MLE Results for Random Effects Tobit Model
Dependent Variable = DYLD Model 1 Model 2 Model 3
Independ. Variables
Coefficient Estimates
Marginal Effects
(%)Coefficient Estimates
Marginal Effects
(%)CoefficientEstimates
Marginal Effects
(%) Constant -0.266*** (-4.10)
-0.278*** (-5.43)
-0.300*** (-6.25)
STOCK -0.008 (-1.01) -0.241
INSD -0.060*** (-2.61) -1.795 -0.041** (-2.04) -1.237
-0.043**(-2.29) -1.273
FAML -0.004 (-0.43) -0.130
STATE 0.024* (1.93) 0.770 0.021*(1.90) 0.672
0.019*(1.83) 0.610
INST 0.008 (0.76) 0.237
MULT 0.004 (0.37) 0.118
AGE 0.004*** (4.52) 0.131 0.004*** (3.93) 0.128
0.004*** (3.83) 0.117
AGESQ -0.00006*** (-3.67) -0.002 -0.00006***
(-3.64) -0.002 -0.00006***
(-3.83) -0.002
MBR 0.00015 (0.04) 0.005
DER
-0.003** (-2.59) -0.090
-0.003*** (-2.95)
-0.090 -0.002*** (-2.61) -0.066
TURN -0.0002 (-0.02) -0.006
DTAX -0.007 (-1.52) -0.224 -0.006(-1.40)
-0.194
MCAP 0.016*** (3.62) 0.479 0.015***
(4.72)0.464 0.016***
(5.34) 0.480
EPS 0.005** (2.52) 0.136 0.005*** (2.84)
0.149 0.006*** (3.36) 0.167
NONFIN
-0.017 (-1.30) -1.686
-0.017(-0.97)
-0.532
Rho ()a 0.597*** (10.46) 0.588***(10.46) 0.578***(11.77) No. of
obs. 759 759 759 Log LKHD 272.33 270.47 269.15 Wald test 2
(15)=85.75 2 (9)=78.26 2 (7)=78.03 P-value 0.000 0.000 0.000 LR
testb 152.83 175.15 202.85 P-value 0.000 0.000 0.000 Notes: See
Table 1 for definitions. t-statistics are in parentheses. *, ** and
*** denote significance at the 10, 5 and 1 percent levels,
respectively. a proportion of total variance contributed by panel
level variance component. b provides a test for pooled (Tobit)
estimator against the random effects panel estimator.
-
Journal of Economic & Administrative Sciences December
2007
59
This result may be attributed to the special characteristic of
the Jordanian capital market at which firms tend to have high
levels of ownership concentration. In such a case, the
concentration of ownership may lead to a concentration in control.
Therefore, minority shareholders will not be able to exert much
influence on dividend policy. As predicted, the coefficient of the
second proxy of agency costs (INSD) is negative and statistically
significant at the 1 and 5 percent levels. This variable is also
economically significant. From Model 3, other things being equal, a
1 percent change in the proportion held by insiders expected to
account for 1.273 percent change in dividend yield. This indicates
that insider ownership is an important determinant of corporate
dividend policy in Jordan, in particular the level of dividends.
The negative and significant relationship between dividend yield
and insider ownership lends support to Rozeff (1982), who proposed
that the use of dividends as a bonding mechanism to reduce agency
costs is less important when there is higher insider ownership.
From Hypothesis 2, the variables representing the firms
ownership structure are expected to influence its dividend policy.
Four dummy variables are used, FAML, STATE, INST, and MULT. Table 3
shows that in Model 1 the coefficients of those variables have the
expected signs but are not significantly different from zero, with
the exception of STATE. The existence of the government or its
agencies as a controlling shareholder seems to influence the level
of dividends paid. The positive significant relationship between
dividend yield and STATE is consistent with the assertion that,
ceteris paribus, state-controlled firms tend to pay more dividends.
