Dividend policy: Shareholder rights and creditor rights under the impact of the global financial crisis

Economic Modelling xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Economic Modelling journal homepage: www.elsevier.com/locate/econmod

Dividend policy: Shareholder rights and creditor rights under the impact of the global financial crisis ⁎

Quoc Trung Trana,b, , Pascal Alphonsec, Xuan Minh Nguyenb a b c

University of Lille 2, Lille, France Foreign Trade University, Vietnam Skema Business School-LSMRC, Lille, France

A BS T RAC T The extant literature shows that shareholder and creditor rights positively affect corporate payout policy in a static macroeconomic environment. This study examines how the effects of shareholder and creditor rights on dividend policy change under the impact of the global financial crisis. We posit that this exogenous shock increases agency costs of both shareholders and creditors. With a sample of 133,631 firm-year observations from 23,890 firms incorporated in 41 countries, we find that both shareholder and creditor rights are less effective in dividend decisions in the post-crisis period and the extent of shareholder (creditor) expropriation in the post-crisis period is larger when creditors (shareholders) are adequately protected.

1. Introduction The extant literature on agency theory shows that there are two types of agency costs including agency costs of shareholders and agency costs of creditors. La Porta et al. (2000) initially suggest two competing dividend models based on agency costs of shareholders, namely the outcome model and the substitute model. The former predicts that stronger protection of minority shareholders results in higher dividend level whilst the latter argues that dividends are paid as a means to earn a reputation of fair treatment of minority shareholders and weaker minority shareholder protection leads to higher payout ratio. La Porta et al. (2000) find that the outcome model is supported empirically. Then, Brockman and Unlu (2009) extend this line of research with an argument that dividend restriction is a compensation for weak creditor rights under private credit agreements and they find supporting evidence of the substitute model based on agency costs of creditors. Recent studies continue to examine dividend policy with the interactions between shareholder and creditor rights (Byrne and O’Connor, 2012; Shao et al., 2013). While prior studies investigate how shareholder and creditor protection affects payout policy in a static macroeconomic environment, this study examines how the effects of shareholder and creditor rights on dividend policy change under the impact of the global financial crisis. We posit that the financial crisis is an exogenous shock to all firms and it makes agency costs of both equity and debt tend to increase. According to Johnson et al. (2000), a financial crisis is an

⁎

exogenous shock lowering expected return on investment opportunities. Ceteris paribus, this shock to returns reduces the marginal cost that managers incur to divert firm resources away from positive NPV projects and they are more likely to expropriate minority shareholders (Lemmon and Lins, 2003). Therefore, agency costs of equity increase. Besides, we posit that agency costs of debt also increase due to the two following reasons: Firstly, the financial crisis leads to external financial constraints. When firms are not able to raise external funds easily, they are less willing to establish a reputation of fair treatment of creditors. Therefore, ceteris paribus firms tend to pay more dividends and creditors are expropriated more severely. Secondly, when profitable investment opportunities are less available, firms tend to use dividends as a means to build up a good reputation on their performance and prevent rapid decreases in their stock prices. As a result, they are less willing to meet creditor demand to restrict dividends as a substitute of weak creditor rights under private credit agreements. We begin the study by comparing the effects of shareholder and creditor rights on both dividend paying propensity and dividend magnitude across 41 countries between the pre-crisis period from 2003 to 2007 and the post-crisis period from 2008 to 2012. After firm characteristics (i.e. profitability, cash holdings, firm growth, debt ratio, asset tangibility, firm size and maturity) and country-level variables (i.e. GDP growth rate, national culture and legal origin dummy) are controlled, we find that both shareholder and creditor rights are less effective in the post-crisis period. Then, we classify countries into groups based on level of shareholder (creditor) protection and replicate

Correspondence to: Foreign Trade University, 15 D5 Street, Ward 25, Binh Thanh District, Ho Chi Minh City, Vietnam. E-mail addresses: [email protected], [email protected] (Q.T. Tran).

http://dx.doi.org/10.1016/j.econmod.2017.03.010 Received 8 October 2016; Received in revised form 11 February 2017; Accepted 13 March 2017 0264-9993/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Tran, Q.T., Economic Modelling (2017), http://dx.doi.org/10.1016/j.econmod.2017.03.010

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

the regression models for each group. The findings show that creditors (shareholders) are expropriated more severely in the post-crisis period if shareholders (creditors) are sufficiently protected. In addition, our robustness checks with panel data regression, the reduced sample, alternative measures of shareholder and creditor protection and additional country-level control variables also indicate consistent findings. This paper contributes to the extant literature by showing that the financial crisis leads to more expropriation of both shareholders and creditors and the extent of expropriation of shareholders (creditors) is larger when creditors (shareholders) are adequately protected. The remaining of this paper is organized as follows: Section 2 presents relevant literature review, Section 3 shows research methodology, Section 4 reports empirical findings, Section 5 is robustnest checks and Section 6 presents conclusions.

that creditor rights are positively related to both the likelihood of dividend payment and payout ratio. Byrne and O’Connor (2012) investigate whether shareholder rights at both country-level and firm-level affect corporate dividend policy when creditors demand. With a sample of 22,374 firms listed in 35 countries, they find that creditor rights, both country-level and firm-level shareholder rights play a significant role in payout policy. Creditors have the greatest impact dividend payment decisions. The outcome model based on agency costs of equity is most effective when creditor protection is strong. In countries of weak creditor protection, creditors are demanding and firms pay lower dividends. Shao et al. (2013) continue to examine the impact of power balance between equity and debt claimants on the optimal dividend policy around the world. They argue that optimal dividend policy is obtained when the marginal agency cost of equity is equal to the marginal agency cost of debt. When shareholder rights are sufficiently strong, firms have more discretion to use dividends as means to improve the relationship between firms and creditors and dividend policy is more sensitive to creditor rights. Conversely, when creditors are adequately protected, firms are more flexible to adjust dividends by the quality of shareholder protection. With a sample of 139,168 observations collected from 39 countries from 1991 to 2010, their empirical research shows that creditor (shareholder) rights are more effective when shareholders (creditors) are strongly protected. Furthermore, Boţoc and Pirtea (2014) analyze behavior of 2636 companies listed in 16 emerging stock markets to examine drivers of dividend policy. Their sensitivity analysis shows that cash needs are more significant to explain payout policy in countries of strong investor protection while liquidity is more important in explaining dividend policy in countries of weak investor protection. Moreover, the extant literature also shows that a financial crisis is a good opportunity to examine corporate governance. Johnson et al. (2000) argue that the Asian financial crisis reduces the available return on investment opportunities and insiders incur lower marginal costs of diverting resources away from profitable investment projects. Therefore, expropriation of minority shareholders becomes more severe. With a sample of 25 emerging markets, they find that more expropriation of shareholders by managers during the crisis can explain the decline in stock markets more effectively than standard macroeconomic measures. Mitton (2002) finds that individual firms can preclude expropriation of minority shareholders when they are not sufficiently protected by law during the East Asian financial crisis. Firms with higher disclosure quality and outside ownership concentration tend to have better performance over the crisis period. Lemmon and Lins (2003) examining the impact of ownership structure on firm value during the East Asian financial crisis and also find that firms where managers and their families separate their control rights and cash flow ownership have lower cumulative stock returns than other firms. Recently, Al-Malkawi et al. (2014) investigate dividend smoothing of firms listed on the Muscat Securities Market under the impact of the global financial crisis. They add a dummy variable to the Lintner's (1956) partial adjustment model to compare dividend smoothing in the pre-crisis period and the post-crisis period. The research findings show that there is no significant effect of the financial crisis on dividend stability. While prior studies examine how legal protection of shareholders and creditors affect dividend policy in a static macroeconomic environment, this paper investigates this relationship under the impact of the global financial crisis which is considered as an exogenous shock increasing agency costs of both shareholders and creditors. We hypothesize that shareholder and creditor rights are less effective in the post-crisis period.

