SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

Dividend policies provide an opportune setting to examine how firms simultaneously manage the diverging interests of shareholders and creditors. Dividends ease shareholders' concerns about expropriation by insiders while exacerbating creditors' concerns about expropriation by shareholders. Firm insiders should set dividend policies to minimize the agency costs of equity and debt. Using a sample of 39 countries for 1991–2010, we find strong evidence that dividends are more positively sensitive to creditor (shareholder) rights when shareholders (creditors) are adequately protected. Our research emphasizes the importance of accounting for the interactions between both agency relationships when studying corporate policies.


I. Introduction

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

The principal–agent model of Jensen and Meckling (1976) distinguishes between two types of agency costs: the agency costs of equity arising from conflicts of interests between insiders and outside equity holders, and the agency costs of debt arising from conflicts between equity holders and debt holders. Both types of agency costs jeopardize the value of equity. This agency costs model has been extensively used in the finance literature to shed light on corporate decisions. However, although insightful, prior studies tend to examine one or the other of these two agency relationships, with little research done on the interactions between the two types of agency conflicts. We seek to fill this gap. We argue that firm insiders are likely to balance their decisions in favor of minority shareholders (creditors) if the agency costs of equity (debt) are more pronounced. To examine empirically the extent to which firms employ such a balancing strategy, we study firms' dividend policy.

In the context of corporate governance, dividend policy is a firm decision over which minority shareholders and creditors have conflicting interests. Simply speaking, minority shareholders would like to see more dividends, as dividends decrease minority shareholders' risk of expropriation by insiders, while creditors prefer less dividends, as dividends increase creditors' risk of expropriation by shareholders. Any change in dividend policy (either an increase or a decrease) is thus likely to be viewed favorably by one side while putting the other side on alert. This implies that firm insiders are unable to simultaneously reduce the agency costs of equity and the agency costs of debt merely by adjusting dividends. To minimize the total agency costs of equity and debt, insiders should balance shareholders' and creditors' interests by employing an optimal dividend policy. Under such a balancing strategy, when the insider–minority shareholder relationship is sound, insiders are expected to employ dividend policies in favor of creditors to reduce the agency costs of debt; in contrast, when the shareholder–creditor relationship is sound, insiders are expected to choose dividend policies that reduce the agency costs of equity.

To test for this balancing strategy, the researcher faces the question of how to capture the soundness of the two agency relationships. One solution is to rely on shareholder protection and creditor protection as proxies for the soundness of the agency relationships. Recent law-and-finance research documents that agency costs of equity and debt, and in turn dividend payouts, are shaped by legal protection.1 Two cross-country empirical studies are particularly influential in this line of research. La Porta et al. (2000) find that dividends are an outcome of the extent of shareholders' legal protection. This “outcome model” of dividends implies that shareholder rights establish the country-level balance of power between insiders and outside shareholders by allowing minority shareholders to extract more dividends out of firm insiders; that is, firms in countries with strong shareholder protections are more likely to issue dividends.2 In addition, Brockman and Unlu (2009) find that to reduce the agency costs of debt, managers pay lower dividends to substitute for weak creditor rights. This “substitute model” of dividends implies that creditor rights establish the country-level balance of power between insiders and creditors by loosening creditors' restrictions on dividend payouts; that is, firms in countries with strong creditor protection are more likely to issue dividends. These studies together suggest that shareholder protection and creditor protection are indicative of the soundness of the two agency relationships and that dividends are sensitive to variation in legal protection. We therefore expect that if a firm employs the balancing strategy, dividends are more sensitive to variation in legal protection of one agency relationship when the other relationship is more strongly protected by law, implying that dividends are employed to cater to the side for which agency problems are more pronounced.

Two fundamental questions arise in light of the findings of La Porta et al. (2000) and Brockman and Unlu (2009). First, if strong creditor protection allows firms to pay higher dividends as shown by Brockman and Unlu, do firms necessarily choose to exercise this discretion when facing less pressure for dividends from outside shareholders (e.g., due to weak shareholder rights)? In other words, does the substitute model hold under weak shareholder protection? Second, if strong shareholder protection leads to more pressure from outside shareholders for dividends as shown by La Porta et al., will firms cater to this pressure and pay more dividends when facing stringent constraints from creditors (e.g., due to weak creditor protection)? In other words, does the outcome model hold under poor creditor protection?

To address these questions, we empirically test a model of dividends that accounts for the interaction between the agency costs of equity and debt:3 shareholder rights (creditor rights) should shape the effect of creditor rights (shareholder rights) on dividends. We suggest that neither the substitute model nor the outcome model provides a full picture of dividends and agency costs. Rather, the substitute model (outcome model) should be more significant for firms under strong minority shareholder (creditor) protection. To elaborate, given an improvement in creditor rights, while firms gain more discretion to pay dividends (substitute model), firms are more likely to exercise this discretion when, ceteris paribus, shareholders are sufficiently protected that firms are pressured to disgorge free cash flows; otherwise, firms might choose not to increase dividends. That is, the quality of shareholder protection determines whether firms exercise their discretion, and hence the substitute model is more effective in explaining dividends under strong shareholder rights (Hypothesis 1). Similarly, under strong shareholder protection, although firms bear more pressure to disgorge cash to reduce the agency costs of equity (outcome model), they will cater to this pressure only if, ceteris paribus, creditors are sufficiently protected that they have flexibility to adjust dividend payments; otherwise, firms might choose not to increase dividends to avoid incurring higher agency costs of debt. That is, the outcome model is more effective in explaining dividends under strong creditor protection (Hypothesis 2).

The two hypotheses are illustrated in Figure I. The interaction effects of shareholder and creditor protection on dividends imply that when setting their dividend policies, firm insiders balance the interests of minority shareholders and creditors. More specifically, when the firm–minority shareholder relationship is sound due to strong shareholder protection, firms are more likely to use dividend policy as a means to improve the firm–creditor relationship, in which case dividend payments will be more sensitive to the quality of creditor protection. Conversely, when the shareholder–creditor relationship is sound due to strong creditor protection, firms are more likely to use dividends to ease the agency costs of equity, in which case dividend payments will be more sensitive to the quality of shareholder protection.

image

Figure I. Effectiveness of the Outcome Model and the Substitute Model. This figure graphs our arguments that the outcome model (the substitute model) is more effective when creditors (minority shareholders) are strongly protected.

Download figure to PowerPoint

To test the hypotheses on the interaction effects between shareholder and creditor rights, we use a sample of 139,168 firm-year observations from 39 countries for 1991–2010. The sample countries show wide variation in the degree of shareholder and creditor rights.4 For instance, 9 countries (Australia, Hong Kong, Israel, Malaysia, New Zealand, Singapore, South Africa, United Kingdom, and Zimbabwe) are of common law origin with creditor rights greater than or equal to the sample median (3), 7 countries (Canada, India, Ireland, Pakistan, Sri Lanka, Thailand, and United States) are of common law origin with creditor rights lower than the sample median, and 5 countries (Austria, Denmark, Germany, Netherlands, and South Korea) are of civil law origin with creditor rights greater than or equal to the sample median.5 It is this variation in shareholder and creditor rights that allows us to empirically test the interaction effects between shareholder and creditor rights.

We begin by examining the validity of the substitute model under different shareholder rights. Controlling for firm leverage, sales growth, firm maturity, profitability, size, and cash holdings, we find that the positive effect of creditor rights on the likelihood of paying dividends and on dividend payout ratios is more significant for firms from countries where minority shareholders have stronger rights. This implies that the validity of the substitute model predicted by Brockman and Unlu (2009) is conditional on strong shareholder protection. These results are robust to additional controls, potential sample selection biases, and alternative estimation procedures including annual regressions and country-mean regressions.

We further analyze the role of shareholder rights in determining the validity of the substitute model by examining whether the effect of creditor rights on dividend initiations and omissions varies across different shareholder rights. According to the substitute model, strong creditor rights should increase the likelihood of nonpayers to initiate a future dividend payment but, as shown by Brockman and Unlu (2009), reduce the likelihood of dividend payers to omit a future dividend payment. Our results show that strong creditor rights more dramatically increase the possibility of dividend initiations under strong shareholder rights and reduce the incidence of dividend omissions only under strong shareholder rights. These findings provide further evidence that the validity of the substitute model is conditional on strong shareholder protection.

