Dividend, share repurchases, investor protection,
creditor rights and earnings
10. January 2020
Name: Philipp Hauri
Student number: s3620808 (RUG) / phha6996 (UU)
Study Programme: MSc IFM Double Degree with Uppsala University Supervisor: Dr. Adri de Ridder
Co-assessor: Dr. Peter Smid Abstract
This study examines how the relation of earnings and payout is influenced by country-level moderators of investor protection and creditor rights. By using an international sample with 42,148 firm-year observations in 3,658 firms during the years 2001-2017, I find that earnings-payout relation is strong. Further, I present evidence that the earnings-earnings-payout relation is alleviated in weak investor protection and creditor right countries due to the negative influence of dividends. Although share repurchases positively impact the strong earnings-payout relation, the positive effect of share repurchases is weaker compared to the negative one of dividends. As a consequence, in weak investor protection and creditor right countries, the strong earnings-payout relation is alleviated when it is tested with the logit regression.
1. Introduction
In an intriguing study, Modigliani and Miller (1961) present evidence that dividends do not influence the value of a company in frictionless markets. Many scholars have tried to disentangle the characteristics of dividends since then. For example, previous studies (e.g. Denis and Osobov, 2008; Fama and French, 2001) show association between firm size, growth opportunities and profitability with dividends. Jensen (1986) shows that firms make use of dividends to mitigate agency costs with respect to excess free cashflows. Since the publication of Modigliani and Miller (1961), the use of dividends has significantly decreased across firms (Fama and French, 2001; Skinner, 2008). As shown by Banyi and Kahle (2014), firms that recently have gone public are less likely to pay dividends. Interestingly however, the aggregate dividends have not declined and they seem to be concentrated among the largest and most profitable firms (DeAngelo et al., 2004; Denis and Osobov, 2008). As a consequence, the traditional strong relation between dividends and earnings (Fama and Babiak, 1968), has decreased over time (Brav et al., 2005).
stock options (Dittmar, 2000; Kahle, 2002; Wansley et al., 1989). Due to the increasing use of share repurchases and their proven economic significance (Manconi et al., 2019), scholars have concluded that share repurchases are substitutes to dividends (Grullon and Michaely, 2002). Consistent with this view, Skinner (2008) shows a stronger relation between share repurchases and earnings than with dividends and conclude that share repurchases are now the dominant payout mechanism. Prior studies such as Skinner (2008) investigating the relation between dividends, share repurchases and total payout base their results on a US sample. Following this introduction, my first research question is: Does the relation between earnings and payout hold in an international setting?
Ever since the publication of La Porta et al. (1998), scholars have been aware about the fact that the law matters in terms of mitigating agency conflicts between insiders (managers) and outsiders (shareholders). Thereafter, La Porta et al. (2000) present two different models how the law in terms of investor protection can impact dividend payouts. The premise of the outcome model is that higher investor protection leads to higher dividend payouts. In contrast, the proposition of the substitute model is that in weak investor protection countries, firms use dividends as “reputational mechanism” to increase investor’s confidence (La Porta et al., 2000, p. 7) and therefore dividend payout is higher. This study examines the moderating effect of investor protection on the main relation between earnings and total payout. Thus, my second research question is: What is the relation between earnings and payout by introducing investor protection as country-level moderator?
the final research question in this study concerns the moderating effect of creditor rights: What is the relation between earnings and payout by introducing creditor rights as country-level moderator?
Despite the increasing worldwide use of share repurchase programs, studies that investigate share repurchase programs on global scale are still scarce compared to the US. For instance, Lee and Suh (2011) investigate a worldwide sample and find that cash holdings is positively related to share repurchases. Further, Manconi et al. (2019) find a relation between share repurchases and short-term and long-term excess returns. However, to the best of my knowledge, no study has ever examined the relation between payout and earnings with moderating country-level factors (investor protection and creditor rights). Therefore, I aim to fill this research gap.
The hypothesized relations are tested on an international sample by using 42,148 firm-year observations from 2001-2017 in 3,658 firms across 39 countries. The relation between earnings and payout as well as the moderating effects of investor protection and creditor rights on payout is tested by applying both ordinary least squares regression (OLS) and logit regressions. OLS bears the benefit that it shows the degree of impact in form of the estimated coefficients while logit allows to regularize the sample and present the propensities in the prediction results of the analysis.
by La Porta et al. (2000) which states that in weak investor protection countries, payouts are higher in order to increase the reputation of the firm.
In weak creditor rights countries a less positive relation between dividends and earnings is observed. As Brockman and Unlu (2009) indicate, dividends are less used in weak creditor rights due to restrictive debt covenants on dividends. In contrast to dividends, share repurchases in weak creditor right countries have a more positive relation with earnings. I conclude that the relation of earnings and share repurchases is stronger in weak creditor right protection countries since debt covenants reduce the likelihood to pay dividends (Brockman and Unlu, 2009).
When tested with the logit regression, the results support the argument that the earnings-payout relation is less positive in weak investor protection and creditor right countries since the negative effect of dividends is stronger than the positive effect of share repurchases.
This study makes several contributions to the literature. First, I extend the literature on the earnings-payout relation using an international sample. Second, the findings corroborate that dividends and share repurchase are used differently in distinctive institutional settings. These findings are not only contributions to literature, but also they are also crucial from a managerial point of view. Managers can use this information by adapting the optimal payout with respect to the institutional setting.
2. Literature Review
2.1. The basics of dividends
2.2. The basics of share repurchases
Share repurchases or also referred as share buybacks are an alternative mechanism to distribute cash to shareholders by buying back its own stock. Important to note is that share repurchases are an option to distribute cash and are therefore not sticky as dividends (Dittmar, 2008). Due to this, firms do not have to comply with any corporate obligation in contrast to dividends (Grullon and Michaely, 2002). There are three popular ways to repurchase shares: (i) the fixed price tender offer, (ii) the Dutch-auction tender offer and (iii) the open market repurchase program. Among the three options, firms prefer the open market repurchase program. In this study, I focus on the open market repurchase program (OMR).
A significant increase in the use of share repurchases has been documented in the United States (e.g. Fama and French, 2001; Grullon and Michaely, 2002) and in the United Kingdom (Ferris et al., 2006). In addition, De Ridder (2015) states how several countries in the European Union as well as other countries around the world removed restrictions of share repurchases. For example, Australia in 1989, Hongkong in 1991, Denmark in 1995, Finland in 1997, France, Germany and India in 1998, and Sweden in 2000. Although share repurchases have become more prominent, the underlying reason of why companies use share repurchases is still debated. The most prominent theories are explained below.
2.2.1. Flexible distribution policy
2.2.2. Agency cost theory and the free cash flow hypothesis
The agency conflict between managers (agent) and shareholders (principal) is covered in the finance and economic literature. Since managers are acting on behalf of shareholders, potential conflict of interest can arise, especially when companies have excess cash. Firms can then decide if they want to retain the cash or distribute to shareholders (Easterbrook, 1984; Jensen, 1986). Companies that decide to retain the cash have more cash than they actually need to make optimal investments and therefore the likelihood of agency conflicts increases (Jensen, 1986). There are several ways how shareholder limit the potential risk of agency conflicts. One way is the managers commitment to pay out dividend and share repurchase. Supporting this theory, Grullon and Michaely (2004) find that repurchasing firms reduce the free cash flow problem and have a lower amount of cash on their balance sheets. Also, Brav et al. (2005) find support in their extensive survey for the argument of excess cash holding.