2. Literature review Jensen and Meckling (1976) define agency relationship as an agreement under which agents perform some service on behalf of principals and posit that the principal-agent relation leads to agency costs. Later studies develop the agency model in broad terms with two kinds of interest conflicts: the conflicts of interest among equity claimants, and the conflicts of interest between equity and debt holders (Brockman and Unlu, 2009). Accordingly, there are two types of agency costs including agency costs of equity and agency costs of debt. The agency costs of equity are commonly utilized to explain incentive problems in corporate dividend policy at firm level. Jensen (1986) posits that excessive funds which are available to managers are a source of agency costs. If firms’ cash flow exceeds that required to finance profitable business projects, corporate managers are motivated to invest in negative net present value projects. Therefore, firms pay cash dividend to mitigate agency costs. When investigating the impact of shareholder rights on dividend policies around the world, La Porta et al. (2000) develop two opposite agency models including outcome model and substitute model. The former considers dividend payment as an outcome of effective legal protection of shareholders. When legal protection of minority shareholders is strong, outsiders use their legal rights to force insiders to pay dividends. The latter argues that if firms want to raise external funds with favorable conditions, they must earn a reputation of fair treatment of minority shareholders. One way to establish a good reputation is reducing excessive free cash flow which is available to insiders by paying dividends. Consequently, weaker minority shareholder protection results in higher payout ratio. Using a sample of 4000 companies from 33 countries, La Porta et al. (2000) find that the outcome model is supported, stronger shareholder rights result in higher dividend payout. In addition, Jiraporn et al. (2011) and Mitton (2004) find support in favor of the outcome model based on agency cost of shareholders at firm level in the U.S market and 33 emerging markets respectively. O’Connor (2013) also finds that dividend payment is the outcome of effective corporate governance with a research sample of 220 firms listed in 21 emerging stock markets. They also find that dividend payment is likely to be a substitute for a lack of transparency in these markets. Furthermore, Brockman and Unlu (2009) extend the line of research at country level with an argument that country-level creditor protection also affects dividend policies. They claim that beside the reputation-building mechanism, dividend constraints are also determined by credit contracts. Under a legal regime of weak creditor protection, creditors have stronger incentives to demand control over the corporate decision-making process via private credit agreements. Consequently, the two contracting parties are more likely to consent to restrict dividends as a compensation for poor creditor rights. Using a research sample of 120,507 observations across 52 countries, they find

2

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

(Shao et al., 2013). Creditor protection is proxied by the revised creditor right aggregate score (CRE) from Djankov et al. (2007). Annual GDP growth rate (GDP) is from the IMF database. National culture variable is the uncertainty avoidance dimension (UAD) from Hofstede et al. (2010).2 The legal origin dummy (LAW) takes 1 for civil law countries and 0 for common law countries. Dividend payout ratio is also a function of independent variables, a period dummy variable and their interactions. Since the dependent variable is continuous to the right of zero, OLS regression for the full sample including non-payers and payers or the sub-sample of payers leads to biased results. Consequently, pooled tobit regression model is employed instead (Huang, 2001; Kim and Maddala, 1992).

3. Methodology 3.1. Research models Following Brockman and Unlu (2009) and Shao et al. (2013), we use two pooled logit and tobit regression models for the likelihood of paying dividends and dividend payout ratios with independent variables at both country-level and firm-level. In order to investigate how the effects of shareholder and creditor rights on dividend policy across countries are different between the pre-crisis period 2003 − 2007 and the post-crisis period 2008 − 2012,1 we add interactive terms between each independent variable with the period dummy variable (PER) to both regression models. The period dummy variable takes the value of 1 for the post-crisis period and 0 otherwise. If the effects of shareholder and creditor protection on dividend policy are identical between the two periods, their interactive terms’ coefficients are statistically insignificant. The likelihood to pay dividends is developed as a logit function of firm-specific variables, country-specific variables, a period dummy variable and their interactions as follows:

DTE = α + β1PRO + γ1PRO*PER + β2CAS + γ2CAS*PER + β3GRO + γ3GRO*PER + β4DEB + γ4DEB*PER + β5TAN + γ5TAN*PER + β6RET + γ6RET*PER + β7SIZ + γ7SIZ*PER + β8ASD + γ8ASD*PER + β9CRE + γ9CRE*PER + β10GDP + γ10GDP*PER + β11UAD + γ11UAD*PER + β12LAW + γ12LAW*PER + δPER + ε

PAY* = α + β1PRO + γ1PRO*PER + β2CAS + γ2CAS*PER + β3GRO

Where: Dividends to earnings (DTE) is measured by total cash dividends scaled by net income. We also run the regression models with dividends to total assets (DTA) and dividends to net sales (DSA) as robustness checks. According to Shao et al. (2013), when making decisions on dividend policy, firm managers balance the benefits of minority shareholders and creditors. Therefore, this study continues to examine how the financial crisis affects the effectiveness of creditor rights (shareholder rights) by level of shareholder rights (creditor rights). There are two pairs of sub-samples: The first pair includes countries of weak and strong shareholder protection. Countries with ASD lower than or equal to the full sample median are of weak shareholder protection, the others are of strong shareholder protection. The second pair includes countries of weak and strong creditor protection. Countries with CRE lower than the full sample median are of weak creditor protection, the others are of strong creditor protection. The regression models for the first pair of sub-samples are as follows:

+ γ3GRO*PER + β4DEB + γ4DEB*PER + β5TAN + γ5TAN*PER + β6RET + γ6RET*PER + β7SIZ + γ7SIZ*PER + β8ASD + γ8ASD*PER + β9CRE + γ9CRE*PER + β10GDP + γ10GDP*PER + β11UAD + γ11UAD*PER + β12LAW + γ12LAW*PER + δPER + ε.

(2)

(1)

Where: PAY is 1 if firms pay dividends and 0 otherwise. Firm-level variables employed in this research are found to affect corporate dividend policy in many prior studies. In line with Fama and French (2001) and Brockman and Unlu (2009), profitability (PRO) is net income divided by total assets. Firms with higher profitability are more likely to pay dividends. Cash holdings (CAS) are measured by cash balance divided by total assets. According to DeAngelo et al. (2006), the expected sign of cash holdings is ambiguous. Cash holding are high due to two causes including accumulated cash flows and the need to invest in future business opportunities. The former leads to high dividends but the latter results in low dividends. Firm growth (GRO) is the ratio of current year's change in net sales. Transaction cost theory, residual theory, and pecking order theory argue that firms prefer internal to external financing; therefore, firms with higher growth rate have less flexibility to pay dividends. Debt ratio (DEB) is the ratio of long-term debt to total assets, asset tangibility (TAN) is net property, plant and equipment scaled by total assets and firm size (SIZ) is the natural logarithm of total assets measured in U.S. dollars. Firms with higher debt ratio, low asset tangibility and smaller size incur higher costs of external financing, hence they are less likely to pay dividends. Firm maturity (RET) is retained earnings to assets. According to life cycle theory developed by Grullon et al. (2002), when firms are more mature, they tend to pay more dividends. The expected signs for firm-specific variables are as follows: PRO (+), CAS (+/-), GRO (-), DEB (-), TAN (+), SIZ (+) and RET (+). Country-level variables include shareholder protection, creditor protection, annual GDP growth rate, national culture and legal origin dummy. Shareholder protection is proxied by the anti-self-dealing index (ASD) from Djankov et al. (2008). The anti-self-dealing index is measured by the first principal component of factors focusing on the strength of minority shareholder protection against the controlling shareholder's self-dealing (i.e. disclosure, approval, and litigation); therefore, it represents the pressure of disgorging cash on insiders