This prediction is based on the argument that state-controlled
firms are often subject to a double set of agency costs since the
ultimate owners of these firms are the citizens. In addition,
government entities may have tax privileges so they prefer to
invest in dividend-paying firms. The marginal effects indicate
that, a 1 percent increase in government ownership in a firm would
lead to an increase in the dividend yield by 0.610 percent. These
results are in line with those obtained by Gul (1999) for China,
and Gugler (2003) for Austria. In testing the information content
of dividends hypothesis (Hypothesis 3), the firm share turnover
ratio (TURN) is used as a proxy for information asymmetry. As
predicted, the result from Model 1 in Table 3 show that the
coefficient on TURN is negative, but the null hypothesis of zero
coefficient for the variable could not be rejected (t-statistic =
-0.02). The evidence here, therefore, does not provide support for
the signaling hypothesis. One
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
60
possible explanation for this result is that the proxy for
information asymmetry is weak10. Hypothesis 4 predicts that firms
with high growth and investment opportunities tend to retain their
income to finance those investments, thus paying less or no
dividends. The variables MBR and AGE are used as proxies for growth
and investment opportunities. From the regression results (Model 1)
in Table 3, contrary to expectations, the coefficient of MBR is
positive and insignificant, whereas, as predicted, the coefficient
of AGE is positive and significantly different from zero. These
findings indicate that the market-to-book value ratio is not
related to dividend yield, while firm age is positively related to
dividend yield. The positive coefficient on MBR is analogous to
that reported by Aivazian et al (2003) for Jordanian firms;
however, they obtain a significant coefficient in their results.
However, the hypothesized relationship between dividends and growth
and investment opportunities could not be rejected since the other
proxy variable for growth (AGE) is highly significant with
t-statistics of 4.52, 3.93 and 3.83 in models 1, 2 and 3,
respectively. The age of the firm is consistently significant at
the 1 percent level, suggesting that mature firms tend to pay
higher levels of dividends. From Model 3, as a firm becomes one
year older, the estimated increase in DYLD is 0.117 percent,
ceteris paribus. These results, reported in Table 3 , are
consistent with prior research of Manos and Green (2001) who found
that the age of the firm and the payout level are positively
related for independent (non-group affiliated) Indian firms. The
result also provides empirical support for the maturity hypothesis
proposed by Grullon et al. (2002). That is, as firms become mature
their growth opportunities tend to decline resulting in lower
capital expenditure needs, and thus more cash flows are available
for dividend payments. When a mature firm has little or no
investment opportunities, a high level of dividends will reduce the
discretionary resources available to managers that could be wasted
in perquisites or unprofitable projects. In other words, dividends
reduce the agency costs associated with free cash flow. The results
hence also provide support for the free cash flow hypothesis
(Easterbrook, 1984, and Jensen, 1986). To the best of the authors
knowledge, this paper provides the first attempt to document that
the age of the firm is quadratically negatively related to dividend
yield as shown by the significant positive coefficients on AGE and
the significant negative coefficients on AGESQ (age squared) in
10 El-Khouri and Almwalla (1997) provided weak or no support for
the signaling hypothesis for Jordanian firms.
-
Journal of Economic & Administrative Sciences December
2007
61
all regressions. This non-linear relationship between the age of
the firm and dividend yield (negative sign of AGESQ) may imply
changes in a firms life cycle, that is, movement from a lower
growth phase to a higher growth phase. In other words, when a firm
finds new investment opportunities (growth phase) it will pay less
or no dividends because paying dividends is no longer optimum.
Again, this evidence reinforces the interpretation above that the
positive association between the age of the firm and dividend yield
is consistent with Grullon et al.s (2002) maturity hypothesis, and
accordingly with the free cash flow hypothesis of Easterbrook
(1984) and Jensen (1986). Another variable found to be a
determinant of corporate dividend policy in Jordan is the firm size
(Hypothesis 5). As expected, Table 3 reports that the coefficients
on size (MCAP) are robustly positively correlated with dividend
yield. The t-statistics of the coefficients on MCAP for models 1, 2
and 3 are 3.62, 4.72 and 5.34, respectively, which are highly
significant at the 1 percent level. Firm size as measured by market
capitalization is positively related to DYLD with marginal effects
of 0.480 percent. If for example a firms market capitalization
increases from JD 10 million to JD 15 million, the expected
increase in its DYLD would be 0.195 percent11. The positive and
significant correlation between dividend yield and size suggests
that large firms are more able to pay dividends. This finding lends
support to the agency costs explanation. The intuition here is that
the larger the firm, the more difficult (costly) is the monitoring
(i.e. the greatest the agency problem). Thus, dividends could play
a role in helping to alleviate the agency problem. Also, the
positive relation between dividend yield and size supports the
generally accepted view proposed by many finance scholars that
