Corporate governance and dividend policy in emerging markets

Corporate governance and dividend policy in emerging markets

Emerging Markets Review 5 (2004) 409 – 426 www.elsevier.com/locate/econbase Corporate governance and dividend policy in emerging markets Todd Mitton*...

153KB Sizes 3 Downloads 383 Views

Emerging Markets Review 5 (2004) 409 – 426 www.elsevier.com/locate/econbase

Corporate governance and dividend policy in emerging markets Todd Mitton* Marriott School Brigham Young University, 684 TNRB, Provo, UT 84602, United States Received 29 January 2004; received in revised form 6 May 2004; accepted 31 May 2004 Available online 2 November 2004

Abstract In a sample of 365 firms from 19 countries, I show that firms with stronger corporate governance have higher dividend payouts, consistent with agency models of dividends. In addition, the negative relationship between dividend payouts and growth opportunities is stronger among firms with better governance. I also show that firms with stronger governance are more profitable, but that greater profitability explains only part of the higher dividend payouts. The positive relationship between corporate governance and dividend payouts is limited primarily to countries with strong investor protection, suggesting that firm-level corporate governance and country-level investor protection are complements rather than substitutes. D 2004 Published by Elsevier B.V. JEL classification: G35; G34 Keywords: Dividend policy; Corporate governance; Emerging markets

1. Introduction In the United States, the debate surrounding dividend policy has traditionally centered on the question of why firms pay dividends, given that the tax disadvantage of dividends appears to be large. But in countries where investor protection is weak, a more * Tel.: +1 801 422 1763. E-mail address: [email protected]. 1566-0141/$ - see front matter D 2004 Published by Elsevier B.V. doi:10.1016/j.ememar.2004.05.003

410

T. Mitton / Emerging Markets Review 5 (2004) 409–426

fundamental question regarding dividend policy might be more relevant: How can shareholders hope to extract dividends from firms, given that the legal environment of the country and the governance mechanisms of individual firms offer investors relatively few protections? Agency theory suggests that outside shareholders have a preference for dividends over retained earnings because insiders might squander cash retained within the firm (see, e.g., Easterbrook, 1984, Jensen, 1986, Myers, 2000). This preference for dividends may be even stronger in emerging markets with weak investor protection if shareholders perceive a greater risk of expropriation by insiders in such countries.1 La Porta et al. (2000) show that dividend payouts are higher, on average, in countries with stronger legal protection of minority shareholders. This finding lends support to what La Porta et al. (2000) call the boutcomeQ agency model of dividends, which hypothesizes that dividends result from minority shareholders using their power to extract dividends from the firm. If protection of minority shareholders does have a positive impact on dividend payouts, then shareholder protection should help explain not just country-level differences in dividend payouts, but also firm-level differences in dividend payouts within countries. Indeed, while country-level investor protection is an important factor in preventing expropriation, firm-level corporate governance could carry equal or greater importance. And corporate governance practices can vary widely even among firms in the same country operating under the same legal regime. This paper uses firm-specific corporate governance ratings developed by Credit Lyonnais Securities Asia (CLSA) for 365 firms from 19 emerging markets to study the impact of firm-level corporate governance on dividend payouts. It is important to note, as do La Porta et al. (2000), that the outcome model does not hinge on investors holding specific rights to dividends. Rather, what is important is that the country’s laws—or the company’s governance practices—allow minority shareholders more rights in general, which rights may then be used to influence dividend policy. For example, observers have noted that Russian firms are more commonly electing independent directors to their boards, despite a legal system that does little to define or enforce board independence (Nicholson, 2003). Mark Mobius, elected as an independent director to the board of Russia’s LUKoil in 2002, later acted on behalf of shareholders to propose a minimum-dividend policy that LUKoil’s board approved in 2003 (Investor Protection Association, 2003). Using the CLSA data, I first show that firms with higher corporate governance ratings have higher dividend payouts. The effect appears to be economically meaningful as well as statistically significant; regression results imply that a one-standard deviation increase in a firm’s corporate governance rating is associated with an average 4% point increase in dividend payouts (the average payout ratio is about 30%). A move from the worst governance (in this dataset, Indonesia’s Indocement) to the best governance (in this dataset, Hong Kong’s HSBC) would imply, on average, a higher dividend payout ratio of some 22% points. This result is consistent with the hypothesis of the outcome agency model that investors that are afforded stronger rights use those rights to extract

1

In a recent paper, Brav et al. (2003) find limited evidence for the agency theory of dividends in the U.S., at least from the perspective of corporate executives.

T. Mitton / Emerging Markets Review 5 (2004) 409–426

411

dividends from the firm. This result is also complementary to the findings of Faccio et al. (2001). Interpreting dividends as a means for limiting expropriation, they study East Asian and Western European firms and find that dividend payouts are significantly impacted by the vulnerability of a firm’s minority shareholders to expropriation by controlling shareholders.2 Even when shareholders are well protected, however, they may not prefer higher dividend payouts if they believe the firm has good investment opportunities available for excess cash. Indeed, La Porta et al. (2000) find that, in countries with strong investor protection, there is a stronger negative relationship between growth opportunities and dividend payouts. That is, it appears that, when shareholders perceive that their rights are well protected, they are more willing to let firms with good growth opportunities retain cash, being confident that they will share in the payoff from good projects later on. In contrast, if shareholders know that investor protection is poor, they may be more haphazard in their desire for dividends, trying to extract whatever value they can— regardless of the firm’s growth opportunities—before being expropriated. Complementary to the country-level finding of La Porta et al. (2000), I find at the firm level that firms with stronger corporate governance also show a stronger negative relationship between dividends and growth opportunities. In other words, the pattern of dividend payouts seems to make more sense among firms with good corporate governance. I next examine how dividend policy is impacted by the interplay of country-level investor protection and firm-level corporate governance. I find, first of all, that across all countries, both country-level investor protection and firm-level corporate governance have explanatory power for dividend payouts. Of the two, the country-level measures have perhaps greater explanatory power. Next, I find that firm-level corporate governance is positively associated with dividend payouts primarily in countries that offer strong investor protection (as measured by legal origin or antidirector rights). This suggests that firm-level corporate governance and country-level investor protection work as complements rather than substitutes. Firm-level improvements in corporate governance may be most effective when the legal regime of the country also offers a higher level of protection for shareholders. The findings of the paper are robust to many different specifications (although I find some specifications where they are not). The results hold when controlling for financial variables that have been shown previously to be correlated with dividend payouts, namely growth, size, and profitability (see Fama and French, 2001). I show separately that firms with stronger governance have higher profitability, but improved profitability explains only part of the connection between governance and dividends. The results also hold when controlling for industry- and country-fixed effects. In addition, I show that the results hold with or without financial firms, with or without mandatory-dividend countries, and when controlling for the tax advantage of dividends. Nevertheless, as will be discussed further in Section 4, because I lack a suitable instrument for firm-level corporate governance, these