PAY* = φ + λ1PRO + η1PRO*PER + λ 2CAS + η2 CAS*PER + λ 3GRO + η3GRO*PER + λ4DEB + η4 DEB*PER + λ5TAN + η5TAN*PER + λ 6RET + η6 RET*PER + λ 7SIZ + η7SIZ*PER + λ 8CRE + η8CRE*PER + λ 9GDP + η9 GDP*PER + λ10UAD + η10 UAD*PER + θPER + ε

(3)

DTE = φ + λ1PRO + η1PRO*PER + λ 2CAS + η2 CAS*PER + λ 3GRO + η3GRO*PER + λ4DEB + η4 DEB*PER + λ5TAN + η5TAN*PER + λ 6RET + η6 RET*PER + λ 7SIZ + η7SIZ*PER + λ 8CRE + η8CRE*PER + λ 9GDP + η9 GDP*PER + λ10UAD + η10 UAD*PER + θPER + ε

(4)

We compare the coefficients of the interactive term CRE*PER in regression results of the strong and weak shareholder protection groups to find whether the extent of creditor expropriation is larger when shareholders are sufficiently protected by law. For the second pair of sub-samples, we replace creditor right index CRE with shareholder right index ASD in the two Eqs. (3) and (4). Then, we compare 2 Bae et al. (2012) posit that three culture proxy variables including uncertainty avoidance (UAD), masculinity (MAS) and long-term orientation (LTO) are highly correlated to each other and suggest that these dimensions should be inserted into regression models separately. We also replicate both logit and tobit models with MAS and LTO separately and find similar core results.

1 The core findings persist when we define the pre-crisis period as five years from 2002 to 2006 and the post-crisis period from 2007 to 2011.

3

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 1 Summary statistics. PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DTE is total cash dividends scaled by net income. PRO is net income divided by total assets. CAS is cash balance divided by total assets. GRO is the ratio of current year's change in net sales. DEB is the ratio of long-term debt to total assets. TAN is net property, plant and equipment scaled by total assets. SIZ is the natural logarithm of total assets measured in U.S. dollars. RET is retained earnings to assets. ASD is anti-self-dealing index from Djankov et al. (2008). CRE is revised creditor right index from Djankov et al. (2007). UAD is uncertainty avoidance dimension from Hofstede et al. (2010). Panel A - Firm‐level data Full sample

DTE (%) PRO (%) CAS (%) GRO (%) DEB (%) TAN (%) RET (%) SIZ Payers (%) No. of observations

Pre-crisis period

Mean

SD

Mean

SD

Mean

SD

44.54 6.27 14.71 14.88 10.95 28.67 18.06 12.26

36.13 4.95 13.37 23.70 12.04 20.38 22.12 1.84

41.57 6.45 14.20 17.42 11.29 28.99 16.68 12.16

34.60 4.98 13.31 23.94 12.17 20.37 21.52 1.84

47.41 6.09 15.23 12.32 10.60 28.35 19.46 12.36

37.33 4.90 13.41 23.18 11.89 20.39 22.62 1.83

65.81 133,631

Panel B - Annual number of firms Year 2003 2004 2005

N 11,522 13,005 13,910

Year 2006 2007 2008

Panel C - Industry distribution SIC industry definition Mineral industries Construction industries Manufacturing Transportation, communications

2‐digit SIC 10–14 15–17 20–39 40–48

Panel D - Country-level data Country Argentina Australia Belgium Canada Hong Kong India Ireland Israel Kenya Malaysia New Zealand Nigeria Pakistan Singapore South Africa Sri Lanka Thailand United Kingdom USA Common law median Austria Denmark Egypt Finland France Germany Indonesia Italy Japan Jordan Mexico Netherlands Norway Peru Philippines Portugal South Korea

No. of firms 53 1022 106 1057 210 2430 52 282 30 925 100 62 233 633 291 177 446 1,400 4,575 282 72 118 70 122 643 646 325 211 3338 103 93 153 200 71 130 49 993

No. of observations 349 4482 626 4363 1230 16,191 260 1405 183 5655 588 379 1281 3821 1762 983 2979 6877 22,259 1405 443 604 225 772 3851 3549 1861 1066 22,229 551 600 886 916 527 777 285 4115

Post-crisis period

64.24 67,233

N 14,450 14,346 12,364

N 6432 4919 74,033 8113

Payer (%) 53.87 65.17 68.85 50.10 68.37 61.31 44.23 54.02 78.69 71.12 80.95 83.38 82.75 69.09 66.12 74.16 81.20 68.43 39.66 68.43 71.33 69.87 79.11 86.79 68.97 56.02 55.02 72.14 90.42 20.87 60.33 69.30 54.48 63.95 57.66 69.12 77.69

4

Year 2009 2010 2011

67.40 66,398

N 12,827 14,211 13,626

SIC industry definition Wholesale trade Retail trade Service industries

DTE (%) 43.17 59.59 46.65 58.97 44.23 27.29 37.71 66.78 47.45 46.56 68.82 52.24 37.42 50.14 44.01 34.88 59.18 49.15 40.31 46.65 46.02 44.82 70.61 67.81 44.13 49.11 41.31 54.29 38.97 77.66 42.99 46.78 56.89 55.14 49.42 54.33 30.89

ASD 0.34 0.76 0.54 0.64 0.96 0.58 0.79 0.73 0.21 0.95 0.95 0.43 0.41 1.00 0.81 0.39 0.81 0.95 0.65 0.73 0.21 0.46 0.20 0.46 0.38 0.28 0.65 0.42 0.50 0.16 0.17 0.20 0.42 0.45 0.22 0.44 0.47

Year 2012

N 13,370

2‐digit SIC 50–51 52–59 > =70

N 7457 8130 24,547

CRE UAD 1 86 3 51 2 94 1 48 4 29 2 40 1 35 3 81 4 50 3 36 4 49 4 55 1 70 3 8 3 49 2 45 2 64 4 35 1 46 3 49 3 70 3 23 2 80 1 59 0 86 3 65 2 48 2 75 1 92 1 65 0 82 3 53 2 50 0 87 1 44 1 104 3 85 (continued on next page)

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 1 (continued) Panel A - Firm‐level data Full sample

Spain Sweden Switzerland Taiwan Turkey Civil law median Full sample median

642 1940 1284 9947 888 887 1066

Pre-crisis period

Post-crisis period

Mean

SD

Mean

SD

Mean

SD

112 367 195 1608 187 170 200

74.45 65.05 66.67 71.38 54.50 69.05 68.85

50.02 56.67 41.39 61.77 53.14 49.72 49.11

0.37 0.33 0.27 0.56 0.43 0.40 0.45

2 1 1 2 2 2 2

86 29 58 69 85 70 58

tion and the average payout ratio are higher in the post-crisis subsample. This can be explained that firms are more willing to pay dividends to establish a good reputation on their business performance in the post-crisis period. Overall, the values of mean and standard deviation in the full sample and the two sub-samples are relatively consistent. These findings indicate no selection bias. Panel B reports number of firms by year. It rises significantly from 2003 and reaches a peak in 2006 with 14,450 firms. This figure remains steady at 14,346 firms in 2007 and falls rapidly in 2008 by about 2000 firms. These fluctuations are consistent with the booming period of stock market in the world before the global financial crisis and the start of the crisis in 2008. From 2009 to 2012, the number of firms varies from 12,800 to 14,300. Panel C illustrates number of firm-year observations by industry. In line with Brockman and Unlu (2009) and Shao et al. (2013), Manufacturing industry accounts for the largest proportion in the research data with 74,033 observations (55.40%), followed by Service industries with 24,547 observations (18.37%). The other industries contain more than 4900 firm-year observations, namely Construction industries (4919), Mineral industries (6432), Wholesale trade (7457), Transportation and communications (8113) and Retail trade (8130). Panel D presents the country-level data. Number of observations and firms varies considerably across countries. Five countries including U.S., Japan, India, Taiwan and U.K. constitute about 55.89% firms and 58.0% firm-year observations of the research data. This sample composition problem is present in many prior studies regardless of data providers. We have more firms from India and Taiwan in this study than in Brockman and Unlu (2009) and Shao et al. (2013) since Compustat Global database is expanded with more firms listed in Asian stock markets in 2007. Japan has the largest percentage of paying firms in the sample at 90.42%. Jordan experiences the lowest at only 20.87%, followed by U.S. with 39.66%. In addition, there is a wide range of shareholder and creditor right indices across countries and this makes research findings more reliable.