2.2.3. Low investment opportunities
Another motive why firms use share repurchase programs is the lack of investment opportunities (Boudry et al., 2013; Wansley et al., 1989). Brav et al. (2005) support this argument by finding evidence that firms that lack sound investment opportunities are more likely to use share repurchase programs. Also, by analysing the investment opportunity set of the Real Estate Investment Trust, Boudry et al. (2013) find a relation between weak investment opportunity and share repurchases.
2.2.4. Undervaluation and signalling hypothesis
However, the empirical evidence on this theory is mixed. For example, Grullon and Michaely (2004) find no abnormal enhancement in operating performance in the three years after the announcement. On the contrary, Lie (2005) finds a significant increase in return on asset (ROA) in three years after by using the same time period and companies as Grullon and Michaely (2004). To be specific, Lie (2005) finds that the share price reacts positively by having a mean and median announcement return in the three days after of 3.0% and 1.9%, respectively. Also, the author shows that a relative performance improvement is observed up to two years after the announcement. Nevertheless, the extent to which firms can use share repurchases to signal undervaluation is still somewhat unclear.
2.2.5. Leverage adjustment
Firms that consider themselves as underleveraged have the possibility to increase leverage to regain an optimal leverage ratio by repurchasing stock. By doing so, repurchasing stock increases the leverage ratio and readjust to the target capital structure (Dittmar, 2000; Wansley et al., 1989).
2.2.6. Compensation policy
this might be the reason why institutional investor prefer share repurchases over dividends (Jain, 2007).
2.2.7. Summary of the theories
Table 1 below summarizes the theories of share repurchases. Table 1. Overview of share repurchase theories
Motives Explanation
Flexible distribution policy (Dittmar, 2008)
The timing of starting share repurchase programs is each firms’ decision.
Free cashflow hypothesis (Jensen, 1986)
Share repurchase can reduce the free cashflow problem by reducing the excess cash on the balance sheet.
Weak investment opportunities
(Boundry et al., 2013; Wansley et al., 1986)
Lack of investment is related to higher engagement in share repurchase programs. Undervaluation and signalling hypothesis
(Dittmar, 2008; Lie, 2005)
Share repurchase programs can be a sign of undervaluation.
Leverage adjustment
(Wansley et al., 1989; Dittmar, 2000)
Repurchasing shares reduces equity and therefore increase the leverage ratio. When manager believe that they are underleveraged, share repurchase can be used.
Compensation policy
(Wansley et al., 1989; Dittmar, 2000; Kahle, 2002)
2.3. Changes in payout policy
Corporate payout policy has undergone changes in the last 30 years. On the one hand, firms which regularly pay dividends have decreased significantly (Fama and French, 2001; Skinner, 2008). For instance, Fama and French (2001) show that the ratio of US firms that pay cash dividends dropped from 66.5% to 20.8% in the period of 1978 to 2000. On the other hand, since share repurchases have become economically significant in the 1980s (Grullon and Michaely, 2002), their importance increased substantially (Skinner, 2008). Some authors even claim that share repurchases are now even more important than dividends, specifically in the US (Lee and Suh, 2011). More recently, a so called “buyback wave” has been observed and the authors refer hereby to the decrease in regulation that has paved the way for firms to introduce share repurchase programs on a global scale (Manconi et al., 2019, p. 1).
2.4. The relation between dividend and earnings
Lintner (1956) shows that firms are unwilling to reduce dividends. Brav et al. (2005) even find that firms prefer to sell assets, lay-off employees or forgo positive NPV projects to avoid reducing dividends. As a payout mechanism, dividends play a crucial in reported earnings. Daniel et al. (2008) even state that dividends are considered to be an earnings threshold. Although there is still a relation between dividends and earnings, the originally very strong connection (Fama and Babiak, 1968) has weakened over time, as shown by Brav et al. (2005).
2.5. The relation between share repurchase and earnings
share repurchases, share repurchases represent the variation in earnings which is a sign of substitution of dividends. Further, the authors show that not only in firms with regular repurchase payments, also in ones with occasional use can a strong link between share repurchases and earnings be detected. The results show convincing evidence for the strong relation between share repurchases and earnings (Skinner, 2008) and additional evidence for the decreasing use of dividends consistent with Fama and French (2001).
2.6. Cross-country legal regimes
Conflict of interest, also referred as agency conflicts, exist between insiders (managers) and outsiders such as minority or controlling shareholders (e.g. La Porta et al., 2000). One way to mitigate potential agency conflicts is the law. Outside investors are given power by the law to reduce the chance of expropriation by insiders. Examples of such rights are the ability to vote on crucial matters, receiving the same share per dividend, or the ability to sue the firm for damages (La Porta et al., 2000, 1998). However, these legal regimes vary significantly across countries. Investors in common law countries (e.g. United States or United Kingdom) are well-protected, especially when compared to civil law countries. Civil law countries are further divided into German, Scandinavian and French civil law countries whereof French civil law countries provide the weakest protection (La Porta et al., 1998). Meanwhile investor protection has been improved in many civil law country. For example, the mandatory adaption of IFRS has substantially improved the information quality of the financial reporting. Especially the forecast accuracy of analysts is better since firms can be easier compared across markets (Houqe et al., 2014).
“reputational mechanism” and therefore the dividend payout in weak protection must be higher since the firms must compensate outsiders for the weak protection (La Porta et al., 2000, p. 7). Nonetheless, consistent with the findings by La Porta et al. (2000), most scholars have found evidence supporting the outcome model of dividends (e.g. Byrne and O’Connor, 2012; Jiraporn et al., 2011; Mitton, 2004).
Table 2, Outcome and substitute model
Models Common law Civil law
Outcome model Payout is higher Payout is lower
Substitute Model Payout is lower Payout is higher
Notes: This table summarizes the key findings by La Porta et al. (2000) by differentiating
between civil and common law countries by taking the payout of firms into account.
2.7. Creditor rights
collateral. That is why, creditors require dividend restrictions in order to offset the weak creditor rights. Also, Brockman and Unlu (2009) show that in weak creditor right countries, the likelihood of dividend omission is higher when debt payments are in jeopardy. Similar results are found by Byrne and O’Connor (2012) which show that firms in weak creditor rights countries have lower amounts and less firms overall that pay dividends.
3. Hypothesis development
The first hypotheses in this study concerns the relation between earnings and the total payout (defined as the sum of share repurchases and dividends) scaled by total assets. Skinner (2008) shows that the relation between earnings and total payout is strong. This pattern should be also true using an international sample. Thus, my first hypothesis is:
Hypothesis 1: Earnings are positively related to payout.