2 In a related finding, Gugler and Yurtoglu (in press) show, using data from Germany, that negative reactions to dividend reductions are stronger among firms where minority shareholders are more susceptible to expropriation.

412

T. Mitton / Emerging Markets Review 5 (2004) 409–426

results are subject to typical problems of endogeneity, and, therefore, any interpretations regarding causality should be made cautiously. I also use additional data to get a sense as to whether the result holds among a broader sample of firms. Lacking governance ratings for a broader sample, I use variables that have been used in previous work as indicators of good governance. These variables (whether or not a firm is diversified, has a Big Five international auditor, or is cross-listed in the U.S.) are available for a much larger sample (over 14,000 firms). Although these indicators are clearly not as refined as the CLSA ratings, they are generally supportive of the results using the CLSA ratings, suggesting that the findings may be applicable to a broader sample. The results presented here add to the current literature in a few ways. These firm-level findings add confidence to previous country-level findings regarding investor protection and dividends. Because country-level measures of investor protection can be correlated with other important variables, some uncertainty remains about which country-level variables impact dividend policy. Showing that differences in shareholder protection at the firm level also impact dividend policy helps confirm that investor protection is a significant factor affecting dividend policy. In addition, the results add to a growing literature that uses firm-level measures of governance to study the impact of corporate governance on corporations around the world. Such papers (too numerous to list briefly) include those that measure firm-level governance with ratings, with various measures of ownership structure, and with other indicators of governance.3 Finally, the firm-level findings suggest that individual firms are not entirely trapped by the legal regimes of their home country. By improving corporate governance at the firm level, firms can demonstrate a commitment to protecting investors that translates into real economic outcomes. Section 2 describes the data used in the study. Section 3 presents the empirical results. Section 4 discusses robustness and alternative interpretations. Section 5 examines the results with a broader sample using governance indicators. Section 6 concludes.

2. Data and methodology To measure the strength of corporate governance at the firm level I use corporate governance ratings developed by Credit Lyonnais Securities Asia (CLSA, 2001). CLSA analysts assess the performance of emerging market firms on 57 issues in seven areas of corporate governance.4 These ratings have been used in several other studies5 to examine the impact of corporate governance on firm performance. In rating the firms, analysts are required to give only binary (yes/no) responses to each of the issues, in an effort to reduce subjectivity. Firms are then given a composite rating based on their scores in the areas of management discipline, transparency, independence, accountability, responsibility, fairness, and social responsibility. The first six areas have a 15% weighting in the composite 3

See Denis and McConnell (2003) for a discussion of many of these papers. See Durnev and Kim (2002) for a complete listing of all 57 issues addressed in the survey. 5 Examples include Durnev and Kim (2002), Khanna et al. (2002), Klapper and Love (2002), Chen et al. (2003), and Friedman et al. (2003). 4

T. Mitton / Emerging Markets Review 5 (2004) 409–426

413

score, and social responsibility has a 10% weighting. The rating is on a scale of 1 to 100, with a higher score indicating stronger corporate governance. I adopt CLSA’s composite score as my primary measure of firm-level corporate governance. A few issues regarding use of the CLSA scores should be addressed. First, it is clear that there is a selection bias in the set of firms covered. CLSA chooses firms to cover based on whether the firms are of interest to international investors. Thus, the results presented should be thought of as applying particularly to larger, more visible, firms. Second, it is not clear that bsocial responsibilityQ should be included as part of the rating as it does not relate directly to minority shareholder protection. In addition, the bmanagement disciplineQ rating is a concern because it includes one issue (out of nine) that is at least partially related to dividend payouts.6 To avoid discarding useful information and to avoid tampering with CLSA’s rating, I stick with the CLSA composite score as a base case. The results presented are robust to exclusion of either the social responsibility or management discipline rating (see Section 4). I match the firms in the CLSA study with financial data from the Worldscope database. I use the October 2002 version of Worldscope, in which the latest data reported for most firms is from 2001, which corresponds with the 2001 date of the CLSA report. The primary measure of dividends I use from Worldscope is the dividend payout ratio, which is defined as dividends per share/earnings per share*100. As secondary measures of dividends I also calculate dividends/cash flow and dividends/sales (as in La Porta et al., 2000). Because dividend payouts naturally change over the life cycle of a firm, I try to account for life cycle properties of firms by controlling for firm size and firm growth rates. To measure size, I take from Worldscope the log of total assets of the firm, measured in billions of US$. To measure growth I take from Worldscope the one-year growth rate in total assets, measured in US$. In addition to size and growth, it has been shown that profitability is positively correlated with dividend payouts (Fama and French, 2001), so I use profitability as an additional control variable. (In Section 4, I discuss the impact of profitability in more detail.) Profitability is measured as return on assets and also comes from Worldscope. To avoid undue influence of outliers, growth and profitability are both winsorized at the 5th and 95th percentile. The Worldscope database does not have a match for every firm rated by CLSA. Of the 495 firms from 25 countries included in the CLSA ratings, I am able to match 447 (90%) of the firms with Worldscope. Of that number, I exclude 45 firms from 4 countries identified by La Porta et al. (2000) as mandatory dividend countries—Brazil, Chile, Colombia, and Greece. (The results are robust to including these countries; see Section 4.) Finally, from this number, I exclude 37 firms that are missing necessary financial data in Worldscope. The most common missing item is the growth rate, as it requires two years of available data rather than one. After these adjustments, 365 firms from 19 countries constitute the base sample. Table 1 presents descriptive statistics of the data by country. The median corporate governance rating of all firms in the sample is 55.4. The highest rating in the sample is