the coefficients of the interactive term ASD*PER in the subsamples of strong and weak creditor protection to find whether the extent of shareholder expropriation is larger in strong creditor protection countries. All regression models include year dummies, industry dummies and interactive terms between industry dummies and the crisis period dummy to control year and industry effects. They are also clustered by firm to control within-firm correlated residuals. Dividends to earnings (DTE), dividends to total assets (DTA), dividends to sales (DSA) and all firm-level variables are winsorized at 5%.3 3.2. Data sources and sample selection To construct the research sample, we collect annual financial and accounting information from Compustat Global and Compustat North America for firms incorporated in countries covered in La Porta et al. (2006). Then, we eliminate firm-year observations meeting the following criteria: (1) firms listed in countries with mandatory dividend payment, namely Brazil, Chile, Colombia, Greece, and Venezuela (La Porta et al., 2000), (2) firms in utilities industry (SIC codes 4900–4999) and financial sector (SIC codes 6000–6999) (Fama and French, 2001), (3) firms with multiple issues of common stocks (Ferris et al., 2009), (4) firms with unconsolidated financial statements (Mahajan and Tartaroglu, 2008), (5) firm-years with missing or incomplete information, (6) firm-years with abnormal data for subsequent analysis including negative net income, negative book equity, negative total assets and dividends greater than total assets. The final sample contains 133,631 firm-year observations that correspond to 23,890 firms incorporated in 41 countries during the period from 2003 to 2012. 4. Empirical results 4.1. Summary statistics Table 1 presents summary of descriptive statistics of the research sample including firm level data, annual number of firms, industry distribution and country-level data. Panel A shows that paying firms constitute a large proportion in the full sample with 65.81%. The paying proportion in this research sample is slightly higher than 64.4% in Brockman and Unlu (2009) and lower than 67.3% in Shao et al. (2013). The two sub-samples for the pre- and post-crisis periods contain similar number of firm-year observations (i.e. 67,233 and 66,398). Under the effect of the global financial crisis, firms have fewer profitable investment projects; therefore, profitability, firm growth and debt ratio are lower but retained earnings and cash holdings are higher in the post-crisis period. In addition, both the average paying propor-

4.2. Differences in the effects of shareholder rights and creditor rights on dividend policy between the pre- and post-crisis periods Table 2 reports estimation results of dividend policy to compare the effects of shareholder rights, creditor rights between the pre-crisis period and the post-crisis period. In line with Brockman and Unlu (2009), the anti-self-dealing index (ASD) and the revised creditor right index (CRE) are positively associated with both the likelihood of paying dividends and dividend amounts across countries. These results support the outcome model for the agency costs of equity (stronger shareholder rights pressure insiders to pay more dividends) and the substitute model for the agency costs of debt (firms pay less dividends to earn a reputation of fair treatment of creditors and meet creditor demand to restrict dividends). Remarkably, the interactive terms between the crisis period dummy (i.e. it takes 1 for the post-crisis period and 0 otherwise) and the two indices representing shareholder

3 We also replicate all regression models with the sample winsorized at 3% and 10% and find that the statistical and economic significance of core variables is similar to the sample of 5% winsorization. This indicates that outliers fail to affect the research findings significantly.

5

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 2 (continued)

Table 2 Differences in the effects of shareholder rights and creditor rights on dividend policy between the pre- and post-crisis periods. PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DTE is total cash dividends scaled by net income. DTA is total cash dividends scaled by total assets. DSA is total cash dividends scaled by sales. PRO is net income divided by total assets. CAS is cash balance divided by total assets. GRO is the ratio of current year's change in net sales. DEB is the ratio of long-term debt to total assets. TAN is net property, plant and equipment scaled by total assets. SIZ is the natural logarithm of total assets measured in U.S. dollars. RET is retained earnings to assets. ASD is anti-self-dealing index from Djankov et al. (2008). CRE is revised creditor right index from Djankov et al. (2007). GDP is annual GDP growth rate. UAD is uncertainty avoidance dimension from Hofstede et al. (2010). LAW is legal origin dummy which is assigned 1 for civil law countries and 0 for common law countries. PER is a dummy variable which takes the value of 1 for the post-crisis period. t-statistics are in parentheses. Dependent variable = PAY

Dependent variable = DTE

Dependent variable = DTA

Dependent variable = DSA

Intercept

−8.3262*** (−42.72)

−101.5056*** (−27.22)

−7.1566*** (−33.03)

−10.5027*** (−37.48)

PRO

0.0365*** (11.51)

−0.4676*** (−6.48)

0.2144*** (48.91)

0.2127*** (38.97)

PRO*PER

0.0120*** (3.00)

0.1766** (2.12)

0.0383*** (7.80)

0.0480*** (7.82)

CAS

−0.0131*** (−10.24)

−0.1434*** (−5.32)

−0.0085*** (−5.25)

0.0021 (0.99)

Dependent variable = PAY

Dependent variable = DTE

Dependent variable = DTA

Dependent variable = DSA

(18.74)

(11.15)

(13.60)

(18.90)

GDP*PER

−0.0622*** (−8.65)

−1.1145*** (−7.75)

−0.0692*** (−8.46)

−0.1204*** (−11.69)

UAD

0.0268*** (23.03)

0.2659*** (10.92)

0.0125*** (8.68)

0.0165*** (9.23)

UAD*PER

−0.0041*** (−3.23)

−0.0567** (−2.21)

−0.0052*** (−3.67)

−0.0092*** (−5.23)

LAW

0.9101*** (16.13)

17.6316*** (14.90)

0.9281*** (13.00)

1.2411*** (13.73)

LAW*PER

−0.2076*** (−3.49)

1.1872 (0.97)

0.0864 (1.25)

0.0285 (0.32)

PER

0.9613*** (4.67)

14.7224*** (3.75)

1.2182*** (5.69)

1.9432*** (7.12)

No. of observations Left-censored

133,631

133,631

133,631

133,631

45,687

45,687

45,687

*** **

CAS*PER

0.0067*** (4.50)

0.2175*** (6.97)

0.0106*** (5.79)

0.0117*** (4.92)

GRO

−0.9333*** (−19.72)

−34.8767*** (−31.86)

−2.0648*** (−32.60)

−2.5820*** (−31.02)

GRO*PER

−0.1985*** (−3.07)

−10.0687*** (−6.90)

−0.4342*** (−5.30)

−0.6292*** (−5.81)

DEB

0.0027* (1.94)

0.0229 (0.79)

0.0000 (0.02)

0.0100*** (4.61)

DEB*PER

−0.0016 (−0.94)

−0.0173 (−0.49)

0.0014 (0.72)

0.0022 (0.87)

TAN

0.0039*** (4.39)

0.0896*** (4.90)

0.0050*** (4.87)

0.0130*** (8.85)

TAN*PER

0.0027*** (2.74)

0.0292 (1.45)

0.0002 (0.21)

0.0004 (0.27)

RET

0.0359*** (41.30)

0.5145*** (31.26)

0.0272*** (29.06)

0.0338*** (28.28)

RET*PER

0.0005 (0.48)

−0.0057 (−0.31)

−0.0004 (−0.40)

0.0001 (0.10)

SIZ

0.3266*** (31.17)