larger firms have easier access to capital markets (see, among
others, Lloyd et al., 1985, Holder et al., 1998, and Fama and
French, 2002), and have lower transaction costs associated with
acquiring new financing as compared to small firms (Alli et al.,
1993). Hypothesis 6 predicts a negative relationship between a
firms financial leverage and dividend yield. Table 3 shows that the
coefficients on debt-to-equity ratio (DER) are negative and
statistically significant at the 5 and 1 percent levels. The
t-statistics of the coefficients on DER for models 1, 2 and 3 are
2.59, 2.95 and 2.61, respectively. This suggests that firms with
high debt ratios tend to pay fewer dividends, and the level of
dividend payments
11 [(LN 15,000,000 LN 10,000,000) * 0.480], see Washer and Casey
(2004) for a similar calculation.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
62
thus seems to be negatively correlated with the level of
financial leverage12. From Model 3, if a firms financial leverage
increases by 1 percent, the estimated decrease in dividend yield
will be 0.066. In relation to emerging equity markets including
Jordan, Aivazian et al. (2003) provided evidence consistent with
findings presented here. However, a word of caution should be
added. The sample used in this study consists of all companies
listed on the ASE including financial companies, which are highly
levered. Turning to the firms profitability (Hypothesis 7), the
estimates of earnings per share (EPS) are all positive and
significant at the 1 percent level, with the exception of Model 1
which is significant at the 5 percent level. This suggests that
profitability is a critical determinant of the level of dividends
paid by Jordanian firms. However, contrary to expectation, the
profitability variable does not seem to have large economic
importance. An increase in earnings per share of 1 percent would
lead to an increase in dividend yield of only 0.167 percent. The
positive relationship between profitability and dividends is well
documented in the literature (see, for example, Jensen et al, 1992,
Han et al., 1999, and Fama and French, 2001, 2002). The results
from Table 3 show that the implementation of a 10 percent tax rate
on dividends in 1996 (Hypothesis 8) seems to have had no
significant influence on corporate dividend policy in Jordan.
Consistent with the tax-preference theory, the coefficients on DTAX
are negative in models 1 and 2, but the hypothesis of zero
coefficients for tax variable could not be rejected. This result
lends support for the findings of Omet (2004) that the
implementation of the tax had no impact on corporate dividend
behavior in Jordan13. 6. Summary and Conclusion This paper has
examined the main determinants of corporate dividend policy in
Jordan. Tobit specifications were used to examine the determinants
of the level or the amount of dividends paid. The results were
obtained using the maximum likelihood estimations of the random
effects Tobit regressions.
12 Another variable (for non-financial firms) that was found to
be significantly and negatively related to dividend policy in
Jordan is the asset tangibility. The results are not reported here
in order to save space but are available upon request. 13 It is
worth mentioning here that when we performed the Wilcoxon test to
see whether dividend payout ratios differ between pre-tax
(1989-1995) and post-tax (1996-2000) periods, the null hypothesis
that payout ratios for both periods have the same distribution is
rejected. The Wilcoxon test statistic yields a Z-statistic of 7.84
(P-value = 0.000).
-
Journal of Economic & Administrative Sciences December
2007
63
Nested procedures were followed to arrive at the best-fitted
model. In order to ensure that our results are robust, several
diagnostic tests were performed. The data showed that ownership
dispersion as measured by the natural log of the number of
stockholders (STOCK) seems to not be related to dividend policy in
Jordan. The fraction held by insiders (INSD), the second proxy for
the agency costs hypothesis, has negative impact on the level of
dividends paid. Similarly, the existence of government or its
agencies (STATE) in a firms ownership structure (controlling
shareholder) affects the amount of dividends (positively). Other
variables of ownership structure seem to have no influence on
dividend policy. By and large, the evidence is consistent with the
agency costs explanation. The firms age, size, and profitability
positively and significantly affect its dividend policy. The use of
age, and especially age squared as proxies for growth has not been
used in empirical testing of dividend policy to the best of the
authors knowledge, and these results therefore suggest interesting
avenues for future research. Again, these findings are generally
consistent with the agency costs hypothesis and lend support to the
pecking order hypothesis. The analysis also found that a firms
financial leverage is significantly and negatively related to its
dividend policy. The study demonstrated that much of the existing
theoretical literature on dividend policy can be applied to an
emerging capital market such as Jordan. Many of the factors that
were found to be significant in the determination of dividend
policy are the same as those found in developed capital
markets.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
64
References Abrutyn, Stephanie, and Robert W. Turner (1990),
Taxes and Firm's
Dividend Policies: Survey Results, National Tax Journal 43,
491-496. Aivazian, Varouj, Laurence Booth, and Sean Cleary (2003),
Do Emerging
Market Firms Follow Different Dividend Policies From U.S.
Firms?, The Journal of Financial Research 26, 371-387.