6

The exact wording of the question is as follows: bOver the past 5 years, is it true that the company has not built up cash levels, through retained earnings or cash calls, that has brought down ROE?Q

414

Table 1 Descriptive statistics by country Country

Observations Corporate governance rating

ARGENTINA 1 CHINA 12 HONG KONG 39 HUNGARY 1 INDIA 68 INDONESIA 17 KOREA (SOUTH) 22 MALAYSIA 40 MEXICO 7 PAKISTAN 9 PERU 1 PHILIPPINES 20 POLAND 2 RUSSIA 1 SINGAPORE 30 SOUTH AFRICA 29 TAIWAN 40 THAILAND 17 TURKEY 9 Total 365

66.70 45.99 62.46 60.40 56.47 38.39 47.71 56.19 64.50 33.90 75.50 43.87 36.20 15.40 65.78 67.55 54.06 55.15 42.13 55.10

66.70 48.70 67.30 60.40 54.35 36.40 46.50 59.40 67.10 30.70 75.50 38.65 36.20 15.40 64.65 67.40 52.90 54.80 38.40 55.40

Total asset Total assets Profitability Dividends/ Dividends/ Antidirector Legal growth (billion US$) (ROA) cash sales rights origin (US$; %) flow (%) (%)

Mean Median S.D.

Mean

Mean

Mean

Mean

Mean

7.34 8.35 0.16 51.74 33.47 8.84 8.44 9.61 10.71 0.49 24.06 1.83 9.29 32.28 6.61 0.55 10.25 4.99 3.48 10.89

6.633 6.861 27.389 4.005 2.493 1.615 17.790 4.370 17.654 0.869 0.540 2.334 6.120 15.673 8.436 5.661 6.689 4.510 4.694 9.916

0.71 5.74 5.73 7.50 8.25 8.29 2.85 6.20 5.33 7.14 11.50 1.85 2.09 18.31 4.36 5.71 5.04 4.23 4.75 5.75

0.00 23.00 39.34 0.00 19.76 28.64 7.15 21.00 8.98 37.66 37.56 12.21 1.86 12.00 30.97 14.92 9.73 16.55 7.19 20.42

0.00 10.50 11.00 0.00 2.83 4.23 1.09 5.10 2.48 6.71 9.13 2.19 0.15 2.94 5.16 7.52 2.62 3.98 1.04 4.76

0.00 0.00 7.66 40.38 45.62 13.20 42.52 43.14 0.00 0.00 10.06 30.25 24.58 11.43 25.25 27.65 6.91 15.76 12.43 14.47 33.49 30.52 8.61 16.68 10.77 14.45 52.06 63.49 21.60 21.60 12.67 17.39 5.29 3.11 6.74 6.74 14.58 14.58 9.24 30.97 28.18 8.55 30.87 30.97 9.20 27.67 32.44 12.22 29.98 28.78 8.33 20.60 0.00 13.87 30.07 28.50

25.00 27.40 22.54 22.60 16.65 25.39 18.29 28.65 20.01 9.53 23.38 28.59 25.68 34.11 37.76 25.95

4 1 5 3 5 2 2 3 1 5 3 3 3 4 5 3 2 2 3.11

civil civil common civil common civil civil common civil common civil civil civil civil common common civil civil civil

The table reports descriptive statistics of variables used in subsequent tables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Country-level variables are compiled from La Porta et al. (2000) and Claessens et al. (2000).

T. Mitton / Emerging Markets Review 5 (2004) 409–426

Mean Median S.D.

Dividend payout ratio (%)

T. Mitton / Emerging Markets Review 5 (2004) 409–426

415

93.5 (Hong Kong’s HSBC Holdings), and the lowest rating in the sample is 13.9 (Indonesia’s Indocement). The average dividend payout ratio for all firms in the sample is 30.1%. In comparison, the average dividend payout ratio for listed U.S. firms from 1993– 98 was 39.3% (Fama and French, 2001). In addition to the financial data described above, Table 1 also shows measures of country-level investor protection. The two measures presented are antidirector rights and legal origin. As explained in La Porta et al. (1997, 1998), antidirector rights is a composite measure, on a scale of 1 to 5, where a higher number indicates that the country offers more legal protections for equity investors. Legal origin is also important in that common law countries have been shown to offer greater shareholder protection than civil law countries (La Porta et al., 1997, 1998). I compile the country-level data from La Porta et al. (2000) and Claessens et al. (2000). In the results that follow, I estimate the effect of corporate governance on dividend payouts using ordinary least squares regression. I present specifications with and without industry-fixed effects and country-fixed effects, but the specification that I ultimately focus on includes both industry- and country-fixed effects. Because tests detect heteroskedasticity of errors in some specifications, I report heteroskedasticity-adjusted standard errors (White, 1980) throughout the analysis.