3.7090*** (19.69)

0.2211*** (20.49)

0.3519*** (24.60)

SIZ*PER

−0.0077 (−0.70)

0.0301 (0.15)

−0.0302*** (−2.76)

−0.0333** (−2.39)

ASD

1.8515*** (14.95)

38.1365*** (14.85)

1.8088*** (12.09)

2.7951*** (14.30)

ASD*PER

−0.4063*** (−2.99)

−6.6888** (−2.47)

−0.3516** (−2.38)

−0.7812*** (−4.11)

CRE

0.3833*** (19.08)

8.1175*** (20.32)

0.4806*** (20.71)

0.5411*** (18.62)

CRE*PER

−0.0746*** (−3.36)

−2.2697*** (11.15)

−0.1040*** (−4.52)

−0.0699** (18.90)

GDP

0.1402***

1.6541***

0.1186***

0.2109***

*

denotes significance at the 1% level. denotes significance at the 5% level. denotes significance at the 10% level.

and creditor protection are negatively related to dividend policy in all regression results. These findings indicate that shareholder rights and creditor rights become less effective in the post-crisis period. Under the impact of the global financial crisis both shareholders and creditors are more severely expropriated. Moreover, similar to Brockman and Unlu (2009) and Shao et al. (2013) the regression results illustrate that firms with higher profitability have higher propensity to pay dividends and dividend payout ratio (i.e. dividends to sales and dividends to total assets). The relationship between profitability and dividends to net income ratio is negative since net income is used as a deflator to measure this payout ratio. These findings imply that when net income is larger, dividend amount is also larger but the proportion of retained earnings is higher and the percentage of income paid as dividends is lower. Moreover, the interactive term of profitability has a significantly positive relationship with both dividend payment propensity and dividend payout ratio. This can be explained that firms maintain a steady stream of dividends or increase dividends to earn good reputation although they have lower profitability in the post-crisis period as presented in Table 1. Cash holdings are negatively related to dividend payment but its interactive terms have positive coefficients. This indicates that before the crisis firms hold more cash to finance future investment and are less likely to pay dividends; however, there are fewer investment opportunities in the post-crisis period, accumulated cash flows are higher and firms pay more dividends (DeAngelo et al., 2006). In line with the transaction cost theory, residual theory and pecking order theory, firm growth and its interactive terms are negatively related to both dividend paying probability and payout ratio. Firms with more investment opportunities tend to retain more earnings for internal financing and they are less willing to pay dividends when experiencing more financial constraint under the impact of the financial crisis. In addition, the positive relationship between debt ratio and payment propensity is inconsistent with the expected sign. This can be explained that debt ratio is a proxy of transaction costs of external financing. Firms with higher debt ratio experience lower transaction costs and are more likely to pay dividends. Furthermore, the regression results show that firms with higher asset tangibility and larger size have higher likelihood to 6

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 3 (continued)

Table 3 Differences in the effects of creditor rights on dividend policy between the pre- and postcrisis periods by level of shareholder rights. PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DTE is total cash dividends scaled by net income. PRO is net income divided by total assets. CAS is cash balance divided by total assets. GRO is the ratio of current year's change in net sales. DEB is the ratio of long-term debt to total assets. TAN is net property, plant and equipment scaled by total assets. SIZ is the natural logarithm of total assets measured in U.S. dollars. RET is retained earnings to assets. ASD is anti-selfdealing index from Djankov et al. (2008). CRE is revised creditor right index from Djankov et al. (2007). GDP is annual GDP growth rate. UAD is uncertainty avoidance dimension from Hofstede et al. (2010). LAW is legal origin dummy which is assigned 1 for civil law countries and 0 for common law countries. PER is a dummy variable which takes the value of 1 for the post-crisis period. t-statistics are in parentheses. Below median shareholder rights (ASD ≤ 0.45)

Intercept

Below median shareholder rights (ASD ≤ 0.45)

Above median shareholder rights (ASD > 0.45)

Dependent variable = PAY

Dependent variable = DTE

Dependent variable = PAY

Dependent variable = DTE

−5.3289*** (−15.66)

−68.4220*** (−9.70)

−8.5251*** (−39.71)

−91.4775*** (−25.16)

Dependent variable = PAY

Dependent variable = DTE

Dependent variable = PAY

Dependent variable = DTE

UAD

−0.0001 (−0.06)

−0.0627 (−1.43)

0.0412*** (38.20)

0.4690*** (24.45)

UAD*PER

−0.0029 (−1.12)

0.0052 (0.10)

−0.0070*** (−6.13)

−0.0338* (−1.71)

PER

0.7799** (2.07)

17.3417** (2.19)

0.7244*** (3.24)

3.7175 (0.99)

No. of observations Left-censored

21,605

21,605

112,026

112,026

−0.2162 (−1.19)

0.0288 (8.16)

0.0205** (2.10)

0.2279 (1.07)

0.0114*** (2.60)

0.1355 (1.48)

CAS

−0.0039 (−1.28)

0.0967 (1.36)

−0.0117*** (−8.28)

−0.1043*** (−3.58)

CAS*PER

−0.0026 (−0.69)

−0.0145 (−0.17)

0.0096*** (5.88)

0.3221*** (9.63)

GRO

−1.1888*** (−9.85)

−38.1988*** (−13.52)

−0.8960*** (−17.38)

−33.8910*** (−28.54)

GRO*PER

−0.0248 (−0.15)

−12.1967*** (−3.16)

−0.2698*** (−3.81)

−11.3942*** (−7.22)

DEB

−0.0027 (−0.80)

−0.1745** (−2.49)

−0.0005 (−0.36)

−0.0218 (−0.70)

DEB PER

−0.0062 (−1.46)

−0.1053 (−1.17)

0.0023 (1.25)

0.0335 (0.87)

TAN

0.0075*** (3.49)

0.1665*** (3.83)

0.0048*** (4.86)

0.1024*** (5.09)

TAN*PER

−0.0008 (−0.33)

0.0006 (0.01)

0.0026** (2.25)

0.0343 (1.53)

RET

0.0353*** (13.94)

0.5902*** (12.94)

0.0367*** (37.08)

0.5247*** (29.07)

RET*PER

−0.0020 (−0.68)

−0.0113 (−0.21)

0.0001 (0.09)

−0.0271 (−1.36)

SIZ

0.4248*** (17.81)

6.0151*** (13.60)

0.3511*** (28.99)

3.7100*** (17.90)

SIZ*PER

−0.0291 (−1.11)

−0.2632 (−0.52)

−0.0062 (−0.49)

0.2908 (1.32)

CRE

−0.0486 (−1.45)

0.1505 (0.22)

0.7141*** (30.77)

14.1706*** (34.33)

CRE*PER

−0.0606* (−1.66)

−1.3330* (−1.77)

−0.1092*** (−4.22)

−2.8601*** (−6.50)

GDP

−0.0439** (−2.43)

−0.5606 (−1.48)

0.1487*** (18.22)

1.4183*** (9.00)

GDP*PER

0.0411** (2.26)

−0.2166 (−0.56)

−0.0729*** (−9.42)

−1.0274*** (24.45)

PRO

0.0487 (6.41)

PRO*PER

*

***

−0.7407 (−9.27)

7609

38,078

*** **

***

Above median shareholder rights (ASD > 0.45)

***

*

denotes significance at the 1% level. denotes significance at the 5% level. denotes significance at the 10% level.