Alli, Kasim L., A.Qayyum Khan, and Gabriel G. Ramirez (1993),
Determinants of Corporate Dividend Policy: A Factorial Analysis,
The Financial Review 28, 523-547.
Anderson, G. J. (1986), An Application of the Tobit Model to
Panel Data: Modelling Dividend Behaviour in Canada, Working Paper,
MacMaster University.
Bali, Rakesh (2003), An Empirical Analysis of Stock Returns
Around Dividend Changes, Applied Economics 35, 51-61.
Barclay, Michael 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 19, 31-44.
Barclay, Michael J., Clifford W. Smith, and Ross L. Watts
(1995), The Determinants of Corporate Leverage and Dividend
Policies, Journal of Applied Corporate Finance 7, 4-19.
Bartov, Eli, and Gordon M. Bodnar (1996), Alternative Accounting
Methods, Information Asymmetry and Liquidity: Theory and Evidence,
The Accounting Review 71, 397-418.
Bhattacharya, Sudipto (1979), Imperfect Information, Dividend
Policy, and "the Bird in the Hand" Fallacy, Bell Journal of
Economics 10, 259-270.
Black, Fischer, and Myron S. Scholes (1974), The Effects of
Dividend Yield and Dividend Policy on Common Stock Prices and
Returns, Journal of Financial Economics 1, 1-22.
Booth, Laurence, Varouj Aivazian, Asli Demirguc-Kunt, and
Vojislav Maksimovic (2001), Capital Structure in Developing
Countries, Journal of Finance 56, 87-130.
Brennan, Michael J. (1970), Taxes, Market Valuation and
Corporate Financial Policy, National Tax Journal 23, 417-427.
Breusch, T. S., and A. R. Pagan (1979), A Simple Test for
Heteroscedasticity and Random Coefficient Variation, Econometrica
47, 1287-1294.
Conroy, Robert M., Kenneth M. Eades, and Robert S. Harris
(2000), A Test of the Relative Pricing Effects of Dividends and
Earnings: Evidence from Simultaneous Announcements in Japan,
Journal of Finance 55, 1199-1227.
-
Journal of Economic & Administrative Sciences December
2007
65
Denis, David J., Diane K. Denis, and Atulya Sarin (1994), The
Information Content of Dividend Changes: Cash Flow Signaling,
Overinvestment, and Dividend Clienteles, Journal of Financial and
Quantitative Analysis 29, 567-587.
Deshmukh, Sanjay (2003), Dividend Initiations and Asymmetric
Information: A Hazard Model, Financial Review 38, 351-368.
Dhaliwal, Dan S., Merle Erickson, and Robert Trezevant (1999), A
Test of the Theory of Tax Clienteles for Dividend Policies,
National Tax Journal 52, 179-194.
Easterbrook, Frank H. (1984), Two Agency Costs Explanations of
Dividends, American Economic Review 74, 650-659.
El-Khouri, Ritab, and Mona Almwalla (1997), The Effect of
Dividend Changes on Security Prices: The case of Jordanian
Companies, Abhath Al-Yarmouk 13, 87-94.
Fama, Eugene F., and Kenneth R. French (2001), Disappearing
Dividends: Changing Firm Characteristics or Lower Propensity to
Pay?, Journal of Financial Economics 60, 3-43.
Fama, Eugene F., and Kenneth R. French (2002), Testing Trade-Off
and Pecking Order Predictions About Dividends and Debt, The Review
of Financial Studies 15, 1-33.
Farinas, Jose C., and Lourdes Moreno (2000), Firms' Growth, Size
and Age: A Nonparametric Approach, Review of Industrial
Organization 17, 249-265.
Godfrey, Leslie G. (1978), Testing for Multiplication
Heteroskedasticity, Journal of Econometrics 8, 227-236.
Gonedes, Nicholas J. (1978), corporate Signalling, External
Accounting, and Capital Market Equilibrium: Evidence on Dividends,
Income, and Extraordinary Items, Journal of Accounting Research 16,
26-79.
Gordon, Myron J. (1959), Dividends, Earnings, and Stock Prices,
Review of Economics and Statistics 41, 99-105.
Greene, William H. (1999), Marginal Effects in the Censored
Regression Model, Economics Letters 64, 43-49.
Grullon, Gustavo, Roni Michaely, and Bhaskaran Swaminathan
(2002), Are Dividend Changes a Sign of Firm Maturity?, Journal of
Business 75, 387-424.