3. Results I first examine whether firms with stronger governance have higher dividend payouts. Regressions results are reported in Table 2. In Table 2, the dependent

Table 2 Firms with stronger governance have higher dividend payouts (1)

(2)

Dependent variable is dividend payout ratio Corporate governance 0.302*** 0.323*** (0.105) (0.105) Growth 0.135** (0.062) Profitability

(3) 0.261*** (0.098) 0.237*** (0.062) 1.397*** (0.211)

Size Industry dummies Country dummies N R2

No No 365 0.026

No No 365 0.036

No No 365 0.144

(4)

(5)

(6)

0.245** (0.097) 0.252*** (0.062) 1.572*** (0.219) 1.558* (0.853) No No 365 0.153

0.271*** (0.104) 0.170*** (0.065) 1.537*** (0.251) 1.147 (1.098) Yes No 365 0.292

0.271** (0.133) 0.129* (0.068) 1.527*** (0.269) 1.327 (1.313) Yes Yes 365 0.352

The table reports regression coefficients of dividend payout ratios on corporate governance ratings and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

416

T. Mitton / Emerging Markets Review 5 (2004) 409–426

variable is the dividend payout ratio and the independent variable of interest is the corporate governance rating. Column 1 of Table 2 shows the coefficient on corporate governance with no control variables. The coefficient is 0.302, meaning that a onepoint increase in the corporate governance rating is associated with an increased dividend payout ratio of 0.3% points. The coefficient on corporate governance is significant at the 1% level of confidence. In subsequent columns of Table 2, I add control variables in turn, adding first growth, then profitability, size, industry dummies, and country dummies. The magnitude of the coefficient on corporate governance decreases somewhat as controls are added, but always remains significant at the 5% level or higher. The largest decrease in the magnitude of the coefficient comes with the addition of profitability as a control. (The relative impact of profitability and governance will be discussed further in Section 4.) In the final column, with all controls included, the coefficient on corporate governance is 0.271 and is significant at the 5% level. As mentioned earlier, the coefficient suggests that a onestandard deviation increase in corporate governance is associated with a higher dividend payout of 4% points. In summary, Table 2 demonstrates that stronger corporate governance is associated with higher dividend payouts, possibly reflecting the increased ability of shareholders to limit expropriation by insiders. As will be typical in the tables that follow, in the final column of Table 2, the control variables also have some explanatory power for dividend payouts, as would be expected. The coefficient on growth is negative and significant, probably reflecting that firms with higher growth retain cash for investment. The coefficient on profitability is positive and highly significant, probably reflecting that profitable firms have more cash available for dividends, all else equal. Finally, the coefficient on size is positive and marginally significant in some specifications, indicating that larger firms have higher dividend payouts. In Table 3, I turn to the secondary question of how dividends are affected by the interaction of corporate governance and growth. As put forth by La Porta et al. (2000), when investors are well protected, we should see a stronger negative relationship between dividends and growth. La Porta et al. (2000) demonstrate this with country-level variables, and, in Table 3, we examine the same with variables that vary at the firm level. Our independent variable of interest in Table 3 is corporate governance interacted with growth. As in La Porta et al. (2000), I implicitly view past growth rates as a proxy for future growth opportunities. Column 1 of Table 3 presents the coefficient on this interaction term along with the main effects on corporate governance and growth and with no other control variables. The coefficient on corporate governance interacted with growth is 0.012 and is significant at the 5% level of confidence. The negative coefficient is as expected and indicates that among firms with stronger corporate governance, the negative relationship between growth and dividends is stronger. In subsequent columns, I progressively add control variables as before. In the intermediate columns the coefficient on the interaction between governance and growth declines and even loses significance in Columns 3 and 4. However, once all controls are included in Column 5, the coefficient is again significant at the 5% level with a magnitude of 0.009. In addition, the separate coefficients on corporate governance and growth remain significant and of the expected sign when the interaction term is included. In short,

T. Mitton / Emerging Markets Review 5 (2004) 409–426

417

Table 3 Stronger governance is associated with a stronger negative relationship between dividends and growth (1) Dependent variable is dividend payout ratio Corporate governance 0.293*** (0.102) Growth 0.121* (0.063) Corporate governancegrowth 0.012** (0.005) Profitability

(2) 0.243** (0.098) 0.225*** (0.063) 0.008* (0.005) 1.358*** (0.215)

Size Industry dummies Country dummies N R2

No No 365 0.049

No No 365 0.150

(3)

(4)

(5)

0.230** (0.098) 0.240*** (0.063) 0.007 (0.005) 1.526*** (0.226) 1.457* (0.863) No No 365 0.158

0.258** (0.104) 0.162** (0.066) 0.006 (0.005) 1.500*** (0.258) 1.085 (1.104) Yes No 365 0.295

0.260** (0.130) 0.120* (0.068) 0.009 (0.004) 1.491*** (0.269) 1.443 (1.325) Yes Yes 365 0.358

The table reports regression coefficients of dividend payout ratios on the interaction of corporate governance with growth and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

Table 3 shows that the pattern of dividend payouts makes more sense among firms with stronger corporate governance. Table 4 investigates the interaction of investor protection and corporate governance. First, I investigate whether country-level variables or firm-level variables are more dominant in their impact on dividend policy. Table 4 addresses this question by simply including investor protection and corporate governance as right-hand-side variables simultaneously. Of course, this calls for modified specifications, as country-fixed effects are now excluded (but industry-fixed effects are retained throughout). Column 1 of Table 4 presents the coefficient on antidirector rights without corporate governance. The coefficient on antidirector rights is 3.442 and is significant at the 1% level. The magnitude of the coefficient implies that firms in countries with a higher score of one point (the scale is from 1 to 5) have higher dividend payout ratios, on average, of more than 3% points. This result is essentially a confirmation of the results in La Porta et al. (2000). In Column 2, corporate governance is also included in the regression. Here, the coefficient on antidirector rights falls to 2.396, but it remains significant, although now, only at the 5% level. Meanwhile, the coefficient on corporate governance is 0.282, also significant at the 5% level. In this case, it does not appear that either effect dominates the other; both country-level investor protection and industry-level corporate governance have strong explanatory power for dividend payouts. Columns 3 and 4 repeat the regressions of Columns 1 and 2 but with profitability included as a control. Here, similar relationships are seen, but the coefficients on antidirector rights and corporate governance are smaller and of a lower significance level.