Next, we extend the preceding analysis by examining the validity of the substitute model across firms with different characteristics. As in the case of strong shareholder protection, we hypothesize that the substitute model is more effective for firms facing higher demand for dividends from outside shareholders. Evidence from the prior dividend literature suggests that outside shareholders have more incentives to demand dividends from firms with a high earned-to-contributed equity mix or a high profitability. Consistent with our predictions, we find that firm maturity and profitability strengthen creditor rights' effect on dividend policy. We interpret these findings as implying that better creditor rights (i.e., lower agency costs of debt) are more likely to lead to higher dividend payments when firms face more pressure for dividends from minority shareholders.

Finally, using the same methodology as previously described, we examine the validity of the outcome model under different levels of creditor rights. We find that the positive effect of shareholder rights on the likelihood of paying dividends and on dividend ratios is more valid for countries with strong creditor rights protection. These findings are robust to a series of robustness tests and are further reinforced by analysis of dividend initiations and omissions.

In sum, by combining the agency costs of equity and debt, this study takes an important step toward a more comprehensive understanding of corporate dividend policies. We contribute to the dividend literature by showing that firms' dividend polices balance the interests of shareholders and creditors. Research focusing on only one agency relationship may therefore lead to inaccurate conclusions.

II. Data and Descriptive Statistics

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

Data Sources and Sample Selection

To construct our sample, we first obtain all firm-year observations from the Worldscope database available on Datastream, which we use to collect financial information for countries covered in La Porta et al. (1998) and Djankov, McLiesh, and Shleifer (2007)—our main sources for country-level shareholder and creditor protection variables. We then exclude (1) firms from countries with regulatory statutes on corporate dividend policies (La Porta et al., 1998), (2) firm-years with missing or abnormal data required for subsequent analyses,6 and (3) financial (6,000 ≤ Standard Industrial Classification [SIC] code ≤ 6,999) or utility (4,900 ≤ SIC code ≤ 4,999) firms. The sample contains 139,168 firm-year observations that correspond to 17,025 unique firms from 39 countries over the 1991–2010 period.

Variables and Descriptive Statistics

In this section, we define all the variables used in our empirical analysis. Numeric codes in parentheses are based on the Worldscope data manual.

Dependent Variables

We examine the effect of investor protection on the likelihood of paying dividends and on dividend payout ratios using logit and ordinary least squares (OLS) regressions. In the logit analysis, the dependent variable is a dividend-payer dummy, Payer, which equals 1 if total dividends paid are positive, and 0 otherwise. Total dividends are defined as the sum of cash common dividends (05376) and cash preferred dividends (05401) over the fiscal year.7 In the OLS analysis, the dependent variable is the dividend payout ratio, Div/sales, which we measure by scaling total dividends by net sales (01001).8 We also run Tobit regressions and OLS regressions removing firms with negative profit or cash flows, generating qualitatively same results.

Independent Variables

Our main country-level variables are indexes of shareholder rights and creditor rights. We employ two proxies for shareholder rights, namely, the legal origin dummy (Law) and the anti-self-dealing index (ASD) derived from Djankov et al. (2008). The legal origin captures the overall quality of shareholder protection and is employed by La Porta et al. (2000). We include ASD because it focuses on private enforcement mechanisms protecting outside shareholders, such as disclosure, approval, and litigation, governing a specific self-dealing transaction, and thus proxies the pressure on managers to disgorge excess cash.9 We measure creditor rights using creditor protection index (CR) from Djankov, McLiesh, and Shleifer (2007).

In all our regressions, we control for firm-level variables that are shown by prior studies to affect dividend policy (e.g., Brockman and Unlu 2009): (1) leverage (Debt), the book value of total liabilities (03351) scaled by the book value of total assets (02999); (2) firm size (Size), the natural logarithm of the dollar value of total book value assets (07230); (3) profitability (Profit), EBIT (18191) scaled by net sales (01001); (4) sales growth (Growth), the logarithmic value of the growth in net sales (01001), calculated as log (datat/datat–1); (5) retained earnings (RE/TE), the ratio of retained earnings (03495) to common equity (03501); and (6) cash holdings (Cash), cash and short-term investments (02001) scaled by the book value of total assets (02999). The predicted signs for the above control variables are as follows: Debt (+/−), Size (+), Growth (−), Profit (+), RE/TE (+) and Cash (+/−).

Summary Statistics

Panel A of Table 1 presents summary statistics. To reduce the influence of outliers, all firm-level variables are winsorized at the 1% and 99% levels. We observe that 67.3% of the firm-years are dividend payers.10 Variation in the other variables does not suggest a selection bias. For example, the 25%, 50%, and 75% percentiles of Debt are 0.347, 0.511, and 0.652, respectively.

Table 1. Summary Statistics
Panel A. Firm-Level Data
 NMeanMedian25%75%
Payer139,1680.6731.0000.0001.000
Div/sales139,1680.0200.0070.0000.021
Debt139,1680.5000.5110.3470.652
Size139,16812.59112.48211.16413.921
Growth139,1680.0850.061−0.0360.180
Profit139,168−0.0480.0650.0200.129
RE/TE139,168−0.0540.3090.0340.608
Cash139,1680.1460.1000.0410.199
Owner139,16832.8431.641.1855.00
Panel B. Number of Observations by Year
YearNYearNYearNYearN
19912,27619963,81520018,225200610,167
19922,54719974,36820029,22620079,465
19932,67219984,92520039,662200812,314
19942,86819995,559200410,092200912,342
19953,09320007,011200510,63720107,904
Panel C. Industry Distribution.
SIC Industry Definition2-Digit SICNSIC Industry Definition2-Digit SICN
Agriculture, forestry, and fisheries<101,448Transportation, communications40–489,520
Mineral industries10–147,391Wholesale trade50–5110,129
Construction industries15–176,357Retail trade52–599,305
Manufacturing20–3973,345Service industries>= 7021,673
Panel D. Country-Level Data
CountryASDCRNPayerDiv/sales
Note
  1. Panel A presents summary statistics for the firm-level variables. Panels B and C present the distribution of observations by year and industry, respectively. Panel D presents country-level summary statistics for shareholder rights, creditor rights (as of 2010), the proportion of payers and the country mean Div/sales ratio. The sample spans 1991–2010. Dividend is the sum of cash common dividends (05376) and cash preferred dividends (05401) issued over the calendar year. Payer equals 1 if the firm pays a Dividend; otherwise, it is equal to 0. Div/sales is the ratio of Dividend to net sales (01001). Debt is the book value of total liabilities (03351) scaled by the book value of total assets (02999). Size is the natural logarithm of dollar value of book value of total assets (07230). Growth is the logarithmic value of net sales (01001) growth calculated as log (datat/datat–1). Profit is EBIT (18191) scaled by net sales (01001). RE/TE is retained earnings (03495) scaled by common equity (03501). Cash is cash and short-term investments (02001) scaled by the book value of total assets (02999). Owner is the percentage of closely held shares (08021). CR refers to creditor rights index from Djankov et al. (2007). ASD is the anti-self-dealing index from Djankov et al. (2008).

Australia0.7636,51751.19%2.91%
Canada0.6414,60637.21%1.47%
Hong Kong0.9647,63955.52%3.07%
India0.5821,08677.72%1.82%
Ireland0.79145867.90%1.33%
Israel0.76391656.99%3.46%
Malaysia0.9536,74162.26%2.46%
New Zealand0.95475874.14%5.11%
Pakistan0.41156374.78%3.02%
Singapore1.0033,82165.77%3.19%
South Africa0.8132,48969.95%2.73%
Sri Lanka0.39238173.49%2.90%
Thailand0.81280061.88%2.97%
United Kingdom0.95411,81966.05%2.09%
United States0.65117,05467.09%1.98%
Zimbabwe0.39412568.00%1.60%
Common law median0.7731,78866.50%2.81%
Argentina0.34154446.51%2.30%
Austria0.21370476.28%1.78%
Belgium0.54261965.75%2.39%
Denmark0.4631,33465.59%1.73%
Egypt0.20228884.03%10.57%
Finland0.4611,59584.64%2.89%
France0.3806,26466.17%1.47%
Germany0.2837,11957.28%1.28%
Indonesia0.6522,12953.78%1.93%
Italy0.4222,45571.20%2.18%
Japan0.50226,63488.91%0.82%
Mexico0.1701,25557.37%2.29%
Netherlands0.20367573.48%1.68%
Norway0.42287645.43%2.82%
Peru0.45081962.03%4.91%
Philippines0.2211,20939.54%1.95%
Portugal0.44154067.78%2.07%
South Korea0.4734,73068.33%0.85%
Spain0.3721,14971.45%2.79%
Sweden0.3313,02661.07%2.06%
Switzerland0.2712,20073.55%1.99%
Taiwan0.5625,54661.59%2.89%
Turkey0.4321,68546.35%2.50%
Civil law median0.4221,33465.75%2.06%

Panel B of Table 1 presents the sample distribution by year. The panel shows that the sample coverage improves over time, peaking in 2009. This evidence is consistent with Worldscope mainly including profitable firms in 1990s and expanding coverage to include nonprofitable firms in more recent years (Denis and Osobov 2008).