Gugler, Klaus (2003), Corporate Governance, Dividend Payout
Policy, and the Interrelation Between Dividends, R&D, and
Capital Investment, Journal of Banking and Finance 27,
1297-1321.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
66
Gul, Ferdinand A. (1999), Government Share Ownership, Investment
Opportunity Set and Corporate Choices in China, Pacific-Basin
Finance Journal 7, 157-172.
Han, Ki C., Suk Hun Lee, and David Y. Suk (1999), Institutional
Shareholders and Dividends, Journal of financial and Strategic
Decisions 12, 53-62.
Hendry, David F. (1995). Dynamic Econometrics (Oxford University
Press, New York).
Holder, Mark E., Frederick W. Langrehr, and J. Lawrence Hexter
(1998), Dividend Policy Determinants: An Investigation of the
Influences of Stakeholder Theory, Financial Management 27,
73-82.
Huang, Ho-Chuan River (2001), Bayesian Analysis of the Dividend
Behaviour, Applied Financial Economics 11, 333-339.
Huergo, Elena, and Jordi Jaumandreu (2004), Firms' Age, Process
Innovation and Productivity Growth, International journal of
Industrial Organization 22, 541-559.
Jensen, Gerald R., Donald P. Solberg, and Thomas S. Zorn (1992),
Simultaneous Determination of Insider Ownership, Debt, and Dividend
Policies, Journal of Financial and Quantitative Analysis 27,
274-263.
Jensen, Michael C. (1986), Agency Costs of Free Cash Flow,
Corporate Finance, and Takeovers, American Economic Review 76,
323-329.
Jensen, Michael C., and William H. Meckling (1976), Theory of
the Firm: Managerial Behavior, Agency Costs and Ownership
Structure, Journal of Financial Economics 3, 305-360.
John, Kose, and Joseph Williams (1985), Dividends, Dillution,
and Taxes: A Signalling Equilibrim, Journal of Finance 40,
1053-1070.
Kim, Byeong Soo, and G. S. Maddala (1992), Estimation and
Specification Analysis of Models of Dividend Behavior Based on
Censored Panel Data, Empirical Economics 17, 111-124.
Lease, Ronald C., Kose John, Avner Kalay, Uri Loewenstein, and
Oded H. Sarig (2000). Dividend Policy: Its Impact on Firm Value
(Harvard Business School Press, Boston, Massachusttes).
Leuz, Christian, and Robert E. Verrecchia (2000), The Economic
Consequences of Increased Disclosure, Journal of Accounting
Research 38, Supplement, 91-124.
Lewellen, Wilbur G., Kenneth L. Stanley, Ronald C. Lease, and
Gary G. Schlarbaum (1978), Some Direct Evidence on the Dividend
Clientele Phenomenon, Journal of Finance 33, 1385-1399.
-
Journal of Economic & Administrative Sciences December
2007
67
Lie, Erik (2000), Excess Funds and Agency Problems: An Empirical
Study of Incremental Cash Disbursements, Review of Financial
Studies 13, 219-248.
Lintner, John (1956), Distribution of Incomes of Corporations
Among Dividends, Retained Earnings, and Taxes, American Economic
Review 46, 97-113.
Litzenberger, Robert H., and Krishna Ramaswamy (1979), The
Effect of Personal Taxes and Dividends on Capital Asset Prices,
Journal of Financial Economics 7, 163-195.
Lloyd, William P., John S. Jahera, and Daniel E. Page (1985),
Agency Costs and Dividend Payout Ratios, Quarterly Journal of
Business and Economics 24, 19-29.
Manos, Ronny, and Christopher Green (2001), Business Group and
Dividend Policy: Evidence on Indian Firms, Working Paper No.33,
(Institute for Development Policy and Management, University of
Manchester).
Michaely, Roni, Richard H. Thaler, and Kent L. Womack (1995),
Price Reactions to Dividend Initiations and Omissions: Overreaction
or Drift?, Journal of Finance 50, 573-608.
Miller, Merton H., and Franco Modigliani (1961), Dividend
Policy, Growth, and the Valuation of Shares, Journal of Business
34, 411-433.
Miller, Merton H., and Kevin Rock (1985), Dividend Policy Under
Asymmetric Information, Journal of Finance 40, 1031-1051.
Miller, Merton H., and Myron S. Scholes (1982), Dividend and
Taxes: Some Empirical Evidence, Journal of Political Economy 90,
1118-1141.
Morgan, Gareth, and Stephen Thomas (1998), Taxes, Dividend
Yields and Returns in the UK Equity Market, Journal of Banking and
Finance 22, 405-423.