418

T. Mitton / Emerging Markets Review 5 (2004) 409–426

Table 4 Country-level measures of investor protection and firm-level measures of corporate governance both have explanatory power for dividend payouts Antidirector rights (1)

(2)

Legal origin (3)

(4)

(5)

(6)

(7)

(8)

Dependent variable is dividend payout ratio Antidirector rights 3.442*** 2.396** 2.853*** 2.186* (1.160) (1.219) (1.050) (1.123) Common law Corporate governance Growth Size

0.028 (0.067) 0.663 (1.182)

Profitability Country dummies Industry dummies N R2

No Yes 364 0.199

10.672*** 7.551** 9.204*** 7.126** (3.266) (3.422) (2.944) (3.075) 0.282** 0.186* 0.257** 0.175 (0.118) (0.111) (0.115) (0.107) 0.058 0.140** 0.156** 0.039 0.063 0.152** 0.164*** (0.067) (0.063) (0.065) (0.066) (0.066) (0.062) (0.063) 0.405 1.734 1.521 0.884 0.591 1.847 1.616 (1.157) (1.138) (1.131) (1.162) (1.135) (1.127) (1.117) 1.624*** 1.558*** 1.576*** 1.525*** (0.250) (0.255) (0.243) (0.248) No No No No No No No Yes Yes Yes Yes Yes Yes Yes 364 364 364 365 365 365 365 0.215 0.295 0.302 0.205 0.219 0.297 0.303

The table reports regression coefficients of dividend payout ratios on country-level measures of investor protection, corporate governance ratings, and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Country-level variables are compiled from La Porta et al. (2000) and Claessens et al. (2000). Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

Columns 5 through 8 repeat the analysis of Columns 1 through 4, but with legal origin as the country-level variable. The results are similar, except that, now, legal origin appears to have stronger explanatory power than corporate governance, to the extent that corporate governance loses significance in Column 8. On balance then, the evidence suggests that both country-level investor protection and firm-level corporate governance have significant explanatory power for dividends, but that the explanatory power of countrylevel investor protection may be somewhat greater. In Table 5, I turn to the next issue of the interaction of country-level investor protection and firm-level corporate governance. The question I want to address here is whether country-level investor protection and firm-level corporate governance act as complements or substitutes. A priori, it is not obvious which to expect. On the one hand, investor protection and corporate governance may be substitutes, meaning that, when investor protection is weak, firms can have a greater impact on strengthening the rights of their shareholders when they improve corporate governance. On the other hand, investor protection and corporate governance may be complements, meaning that the efforts of individual firms to improve corporate governance may have little

T. Mitton / Emerging Markets Review 5 (2004) 409–426

419

Table 5 Are country-level investor protection and firm-level corporate governance complements or substitutes? Legal origin

Antidirector rights

Common law countries

Civil law countries

All countries

Above median

Below median

All countries

(1)

(2)

(3)

(4)

(5)

(6)

Dependent variable is dividend payout ratio Corporate governance 0.515*** 0.226 (0.153) (0.222) Corporate governance Common law Corporate governance Antidirector rights Growth 0.043 0.271 (0.079) (0.166) Profitability 1.093*** 2.328*** (0.365) (0.413) Size 0.975 3.823* (1.659) (2.304) Country dummies Yes Yes Industry dummies Yes Yes N 232 133 R2 0.391 0.549

0.223* (0.129) 0.583** (0.256)

0.128* (0.069) 1.536*** (0.270) 1.395 (1.316) Yes Yes 365 0.363

0.451** (0.227)

0.035 (0.090) 0.868** (0.427) 1.938 (1.942) Yes Yes 176 0.368

0.011 (0.178)

0.213 (0.130) 2.277*** (0.387) 0.203 (2.137) Yes Yes 188 0.453

0.260** (0.130)

0.164 (0.107) 0.122* (0.069) 1.512*** (0.267) 1.402 (1.321) Yes Yes 364 0.357

The table reports regression coefficients of dividend payout ratios on corporate governance ratings, the interaction of governance with country-level measures of investor protection, and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Country-level variables are compiled from La Porta et al. (2000) and Claessens et al. (2000). Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

effectiveness if the country does not provide a legal environment that respects shareholder rights. I address this question in Table 5 in two ways, by partitioning the sample, and with interaction terms. First, I split the sample between countries with strong investor protection and weak investor protection. As discussed previously, I use two measures of investor protection, legal origin (with common law countries offering better protection), and antidirector rights (with a higher score indicating better protection). In Columns 1 and 2, I split the sample between common law countries and civil law countries. Throughout Table 5, I include all control variables. Column 1 shows that, among firms in common law countries, the coefficient on corporate governance is much greater than it is among all firms. The coefficient is 0.515 (compared to 0.271 in Table 2), and is significant at the 1% level. Meanwhile, Column 2 shows that, among firms in common law countries, the relationship between governance and dividends is actually negative (but not significantly different from zero). Columns 4 and 5 tell a similar story with antidirector rights. Among firms in countries with above-median antidirector rights, the coefficient on corporate

420

T. Mitton / Emerging Markets Review 5 (2004) 409–426

governance is 0.451, but among firms with below-median antidirector rights, the coefficient on corporate governance is 0.011. These results strongly suggest that, regarding dividend policy, country-level investor protection and firm-level corporate governance are complements rather than substitutes. The second approach in Table 5 is to create interactions of investor protection and corporate governance. In Column 3, corporate governance is interacted with a dummy variable equal to one for common law countries. The coefficient on the interaction term is positive and significant at the 5% level. Column 6 confirms this results with corporate governance interacted with antidirector rights. Again the coefficient on the interaction term is positive, although not significant at standard levels. The coefficients on the interaction terms confirm what is demonstrated with the partitioned sample, that investor protection and corporate governance are complements in their impact on dividend policy. This result stands in contrast to the results of Klapper and Love (2002) and Durnev and Kim (2002) who find that firm-level corporate governance and country-level investor protection are substitutes in terms of their impact on firm value.