Panel C of Table 1 presents the sample distribution by industry sector. Manufacturing industries account for the largest number of observations (73,345), followed by service, trade, transportation, mineral, and construction industries, each with more than 6,000 firm-year observations.

Finally, Panel D of Table 1 provides country-level summary information. As is common in multinational studies, the United States, the United Kingdom, and Japan together account for more than 40% of the total observations. In the robustness section below, we show that our results are not driven by sample composition. We also observe that common law countries, as opposed to civil law countries, have higher payout ratios but indistinctive payout propensities based on their respective medians. Furthermore, we do not observe a consistent pattern of dividend payouts across countries. For example, Australia, which ranks high in terms of both shareholder rights and creditor rights, has a higher payout propensity and ratio than Canada, which ranks high on shareholder rights but low on creditor rights. However, India, which has strong shareholder rights and weak creditor rights, has a higher payout propensity but lower payout ratio than Australia. These complicated dividend patterns require further investigation using multivariate analysis.

III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

In this section, we examine Hypothesis 1, which posits that the relevance of the substitute model depends on the quality of shareholder protection.

The Substitute Model under Different Shareholder Rights: Main Evidence

In this section, we examine under different levels of shareholder rights the relation between creditor rights and two dividend policy variables: the propensity to pay cash dividends (Payer) and the dividend payout ratio (Div/sales). Our empirical strategy consists of estimating logit and OLS regressions of Payer and Div/sales, respectively, on creditor rights (CR) for subsamples of strong and weak shareholder protection. We split the subsamples according to legal origin and at the median score of ASD (0.5). Our first hypothesis predicts that the positive relation between creditor rights and dividend policy described by the substitute model is more likely to hold for firms operating under strong shareholder rights. In addition to the firm-level control variables discussed previously, all regressions include year and industry fixed effects (defined by the first SIC digit) with robust standard errors adjusted for clustering at the firm level. For robustness, we also cluster the errors at the country-year level to address possible correlations among firms that share the same country and year; the results of the tests that follow remain qualitatively unchanged.

Creditor Rights and the Likelihood of Paying Dividends

Table 2, Models 1 through 5, presents the results from estimating the likelihood of dividend payment using the following logit model (firm and time subscripts are omitted):

  • display math(1)
Table 2. Creditor Rights and Dividend Policy under Different Shareholder Rights
Independent VariablesDependent Variable = PayerDependent Variable = Div/sales
(Predicted Sign)(1) Full Sample(2) Law= 1(3) Law= 0(4) ASD > 0.5(5) ASD < = 0.5(6) Full Sample(7) Law= 1(8) Law= 0(9) ASD > 0.5(10) ASD < = 0.5
Note
  1. Models 1 through 5 are logit regressions with Payer as the dependent variable, where Model 1 is run on the full sample and Models 2 through 5 are run on subsamples differentiated by legal origin and the sample medians of ASD. Models 6 through 10 are ordinary least squares regressions with Div/sales as the dependent variable, where Model 6 is run on the full sample and Models 7 through 10 are run on subsamples differentiated by legal origin and the sample median of ASD. Payer equals 1 if the firm pays a Dividend, and 0 otherwise. Div/sales is the ratio of Dividend to net sales (01001). Dividend is the sum of cash common dividends (05376) and cash preferred dividends (05401) issued over the calendar year. Debt is the book value of total liabilities (03351) scaled by the book value of total assets (02999). Size is the natural logarithm of dollar value of book value of total assets (07230). Growth is the logarithmic value of net sales (01001) growth calculated as log (datat/datat–1). Profit is EBIT (18191) scaled by net sales (01001). RE/TE is retained earnings (03495) scaled by common equity (03501). Cash is cash and short-term investments (02001) scaled by the book value of total assets (02999). Law equals 1 for common law countries and 0 for civil law countries. CR is creditor rights from Djankov et al. (2007). ASD is the anti-self-dealing index from Djankov et al. (2008). The p-values reported in parentheses control for firm-level clustering effects.

  2. a

    Significant at the 1% level.

  3. b

    Significant at the 5% level.

N139,16866,3177285172,99866,170139,16866,3177285172,99866,170
CR0.316a (<.001)0.481a (<.001)0.044a (.068)0.497a (<.001)0.052b (.033)0.001a (<.001)0.005a (<.001)−0.003a (<.001)0.005a (<.001)−0.003a (<.001)
Law−0.533a (<.001)    0.006a (<.001)    
Debt (+/−)−0.729a (<.001)−0.013 (.904)−1.307a (<.001)−0.161 (.135)−1.504a (<.001)−0.042a (<.001)−0.044a (<.001)−0.041a (<.001)−0.047a (<.001)−0.037a (<.001)
Size (+)0.353a (<.001)0.341a (<.001)0.432a (<.001)0.358a (<.001)0.441a (<.001)0.002a (<.001)0.004a (<.001)0.001a (<.001)0.004a (<.001)0.001a (<.001)
Growth (−)−0.496a (<.001)−0.493a (<.001)−0.569a (<.001)−0.486a (<.001)−0.581a (<.001)−0.008a (<.001)−0.009a (<.001)−0.005a (<.001)−0.010a (<.001)−0.006a (<.001)
Profit (+)1.219a (<.001)1.265a (<.001)1.119a (<.001)1.337a (<.001)1.029a (<.001)0.006a (<.001)0.005a (<.001)0.008a (<.001)0.005a (<.001)0.007a (<.001)
Cash (+/−)−0.090a (.366)−0.651a (<.001)0.491a (.001)−0.291b (.021)0.145 (.378)0.013a (<.001)0.014a (<.001)0.015a (<.001)0.018a (<.001)0.000a (<.001)
RE/TE (+)1.054a (<.001)0.735a (<.001)1.883a (<.001)0.767a (<.001)1.782a (<.001)0.001a (<.001)0.001a (<.001)−0.000 (.825)0.001a (.004)0.000 (.305)
Intercept, year, & ind. fixed effectsYesYesYesYesYesYesYesYesYesYes
R20.3930.4070.4100.3960.4120.1190.1220.1320.1310.121

We begin in Model 1 by estimating the likelihood of paying dividends according to equation (1). In this regression, we include a set of firm-level control variables, industry dummies, year dummies, and the proxies for shareholder rights (Law) and creditor rights (CR). Consistent with Brockman and Unlu (2009), we find that CR has a positive and statistically significant coefficient (at the 1% level). This finding suggests that better creditor protection eases constraints on corporate dividend payments from creditors, as predicted by the substitute model. Turning to the shareholder rights index, we find that inconsistent with prior studies (e.g., Brockman and Unlu 2009), Law has a negative and significant coefficient (at the 1% level). We show in Section IV that shareholder rights only positively affect dividend policies under strong creditor rights (Hypothesis 2).