This study also takes a country-level moderator into account in terms of investor protection. There are several studies that present evidence that in strong investor protection countries, the propensity of firms paying dividends or share repurchases is higher compared to weak protection countries (e.g. Alzahrani and Lasfer, 2012; La Porta et al., 2000). This can be explained by the lower likelihood of expropriation of minority shareholders (Djankov et al., 2008; La Porta et al., 2000). Potential expropriation in weak protection countries might alleviate the relation between total payout and earnings since additional earnings are more likely to be expropriated instead of paid to shareholders. My second hypothesis is therefore:
Hypothesis 2: The positive relation between earnings and payout is alleviated in weak
In addition, I introduce another country-level variable, namely country-level creditor rights. Brockman and Unlu (2009) provide evidence that in weak creditor right protection countries, the amount and the propensity to pay dividends is lower. Although dividends play a less significant role than they used to and are often substituted by share repurchases (Grullon and Michaely, 2002; Skinner, 2008), dividends are still relevant especially among the largest and most profitable companies (DeAngelo et al., 2004; Denis and Osobov, 2008). Thus, I hypothesize that the enforcement of these dividend restriction in weak creditor right countries alleviates the relation between total payout and earnings. Thus, I arrive at the following hypothesis:
Hypothesis 3: The positive relation between earnings and payout is alleviated in weak
creditor protection countries.
4. Data
4.1. Data and sample selection
5. Methodology
5.1. Dependent variables
5.1.1. Payout
The first dependent variable in this study is total payout (PAYOUT), defined as the sum of dividends and the share repurchases scaled by total assets at year-end.
5.1.2. Dividend
Data for dividends (DIVIDENDS) is also collected from Eikon Thomson Reuters database. Common dividends are considered that represent the cash dividends payed by the firm scaled by total assets at year-end.
5.1.3. Share repurchases
Most previous studies have used Compustat to measure the actual amount of share repurchases by deducting any reduction in the value of the net number of preferred stock outstanding (Compustat item #56) from the total expenditures on the purchase of preferred and common stocks (Compustat item #115). However, this is only available for US and Canadian companies. Lee and Suh (2011) present an alternative solution for global studies on share repurchases and use Worldscope for their analysis and I use the same approach in my study1. Share repurchases (REPURCHASES) are then scaled by total assets at year-end.
1 The approach by Lee and Suh (2011) uses Worldscope data item #04751 (Common/Preferred
5.2. Major independent variables
5.2.1. Earnings
The level of earnings (EARNINGS) is measured by the net income scaled by total assets at year-end. EARNINGS is a continuous variable that also includes firms that experiences negative earnings.
5.2.2. Investor protection
Investor protection can be measured in various ways. Investor protection is measured by taking the revised anti-self-dealing index by Djankov et al. (2008) which ranks the countries from zero (weak investor protection) to five (strong investor protection). In order to differentiate strong and weak investor protection rights, I follow Byrne and O’Connor (2012) that categorizes strong investor countries if they are above the median and weak if they are below. Since weak investor protection is of interest in my study, it is referred as WIP in the regression models which is a binary variable that takes 1 for weak investor protection and 0 otherwise.
5.2.3. Creditor rights
in my study, it is referred as WCRED in the regression models which is a binary variable that takes 1 for weak creditor rights and 0 otherwise.
5.3. Control variables
5.4. Multivariate analysis
5.4.1. Earnings and payout ratios
In order to test the first hypothesis, the following regression model is used. !"#$% '$(")#*,,
= ./+ .123456578*,,91+ .:86;2*,,91+ .<=/?*,,91 + .@42*,,91 + .AB2C24372*,,91+ .D EFCGB*,,91+ .HE38I*,,91+ .J7KLM,,91 + N.ME")O#P( E"O#P"% + N.Q6OR)S#P( E"O#P"% + N.,TU$P E"O#P"% + V* ,
where i is a subscript for each firm, j is a subscript for each industry, k for each country and t for each year. All independent variables are lagged by one year to reduce possible problems of simultaneous causality. Further, I introduce country, year, and industry fixed effects to reduce endogeneity issues. Also, all continuous variables are winsorized at 1% and 99% to reduce the impact of outliers. Based on H1, I expect level of earnings (.1) to be positive.
5.4.2. Investor protection as moderating effect
Equation (2) introduces the country-level moderator investor protection as interaction variable. In order to do so, I use the revised anti-self-dealing index by Djankov et al. (2008). !"#$% '$(")#*,,
= ./+ .123456578*,,91+ .:W6LM + .<23456578*,,91∗ W6LM + .@86;2*,,91+ .A=/?*,,91 + .D42*,,91+ .HB2C24372*,,91
+ .J EFCGB*,,91+ .YE38I*,,91+ .1/7KLM,,91+ N.Q6OR)S#P( E"O#P"%
+ N.,TU$P E"O#P"% + V* , (2)
in equation (2), the variable of interest is .<, the interaction term of EARNINGS and
WIP. I expect the interaction term to be negative meaning that the relation between EARNINGS
and PAYOUT is less positive in countries with weaker investor protection.
5.4.3. Creditor rights as moderating effect
Lastly, to better understand the influence of country-level creditor rights, this variable is added to the regression model. The regression is denoted as follows:
!"#$% '$(")#*,,
= ./+ .123456578*,,91+ .:WE42KM+ .<23456578*,,91∗ WE42KM + .@86;2*,,91+ .A=/?*,,91 + .D42*,,91+ .HB2C24372*,,91
+ .J EFCGB*,,91+ .YE38I*,,91+ .1/7KLM,,91+ N.Q6OR)S#P( E"O#P"% + N.,TU$P E"O#P"% + V* ,
in equation three, the variable of interest is .<, the interaction term of EARNINGS and
WCRED. According to H3, I expect the interaction term to be negative meaning that the
relation between EARNINGS and PAYOUT is alleviated in weak creditor right countries.
5.5. Logistic regression
In addition, I employ logit test the propensities of PAYOUT, DIVIDENDS and
REPURCHASES. By changing the dependent variables into binary variables, the benefit is that
it allows to regularize the sample and presents the propensities in the prediction results. In order to do so, DIVIDENDS and REPURUCHASES are denoted with 1 if firms pay dividends or make use of repurchase programs and with 0 otherwise. PAYOUT is denoted with 1 if firms have dividends or share repurchase payouts or both together and with 0 otherwise. All other control variables used for the OLS regression remain the same.
6. Results
6.1. Descriptive statistics
Table 3 presents the summary statistics of variables used in this study and reports the total observations, median, mean, standard deviation, minimum and maximum of all relevant variables in this study. Table 4 shows the mean of the relevant variables, namely PAYOUT,
DIVIDENDS, REPURCHAES as well as the country-level variables of investor protection and
creditor rights. In addition, the number of firms by country is presented. Generally, I observe Table 3. Summary statistics
Variable Median Mean Std. Dev. Min Max
L3TGZ!,91 0.016 0.038 0.059 0 0.331 23456578,91 0.049 0.040 0.117 -0.597 0.320 K6C6K25K8,91 0.0097 0.022 0.036 0 0.219 42LZ4EI3828,91 0 0.015 0.039 0 0.238 86;2,91 14.212 14.029 2.033 6.434 18.162 =/?,91 0.520 0.515 0.219 0.046 1.155 42,91 0.550 0.308 1.433 -8.608 4.637 B2C24372,91 0.149 0.174 0.162 0 0.724 E38I,91 0.101 0.156 0.161 0 0.794 EFCGB,91 0.027 0.055 0.083 0.002 0.539 7KL,91 10.694 10.480 0.707 6.112 11.543 Investor Protection 3.5 3.658 0.777 2 5 Creditor Rights 1 1.703 1.040 0 4
Notes: This table reports the summary statistics of all variables. There are 42,148
firm-year observations for all variables. PAYOUT is the sum of DIVIDENDS and
REPURCHASES. DIVIDENDS are the total dividends paid by the firm scaled by total
assets. REPURCHASES are the share repurchases by the firm scaled by total assets.