4. Alternative Interpretations and Robustness If we interpret the results presented in the previous section in the context of the outcome agency model of dividends, then the interpretation is as has been discussed previously. That is, when shareholders are afforded stronger rights by companies that have stronger corporate governance, shareholders use that power to extract dividends. But other interpretations of the results are possible. The first alternative to consider is that firms with stronger governance have improved operating performance which then allows these firms to pay higher dividends (irrespective of the power or wishes of shareholders). In particular, if stronger governance leads to greater profitability, then it may be greater profitability that leads to higher dividend payouts, not the exertion of influence by shareholders. Because a positive correlation between governance and profitability has not yet been established, I first address this issue in Table 6. In Table 6, the dependent variable is profitability. Column 1 shows that corporate governance does indeed have a positive effect on profitability, with a coefficient of 0.058 which is significant at the 5% level. Controls for growth and size are also highly significant, and industry dummies are included throughout. The significance of corporate governance remains if we add either country-fixed effects or indicators of country-level investor protection. (Neither antidirector rights nor legal origin has significant explanatory power for profitability.) In sum, Table 6 establishes a positive association between firm-level corporate governance and profitability. (Again, no causality can be inferred, it could be, for example, that when firms become more profitable, they have the luxury of improving governance.) Returning then to the question of whether improved profitability explains the relationship between governance and dividends, previous tables already demonstrate that this explanation may have some relevance because the coefficient on corporate governance declines when profitability is included as a control variable. However, previous tables also demonstrate that profitability is not the entire explanation, because corporate governance retains significance even when profitability is included as a control. In sum, both the

T. Mitton / Emerging Markets Review 5 (2004) 409–426

421

Table 6 Firms with stronger governance have greater profitability (1) Dependent variable is profitability ratio Corporate governance 0.058** (0.023) Antidirector rights

(2) 0.074*** (0.027)

(3) 0.061** (0.024) 0.135 (0.282)

Common law Growth Size Country dummies Industry dummies N R2

0.066*** (0.020) 0.691*** (0.220) No Yes 365 0.387

0.057*** (0.021) 0.703*** (0.273) Yes Yes 365 0.422

0.063*** (0.020) 0.716*** (0.223) No Yes 364 0.391

(4) 0.054** (0.026)

0.278 (0.757) 0.066*** (0.020) 0.672*** (0.228) No Yes 365 0.388

The table reports regression coefficients of profitability (return on assets) on corporate governance ratings and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets and size is the log of total assets. Country-level variables are compiled from La Porta et al. (2000) and Claessens et al. (2000). Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

indirect effect of profitability and the direct effect of governance have explanatory power for dividend payouts. In subsequent robustness checks, I present results with and without profitability as a control variable. An additional alternative interpretation of the results is that causality is reversed or that both governance and dividends are jointly influenced by an omitted variable. Ideally, I would like to use standard econometric techniques to deal with potential endogeneity. This would involve identifying an instrument, that is, a variable that is correlated with the key independent variable (in this case, the corporate governance rating), but that is otherwise uncorrelated with the dependent variable (in this case, dividend payout ratios). Unfortunately, I am unable to identify an appropriate instrument for this situation. This type of problem is a recurring issue in studies of corporate governance. In one atypical case, Black et al. (2002) are able to use an instrumental variables approach to establish causation running from corporate governance to firm value, but their instrument is specific to institutional details of Korea and is not suited to this cross-country analysis. The bottom line for this study is that, lacking a suitable instrument for corporate governance, it is not possible to rule out alternative explanations based on endogeneity. In Table 7, I turn to some additional regressions to assess the robustness of the main result along different dimensions. In each case, I present results with and without profitability as a control variable. In the first two robustness checks I alter the corporate governance rating to exclude the discipline measure and the social responsibility measure respectively. The reasons for potentially wanting to omit these variables are discussed in

422

Table 7 Robustness checks Exclude social responsibility measure from rating

Dividends/cash flow as dependent variable

Dividends/sales as dependent variable

Include mandatory dividend countries

Exclude financial firms

Control for country’s tax advantage of dividends

(1)

(3)

(5)

(7)

(9)

(11)

(13)

(2)

(4)

(6)

(8)

(10)

(12)

(14)

Dependent variable is dividend payout ratio (except where noted) Corporate 0.336** 0.247* 0.345*** 0.239* 0.278** 0.172 0.056 0.024 0.337** 0.224* 0.471*** 0.328** 0.456*** 0.339*** governance (0.131) (0.132) (0.123) (0.124) (0.135) (0.133) (0.035) (0.036) (0.134) (0.134) (0.155) (0.155) (0.114) (0.109) Growth 0.036 0.127* 0.039 0.127* 0.028 0.105* 0.009 0.016 0.037 0.135** 0.028 0.135* 0.069 0.158** (0.071) (0.068) (0.071) (0.068) (0.057) (0.054) (0.020) (0.018) (0.069) (0.066) (0.079) (0.076) (0.072) (0.069) Size 0.274 1.364 0.234 1.315 0.012 0.930 0.133 0.175 0.541 0.813 1.025 1.954 0.487 1.499 (1.398) (1.321) (1.387) (1.316) (1.178) (1.153) (0.632) (0.584) (1.415) (1.284) (1.489) (1.402) (1.224) (1.181) Profitability 1.550*** 1.533*** 1.350*** 0.438*** 1.642*** 1.539*** 1.457*** (0.268) (0.269) (0.256) (0.077) (0.262) (0.272) (0.257) Dividend tax 7.126 3.190 advantage (12.186) (11.766) Country Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No dummies Industry Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes dummies N 365 365 365 365 359 359 365 365 402 402 278 278 339 339 R2 0.268 0.351 0.271 0.351 0.373 0.436 0.337 0.409 0.259 0.349 0.313 0.408 0.226 0.303 The table reports regression coefficients of measures of dividends on corporate governance ratings and control variables. Corporate governance ratings come from CLSA (2001). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Mandatory dividend countries added to the sample include Brazil, Chile, Colombia, and Greece. Financial firms are defined as those with primary SIC in the range 6000–6999. Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