In Models 2 through 5 we conduct subsample analyses. Specifically, we bisect the full sample according to legal origin (Law) and at the median value of ASD (0.5) to test whether the validity of the substitute model depends on the quality of shareholder protection. Models 2 and 3 estimate the likelihood of paying dividends as a function of CR for strong shareholder rights (Model 2) and weak shareholder rights (Model 3) subsamples. For the subsample of firms from common law countries (Law= 1), CR continues to have a positive and statistically significant coefficient. In comparison, the effect of CR on dividends is substantially smaller in the subsample of firms from civil law countries: the coefficient on CR is lower (coefficient = 0.044 in Model 3 vs. 0.481 in Model 2), and it is both statistically and economically less significant (p-value = .068 in Model 3 vs. < .001 in Model 2; an increase in CR from 0 to 4 increases the likelihood of dividend payment from 77% to 80% in Model 3 vs. from 47% to 86% in Model 2, with all the other variables set to their median values). Additional tests indicate that the difference in the coefficient estimates on CR between Models 2 and 3 is statistically significant at the 1% level.

Models 4 and 5 of Table 2 present similar results using the alternative measure of shareholder rights, ASD. In particular, for the strong shareholder protection subsample (Model 4; ASD > 0.5), we find that the coefficient on CR is again positive and statistically significant at the 1% level, suggesting that the likelihood of paying dividends increases with creditor rights as predicted by the substitute model. However, for the subsample of firms with poor shareholder rights (Model 5; ASD ≤ 0.5), the effect of CR is weaker and less significant (coefficient = 0.052 in Model 5 vs. 0.497 in Model 4; p-value = .033 in Model 5 vs. < .001 in Model 4). The difference between the CR coefficients in Models 4 and 5 is also statistically significant at the 1% level.

Taken together, the results in Models 1 through 5 of Table 2 provide strong support for the prediction in our first hypothesis that the positive effect of creditor rights on dividends as described by the substitute model is more significant under strong shareholder rights.11 We interpret these findings as implying that although creditor rights mitigate the agency costs of debt and provide firms more discretion to pay dividends, firms choose to exercise this discretion when shareholder rights are strong.

Creditor Rights and Dividend Payouts

In the rest of Table 2 we replicate the previous analysis using equation (1) with Div/sales as the dependent variable. In Model 6 of Table 2, we estimate the dividend payout ratio as a function of CR, Law, firm-level control variables, and industry–year dummies. Consistent with the substitute model, CR exhibits a significant and positive effect on Div/sales, with the coefficient indicating that an increase in CR from 0 to 4 translates into a 0.4% increase in dividends as a portion of net sales. The subsample results reported in Models 7 through 10 generally reinforce our earlier findings. In Models 7 and 9, which correspond to the strong shareholder protection subsample (Law = 1 for Model 7; ASD > 0.5 for Model 9), we continue to estimate a positive and highly significant relation between CR and the dividend payout ratio. Turning to the weak shareholder protection subsample (Law = 0 for Model 8; ASD ≤ 0.5 for Model 10), CR loads negatively at the 1% level in Models 8 and 10, which is in sharp contrast to the prediction of the substitute model.

Overall, the findings using dividend payouts together with those using the propensity to pay dividends lend strong support to the prediction in Hypothesis 1 that the substitute model better explains the relation between creditor rights and dividend payouts when shareholder protection is strong.

Robustness Checks

In this subsection, we verify the robustness of our preceding findings. First, we control for additional firm- and country-level variables that have been shown to be related to dividend policy. Second, we examine whether our results are driven by some countries contributing disproportionately to our sample (e.g., Japan, United States, and United Kingdom). Third, we test whether our results vary over time using annual regressions. Fourth, we repeat the regressions in Table 2 using country means with clustering at the country level. The results of these robustness tests are reported in Table 3. For brevity, we only report coefficient estimates (p-values) for CR.

Table 3. Coefficient Estimates (p-values) of CR in Robustness Checks on the Results in Table 2
 Dependent Variable = PayerDependent Variable = Div/sales
 Model 2 Table 2Model 3 Table 2Model 4 Table 2Model 5 Table 2Model 7 Table 2Model 8 Table 2Model 9 Table 2Model 10 Table 2
Note
  1. Panel A reports the coefficients on CR in Models 2 through 5 and 7 through 10 of Table 2 after adding a number of firm- and country-level controls. Stock_cap is a country's stock market capitalization scaled by GDP. Dividend Premium, proposed by Baker and Wurgler (2004), is calculated as the difference in the logs of the asset-weighted average market-to-book ratios of dividend payers and nonpayers. Tax is the tax advantage of dividends over capital gains from La Porta et al. (2000). Fin_arch (Kwok and Tadesse 2006) indexes financial architecture, with larger values representing more stock-market-oriented financial systems. Owner is the percentage of closely held shares (08021). Panel B reports the coefficients on CR in Models 2 through 5 and 7 through 10 of Table 2 excluding firms from the U.S., the U.K., and Japan. Panel C reports the distribution of CR across annual regressions for Models 2 through 5 and 7 through 10 of Table 2. Panel D reports the coefficients on CR using country means with country-level error clustering and year fixed effects.

  2. a

    Significant at the 1% level.

  3. b

    Significant at the 5% level.

  4. c

    Significant at the 10% level.

Panel A. Coefficient Estimates (p-value) for CR, Including Different Firm- and Country-Level Controls
         
Stock_cap0.714a0.073a0.710a0.090a0.006a−0.002a0.006a−0.003a
 (<.001)(<.001)(<.001)(<.001)(<.001)(<.001)(<.001)(<.001)
Premium0.479a0.0290.493a0.0310.005a−0.003a0.005a−0.003a
 (<.001)(.234)(<.001)(.201)(<.001)(<.001)(<.001)(<.001)
Tax0.537a0.051b0.543a0.064b0.004a−0.002a0.004a−0.003a
 (<.001)(.040)(<.001)(.012)(<.001)(<.001)(<.001)(<.001)
Fin_arch0.502a0.039c0.529a0.043c0.005a−0.003a0.005a−0.003a
 (<.001)(.100)(<.001)(.070)(<.001)(<.001)(<.001)(<.001)
Owner0.472a0.047c0.487a0.054b0.004a−0.003a0.005a−0.003a
 (<.001)(.051)(<.001)(.026)(<.001)(<.001)(<.001)(<.001)
Panel B. Coefficient Estimates (p-value) for CR, Excluding Firms from the U.S., U.K., and Japan
Coefficient estimates0.220a−0.0100.253a−0.0070.004a−0.001a0.005a−0.002a
p-value(<.001)(.656)(<.001)(.753)(<.001)(<.001)(<.001)(<.001)
Panel C. Coefficient Estimate Distribution for CR across Annual Regressions
No. of positive & 10% significant206203201200
No. of positive & insignificant060110301
No. of negative & 10% significant0001014015
No. of negative & insignificant08050204
Panel D. Coefficient Estimates (p-value) for CR for Regressions with Country Means
Coefficient estimates0.044b0.0080.056b0.0250.005a−0.0010.007a−0.001b
p-value(.027)(.638)(.012)(.155)(<.001)(.140)(<.001)(.022)
Additional Controls

In Panel A of Table 3, we examine whether our earlier findings are robust to separately adding to Models 2 through 5 and 7 through 10 of Table 2 several controls motivated by prior research:12 (1) the level of stock market development (Stock_cap) from Demirguc-Kunt and Levine (1996) and their subsequent work, defined as the ratio of stock market capitalization to GDP;13 (2) the dividend catering Premium14 proposed by Baker and Wurgler (2004), calculated as the difference in the logs of the asset-weighted average market-to-book ratios of dividend payers and nonpayers (for any country-year, we require that there be at least five dividend payers and five nonpayers); (3) the tax advantage of dividends over capital gains (Tax) derived from La Porta et al. (2000), defined as the ratio of the value, to an outside investor, of US$1 distributed as dividend income to the value of US$1 when kept inside the firm as retained gains; (4) the financial architecture index (Fin_arch) of Kwok and Tadesse (2006), with larger values representing more stock-market-oriented financial systems; and (5) the level of insider ownership (Owner), defined as the percentage of closely held shares.

The results reported in Panel A indicate that the positive coefficient of creditor rights is larger and more statistically significant under strong shareholder rights (i.e., Models 2, 4, 7, and 9). For firms under weak shareholder rights, the coefficient of creditor rights is either statistically less significant (Models 3 and 5) or even negative (Models 8 and 10). In unreported tests, we include all of the country-level controls simultaneously and find the same patterns. Overall, these results suggest that the results in Table 2 are not spuriously driven by the omission of other important factors.