pay and dividend payout ratio. Firms with more tangible assets and larger size have easier access to external funds and they are more willing to pay dividends. Moreover, the life-cycle hypothesis of dividends is also supported with the significantly positive relationship between retained earnings to assets ratio and dividend policy. Moreover, economic growth affects dividend payment positively and this impact declines in the post-crisis period. In contrary to Bae et al. (2012) and Zheng and Ashraf (2014), uncertainty avoidance index is positively related to dividend policy. However, this relationship is consistent with the bird-in-the-hand theory and catering theory of dividends. Investors prefer high to low dividends; hence firms pay more dividends to cater investors when the uncertainty avoidance culture is high (Bae et al., 2012). 4.3. The differences in the effects of creditor rights on dividend policy between the pre- and post-crisis periods by level of shareholder rights In this section, we investigate how the financial crisis affects the effect of creditor rights on dividend policy between two groups of countries including weak shareholder protection group (ASD is lower than or equal to the sample median of 0.45) and strong shareholder protection group (ASD is higher than 0.45). Table 3 illustrates that the strong shareholder protection sub-sample has significantly positive coefficients of CRE for both logit and tobit regression models whilst the corresponding coefficients of the weak shareholder protection group are not significant. These findings are in line with Shao et al. (2013), when shareholders are sufficiently protected, firms have more discretion to use dividends as a substitute of weak creditor rights. Remarkably, the interactive terms of the creditor right variable in the strong shareholder protection group are both statistically and economically more significant. Although the financial crisis results in more expropriation of both shareholders and creditors, managers cannot expropriate minority shareholders easily when shareholder protection is strong. Therefore, ceteris paribus firms in strong shareholder protection countries are less likely to use dividend policy to improve the firm-creditor relationship and creditors are more expropriated. 4.4. The differences in the effects of shareholder rights on dividend policy between the pre- and post-crisis periods by level of creditor rights In this section, we compare how the financial crisis affects the effect of shareholder rights on dividend policy between two groups of countries: the weak creditor protection group with CRE lower than 7

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 4 (continued)

Table 4 Differences in the effects of creditor rights on dividend policy between the pre- and postcrisis periods by level of creditor rights. PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DTE is total cash dividends scaled by net income. PRO is net income divided by total assets. CAS is cash balance divided by total assets. GRO is the ratio of current year's change in net sales. DEB is the ratio of long-term debt to total assets. TAN is net property, plant and equipment scaled by total assets. SIZ is the natural logarithm of total assets measured in U.S. dollars. RET is retained earnings to assets. ASD is anti-selfdealing index from Djankov et al. (2008). CRE is revised creditor right index from Djankov et al. (2007). GDP is annual GDP growth rate. UAD is uncertainty avoidance dimension from Hofstede et al. (2010). LAW is legal origin dummy which is assigned 1 for civil law countries and 0 for common law countries. PER is a dummy variable which takes the value of 1 for the post-crisis period. t-statistics are in parentheses. Below median creditor rights (CRE < 2)

Intercept

Below median creditor rights (CRE < 2)

Above median creditor rights (CRE ≥ 2)

Dependent variable = PAY

Dependent variable = DTE

Dependent variable = PAY

Dependent variable = DTE

−5.1085*** (−15.59)

−14.5503** (−2.38)

−6.5582*** (−26.17)

−87.7919*** (−18.49)

Dependent variable = PAY

Dependent variable = DTE

Dependent variable = PAY

Dependent variable = DTE

UAD

0.0373*** (24.34)

0.2937*** (10.07)

0.0077*** (5.19)

0.2869*** (9.64)

UAD*PER

0.0007 (0.46)

−0.0064 (−0.21)

0.0043** (2.48)

0.1674*** (5.05)

PER

0.2113 (0.64)

8.7198 (1.33)

−0.7070*** (−2.56)

−15.0175*** (−2.99)

No. of observations Left censored

61,328

61,328

72,303

72,303

0.0428 (10.83)

−0.6301 (−7.14)

−0.0836 (−0.57)

0.0240*** (4.49)

0.1594 (1.48)

−0.0140*** (−7.00)

−0.2160*** (−5.39)

−0.0045*** (−2.69)

0.1009*** (2.85)

CAS*PER

0.0010 (0.45)

0.0671 (1.44)

0.0102*** (5.03)

0.3261*** (7.86)

GRO

−1.2064*** (−13.34)

−37.7017*** (−18.16)

−0.8422*** (−15.05)

−32.6717*** (−25.96)

GRO*PER

−0.2668** (−2.22)

−11.6796*** (−4.20)

−0.1794** (−2.30)

−9.6900*** (−5.71)

DEB

0.0054** (2.41)

0.1727*** (3.92)

−0.0043** (−2.49)

−0.1612*** (−4.38)

DEB*PER

−0.0007 (−0.24)

−0.0432 (−0.78)

−0.0017 (−0.77)

0.0333 (0.73)

TAN

0.0048*** (3.17)

0.0691** (2.42)

0.0043*** (3.81)

0.0971*** (4.23)

TAN PER

−0.0012 (−0.75)

−0.0307 (−0.96)

0.0031 (2.34)

RET

0.0340*** (25.55)

0.5009*** (20.97)

0.0381*** (28.91)

0.6000*** (25.78)

RET*PER

−0.0040*** (−2.87)

−0.0156 (−0.58)

0.0014 (0.92)

−0.0012 (−0.04)

SIZ

0.3435*** (20.74)

2.9949*** (10.73)

0.4530*** (29.95)

5.6943*** (22.44)

SIZ*PER

−0.0222 (−1.29)

0.0613 (0.20)

0.0153 (0.91)

0.1088 (0.39)

ASD

−2.8975*** (−10.29)

−59.9395*** (−12.16)

1.0244*** (8.54)

38.7820*** (15.59)

ASD*PER

0.2213 (0.78)

−4.1716 (−0.81)

−0.0132 (−0.09)

−4.9947* (−1.83)

GDP

−0.1718*** (−8.42)

−1.6700*** (−3.91)

0.0418*** (4.62)

−0.5826*** (−3.27)

GDP*PER

0.0824*** (4.09)

0.6312 (10.07)

−0.0079 (−0.88)

0.1268 (9.64)

PRO

0.0360 (6.45)

PRO*PER

0.0043 (0.64)

CAS

*

−0.8618 (−6.69)

***

***

**

0.0521 (2.01)

21,949

23,738

*** **

***

Above median creditor rights (CRE ≥ 2)

***

*

denotes significance at the 1% level. denotes significance at the 5% level. denotes significance at the 10% level.

the full sample median of 2 and the strong creditor protection group with CRE from 2 to 4. Table 4 shows that the shareholder right index is positively related to dividend policy in the strong creditor protection group while this relationship is significantly opposite in the weak creditor right group. This is consistent with Shao et al. (2013), firms have more flexibility to use dividends to reduce agency costs of equity when creditors are strongly protected. Interestingly, the interactive term of shareholder protection index has a significant negative relationship with dividend payout ratio in the strong creditor protection group while this relationship is insignificant in the weak creditor right group. Under the impact of the global financial crisis, managers are less likely to use dividends as a means to improve both the manager-shareholder relaitonship and the shareholder-creditor relationship. However, when managers have less flexibility to adjust dividends due to strong creditor protection, the extent of minority shareholder expropriattion is larger. 5. Robustness checks Table 5 presents robustness tests to verify our results in Table 2. Panel A illustrates that the results of Tobit regression for panel data are consistent with those of Tobit regression for pooled data clustered by firms in Table 2. Both shareholder and creditor rights are less effective in the post-crisis period. However, the pooled OLS regression for payers in Panel B shows insignificant relationship between the interactive term of shareholder protection index and dividend payout ratio. This can be explained by the selection bias. From econometric perspective, dividend payout ratio is left-censored and using OLS regression for the full sample or the sub-sample of payers leads to biased results due to selection problem (Wooldridge, 2010). From Panel C to Panel E, we use Pooled Tobit regression clustered by firms. Panel C shows the estimation results for the reduce sample which is obtained from the full sample excluding U.S., Japan, India, Taiwan and U.K. These five countries account for 55.89% firms and 58.0% observations of the full sample. The findings illustrate that both shareholder and creditor rights become less effective in the post-crisis period regardless of the data composition problem. Panel D shows regression results of the full sample with other measures of creditor and shareholder protection including the original creditor right index (LCR) from La Porta et al. (1998) and the revised anti-director index (RAD) from Djankov et al. (2008) resepectively. The original creditor right index (LCR) and the revised creditor right index (CRE) from Djankov et al. (2007) are measured with the same components but the former is more outdated. The revised anti-director index (RAD) are