T. Mitton / Emerging Markets Review 5 (2004) 409–426

Exclude discipline measure from rating

T. Mitton / Emerging Markets Review 5 (2004) 409–426

423

Section 2. Columns 1 through 4 demonstrate that the results are robust to excluding these measures, as the coefficient on corporate governance retains significance at the 10% level or higher whether or not profitability is included as a control variable. In the next two robustness checks, I use alternative definitions of dividend payouts, dividends/cash flow and dividends/sales. While these are not standard definitions of dividend payouts, they are used in La Porta et al. (2000) as alternative definitions. These results are presented in Columns 5 through 8 of Table 7. Here, the robustness of the results does not fare as well. With dividends/cash flow as the dependent variable, the coefficient on corporate governance is significant in one specification, but, with dividends/sales as the dependent variable, the coefficient on corporate governance is not significant, although it retains the expected sign. In the next two robustness checks presented, I change the sample in two important ways. Columns 9 and 10 repeat the basic results but with countries designated as mandatory dividend countries now included in the regressions. The coefficient on corporate governance is significant in both of these specifications. Columns 11 and 12 repeat the analysis with but with financial firms (SICs in the 6000s) excluded. Without financial firms the results are even stronger than in the full sample. Finally, in Columns 13 and 14, I present one more robustness check, where I control for a country’s tax advantage of dividends. This measure is taken from La Porta et al. (2000). The results are robust to inclusion of this variable, which in itself, has little explanatory power for dividend payouts.

5. Tests using a broader sample of firms As a final test of the robustness of the results, Table 8 presents results using governance indicators in place of the CLSA ratings. These indicators are rougher measures of governance than the CLSA ratings, but they have the advantage that they can be examined for a larger set of firms. The intent from examining this data is to get an indication as to whether the results presented previously apply to a broader set of firms. An alternative, and sharper, way of expanding the sample would be to use as the governance measure (or measure of susceptibility to agency problems) the divergence between cash-flow rights and control rights, as has been done in a number of papers including Lins (2003), Lemmon and Lins (2003), Harvey et al. (in press), and Faccio et al. (2001). Lacking this detailed data, I rely on financial data from Worldscope, but now using indicators that allow the sample to include over 14,000 firms from 50 countries. I examine three indicators of governance that are available for the large sample. The first is a dummy variable that equals one if the firm operates in only one industry, with industries defined at the two-digit SIC level. A focused firm is used as an indicator of good governance both in the CLSA study and in previous research (see, e.g., Mitton, 2002 and Friedman et al., 2003). The second indicator is a dummy variable that equals one if the auditor of the firm is one of the Big Five international accounting firms.7 Having a Big Five auditor has been used previously as an indicator of higher disclosure quality (Titman and Trueman, 1986; Reed et

7

The data come from 2001, prior to the breakup of Arthur Andersen.

424

T. Mitton / Emerging Markets Review 5 (2004) 409–426

Table 8 Governance indicators in an expanded sample (1) Dependent variable is dividend payout ratio Focused 0.912 ** (0.443) Big Five Auditor

(2)

2.113*** (0.542)

Cross-listed Focusedgrowth

5.451*** (1.047) 0.042*** (0.016)

Big Five Auditorgrowth

0.023 (0.016)

Cross-listedgrowth Growth Size Profitability N R2

(3)

0.056*** (0.009) 2.511*** (0.112) 1.487*** (0.034) 14766 0.270

0.057*** (0.009) 2.343*** (0.116) 1.484*** (0.034) 14766 0.271

0.056 (0.041) 0.058*** (0.009) 2.587*** (0.114) 1.489*** (0.034) 14766 0.271

(4) 0.933** (0.442) 2.144*** (0.542) 5.523*** (1.048) 0.042** (0.016) 0.022 (0.016) 0.050 (0.041) 0.056*** (0.009) 2.501*** (0.120) 1.483*** (0.034) 14766 0.272

The table reports regression coefficients of dividend payout ratios on corporate governance indicators, interactions of the indicators with growth, and control variables. Focused means the firm operates in just one two-digit SIC industry. Big Five Auditor means the name of the firm’s auditor is one of the Big Five international firms. Cross-listed means the firm’s stock is listed in the U.S. (directly or as a Level II or III ADR). Financial data come from Worldscope. Growth is the 1-year growth rate in total assets, profitability is return on assets, and size is the log of total assets. Heteroskedasticity–robust standard errors are reported below coefficients in parentheses. Also estimated but not reported are full sets of industry and country dummy variables. * 10% level of significance. ** 5% level of significance. *** 1% level of significance.

al., 2000; Mitton, 2002), which is an important element of corporate governance (see La Porta et al., 1998). The third indicator is a dummy variable that equals one if the firm’s stock is cross-listed on a U.S. exchange, either directly or with a Level II or III depository receipt. Because such a cross-listing subjects the firm to a higher level of disclosure requirements and investor scrutiny, it may bond the firm to maintain a higher standard of governance (see Coffee, 1999; Stulz, 1999 and Reese and Weisbach, 2002 for examples of this view). Table 8 presents results of regressions of dividend payout ratios on these indicators. Column 1 shows that focused firms have higher dividend payout ratios than diversified firms, even after controlling for growth, size, profitability, industry, and country. The difference does not appear large (focused firms have higher payout ratios of just under one percentage point) but it is statistically significant at the 5% level. Column 1 also shows that the negative relationship between growth and dividends is stronger among focused firms, suggesting that focused firms allocate capital more efficiently. This coefficient is significant at the 1% level. Column 2 repeats the analysis for the Big Five auditor indicator. Firms with Big Five auditors have higher payout ratios (by around two

T. Mitton / Emerging Markets Review 5 (2004) 409–426

425

percentage points) and this difference is significant at the 1% level. The negative relationship between growth and dividends is also stronger among firms with Big Five auditors, although this difference is not significant. Column 3 shows the same results for the cross-listed indicator. Here, the results are not as expected, as cross-listed firms have lower dividend payouts. However, the negative relationship between growth and dividends is still stronger among cross-listed firms (though again not significant). Column 4 shows that the same results hold with all three indicators included simultaneously. In summary, although the evidence is mixed for cross-listed firms, on balance, these results seem to support, in a broader sample, that stronger governance is associated with higher dividend payouts and a stronger negative relationship between growth and dividends.