Sample Composition

Brockman and Unlu (2009) note that the United States, United Kingdom, and Japan contribute disproportionately to multinational studies regardless of the data vendor. This is also the case in our sample, where these three countries account for approximately 40% of the sample firm-year observations. Accordingly, we explore whether our results are sensitive to sample composition. We address this issue in two ways. First, we reestimate Models 2 through 5 and 7 through 10 of Table 2 after removing these three countries (United States, United Kingdom, and Japan) sequentially (not reported for brevity) and then as a group (Panel B of Table 3). Excluding these countries from the analysis yields qualitatively similar results, which implies that our results are not driven by sample composition bias. Second, we limit attention to subsamples of the largest 10 (50) firms from each country-year in terms of total assets. Because most of these large firms (> 85%) are dividend payers, in this analysis we repeat the tests only for Models 7 through 10 of Table 2, where Div/sales is the dependent variable. The results (not reported for brevity) using these smaller samples strongly support our earlier evidence that the positive effect of creditor rights on dividend payouts is conditioned by shareholder rights.

Annual Regressions

In Panel C of Table 3, we run annual regressions for Models 2 through 5 and 7 through 10 of Table 2. We report distributions of coefficients on CR across the annual regressions. Because the annual CR coefficients show clearly different patterns between strong and weak shareholder rights, we do not compute the time-series standard deviations (Fama and Macbeth 1973). The results show that CR loads positively and statistically significantly in all 20 annual regressions under strong shareholder rights (Models 2, 4, 7, and 9). For the annual regressions under weak shareholder protection, CR is generally statistically insignificant (Models 3 and 5) or even negative and significant (Models 8 and 10). These findings further confirm that our results in Table 2 are not driven by observations from recent years dominating the sample.

Country-Level Regressions

In Panel D of Table 3, we employ annual country-level means of the firm-specific variables and repeat the subsample regressions in Table 2 to compress within-country correlations and further mitigate the sample composition issue. We adopt an OLS framework including year fixed effects and control for clustering at the country level. Reinforcing our findings in Table 2, the results show a significant and positive coefficient on creditor rights for the subsample of firms under strong shareholder rights (Models 2, 4, 7, and 9). In contrast, the coefficients on CR are indistinguishable from zero (Models 3, 5, and 8) for the set of firms under weak shareholder rights except in Model 10 wehre CR has a significant and negative coefficient.

Prior research suggests that shareholder protection is correlated with stock market development (e.g., La Porta et al. 1998; Djankov et al. 2008). An alternative hypothesis therefore is that the variation in the effect of creditor rights on dividends across shareholder rights is actually driven by other institutional factors that vary across stock markets of different levels of development. We confront this issue by reestimating Models 1 and 6 of Table 2 after bisecting the samples by stock market development (stock market capitalization/gross domestic product [GDP]) instead of shareholder rights. If our prior evidence in Table 2 is driven by the correlation between shareholder rights and stock market development, we should also observe a difference in the effect of creditor rights on dividends across the subsamples based on stock market development. However, we do not obtain this substantial difference (results are not reported for brevity).

Taken together, the results suggest that the positive effect of creditor rights on dividend payment propensity and ratios predicted by the substitute model is more effective under strong shareholder rights. This implies that firms are more likely to exercise their discretion to pay dividends due to better creditor rights when they face more pressure for dividends from minority shareholders.

Additional Implications of the Role of Shareholder Rights in the Substitute Model

In the following three subsections we extend our analysis to examine additional testable implications on the role of shareholder rights in determining the relevance of the substitute model. First, we examine whether there is any difference in the relation between creditor rights and the likelihood of dividend initiations across shareholder rights. The substitute model implies that strong creditor rights might also increase the likelihood of dividend initiations. Second, we investigate the relation between creditor rights and the likelihood of dividend omissions across strong and weak shareholder rights. According to the substitute model, strong creditor rights should not only increase firms' propensity to pay dividends but also reduce the likelihood of dividend omissions (Brockman and Unlu 2009). Based on Hypothesis 1, we expect that the effects of creditor rights on dividend initiations and omissions to be more pronounced under strong shareholder rights. Third, we test Hypothesis 1 with two firm-level variables (maturity and profitability), instead of shareholder rights, as proxies for minority shareholders' pressure for dividends. We expect the substitute model to be more effective for firms with characteristics leading to more pressure for dividends. Tables 4 and 5 present the results of these additional tests.

Table 4. Creditor Rights and Dividend Initiations/Omissions under Different Shareholder Rights
 (1)(2) Law = 1(3) Law = 0(4) ASD > 0.5(5) ASD ≤ 0.5
Note
  1. This table presents fixed effect logit regression results with firm-level clustered errors. The dependent variable in Panel A, Init, equals 1 if a firm that did not pay a Dividend in the previous year chooses to pay a Dividend in the current year, and 0 otherwise. The dependent variable in Panel B, Omit, equals 1 if a firm that paid a Dividend in the previous year fails to pay a Dividend in the current year, and 0 otherwise. Dividend is the sum of cash common dividends (05376) and cash preferred dividends (05401) issued over the calendar year. Model 1 is based on the samples of firms for which dividend initiations or omissions are possible, while Models 2 through 5 are run on subsamples of strong and weak shareholder rights, differentiated by legal origin and the median of ASD. Debt is the book value of total liabilities (03351) scaled by the book value of total assets (02999). Size is the natural logarithm of dollar value of book value of total assets (07230). Growth is the logarithmic value of net sales (01001) growth calculated as log (datat/datat–1). Profit is EBIT (18191) scaled by net sales (01001). RE/TE is retained earnings (03495) scaled by common equity (03501). Cash is cash and short-term investments (02001) scaled by the book value of total assets (02999). Law equals 1 for common law countries and 0 for civil law countries. CR is creditor rights from Djankov et al. (2007). ASD is the anti-self-dealing index from Djankov et al. (2008).

  2. a

    Significant at the 1% level.

Panel A. Dependent Variable = Init
N38,23521,17617,05923,54714,688
CR0.342a (<.001)0.486a (<.001)0.098a (<.001)0.485a (<.001)0.080a (.002)
LawYesNoNoNoNo
Intercept, Debt, Size, Growth, Profit, Cash, RE/TEYesYesYesYesYes
Year & ind. fixed effectsYesYesYesYesYes
Panel B. Dependent Variable = Omit
N83,90636,17047,73639,67844,228
CR−0.100a (<.001)−0.194a (<.001)−0.002 (.921)−0.249a (<.001)−0.017 (.506)
LawYesNoNoNoNo
Intercept, Debt, Size, Growth, Profit, Cash, RE/TEYesYesYesYesYes
Year & ind. fixed effectsYesYesYesYesYes
Table 5. Creditor Rights and Dividend Policy under Different Firm Characteristics
Independent Variables (Predicted Sign)Dependent Variable = PayerDependent Variable = Div/sales
(1) VAR = RE/TE(2) VAR = Profit(3) VAR = RE/TE(4) VAR = Profit
Note
  1. Models 1 and 2 are logit regressions with Payer as the dependent variable, and Models 3 and 4 are ordinary least squares regressions with Div/sales as the dependent variable. Each of the models includes an interaction between a firm-level variable (VAR) and investor rights (Law and CR). Payer equals 1 if a firm pays a Dividend, and 0 otherwise. Dividend is the sum of cash common dividends (05376) and cash preferred dividends (05401) issued over the calendar year. Div/sales is the ratio of Dividend to net sales (01001). All the firm-level independent variables take values of samplewise decile ranks (0–9). Debt is the book value of total liabilities (03351) scaled by the book value of total assets (02999). Size is the natural logarithm of dollar value of book value of total assets (07230). Growth is the logarithmic value of net sales (01001) growth calculated as log (datat/datat–1). Profit is EBIT (18191) scaled by net sales (01001). RE/TE is retained earnings (03495) scaled by common equity (03501). Cash is cash and short-term investments (02001) scaled by the book value of total assets (02999). Law equals 1 for common law countries and 0 for civil law countries. CR is creditor rights from Djankov et al. (2007). The p-values reported in parentheses control for firm-level clustering effects.

  2. a

    Significant at the 1% level.

  3. b

    Significant at the 10% level.