EARNINGS is net income scaled by total assets. SIZE is the natural logarithm of total
Table 4. Mean by country
Country Firm-year
observations Payout Dividends Repurchases Earnings Protection Investor Creditor Rights
Argentina 195 0.023 0.022 0.001 0.057 2 1 Austria 148 0.027 0.022 0.005 0.045 2.5 3 Australia 3491 0.039 0.033 0.005 -0.009 4 3 Belgium 303 0.044 0.034 0.010 0.045 3 2 Brazil 491 0.039 0.034 0.003 0.042 5 1 Canada 2246 0.031 0.024 0.006 0.019 4 1 Chile 315 0.032 0.031 0.000 0.048 4 2 Colombia 80 0.020 0.020 0.000 0.038 3 0 Denmark 360 0.066 0.031 0.035 0.071 4 3 Finland 421 0.048 0.042 0.005 0.057 3.5 1 France 1194 0.023 0.017 0.005 0.035 3.5 0 Germany 1145 0.028 0.022 0.005 0.042 3.5 3 Greece 124 0.043 0.033 0.002 0.057 2 1 India 245 0.036 0.035 0.000 0.095 5 2 Indonesia 377 0.055 0.046 0.002 0.099 4 2 Ireland 145 0.034 0.023 0.011 0.080 5 1 Israel 135 0.046 0.038 0.008 0.069 4 3 Italy 435 0.028 0.024 0.003 0.037 2 2 Japan 5029 0.016 0.011 0.005 0.035 4.5 2 Kenya 9 0.078 0.078 0.000 0.160 2 4 Malaysia 518 0.058 0.048 0.003 0.086 5 3 Mexico 432 0.031 0.023 0.007 0.055 3 0 Netherlands 183 0.026 0.016 0.010 0.038 2.5 3 New Zealand 361 0.053 0.049 0.004 0.070 4 4 Norway 354 0.036 0.026 0.010 0.033 3.5 2 Pakistan 16 0.030 0.030 0.000 0.114 4 1 Peru 269 0.035 0.031 0.002 0.070 3.5 0 Philippines 61 0.032 0.030 0.002 0.074 4 1 Portugal 141 0.033 0.028 0.004 0.030 2.5 1 Singapore 384 0.045 0.040 0.005 0.070 5 3 South Africa 908 0.047 0.039 0.007 0.086 5 3 Spain 342 0.047 0.036 0.008 0.053 5 2 Sri Lanka 13 0.009 0.009 0.000 0.054 4 2 Sweden 900 0.044 0.034 0.009 0.065 3.5 1 Switzerland 698 0.044 0.029 0.015 0.054 3 1 Thailand 273 0.049 0.045 0.000 0.079 4 2 Turkey 231 0.028 0.028 0.000 0.071 3 2 United Kingdom 2687 0.043 0.032 0.010 0.061 5 4 United States 16489 0.045 0.014 0.030 0.038 3 1
Notes: This table reports the mean by country. All company financial variables are winsorized
that in countries with strong investor protection, creditor rights are also strong and vice versa, suggesting a positive correlation. However, consistent with findings in previous studies (Byrne and O’Connor, 2012), there are some exceptions. For instance, Canada has strong investor protection but weak creditor rights. Similarly, France has also rather strong investor protection while having weak creditor rights.
Table 5 presents the univariate analysis that tests the significance level of the difference of strong and weak investor protection in form of a t-test. While strong investor protection countries have significantly higher DIVIDEND payout, they report a lower payout through
REPURCHASES which makes the PAYOUT significantly higher for weak investor protection
countries. Also, firms in weak investor protection countries are larger in SIZE, have a higher
M/B and RE. Also, in Table 5, strong and weak creditor rights are tested (t-test) and a similar
Table 5. Univariate analysis of investor protection and creditor rights Panel A. T-test of strong and weak investor protection
Variables Strong investor
protection Weak investor protection Mean difference (t-test) L3TGZ!,91 0.034 0.041 -0.007*** 23456578,91 0.038 0.041 -0.002* K6C6K25K8,91 0.027 0.017 0.009*** 42LZ4EI3828,91 0.006 0.022 -0.016*** 86;2,91 13.795 14.210 -0.415*** =/?,91 0.498 0.528 -0.030*** 42,91 0.255 0.349 -0.094*** B2C24372,91 0.155 0.189 -0.034*** E38I,91 0.144 0.165 -0.021*** EFCGB,91 0.065 0.048 0.017*** 7KL,91 10.252 10.657 -0.405***
Panel B. T-test of strong and weak creditor rights
Variables Strong creditor
rights Weak creditor rights Mean difference (t-test) L3TGZ!,91 0.034 0.041 -0.007*** 23456578,91 0.041 0.039 0.002 K6C6K25K8,91 0.027 0.018 0.009*** 42LZ4EI3828,91 0.006 0.022 -0.016*** 86;2,91 14.027 14.030 -0.003 =/?,91 0.517 0.514 0.003 42,91 0.330 0.291 0.040** B2C24372,91 0.155 0.189 -0.035*** E38I,91 0.146 0.163 -0.016*** EFCGB,91 0.059 0.052 0.007*** 7KL,91 10.286 10.629 -0.343***
Notes: This table reports the two-tailed t-test investor protection and creditor rights.