**

8

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

Table 5 (continued)

Table 5 Robustness checks for the differences in the effects of shareholder rights, creditor rights on dividend policy between the pre- and post-crisis periods. PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DTA is total cash dividends scaled by net income. CRE is revised creditor right index from Djankov et al. (2007). ASD is anti-self-dealing index from Djankov et al. (2008). LCR is original creditor right index is from La Porta et al. (1998). RAD is revised anti-director index from Djankov et al. (2008). PER is a dummy variable which takes the value of 1 for the post-crisis period. t-statistics are in parentheses. Dependent variable = PAY Panel A - Tobit regression for panel data ASD 2.9906*** (13.25)

Dependent variable = PAY

Panel E - Additional controls (Financial architecture index, Tax advantage of dividend and their interactive terms with the period dummy are added) ASD 3.6425*** 42.5412*** (21.20) (−3.27)

Dependent variable = DTE

ASD*PER

−0.4433** (−2.34)

−10.8296*** (−3.27)

38.0707*** (13.93)

CRE

0.2277*** (9.98)

4.3150*** (9.75)

−0.0422* (−1.68)

−2.1495*** (−4.55)

ASD*PER

−0.3719** (−2.07)

−5.9436*** (−3.17)

CRE*PER

CRE

0.6553*** (19.30)

9.3817*** (23.24)

***

−0.0609** (−2.14)

−1.5762*** (3.12)

** *

CRE*PER

Panel B - Pooled OLS regression for payers only ASD

2.3111 (1.07)

CRE

2.1938*** (6.82)

CRE*PER

−1.8574*** (−5.30)

Panel C - Regression results for the reduced sample (U.S., Japan, India, Taiwan and U.K. are excluded) ASD 1.1860*** 22.5120*** (7.20) (6.25) ASD*PER

−0.7040*** (−4.07)

−13.6189*** (−3.72)

CRE

0.1215*** (5.01)

1.3662** (2.56)

CRE*PER

−0.0840*** (−3.17)

−2.1455*** (−3.83)

Panel D - Other measures of shareholder rights and creditor rights

6. Conclusion

Creditor right is measured as original creditor right index 54.4122*** ASD 1.9403*** (16.64) (20.69) ASD PER

−0.3919 (−3.06)

LCR

0.3977*** (24.14)

3.5855*** (9.89)

LCR*PER

−0.0299* (−1.69)

−1.0466*** (−2.77)

***

−11.0280 (−4.13)

The extant literature shows that shareholder and creditor protection affects corporate payout policy in a static macroeconomic environment. La Porta et al. (2000) and Brockman and Unlu (2009) find that shareholder rights and creditor rights are positively related to dividend policy respectively. Recently, Shao et al. (2013) investigating the balance between the agency costs of equity and debt in dividend decisions across countries and find that creditor rights (shareholder rights) are more effective when minority shareholders (creditors) are strongly protected. In this study, we posit that the global financial crisis results in increases in both types of agency costs and use it as an exogenous shock to examine the effects of shareholder and creditor rights on dividend policy. With a sample of 133,631 firm-year observations from 23,962 firms incorporated in 41 countries, the regression results are consistent with La Porta et al. (2000), Brockman and Unlu (2009) and (Shao et al. (2013)). Remarkably, we find that both shareholder and creditor rights are less effective in the post-crisis period. This implies that under the impact of the financial crisis, both shareholders and creditors are expropriated more severely. Moreover, this paper continues to investigate the differences in the effects of creditor (shareholder) rights on dividend policy between the

***

Shareholder right is measured as revised anti-director index RAD 0.6316*** 3.1850*** (30.67) (7.49) RAD*PER

−0.0574** (−2.46)

−3.8767*** (−8.48)

CRE

0.3093*** (16.12)

9.7756*** (25.08)

CRE*PER

−0.0764*** (−3.61)

−1.5769*** (−3.83)

denotes significance at the 1% level. denotes significance at the 5% level. denotes significance at the 10% level.

criticized that they only predict financial outcome on the basis of investor protection measures across countries and fail to explicitly focus on self-dealing defined as the problem of investor expropriation Djankov et al. (2008). However, these regression models indicate similar core findings as presented in Table 2. Moreover, prior studies show that the preferential tax treatment of dividends to capital gains and the financial structure also affects dividend policy across countries (La Porta et al., 2000). The tax policy with favorable treatment of dividends makes investors prefer a high dividend payout to a low dividend payout. Therefore, if firms want to maximize their shareholders’ wealth, they should pay more dividends (Litzenberger and Ramaswamy, 1979). Besides, the financial structure in which the financial system is bank-based or market-based may play an important role in firms’ access to funds under the impact of the financial crisis. As a result, we add two country-level control variables, namely the preferential tax treatment of dividends to capital gains from La Porta et al. (2000) and the financial architecture index (higher values represent higher stock market orientation of financial systems) from Cy Kwok (2006) and their interactive terms as control variables. Panel E shows that core findings remain stable qualitatively. Table 6 presents robustness checks to verify the research findings in Table 3. The estimation results are from Tobit regression for panel data and Pooled tobit regression clustered by firm for the reduced sample and additional controls. We find that the core findings persist.

4.2662** (2.09)

ASD*PER

*

Dependent variable = DTE

9

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

pre- and post-crisis periods by level of shareholder (creditor) protection. We find that the level of creditor (shareholder) expropriation under the impact of the financial crisis is higher when shareholders (creditors) are strongly protected.

Table 6 Robustness checks for the differences in the effects of shareholder rights (creditor rights) on dividend policy between the pre- and post-crisis periods by level of creditor rights (shareholder rights). PAY is a binary variable which is equal to 1 if total paid dividends are greater than zero and 0 otherwise. DSA is total cash dividends scaled by sales. CRE is revised creditor right index from Djankov et al. (2007). ASD is anti-self-dealing index from Djankov et al. (2008). PER is a dummy variable which takes the value of 1 for the post-crisis period. Additional controls including financial architecture index, tax advantage of dividend and their interactive terms with the period dummy are added. t-statistics are in parentheses.

Acknowledgement We are thankful to the Editor and the anonymous reviewers for their valuable comments on earlier versions of this paper.

Panel A - Differences in the effects of shareholder rights on dividend policy by level of creditor rights Below median shareholder rights (ASD ≤ 0.45)

Above median shareholder rights (ASD > 0.45)

Dependent variable = PAY

Dependent variable = PAY

Dependent variable = DTE

1.1729*** (32.26)

15.6339*** (38.20)

−0.0920*** (−2.97)

−2.1263*** (−6.99)

Dependent variable = DTE

Tobit regression for panel data CRE −0.1422*** −0.5126 (−2.67) (−0.68) CRE*PER

−0.0442 (−0.96)

−0.7552 (−1.34)

