AL Malkawi 2007

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.


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


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.


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


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).


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


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.


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


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.


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.


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


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.


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.


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


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.


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.


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).


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.


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.


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.


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.


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.


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.


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.

Date post:	06-Nov-2015
Category:	Documents
Upload:	yolla-monica-angelia-thenu
View:	225 times
Download:	0 times

AL Malkawi 2007

Documents