Table 6. Correlation matrix VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (1) !"#$%&'() 1.000 (2) *"+,-,./'() 0.394*** 1.000 (3) 0-1-0*,0/'() 0.696*** 0.370*** 1.000 (4) +*!%+ − 34"/*/'() 0.734*** 0.216*** 0.051*** 1.000 (5) /-5*'() 0.066*** 0.301*** 0.017*** 0.090*** 1.000 (6) 6/8'() -0.016** -0.065*** -0.028*** -0.001 0.423*** 1.000 (7 ) +*'() 0.138*** 0.397*** 0.105*** 0.105*** 0.342*** 0.066*** 1.000 (8) 9*1*+".*'() -0.043*** -0.093*** -0.062*** -0.008 0.199*** 0.675*** -0.002 1.000 (9) 3"/4'() 0.082*** -0.131*** 0.005 0.105*** -0.354*** -0.356*** -0.218*** -0.329*** 1.000 (10) 3:1$9'() -0.044*** -0.276*** -0.030*** -0.046*** -0.422*** -0.159*** -0.221*** -0.073*** 0.188*** 1.000 (11) .0!'() 0.020*** -0.086*** -0.111*** 0.138*** 0.028*** -0.027*** -0.018*** 0.042*** 0.068*** 0.052*** 1.000 (12) Weak investor Protection 0.059 *** 0.010* -0.127*** 0.202*** 0.101*** 0.067*** 0.033*** 0.105*** 0.065*** -0.099*** 0.284*** 1.000 (13) Weak creditor Rights 0.062 *** -0.009 -0.122*** 0.201*** 0.001 -0.007 -0.014** 0.107*** 0.049*** -0.043*** 0.240*** 0.722*** 1.000
Notes: Table 6 reports the correlation matrix for all the variables used in the regressions. All company financial variables are winsorized at 1% and 99%
6.2. Payout, earnings, investor protection and creditor rights
Table 7 presents the OLS regression results with PAYOUT as dependent variable. In model (1) the coefficient of EARNINGS is positive (0.191) and significant at 1% level. This supports hypothesis 1 and is also consistent with the findings by Skinner (2008). In model (2) to (5) the interaction with weak creditor rights (WCRED) and weak investor protection (WIP) is however insignificant. I interpret that it is caused due to opposing forces of REPURCHASES (positive) and DIVIDENDS (negative), which is further elaborated later in this study. The control variables are generally in line with expectations. SIZE, RE and CASH have positive coefficients meaning that firms that are larger, have more retained earnings and cash holding pay out more. In addition, M/B is positive which is rather surprising, nevertheless, Alzahrani and Lasfer (2012) finds similar results for M/B. Also, contrary to my expectations, CFVol has a positive coefficient that more cashflow volatility increases PAYOUT. However, once tested with the logit regression in Table 8, CFVol and M/B turn to be negative. Consistent with my expectation, LEVERAGE has a negative coefficient which shows that more indebtedness leads to a lower PAYOUT.
Table 8 shows the results from the logit regression. In model (1) the relation between
EARNINGS and PAYOUT is tested and the test confirms a positive and significant relation
between EARNINGS and PAYOUT. This further supports hypothesis 1. In model (2) the country-level variable of investor protection is introduced. As shown in model (3) the
interaction term of earnings and weak investor protection (WIP) is negative (-2.068) and
significant at the 1% level meaning that the main relation is less positive in countries with weak
investor protection. In other words, this supports the outcome model proposed by La Porta et
al. (2000) and also hypothesis 2 that in countries with weak investor protection the strong
relation between earnings and payout is alleviated. In model (4) the country-level creditor
Table 7. OLS regression earnings on payout
This table shows the OLS regression that tests the impact of earnings on payout. The dependent variable is a continuous variable that is the sum of dividend and share repurchases scaled by total assets. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 0.191*** 0.204*** 0.193*** 0.204*** 0.200*** (0.008) (0.009) (0.012) (0.009) (0.013) WIP 0.006*** 0.005*** (0.001) (0.001) WIP * !"#$%$&'()* (0.014) 0.021 WCRED 0.008*** 0.008*** (0.001) (0.001) WCRED * !"#$%$&'()* (0.015) 0.007 '%+!()* 0.002*** -0.0003 -0.0003 -0.0002 -0.0002 (0.000) (0.000) (0.000) (0.000) (0.000) ,/.()* 0.016*** 0.011*** 0.012*** 0.013*** 0.013*** (0.004) (0.004) (0.004) (0.004) (0.004) #!()* 0.001* 0.0005 0.0005 0.001 0.001 (0.000) (0.000) (0.000) (0.000) (0.000) /!0!#"&!()* -0.032*** -0.019*** -0.019*** -0.022*** -0.022*** (0.005) (0.004) (0.004) (0.005) (0.005) 1"'2()* 0.060*** 0.053*** 0.053*** 0.053*** 0.053*** (0.004) (0.004) (0.004) (0.004) (0.004) 1304/()* 0.034*** 0.046*** 0.046*** 0.048*** 0.048*** (0.008) (0.008) (0.008) (0.008) (0.008) &56()* -0.007*** 0.001 0.001 0.001 0.001 (0.002) (0.001) (0.001) (0.001) (0.001) Constant 0.020 -0.013 -0.011 -0.016 -0.015 (0.021) (0.011) (0.011) (0.011) (0.011) Number of observations 42148 42148 42148 42148 42148 R2 0.272 0.242 0.243 0.245 0.245
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 8. Logit regression earnings on payout
This table shows the logit regression that tests the impact of earnings on payout, defined as the sum of dividends and share repurchases. The dependent variable is a binary variable which takes 1 if the firms report a payout and 0 otherwise. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 5.683*** 5.612*** 6.905*** 5.558*** 7.412*** (0.305) (0.301) (0.525) (0.300) (0.610) WIP -1.000*** -0.934*** (0.079) (0.084) WIP * !"#$%$&'()* -2.068 *** (0.640) WCRED -0.905*** -0.817*** (0.074) (0.080) WCRED * !"#$%$&'()* -2.764 *** (0.699) '%+!()* 0.468*** 0.485*** 0.487*** 0.466*** 0.468*** (0.025) (0.023) (0.023) (0.022) (0.023) ,/.()* -0.573** -0.304 -0.331 -0.517** -0.529** (0.231) (0.230) (0.230) (0.229) (0.229) #!()* 0.164*** 0.173*** 0.174*** 0.159*** 0.161*** (0.020) (0.020) (0.020) (0.020) (0.020) /!0!#"&!()* -0.733*** -1.107*** -1.104*** -0.946*** -0.967*** (0.267) (0.266) (0.266) (0.268) (0.269) 1"'2()* -0.220 -0.183 -0.205 -0.355* -0.370* (0.203) (0.204) (0.203) (0.200) (0.200) 1304/()* -2.336*** -2.143*** -2.145*** -2.476*** -2.471*** (0.434) (0.437) (0.439) (0.432) (0.436) &56()* -0.477*** 0.003 0.010 -0.043 -0.033 (0.171) (0.055) (0.055) (0.054) (0.054) Constant -0.530 -3.941*** -4.050*** -2.987*** -3.184*** (1.723) (1.052) (1.053) (0.931) (0.952) Number of observations 41815 41898 41898 41898 41898 R2 0.321 0.304 0.305 0.302 0.303
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Year dummies Yes Yes Yes Yes Yes
interaction is observed in model (5). The interaction term is significant at the 1% level meaning
that in weak creditor right countries the relation of earnings and payout is less positive. This
supports hypothesis 3 and is consistent with the findings by Brockman and Unlu (2009). All
three hypothesis are supported by the logit regression. Important to note is that overall there
are more firms paying dividends (28,815) than repurchases (19,210). The interaction in the OLS is insignificant and while in the logit regression significant. I interpret that this is due to the opposite forces by the country-level moderators (negative for DIVIDENDS and positive for
REPURCHASES) as shown in Tables 9 and 12. In the OLS the negative and positive effects is
offset while in the logit regression the negative effect of DIVIDENDS is stronger and therefore the interactions remain negative and significant. Since the dependent variable PAYOUT is the sum of share repurchases and dividends, it is of interest to observe the driving forces behind
the negative interactions. Therefore, dividends and share repurchases are tested separately as
dependent variables.