References Al-Malkawi, H.-A.N., Bhatti, M.I., Magableh, S.I., 2014. On the dividend smoothing, signaling and the global financial crisis. Econ. Model. 42, 159–165. http:// dx.doi.org/10.1016/j.econmod.2014.06.007. Bae, S.C., Chang, K., Kang, E., 2012. Culture, corporate governance, and dividend policy: international evidence. J. Financ. Res. 35 (2), 289–316. http://dx.doi.org/10.1111/ j.1475-6803.2012.01318.x. Boţoc, C., Pirtea, M., 2014. Dividend payout-policy drivers: evidence from emerging countries. Emerg. Mark. Financ. Trade 50 (sup4), 95–112. Brockman, P., Unlu, E., 2009. Dividend policy, creditor rights, and the agency costs of debt. J. Financ. Econ. 92 (2), 276–299. http://dx.doi.org/10.1016/ j.jfineco.2008.03.007. Byrne, J., O’Connor, T., 2012. Creditor rights and the outcome model of dividends. Q. Rev. Econ. Financ. 52 (2), 227–242. http://dx.doi.org/10.1016/j.qref.2012.04.002. Cy Kwok, C., 2006. National culture and financial systems. J. Int. Bus. Stud. 37 (2), 227–247. DeAngelo, H., DeAngelo, L., Stulz, R.M., 2006. Dividend policy and the earned/ contributed capital mix: a test of the life-cycle theory. J. Financ. Econ. 81 (2), 227–254. http://dx.doi.org/10.1016/j.jfineco.2005.07.005. Djankov, S., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2008. The law and economics of self-dealing. J. Financ. Econ. 88 (3), 430–465. http://dx.doi.org/10.1016/ j.jfineco.2007.02.007. Djankov, S., McLiesh, C., Shleifer, A., 2007. Private credit in 129 countries. J. Financ. Econ. 84 (2), 299–329. http://dx.doi.org/10.1016/j.jfineco.2006.03.004. Fama, E.F., French, K.R., 2001. Disappearing dividends: changing firm characteristics or lower propensity to pay*. J. Financ. Econ. 60 (1), 3–43. http://dx.doi.org/10.1016/ S0304-405X(01)00038-1. Ferris, S.P., Jayaraman, N., Sabherwal, S., 2009. Catering effects in corporate dividend policy: the international evidence. J. Bank. Financ. 33 (9), 1730–1738. http:// dx.doi.org/10.1016/j.jbankfin.2009.04.005. Grullon, G., Michaely, R., Swaminathan, B., 2002. Are dividend changes a sign of firm maturity*. J. Bus. 75 (3), 387–424. Hofstede, G., Hofstede, G.J., & Minkov, M., 2010. . Cultures and organizations: Software of the mind (3rd ed.): Citeseer. Huang, H.-C.R., 2001. Bayesian analysis of the dividend behaviour. Appl. Financ. Econ. 11 (3), 333–339. http://dx.doi.org/10.1080/096031001300138735. Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance, and takeovers. Am. Econ. Rev. 76 (2), 323. Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. J. Financ. Econ. 3 (4), 305–360. http://dx.doi.org/ 10.1016/0304-405X(76)90026-X. Jiraporn, P., Kim, J.C., Kim, Y.S., 2011. Dividend payouts and corporate governance quality: an empirical investigation. Financ. Rev. 46 (2), 251–279. http://dx.doi.org/ 10.1111/j.1540-6288.2011.00299.x. Johnson, S., Boone, P., Breach, A., Friedman, E., 2000. Corporate governance in the Asian financial crisis. J. Financ. Econ. 58 (1–2), 141–186. http://dx.doi.org/ 10.1016/S0304-405X(00)00069-6. Kim, B.S., Maddala, G.S., 1992. Estimation and Specification Analysis of Models of Dividend Behavior Based on Censored Panel Data. Empir. Econ. 17 (1), 111–124. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., 2006. What Works in Securities Laws*. J. Financ. 61 (1), 1–32. http://dx.doi.org/10.1111/j.1540-6261.2006.00828.x. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. J. Political Econ. 106, 42. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R.W., 2000. Agency problems and dividend policies around the world. J. Financ. 55 (1), 1–33. Lemmon, M.L., Lins, K.V., 2003. Ownership structure, corporate governance, and firm value: evidence from the east asian financial crisis. J. Financ. 58 (4), 1445–1468. http://dx.doi.org/10.1111/1540-6261.00573. Lintner, J., 1956. Distribution of incomes of corporations among dividends, retained earnings, and taxes. Am. Econ. Rev. 46 (2), 97–113. Litzenberger, R.H., Ramaswamy, K., 1979. The effect of personal taxes and dividends on capital asset prices: theory and empirical evidence. J. Financ. Econ. 7 (2), 163–195. http://dx.doi.org/10.1016/0304-405X(79)90012-6. Mahajan, A., Tartaroglu, S., 2008. Equity market timing and capital structure: international evidence. J. Bank. Financ. 32 (5), 754–766. http://dx.doi.org/ 10.1016/j.jbankfin.2007.05.007. Mitton, T., 2002. A cross-firm analysis of the impact of corporate governance on the East Asian financial crisis. J. Financ. Econ. 64 (2), 215–241.

Regression results for the reduced sample (U.S., Japan, India, Taiwan and U.K. are excluded) CRE −0.0486 0.1505 0.3682*** 3.4754*** (−1.45) (0.22) (9.00) (3.57) CRE*PER

−0.0606* (−1.66)

−1.3330* (−1.77)

−0.1980*** (−4.23)

−3.8405*** (−3.92)

Additional controls (Financial architecture index, Tax advantage of dividend and their interactive terms with the period dummy are added) CRE −0.2266*** −3.7684*** 0.5934*** 6.8285*** (−5.95) (−4.75) (21.33) (13.21) CRE*PER

−0.0377 (−0.87)

−0.6908 (−0.75)

−0.0621** (−1.99)

−2.6448*** (−4.74)

Panel B - Differences in the effects of creditor rights on dividend policy by level of shareholder rights Below median creditor rights Above median creditor rights (CRE < 2) (CRE ≥ 2) Dependent Dependent Dependent Dependent variable = PAY variable = DTE variable = PAY variable = DTE Tobit regression for panel data ASD −5.1903*** −68.3790*** (−11.64) (−14.16) ASD*PER

0.9226*** (2.77)

−4.7450 (−1.49)

1.5887*** (8.31)

45.7391*** (18.30)

−0.0756 (−0.44)

−5.9978*** (−3.10)

Regression results for the reduced sample (U.S., Japan, India, Taiwan and U.K. are excluded) ASD −0.6181* 2.7609 1.1201*** 35.6594*** (−1.88) (0.37) (7.25) (10.66) ASD*PER

0.7972** (2.16)

3.3613 (0.42)

−0.0342 (−0.21)

−10.3322*** (−3.01)

Additional controls (Financial architecture index, Tax advantage of dividend and their interactive terms with the period dummy are added) ASD −2.3949*** −46.5518*** 1.8823*** 44.7749*** (−7.95) (−8.68) (9.71) (11.96) ASD*PER

0.7382** (2.21)

5.4756 (0.88)

0.7242*** (3.11)

−2.7985 (−0.65)

*** ** *

denotes significance at the 1% level. denotes significance at the 5% level. denotes significance at the 10% level.

10

Economic Modelling xxx (xxxx) xxx–xxx

Q.T. Tran et al.

j.1475-6803.2013.12002.x. Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data 10th ed.. MIT Press, Cambridge. Zheng, C., Ashraf, B.N., 2014. National culture and dividend policy: international evidence from banking. J. Behav. Exp. Financ. 3 (0), 22–40. http://dx.doi.org/ 10.1016/j.jbef.2014.07.002.

Mitton, T., 2004. Corporate governance and dividend policy in emerging markets. Emerg. Mark. Rev. 5 (4), 409–426. http://dx.doi.org/10.1016/j.ememar.2004.05.003. O’Connor, T., 2013. Dividend payout and corporate governance in emerging markets: which governance provisions matter*. Int. J. Corp. Gov. 4 (2), 181–207. http:// dx.doi.org/10.1504/IJCG.2013.055754. Shao, L., Kwok, C.C.Y., Guedhami, O., 2013. Dividend policy: balancing Shareholders' and Creditors' interests. J. Financ. Res. 36 (1), 43–66. http://dx.doi.org/10.1111/

11