6. Conclusion The ultimate goal of corporate governance is to ensure that suppliers of finance to corporations receive a return on their investment (Shleifer and Vishny, 1997). While suppliers of equity can receive a return through dividends or capital gains, agency theory suggests that shareholders may prefer dividends, particularly when they fear expropriation by insiders. The outcome agency model tells us that when shareholders have greater rights, they can use their power to influence dividend policy. Shareholders can receive greater rights either through a country’s legal protection or through a firm’s governance practices that may not be mandated by government. This paper shows that firm-level corporate governance, in addition to country-level investor protection, is associated with higher dividend payouts, suggesting that both mechanisms help reduce agency problems. In addition, stronger corporate governance is shown to be associated with a stronger negative relationship between growth and dividends, demonstrating that the pattern of dividend payouts makes more sense among firms with stronger corporate governance. The results suggest that, when shareholders are well protected, either by governments or by corporations themselves, capital can be allocated more efficiently.

Acknowledgement I appreciate the helpful comments of Jim Brau, Simon Johnson, Mike Lemmon, Grant McQueen, Sendhil Mullainathan, Mike Pinegar, David Scharfstein, and Jeremy Stein on earlier versions of the paper. All errors are mine.

References Black, B., Jang H., Kim W., 2002, Does corporate governance affect firm value? Evidence from Korea, Working paper, Stanford Law School. Brav, A., Graham, J., Harvey, C., Michaely, R., 2003. Payout policy in the 21st century, NBER Working paper #9657. Chen, K.C.W., Chen, Z., Wei, K.C.J., 2003. Disclosure, corporate governance, and the cost of equity capital: evidence from Asia’s emerging markets, Working paper, Hong Kong University of Science and Technology.

426

T. Mitton / Emerging Markets Review 5 (2004) 409–426

Claessens, S., Djankov S., Nenova T., 2000. Corporate risk around the world, Working Paper, World Bank. Coffee Jr., J., 1999. The future as history: the prospects for global convergence in corporate governance and its implications. Northwestern University Law Review 93, 641 – 707. Credit Lyonnais Securities Asia, 2001. Saints and sinners: who’s got religion? Published report. Denis, D., McConnell, J., 2003. International corporate governance. Journal of Financial and Quantitative Analysis 38, 1 – 36. Durnev, A., Kim E.H., 2002. To steal or not to steal: firm attributes, legal environment, and valuation, Working Paper, University of Michigan. Easterbrook, F., 1984. Two agency-cost explanations of dividends. American Economic Review 74, 650 – 659. Faccio, M., Lang, L.H.P., Young, L., 2001. Dividends and expropriation. American Economic Review 91, 54 – 78. Fama, E.F., French, K.R., 2001. Disappearing dividends: changing firm characteristics or lower propensity to pay. Journal of Financial Economics 60, 3 – 43. Friedman, E., Johnson, S., Mitton, T., 2003. Propping and tunneling. Journal of Comparative Economics 31, 732 – 750. Gugler, K., Yurtoglu, B., 2003. Corporate governance and dividend pay-out policy in Germany. European Economic Review (in press). Harvey, C., Lins, K.V., Roper, A., 2004. The effect of capital structure when expected agency costs are extreme. Journal of Financial Economics (in press). Investor Protection Association, 2003. dLUKOILT fixed the dividend policy it is carrying out, www.corp-gov.org, January 9. Jensen, M.C., 1986. Agency costs of free-cash flow, corporate finance, and takeovers. American Economic Review 76, 323 – 329. Khanna, T., Kogan, J., Palepu, K., 2002. Globalization and similarities in corporate governance: a cross-country analysis, Working Paper, Harvard Business School. Klapper, L.F., Love, I., 2002. Corporate governance, investor protection, and performance in emerging markets. World Bank Policy Research Working Paper 2818. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 1997. Legal determinants of external finance. Journal of Finance 52, 1131 – 1150. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 1998. Law and finance. Journal of Political Economy 106, 1115 – 1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 2000. Agency problems and dividend policies around the world. Journal of Finance 55, 1 – 33. Lemmon, M.L., Lins, K.V., 2003. Ownership structure, corporate governance, and firm value: evidence from the East Asian financial crisis. Journal of Finance 58, 1445 – 1468. Lins, K.V., 2003. Equity ownership and firm value in emerging markets. Journal of Financial and Quantitative Analysis 38, 159 – 184. Mitton, T., 2002. A cross-firm analysis of the impact of corporate governance on the East Asian financial crisis. Journal of Financial Economics 64, 215 – 241. Myers, S.C., 2000. Outside equity. Journal of Finance 55, 1005 – 1037. Nicholson, A., 2003. Independent directors find seat on the board. Moscow Times 2, 14 (October). Reed, B., Trombley, M., Dhaliwal, D., 2000. Demand for audit quality: the case of Laventhol and Horwath’s auditees. Journal of Accounting, Auditing and Finance 15, 183 – 198. Reese Jr., W., Weisbach, M., 2002. Protection of minority shareholder interests, cross-listings in the United States, and subsequent equity offerings. Journal of Financial Economics 66, 65 – 104. Shleifer, A., Vishny, R.W., 1997. A survey of corporate governance. Journal of Finance 52, 737 – 783. Stulz, R., 1999. Globalization of equity markets and the cost of capital, NBER Working Paper #7021. Titman, S., Trueman, B., 1986. Information quality and the valuation of new issues. Journal of Accounting and Economics 8, 159 – 172. White, Hal, 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817 – 838.