CR*VAR0.093a0.035a0.001a0.001a
 (<.001)(<.001)(<.001)(.003)
Law*VAR−0.228a0.023**0.001a0.001a
 (<.001)(.021)(<.001)(.001)
CR−0.068a0.134a−0.00010.001a
 (.003)(<.001)(.815)(.007)
Law0.201a−0.606a0.00020.001
 (<.001)(<.001)(.652)(.171)
Debt (+/−)−0.025a−0.029a−0.001a−0.001a
 (<.001)(<.001)(<.001)(<.001)
Size (+)0.219a0.212a0.001a0.001a
 (<.001)(<.001)(<.001)(<.001)
Growth (−)−0.045a−0.043a−0.001a−0.002a
 (<.001)(<.001)(<.001)(<.001)
Profit (+)0.142a0.075a0.005a0.004a
 (<.001)(<.001)(<.001)(<.001)
Cash (+/−)−0.008b−0.0030.001a0.001a
 (.092)(.481)(.001)(.001)
RE/TE (+)0.265a0.328a−0.001a−0.0001
 (<.001)(<.001)(<.001)(.322)
Intercept, year, & ind. fixed effectsYesYesYesYes
Creditor Rights and Dividend Initiations under Different Shareholder Rights

Panel A of Table 4 presents the results from estimating the likelihood of dividend initiations using several specifications of equation (1). The dependent variable is Init, which equals 1 if a firm initiates a dividend payment in the current year relative to the previous year. In these tests, we use a sample of all the firm-years in which the firm failed to pay dividends in the previous year.

In Model 1, we include a set of firm-level control variables, industry–year dummies, and the proxies for shareholder rights (Law) and creditor rights (CR). Consistent with the substitute model, we observe a significant and positive relation between CR and the likelihood of dividend initiations. The subsample analyses in Models 2 through 5 suggest moderate support for Hypothesis 1. In particular, consistent with the substitute model, we observe a consistently positive coefficient on CR across subsamples under strong and weak shareholder rights, but the coefficient is larger under strong shareholder protection (Models 2 and 4). Overall, the findings in Panel A imply that the substitute model is more effective in explaining dividend initiations under strong shareholder rights. Under better creditor rights, firms have more discretion to initiate a dividend payment, but they are more likely to exercise this discretion when strong shareholder rights force them to disgorge cash.

Creditor Rights and Dividend Omissions under Different Shareholder Rights

Panel B of Table 4 presents the results from estimating the likelihood of dividend omissions using several specifications of equation (1). The dependent variable is Omit, which, defined as in Brockman and Unlu (2009), equals 1 if a firm omits a dividend payment in the current year relative to the previous year. We use all the firm-years in which dividends were paid in the previous year.

In Model 1, we include a set of firm-level control variables, industry–year dummies, and the proxies for shareholder rights (Law) and creditor rights (CR). We find a significant and negative relation between CR and the likelihood of dividend omissions, consistent with the substitute model (Brockman and Unlu 2009). Models 2 through 5 present our subsample analyses, where we bisect the sample used in Model 1 according to legal origin and at the median value of ASD to test whether the effect of creditor rights on dividend omissions depends on the quality of shareholder protection. Models 2 and 4 present the results for the subsamples with strong shareholder protection (Law = 1 and ASD > 0.5, respectively). In these models, the coefficients on CR are negative and highly significant, suggesting that creditor rights decrease the likelihood of dividend omissions, consistent with the substitute model. In contrast, in Models 3 and 5, which are based on subsamples with weak shareholder protection (Law = 0 and ASD ≤ 0.5, respectively), the coefficient on CR is statistically insignificant. This difference in CR coefficients across different shareholder rights indicates that the substitute model is effective at explaining dividend omissions only under strong shareholder protection. Under better creditor rights, firms have more discretion to reduce the incidence of dividend omissions, but they are more likely to exercise this discretion when strong shareholder rights force them to sustain dividend payments.

Firm Characteristics Relevant to Shareholders' Pressure for Dividends

So far we have employed shareholder rights to capture minority shareholders' pressure for dividends. However, prior research also identifies firm characteristics (for our purposes, maturity15 and profitability) that are related to the demand for dividends. In particular, minority shareholders have more incentives to ask for dividends from firms that are mature or highly profitable. Thus, we expect these firm characteristics to affect the relation between creditor rights and dividend policy in a manner similar to strong shareholder rights as documented in Table 2. In Table 5 we report the results from estimating the likelihood of paying cash dividends and dividend payouts using different specifications of equation (1) with CR*VAR and Law*VAR included as additional controls. VAR is given by RE/TE or Profit. We conduct the analysis using annual decile ranks of firm-level independent variables instead of the variables themselves. Using annual decile ranks can accentuate the differential effects of creditor rights on dividends across groups of firms formed according to maturity or profitability. Additionally, as explained by Brockman and Unlu (2009, p. 294), this approach “helps control for possible confounding effects from time-varying firm characteristics on the coefficient estimates.”

The results in Models 1 through 4 of Table 5 show that for both Payer and Div/sales, the coefficient on the interactions CR*RE/TE and CR*Profit are positive and highly significant, suggesting that the positive effect of creditor rights on the dividend variables is stronger for mature or profitable firms.16 This result echoes the finding that high disclosure quality is more likely to lead to dividend payments among high-RE/TE firms (Brockman and Unlu 2011). Therefore, the results are in line with the spirit of Hypothesis 1: the substitute model is more effective when minority shareholders' pressure for dividends is strong.

In sum, the results in Section III confirm our first hypothesis: the effect of creditor rights on dividend policies, as predicted by the substitute model, is more pronounced under strong shareholder protection. If strong creditor protection provides firm insiders more discretion to pay dividends, they are more likely to exercise this discretion when strong shareholder rights exert more pressure for dividends.

IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

In this section, we examine Hypothesis 2, which posits that the relevance of the outcome model depends on the quality of creditor protection. We follow the same empirical approach as in Section III.

The Outcome Model under Different Creditor Rights

In this subsection we run logit and OLS regressions for Payer and Div/sales on strong and weak creditor rights subsamples. We report the results in Table 6. In Models 1 through 5, the dependent variable is the probability of paying cash dividends (Payer), whereas in Models 6 through 10 the dependent variable is the amount of cash dividends paid (Div/sales).

Table 6. Shareholder Rights and Dividend Policy under Different Creditor Rights
Independent Variables (Predicted Sign)Dependent Variable = PayerDependent Variable = Div/sales
(1) Full Sample(2) CR ≥ 3(3) CR < 3(4) CR ≥ 3(5) CR < 3(6) Full Sample(7) CR ≥ 3(8) CR < 3(9) CR ≥ 3(10) CR < 3
Note
  1. Models 1 through 5 are logit regressions with Payer as the dependent variable, where Model 1 is run on the full sample and Models 2 through 5 are run on subsamples differentiated by the sample medians of CR. Models 6 through 10 are ordinary least squares regressions with Div/sales as the dependent variable, where Model 6 is run on the full sample and Models 7 through 10 are run on subsamples differentiated by sample medians of CR. Payer equals 1 if the firm pays Dividend, and 0 otherwise. Div/sales is the ratio of Dividend to net sales (01001). Dividend is the sum of cash common dividends (05376) and cash preferred dividends (05401) issued over the calendar year. Debt is book value total liabilities (03351) scaled by book value total assets (02999). Size is the natural logarithm of dollar value of total book value assets (07230). Growth is the logarithmic value of net sales (01001) growth calculated as log (datat/datat–1). Profit is EBIT (18191) scaled by net sales (01001). RE/TE is retained earnings (03495) scaled by common equity (03501). Cash is cash and short-term investments (02001) scaled by book value total assets (02999). Law equals 1 for common law countries and 0 for civil law countries. CR is creditor rights from Djankov et al. (2007). ASD is anti-self-dealing index from Djankov et al. (2008). The p-values reported in parentheses control for firm-level clustering effects.

  2. a

    Significant at the 1% level.

  3. b

    Significant at the 5% level.