6.3. Dividend, earnings, investor protection and creditor rights
Table 9. OLS regression earnings on dividend
This table shows the OLS regression that tests the impact of earnings on dividends. DIVIDENDS is the dependent variable which is a continuous variable scaled by total assets. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 0.117*** 0.127*** 0.159*** 0.127*** 0.167*** (0.006) (0.006) (0.009) (0.006) (0.010) WIP -0.008*** -0.005*** (0.001) (0.001) WIP * !"#$%$&'()* -0.059 *** (0.010) WCRED -0.006*** -0.004*** (0.001) (0.001) WCRED * !"#$%$&'()* -0.070 *** (0.010) '%+!()* -0.0005* -0.002*** -0.002*** -0.002*** -0.002*** (0.000) (0.000) (0.000) (0.000) (0.000) ,/.()* 0.010*** 0.014*** 0.012*** 0.012*** 0.011*** (0.003) (0.003) (0.003) (0.003) (0.003) #!()* -0.000 -0.0002 -0.0002 -0.0003 -0.0002 (0.000) (0.000) (0.000) (0.000) (0.000) /!0!#"&!()* -0.017*** -0.018*** -0.018*** -0.017*** -0.017*** (0.003) (0.003) (0.003) (0.003) (0.003) 1"'2()* 0.019*** 0.011*** 0.009*** 0.010*** 0.008*** (0.003) (0.003) (0.003) (0.003) (0.003) 1304/()* 0.009 0.018*** 0.019*** 0.017*** 0.018*** (0.006) (0.006) (0.006) (0.006) (0.006) &56()* -0.001 -0.003*** -0.002*** -0.003*** -0.003*** (0.002) (0.001) (0.001) (0.001) (0.001) Constant 0.015 0.060*** 0.054*** 0.067*** 0.059*** (0.017) (0.009) (0.008) (0.009) (0.008) Number of observations 42148 42148 42148 42148 42148 R2 0.284 0.238 0.247 0.236 0.248
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 10. Logit regression earnings on dividend
This table shows the logit regression that tests the impact of earnings on dividends. The dependent variable is a binary variable which takes 1 if the firms report dividend payouts and 0 otherwise. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 7.577*** 7.137*** 9.457*** 7.021*** 10.602*** (0.420) (0.392) (0.666) (0.389) (0.797) WIP -1.768*** -1.599*** (0.087) (0.093) WIP * !"#$%$&'()* -3.698 *** (0.808) WCRED -1.653*** -1.420*** (0.077) (0.085) WCRED * !"#$%$&'()* -5.276 *** (0.901) '%+!()* 0.430*** 0.439*** 0.443*** 0.408*** 0.415*** (0.029) (0.025) (0.026) (0.025) (0.025) ,/.()* -0.256 0.500* 0.453* 0.089 0.074 (0.265) (0.260) (0.262) (0.256) (0.258) #!()* 0.282*** 0.275*** 0.276*** 0.243*** 0.245*** (0.030) (0.028) (0.028) (0.027) (0.027) /!0!#"&!()* -0.464 -1.424*** -1.422*** -1.088*** -1.134*** (0.297) (0.292) (0.293) (0.291) (0.292) 1"'2()* -1.846*** -1.801*** -1.790*** -2.010*** -1.995*** (0.260) (0.252) (0.254) (0.246) (0.250) 1304/()* -3.230*** -2.903*** -2.904*** -3.399*** -3.383*** (0.529) (0.536) (0.545) (0.511) (0.525) &56()* -0.221 -0.181*** -0.178*** -0.271*** -0.263*** (0.163) (0.062) (0.063) (0.063) (0.063) Constant -1.961 -1.386 -1.554 0.484 0.111 (1.592) (1.172) (1.155) (0.991) (1.016) Number of observations 41994 42077 42077 42077 42077 R2 0.373 0.331 0.333 0.328 0.331
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 10 presents the results of the logit regression. Similar to the logit regression and as expected the relation between EARNINGS and DIVIDENDS is positive (7.577) and significant at the 1% level in model (1). In model (3) when the interaction with weak investor protection (WIP) is introduced it shows a negative (-3.698) and significant (at the 1% level) interaction term meaning that the main relation between dividend and earnings is less positive in weak investor protection countries. This again supports the findings by La Porta et al. (2000) that weak investor protection countries have less dividend payouts compared to strong investor protection countries consistent with the outcome model of dividends. In model (5) the country-level creditor rights are introduced and the interaction between EARNINGS and weak creditor rights (WCRED) is negative (-5.276) and significant at the 1% level. This again supports the findings by Brockman and Unlu (2009) that due to dividend restriction in weak creditor rights
countries dividend payouts are less strong. The findings when DIVIDENDS is used as
dependent variable are similar with the results when PAYOUT is used. However, it is important
that the probabilities of the interaction terms are more negative in case of DIVIDENDS as dependent variables.
6.4. Share repurchases, earnings, investor protection and creditor rights
Table 11 illustrates the results using the OLS regression models when REPURCHASES
is used as dependent variable. In model (1) a positive (0.060) and significant (at 1% level)
EARNINGS coefficient is observed. In model (3) a positive (0.083) and significant (at 1% level)
interaction term with weak investor protection (WIP) and EARNINGS is observed. This is
evidence of the substitution model of La Porta et al. (2000) which has the premise that the
payout in weak investor protection is higher since firms in these countries use it as reputation
building. By paying dividends, firms in weak investor protection countries show a willingness
to reduce the chance for expropriation since less cash is available after paying dividends. Firms
Table 11. OLS regression earnings on share repurchases
This table shows the OLS regression that tests the impact of earnings on share repurchases. Share repurchase (REPURCHASES) is the dependent variable which is a continuous variable scaled by total assets. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 0.060*** 0.064*** 0.020*** 0.064*** 0.019*** (0.005) (0.005) (0.005) (0.005) (0.005) WIP 0.013*** 0.009*** (0.001) (0.001) WIP * !"#$%$&'()* 0.083*** (0.007) WCRED 0.014*** 0.011*** (0.001) (0.001) WCRED * !"#$%$&'()* 0.079*** (0.007) '%+!()* 0.003*** 0.002*** 0.002*** 0.002*** 0.002*** (0.000) (0.000) (0.000) (0.000) (0.000) ,/.()* 0.003 -0.005 -0.003 -0.001 0.000 (0.003) (0.003) (0.003) (0.003) (0.003) #!()* 0.001*** 0.001** 0.001** 0.001*** 0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) /!0!#"&!()* -0.012*** 0.001 0.001 -0.003 -0.003 (0.003) (0.003) (0.003) (0.003) (0.003) 1"'2()* 0.036*** 0.036*** 0.039*** 0.038*** 0.040*** (0.003) (0.003) (0.003) (0.003) (0.003) 1304/()* 0.018*** 0.021*** 0.021*** 0.024*** 0.023*** (0.004) (0.004) (0.004) (0.004) (0.004) &56()* -0.007*** 0.005*** 0.004*** 0.005*** 0.004*** (0.001) (0.000) (0.000) (0.000) (0.000) Constant 0.013 -0.077*** -0.069*** -0.086*** -0.078*** (0.011) (0.005) (0.005) (0.006) (0.006) Number of observations 42148 42148 42148 42148 42148 R2 0.201 0.160 0.174 0.166 0.179
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 12. Logit regression earnings on share repurchases