Myers, Stewart C. (1984), The Capital Structure Puzzle, Journal
of Finance 39, 575-592.
Myers, Stewart C., and Nicholas S. Majluf (1984), Corporate
Financing and Investment Decisions When Firms Have Information That
Investors Do Not Have, Journal of Financial Economics 13,
187-221.
Nissim, Doron, and Amir Ziv (2001), Dividend Changes and Future
Profitability, Journal of Finance 56, 2111-2133.
Omet, Ghassan (2004), Dividend Policy Behaviour in the Jordanian
Capital Market, International Journal of Business 9, 287-299.
Pettit, R. Richardson (1972), Dividend Announcements, Security
Performance, and Capital Market Efficiency, Journal of Finance 27,
993-1007.
-
Dr. Husam-Aldin Nizar Al-Malkawi December 2007
68
Pettit, R. Richardson (1977), Taxes, Transactions costs and the
Clientele Effect of Dividends, Journal of Financial Economics 5,
419-436.
Poterba, James M., and Lawrence H. Summers (1984), New Evidence
That Taxes Affect the Valuation of Dividends, Journal of Finance
39, 1397-1415.
Rodriguez, Alicia C., Migule A. Molina, Ana L. G. Perez, and
Urbano M. Hernandez (2003), Size, Age and Activity Sector on the
Growth of the Small and Medium Firm Size, Small Business Economics
21, 289-307.
Rozeff, Michael S. (1982), Growth, Beta and Agency Costs as
Determinants of Dividend payout Ratios, The Journal of Financial
Research 5, 249-259.
Scholz, John Karl (1992), A Direct Examination of the Dividend
Clientele Hypothesis, Journal of Public Economics 49, 261-285.
StataCorp. (2003), Stata Statistical Software: Release 8.0,
(Stata Corporation, College Station, Texas).
Verbeek, Marno (2000). A Guide to Modern Econometrics, (John
Wiley & Sons,Ltd, Brisbane).
Watts, Ross (1973), The Information Content of Dividends,
Journal of Business 46, 191-211.
Washer, Kenneth M., and K. Michael Casey (2004), A Note on the
Determinants of Equity Payouts, Journal of Applied Business
Research 20, 83-90.
Yoon, Pyung Sig, and Laura T. Starks (1995), Signalling,
Investment Opportunities, and Dividend Announcements, Review of
Financial Studies 8, 995-1018.
-
Journal of Economic & Administrative Sciences December
2007
69
Appendix 1 Correlation Matrix and Variance Inflation Factors
(VIF)
for the Explanatory Variables
ST
OC
K
INSD
FAM
L
STA
TE
INST
MU
LT
AG
E
AG
ESQ
MBR
DER
TU
RN
DTA
X
MC
AP
EPS
TA
NG
VIF
STOC
K
1.000
-.362
-.166
-.145
-.156
-.236
-.090
.163
-.148
-.007
.085
-.087
.178
-.020
.010
1.32
INSD
1.000
.107
.484
.405
.480
.241
.170
.215
.098
-.216
.047
.085
-.022
.018
2.35
FAM
L
1.000
-.144
.013
.388
-.024
-.042
-.062
-.053
-.053
.079
-.202
-.069
-.124
1.61
STATE
1.000
-.161
-.171
.402
.365
.117
.040
-.160
-.092
.244
.058
.003
2.20
INST
1.000
.545
-.017
-.084
.018
.012
-.090
.129
-.030
-.055
-.037
2.34
MU
LT
1.000.157
.104
.031
-.058
-.160
.080
.016
-.056
-.100
2.32
AG
E
1.000
.946
.039
.121
-.095
.007
.315
.068
-.113
1.42
AG
ESQ
1.000
.017
.151
-.097
-.004
.368
.042
-.023
MBR
1.000
.434
-.102
-.294
.306
.094
.014
1.81
DE
R
1.00
.014
.037
-.009
-.075
.094
1.45
TUR
N
1.000
-.074
-.160
-.060
.112
1.11
DT
AX
1.000
-.151
-.073
.038
1.21
MC
AP
1.00 0
.178
.080
1.51
EPS
1.00 0
-.084
1.14
TA
NG
1.000
1.11
-
7002 rebmeceD iwaklaM-lA raziN nidlA-masuH .rD
07
:
.
-
atad lenap()
0002 9891
)tiboT(
.
-
Reproduced with permission of the copyright owner. Further
reproduction prohibited without permission.