N139,16863,51975,64963,51975,649139,16863,51975,64963,51975,649
Law−0.533a (<.001)0.418a (<.001)−0.996a (<.001)  0.006a (<.001)0.015a (<.001)−0.0002 (.743)  
ASD   0.687a (<.001)−1.854a (<.001)   0.020a (<.001)−0.015a (<.001)
CR0.316a (<.001)    0.001a (<.001)    
Debt (+/−)−0.729a (<.001)−0.695a (<.001)−0.779a (<.001)−0.693a (<.001)−0.563 (.137)−0.042a (<.001)−0.045a (<.001)−0.039a (<.001)−0.046a (<.001)−0.040a (<.001)
Size (+)0.353a (<.001)0.424a (<.001)0.357a (<.001)0.420a (<.001)0.324a (<.001)0.002a (<.001)0.004a (<.001)0.002a (<.001)0.003a (<.001)0.002a (<.001)
Growth (−)−0.496a (<.001)−0.377a (<.001)−0.616a (<.001)−0.362a (<.001)−0.693a (<.001)−0.008a (<.001)−0.009a (<.001)−0.006a (<.001)−0.009a (<.001)−0.006a (<.001)
Profit (+)1.219a (<.001)1.101a (<.001)1.285a (<.001)1.111a (<.001)1.158a (<.001)0.006a (<.001)0.005a (<.001)0.008a (<.001)0.005a (<.001)0.008a (<.001)
Cash (+/−)−0.090 (.366)−0.299b (.015)0.220 (.167)−0.345a (.005)0.667a (<.001)0.013a (<.001)0.009a (<.001)0.019a (<.001)0.007a (.005)0.018a (<.001)
RE/TE (+)1.054a (<.001)1.127a (<.001)1.055a (<.001)1.119a (<.001)1.041a (<.001)0.001a (<.001)0.001a (<.001)−0.000 (.983)0.001a (.001)−0.000 (.591)
Intercept, year, & ind. fixed effectsYesYesYesYesYesYesYesYesYesYes
R20.3930.4820.3360.4810.3120.1190.1510.1080.1410.111

In Models 1 and 6, we include a set of firm-level control variables, industry–year dummies, and the proxies for shareholder rights (Law) and creditor rights (CR) as in equation (1). Law shows a significant and negative effect on Payer in Model 1, in contrast to the outcome model, but a significant and positive effect on Div/sales in Model 6, consistent with the outcome model.

In Models 2 through 5 and Models 7 through 10, we replicate the analysis in Models 1 and 6 on subsamples with strong and weak creditor protection based on the median CR score (3). The shareholder rights coefficients are significant and positive for firms under strong creditor rights (Models 2, 4, 7, and 9), but are negative in Models 3, 5, 8, and 10, where creditor rights are weak. The evidence in Table 6 generally confirms that the outcome model is more effective in explaining dividend policies under strong creditor rights.

Notice that the results of Models 3, 5, 8, and 10 run in the opposite direction of the prediction of the outcome model. This novelty does not qualitatively go against Hypothesis 2, which posits that the outcome model should be more effective under strong creditor rights. Chae, Kim, and Lee (2009) find that better corporate governance is associated with a higher dividend payout for firms with lower external financial constraints, whereas this relation is reversed for firms with higher external financial constraints. If creditor rights proxy for cross-country differences in external financial constraints (La Porta et al., 1997), their findings may help explain the reversal with respect to the outcome model.

Robustness Checks

Similar to the robustness tests in Table 3, we verify our results in Table 6 with additional tests. First, we control for additional variables. Second, we examine whether our results are driven by the large presence of firms from the United States, United Kingdom, and Japan. Third, we run annual regressions and test whether our results vary across years. Finally, we repeat all the regressions using country means with country-clustered effects. In short, none of these robustness tests qualitatively changes our results in Table 6, which implies that firms are more likely to cater to pressure for dividends when facing fewer constraints from better protected creditors. The results are not reported for brevity.

Additional Implications of the Role of Creditor Rights in the Outcome Model

In this subsection, we further examine whether Hypothesis 2 has implications for dividend initiations and omissions and for firms with different characteristics. The unreported results yield two main insights. First, we examine whether the effect of shareholder rights on dividend initiations and omissions is different across strong and weak creditor rights. According to the outcome model, strong shareholder rights should increase initiations but reduce the likelihood of dividend omissions. We find that the effects predicted by the outcome model are more positive and significant under strong creditor rights. Second, we test Hypothesis 2 using Size and Profit as proxies for creditors' constraints instead of creditor rights. If strong creditor rights enhance the effect of shareholder rights on dividend policy by relaxing constraints from creditors, we expect to observe a stronger effect of shareholder rights for large and profitable firms. Our results confirm this inference (the interactions Law*Size and Law*Profit load with significant and positive coefficients).

In sum, the analysis in Section IV, which is a natural extension of Section III, suggests that the outcome model is more effective under strong creditor rights. This result implies that if firm insiders are facing greater shareholder pressure for dividends, they are more likely to cater to this pressure when better protected creditors place fewer constraints on dividend policy.

Conclusion

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References

The extant international dividend literature focuses on either the agency costs of equity arising from the conflict between the firm and minority shareholders or the agency costs of debt arising from the conflict between the firm and creditors. This study combines both relations and shows that the optimal dividend policy balances the interests of minority shareholders and creditors. In particular, firm insiders are predicted to optimize their dividend policies to the point where, for an extra unit of dividend payment, the marginal savings of agency costs of equity are equal to the marginal increase in the agency costs of debt. Two main inferences follow: first, the substitute model should be more effective under strong shareholder rights, and second, the outcome model should be more effective under strong creditor rights. The intuition is that when one agency relationship is strongly protected by external laws, firm insiders employ dividend policies more to improve the other agency relationship (reflected in the higher sensitivity of dividends to legal changes). The empirical results are in line with our hypotheses: the substitute model (outcome model) is more effective under strong shareholder rights (creditor rights), consistent with the balancing behavior of firms when making dividend decisions. These results have important implications, as they suggest that research focusing on only one agency relationship may lead to inaccurate conclusions. Finally, we note that we find a reversal of the predictions for the outcome model under weak creditor rights, a finding that this study cannot fully explain. Resolving this anomaly would be an interesting topic for future research.

Notes
  1. 1

    Other major theories of dividend policy include tax (e.g., Lasfer 1996; Foley et al. 2007), signaling (e.g., Garrett and Priestley 2000; Caton, 2003; Lee and Yan 2003; Booth and Chang 2011), clientele (e.g., Allen, 2000), and catering (e.g., Baker and Wurgler 2004) theories. In this article, we follow recent research that emphasizes agency theory in explaining corporate dividend policies (e.g., Rozeff 1982; Schooley and Barney 1994; La Porta et al. 2000; Faccio, 2001; DeAngelo, DeAngelo, and Stulz 2006; Denis and Osobov 2008; Eije and Megginson 2008; Brockman and Unlu 2009; Nishikawa, 2011).

  2. 2

    Most subsequent studies support the validity of the outcome model. Mitton (2004) examines the impact of firm-level corporate governance on dividend payouts for a sample of 365 firms from 19 emerging markets and finds that, consistent with the outcome model, firms with stronger corporate governance practices have higher dividend payouts. Faccio, Lang, and Young (2001) compare dividend policies of publicly traded firms from East Asia and Western Europe, where unlike their U.S. and U.K. counterparts, most firms have controlling shareholders (typically a family) that have both the incentives and the ability to expropriate minority shareholders' wealth. They find that, compared to the firms in East Asia, firms in Western Europe pay more dividends due to better shareholder protection.

  3. 3

    Because of space constraints, a theoretical model that examines the interaction between the agency costs of equity and debt in determining dividend policies is not included in this article. It is, however, available from the authors upon request.

  4. 4

    Brockman and Unlu (2009) report that the cross-country correlation between creditor rights and shareholder rights is 18% (p-value = .13) based on the 70 common countries that appear in both Djankov, McLiesh, and Shleifer (2007) and Djankov et al. (2008).

  5. 5

    Although there are only 5 countries from civil law countries with creditor rights greater than or equal to 3, Panel D in Table 1 shows that, except for Japan, firms from these 5 countries account for one-third of the sample population from civil law countries, which has the largest number of firms. Furthermore, Japan had a creditor rights score of 3 before 2000. Even within civil law countries, there is considerable variation in creditor rights: 5 countries with a score of 3, 9 countries with a score of 2, 6 countries with a score of 1, and 3 countries with a score of 0.