This table shows the logit regression that tests the impact of earnings on share repurchases. The dependent variable is a binary variable which takes 1 if the firms report share repurchases payouts and 0 otherwise. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 2.776*** 2.020*** 0.678** 2.024*** 0.674** (0.233) (0.230) (0.304) (0.229) (0.335) WIP 0.181*** 0.062 (0.056) (0.060) WIP * !"#$%$&'()* 2.290 *** (0.446) WCRED 0.403*** 0.286*** (0.054) (0.058) WCRED * !"#$%$&'()* 2.197 *** (0.451) '%+!()* 0.321*** 0.377*** 0.372*** 0.385*** 0.379*** (0.019) (0.019) (0.019) (0.019) (0.018) ,/.()* -0.434** -0.822*** -0.792*** -0.701*** -0.686*** (0.174) (0.172) (0.172) (0.172) (0.172) #!()* 0.086*** 0.107*** 0.104*** 0.111*** 0.108*** (0.016) (0.019) (0.019) (0.019) (0.019) /!0!#"&!()* -0.540*** -0.147 -0.157 -0.364* -0.358* (0.205) (0.201) (0.200) (0.202) (0.202) 1"'2()* 0.705*** 1.153*** 1.186*** 1.148*** 1.177*** (0.160) (0.157) (0.158) (0.157) (0.158) 1304/()* -0.955** -1.400*** -1.395*** -1.330*** -1.324*** (0.407) (0.413) (0.413) (0.413) (0.413) &56()* 0.210 0.610*** 0.594*** 0.580*** 0.565*** (0.139) (0.047) (0.047) (0.045) (0.045) Constant -8.432*** -11.402*** -11.115*** -11.404*** -11.088*** (1.420) (0.796) (0.794) (0.818) (0.819) Number of observations 42094 42132 42132 42132 42132 R2 0.197 0.142 0.143 0.146 0.147
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Model (5) shows a positive (0.079) and significant at 1% level interactions coefficients. To the
best of my knowledge this interaction between creditor rights and share repurchases has never
been done before and I explain this positive term that firms substitute dividends for share
repurchases due to the restrictions on dividends in weak creditor right countries. That is why,
firms in weak creditor right countries are more likely to repurchase shares. Table 12 presents
the results of the logit regression. The results are consistent with the OLS regression. In model (1) the relation is as expected positive (2.776) and significant at the 1% level. In model (2) and (3) the variable of weak investor protection (WIP) is added. The interaction term is positive
(2.290) and significant at the 1% level meaning that in weak investor protection countries the
probability of buying back shares is significantly higher. In model (4) and (5) the country-level
creditor rights (WCRED) are added and again a positive (2.197) and a significant (at the 1%
level) interaction term is observed.
6.5. Robustness tests
In order to test for robustness of my results, I use a balanced panel. In other words, I drop all companies that have missing firm-year observations. I end up with a total of 20,320 firm-year observations representing 1,230 firms. The tables for the robustness tests are in the appendix. Table 15 and 16 show the results for PAYOUT as dependent variable. Again,
EARNINGS is positive significant at the 1% level in the OLS as well as in the logit regression
Consistent with the previous regressions, Tables 17 and 18 in the appendix present similar results when DIVIDENDS is used as dependent variable. In the OLS as well as the logit regression, the interactions of EARNINGS and weak investor protection (WIP) is negative and significant at the 1% level. Also, the interaction of EARNINGS and weak creditor rights (WCRED) is negative and significant at the 1% level. Using REPURCHASES as the dependent variable, the balanced panel further confirms the positive and significant interactions of
EARNINGS with WIP and WCRED in Tables 19 and 20 in the appendix. Both interactions are
in the OLS as well as the logit regression positive and significant at the 1% level.
In sum, the balanced panel further supports the findings in this study by presenting a positive relation of EARNINGS with PAYOUT, DIVIDENDS and REPURCHASES. Further, in weak investor protection and creditor right countries, the relation between EARNINGS and
DIVIDENDS is less positive while the relation between EARNINGS and REPURCHASES is
more positive, consistent with previous results.
7. Conclusion
After a significant increase in the use of share repurchases on a global scale (Manconi et al., 2019), this study explores the relation of earnings on payout, dividends and repurchases. Consistent with previous studies (Skinner, 2008), a positive relation is observed when it is tested with logit as well as OLS. I extend previous studies by introducing two country-level moderators in form of weak investor protection and creditor rights. The results show that in weak investor protection countries, the relation of earnings with dividends is less positive, consistent with the outcome model by La Porta et al. (2000). In contrast, the relation of earnings and repurchases is positive and significant which is a sign that firms in weak investor protection countries are more likely to use repurchases.
protection countries that indicates that the restrictions put on payout lead to a less frequent use of dividends to pay out earnings. In order to compensate for this restriction, firms are more likely to use repurchases which is supported by the positive and significant relation. The results are also economically significant since they show a large difference on how share repurchases and dividends are used depending on the institutional setting.
Appendix
Table 13. Firm-level variables Firm-level
variable Symbol Description Source
Payout 6"849:()* Common dividend plus share
repurchases scaled by total assets at year-end
Worldscope
Dividends 5%0%5!$5'()* Total amount of dividends paid
by the firm scaled by total assets at year-end
Worldscope
Share repurchases #!69#12"'!'()* Total amount of repurchases
scaled by total assets at year-end
Worldscope
Earnings !"#$%$&'()* Net income scaled by total
assets at year-end
Worldscope
Size '%+!()* The natural logarithm of total
sales in US dollar at year-end
Worldscope
Market to book ratio
,/.()* Total assets−book value of
equity + market value of equity) divided by total assets at year-end
Worldscope & Datastream
Retained Earnings #!()* Retained earnings scaled by total equity at year-end
Worldscope Leverage /!0!#"&!()* The ratio of long-term debt
scaled with total assets at year-end
Worldscope & Datastream Cashflow
Volatility
1304/()* Standard deviation of operating
cash flow to total assets at year-end
Worldscope & Datastream
Cash 1"'2()* Ratio of cash and short-term
investments to total assets at year-end
Table 14. Country-level variables Country-level
variable Symbol Description Source
Weak investor protection
WIP To measure investor protection the revised anti-self-dealing index is taken into account. This index shows the level of investor protection by country. If the value is above median, it means that the country has strong investor protection and if it is below the median it means weak investor protection.
Djankov et al. (2008)
Weak creditor rights
WCRED This index shows the level of creditor rights by country. If the value is above median, it means that the country has strong creditor rights and if it is below the median it means weak creditor rights.