  6. 6

    In the original sample, we detect 7,667 firm-years with negative total liabilities, negative total assets, or total liabilities greater than total assets.

  7. 7

    La Porta et al. (2000) and Brockman and Unlu (2009) also employ total cash dividends. Our core findings remain robust to using common dividends instead of total dividends.

  8. 8

    We scale dividends by net sales instead of earnings for two reasons. First, Avizian, Booth, and Cleary (2003) argue that Div/earnings is not a good measure of dividend policy because it becomes unstable when earnings are low. Second, Leuz, Nanda, and Wysocki (2003) find significant cross-country variation in earnings management, which is influenced by the strength of investor protection. However, all our core results continue to hold when we measure dividend payouts using dividends to earnings, dividends to free cash flows, or dividends to total assets.

  9. 9

    Our core findings persist when we employ the original antidirector scores (La Porta et al. 1998) and revised antidirector scores (Djankov et al. 2008).

  10. 10

    Using Compustat Global, Brockman and Unlu (2009) find that dividend-paying firms account for 64.4% of the total sample.

  11. 11

    We also replicate the tests in this section using the four components of the creditor rights index from Djankov, McLiesh, and Shleifer (2007). The results using three of the four components strongly support the results in Table 2, whereas the other component (SECURED_FIRST) does not generate supportive results because of a lack of variation, consistent with Brockman and Unlu (2009).

  12. 12

    We also control rule of law (La Porta et al. 1998) and document similar results.

  13. 13

    We also control for Stock_turn, the ratio of stock-market-traded value to gross domestic product, and Turnover, the ratio of stock-market-traded value to market capitalization, and find similar results.

  14. 14

    Baker and Wurglar (2004) show that the propensity to pay dividends is related to the market demand for dividends as proxied by Premium. Li and Lie (2006) present further evidence on the positive relation between market demand and dividend payouts.

  15. 15

    DeAngelo, DeAngelo, and Stulz (2006) find a strong, positive relation between the propensity to pay dividends and firm maturity as proxied by the earned-to-contributed equity mix. They argue that as the proportion of common equity that comes from accumulated profit increases, a firm has fewer good investment opportunities and thus shareholder pressure for dividends increases.

  16. 16

    Prior studies suggest that other firm characteristics, such as leverage (Jensen 1986; Easterbrook 1984) and size (Fama and French 2002), are also related to the extent of agency conflicts and thus can lead to more pressure for dividends. In unreported tests, we find that among low-leverage and large firms, the effect of creditor rights on dividends is stronger. Furthermore, our hypotheses are based on the assumption that firm insiders are facing pressures from both minority shareholders and creditors. Therefore, one may argue that our results do not hold for unlevered or low-leverage firms (i.e., firms without pressure from creditors). We repeat our analysis for low-leverage firms (debt ratio less than 20%) and do find that shareholder (creditor) rights' effect on dividends does not differ across creditor (shareholder) rights, as predicted by our hypotheses. However, it is important to note that leverage may not necessarily proxy for pressure from creditors. For example, DeAngelo, DeAngelo, and Stulz (2006) argue that a low-leverage firm can be either a safe firm or a firm that is poorly assessed by creditors and fails to raise much debt capital (e.g., start-up firms).

References

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Data and Descriptive Statistics
  5. III. The Effect of Shareholder Rights on the Relation between Creditor Rights and Dividend Policy
  6. IV. The Impact of Creditor Rights on the Relation between Shareholder Rights and Dividend Policy
  7. Conclusion
  8. References
  • Allen, F., A. Bernardo, and I. Welch, 2000, A theory of dividends based on tax clienteles, Journal of Finance 55, 24992536.
  • Avizian, V., L. Booth, and S. Cleary, 2003, Do emerging market firms follow different dividend policies from U.S. firms? Journal of Financial Research 106, 37187.
  • Baker, M., and J. Wurgler, 2004, A catering theory of dividends, Journal of Finance 59, 12565.
  • Booth, L., and B. Chang, 2011, Information asymmetry, dividend status, and SEO announcement day returns, Journal of Financial Research 34, 15577.
  • Brockman, P., and E. Unlu, 2009, Dividend policy, creditor rights, and the agency costs of debt, Journal of Financial Economics 92, 27699.
  • Brockman, P., and E. Unlu, 2011, Earned/contributed capital, dividend policy, and disclosure quality: An international study, Journal of Banking and Finance 35, 161025.
  • Caton, G., J. Goh, and N. Kohers, 2003, Dividend omissions and intraindustry information transfers, Journal of Financial Research 26, 5164.
  • Chae, J., S. Kim, and E. Lee, 2009, How corporate governance affects payout policy under agency problems and external financial constraints, Journal of Banking and Finance 33, 20932101.
  • DeAngelo, H., L. DeAngelo, and R. Stulz, 2006, Dividend policy and the earned/contributed capital mix: A test of the lifecycle theory, Journal of Financial Economics 81, 22754.
  • Demirguc-Kunt, A., and R. Levine, 1996, Stock market development and financial intermediaries: Stylized facts, World Bank Economic Review 10, 291321.
  • Denis, D., and I. Osobov, 2008, Why do firms pay dividends? International evidence on the determinants of dividend policy, Journal of Financial Economics 89, 6282.
  • Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer, 2008, The law and economics of self-dealing, Journal of Financial Economics 88, 43065.
  • Djankov, S., C. McLiesh, and A. Shleifer, 2007, Private credit in 129 countries, Journal of Financial Economics 84, 299329.
  • Easterbrook, F., 1984, Two agency-cost explanations of dividends, American Economic Review 74, 65059.
  • Eije, H., and W. Megginson, 2008, Dividends and share repurchases in the European Union, Journal of Financial Economics 89, 34774.
  • Faccio, M., L. Lang, and L. Young, 2001, Dividends and expropriation, American Economic Review 91, 5478.
  • Fama, E., and K. French, 2002, Testing trade-off and pecking order predictions about dividends and debt, Review of Financial Studies 15, 133.
  • Fama, E., and J. MacBeth, 1973, Risk, return, and equilibrium: Empirical tests, Journal of Political Economy 81, 60736.
  • Foley, F., J. Hartzell, S. Titman, and G. Twite, 2007, Why do firms hold so much cash? A taxed-based explanation, Journal of Financial Economics 86, 579607.
  • Garrett, I., and R. Priestley, 2000, Dividend behavior and dividend signaling, Journal of Financial and Quantitative Analysis 35, 17389.
  • Jensen, M., 1986, Agency costs of free cash flow, corporate finance and takeovers, American Economic Review 76, 32329.
  • Jensen, M., and W. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs, and ownership structure, Journal of Financial Economics 3, 30560.
  • Kwok, C., and S. Tadesse, 2006, National culture and financial systems, Journal of International Business Studies 37, 22747.
  • La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R. Vishny, 1997, Legal determinants of external finance, Journal of Finance 52, 113150.
  • La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R. Vishny, 1998, Law and finance, Journal of Political Economy 106, 111355.
  • La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R. Vishny, 2000, Agency problems and dividend policies around the world, Journal of Finance 55, 133.
  • Lasfer, M., 1996, Taxes and dividends: The UK evidence, Journal of Banking and Finance 20, 45572.
  • Lee, B., and N. Yan, 2003, The market's differential reactions to forward-looking and backward-looking dividend changes, Journal of Financial Research 26, 44968.
  • Leuz, C., D. Nanda, and P. Wysocki, 2003, Earnings management and investor protection: An international comparison, Journal of Financial Economics 69, 50527.
  • Li, W., and E. Lie, 2006, Dividend changes and catering incentives, Journal of Financial Economics 80, 293308.
  • Mitton, T., 2004, Corporate governance and dividend policy in emerging markets, Emerging Markets Review 5, 40926.
  • Nishikawa, T., A. Prevost, and R. Rao, 2011, Bond market reaction to stock repurchases: Is there a wealth transfer effect? Journal of Financial Research 34, 50322.
  • Rozeff, M., 1982, Growth, beta and agency costs as determinants of dividend payout ratios, Journal of Financial Research 5, 24959.
  • Schooley, D., and D. Barney, 1994, Using dividend policy and managerial ownership to reduce agency costs, Journal of Financial Research 17, 36373.