Djankov et al. (2007)
Table 15. Robustness OLS earnings on payout
This table shows the OLS regression that tests the impact of earnings on payout using a balanced panel. The dependent variable is a continuous variable that is the sum of dividend and share repurchases scaled by total assets. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 0.245*** 0.268*** 0.261*** 0.267*** 0.278*** (0.015) (0.017) (0.026) (0.017) (0.027) WIP 0.011*** 0.010*** (0.002) (0.002) WIP * !"#$%$&'()* (0.029) 0.013 WCRED 0.011*** 0.012*** (0.002) (0.002) WCRED * !"#$%$&'()* (0.031) -0.019 '%+!()* 0.002*** -0.001 -0.001 -0.0005 -0.0005 (0.001) (0.001) (0.001) (0.001) (0.001) ,/.()* 0.028*** 0.024*** 0.025*** 0.026*** 0.026*** (0.007) (0.007) (0.007) (0.007) (0.007) #!()* 0.001 0.001 0.001 0.001 0.001 (0.001) (0.001) (0.001) (0.001) (0.001) /!0!#"&!()* -0.038*** -0.024*** -0.024*** -0.026*** -0.026*** (0.008) (0.008) (0.008) (0.008) (0.008) 1"'2()* 0.066*** 0.057*** 0.057*** 0.058*** 0.057*** (0.007) (0.007) (0.007) (0.007) (0.007) 1304/()* 0.049*** 0.076*** 0.076*** 0.080*** 0.081*** (0.017) (0.018) (0.018) (0.018) (0.018) &56()* -0.005 -0.001 -0.001 -0.001 -0.001 (0.003) (0.002) (0.002) (0.002) (0.002) Constant -0.008 0.013 0.014 0.009 0.007 (0.033) (0.021) (0.020) (0.021) (0.020) Number of observations 20320 20320 20320 20320 20320 R2 0.321 0.284 0.284 0.285 0.285
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 16. Robustness logit earnings on payout
This table shows the logit regression using a balanced panel that tests the impact of earnings on payout, defined as the sum of dividends and share repurchases. The dependent variable is a binary variable which takes 1 if the firms report a payout and 0 otherwise. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 6.307*** 6.213*** 8.256*** 6.137*** 8.745*** (0.580) (0.570) (1.163) (0.571) (1.343) WIP -1.094*** -1.000*** (0.130) (0.141) WIP * !"#$%$&'()* -3.044 ** (1.322) WCRED -1.110*** -0.995*** (0.133) (0.146) WCRED * !"#$%$&'()* -3.663 ** (1.468) '%+!()* 0.432*** 0.471*** 0.472*** 0.459*** 0.460*** (0.044) (0.042) (0.042) (0.041) (0.042) ,/.()* -0.211 0.040 0.015 -0.132 -0.143 (0.474) (0.496) (0.495) (0.489) (0.490) #!()* 0.193*** 0.198*** 0.196*** 0.190*** 0.189*** (0.034) (0.035) (0.035) (0.034) (0.034) /!0!#"&!()* -1.491*** -1.935*** -1.943*** -1.777*** -1.808*** (0.529) (0.539) (0.538) (0.544) (0.544) 1"'2()* -0.025 0.045 0.015 -0.067 -0.094 (0.367) (0.371) (0.370) (0.361) (0.362) 1304/()* -3.104*** -3.162*** -3.101*** -3.703*** -3.622*** (0.925) (0.958) (0.970) (0.952) (0.968) &56()* -0.270 0.128 0.143 0.125 0.140 (0.297) (0.113) (0.113) (0.111) (0.111) Constant -3.406 -5.768*** -6.012*** -5.494*** -5.755*** (2.638) (1.289) (1.299) (1.282) (1.292) Number of observations 19728 19984 19984 19984 19984 R2 0.331 0.298 0.300 0.298 0.300
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 17. Robustness OLS earnings on dividends
This table shows the OLS regression that tests the impact of earnings on dividends using a balanced panel. DIVIDENDS is the dependent variable which is a continuous variable scaled by total assets. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 0.138*** 0.154*** 0.207*** 0.155*** 0.221*** (0.010) (0.011) (0.019) (0.011) (0.020) WIP -0.007*** -0.002 (0.001) (0.001) WIP * !"#$%$&'()* -0.103 *** (0.020) WCRED -0.008*** -0.002 (0.001) (0.001) WCRED * !"#$%$&'()* -0.123 *** (0.021) '%+!()* -0.001* -0.003*** -0.002*** -0.003*** -0.002*** (0.000) (0.000) (0.000) (0.000) (0.000) ,/.()* 0.020*** 0.023*** 0.021*** 0.021*** 0.020*** (0.004) (0.005) (0.005) (0.005) (0.004) #!()* 0.001 -0.000 -0.000 -0.0001 -0.000 (0.000) (0.000) (0.000) (0.000) (0.000) /!0!#"&!()* -0.023*** -0.021*** -0.022*** -0.019*** -0.021*** (0.005) (0.005) (0.005) (0.005) (0.005) 1"'2()* 0.023*** 0.014*** 0.013*** 0.014*** 0.012** (0.004) (0.005) (0.005) (0.005) (0.005) 1304/()* -0.004 0.021* 0.028** 0.018 0.027** (0.011) (0.011) (0.011) (0.011) (0.012) &56()* 0.001 -0.005*** -0.004*** -0.005*** -0.004*** (0.003) (0.002) (0.001) (0.002) (0.001) Constant -0.003 0.087*** 0.075*** 0.088*** 0.074*** (0.027) (0.015) (0.014) (0.015) (0.013) Number of observations 20320 20320 20320 20320 20320 R2 0.355 0.279 0.299 0.282 0.311
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
Table 18. Robustness logit earnings on dividends
This table shows the logit regression that tests the impact of earnings on dividends using a balanced panel. The dependent variable is a binary variable which takes 1 if the firms report dividend payouts and 0 otherwise. Model 2 and 3 test the impact of weak investor protection while model 4 and 5 test the influence of weak creditor rights. All explanatory variables are lagged by one year. Standard errors are shown in parentheses and are clustered at firm-level. The sample period is 2001-2017. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
!"#$%$&'()* 7.372*** 6.790*** 12.354*** 6.697*** 14.093*** (0.731) (0.674) (1.290) (0.678) (1.518) WIP -1.932*** -1.595*** (0.138) (0.149) WIP * !"#$%$&'()* -7.909 *** (1.483) WCRED -1.974*** -1.562*** (0.135) (0.151) WCRED * !"#$%$&'()* -9.933 *** (1.671) '%+!()* 0.420*** 0.436*** 0.449*** 0.416*** 0.431*** (0.047) (0.043) (0.044) (0.042) (0.044) ,/.()* 0.354 0.902* 0.812* 0.573 0.516 (0.475) (0.482) (0.489) (0.476) (0.488) #!()* 0.329*** 0.309*** 0.305*** 0.293*** 0.292*** (0.056) (0.051) (0.052) (0.050) (0.052) /!0!#"&!()* -1.098** -1.894*** -1.878*** -1.615*** -1.655*** (0.509) (0.515) (0.518) (0.520) (0.525) 1"'2()* -1.678*** -1.597*** -1.640*** -1.805*** -1.852*** (0.440) (0.429) (0.444) (0.418) (0.439) 1304/()* -4.524*** -4.218*** -4.070*** -5.135*** -4.981*** (1.079) (1.117) (1.147) (1.046) (1.093) &56()* 0.059 -0.112 -0.093 -0.134 -0.120 (0.274) (0.134) (0.136) (0.133) (0.135) Constant -5.103** -2.511* -3.133** -1.822 -2.518* (2.412) (1.385) (1.416) (1.391) (1.431) Number of observations 20032 20208 20208 20208 20208 R2 0.382 0.331 0.338 0.332 0.342
2-digit SIC dummies Yes Yes Yes Yes Yes
Country dummies Yes No No No No
