Finance Research Letters 17 (2016) 197–210 Contents lists available at ScienceDirect Finance Research Letters journal homepage: www.elsevier.com/locate/frl Dividend payout policies: Evidence from Latin America Julian Benavides a,∗, Luis Berggrun b, Hector Perafan b a b Director of the Accounting and Financial Studies Department, Universidad Icesi Cali Colombia Department of Accounting and Financial Studies, Universidad Icesi Cali Colombia a r t i c l e i n f o Article history: Received 12 February 2016 Accepted March 2016 Available online 17 March 2016 JEL classification: C33 G32 C34 G35 Keywords: Dividends Pecking order model Trade-off model Lifecycle theory of dividends Latin America a b s t r a c t This paper examines dividend payout policies for firms in six Latin American countries from 1995 to 2013 As predicted by the pecking order and trade-off models, the dividend payout is positively linked to profitability and negatively related to past indebtedness and investment opportunities We also find that the target dividend payout ratio is positively related to governance indicators at the country level In addition, the speed to which firms adjust their dividends to changes in earnings is lower in high governance countries in the region Thus, firms smooth dividends more in countries with higher governance scores We not find evidence supporting the lifecycle theory nor illiquidity effects on dividends levels © 2016 Elsevier Inc All rights reserved Introduction Means to give firm shareholders their money back, have always been (and always will be) a contentious issue The financial literature has studied the ways, dividends and repurchases, and the motivations behind giving back cash to shareholders Floyd et al., (2015), for example, study how payout policies evolve over the last 30 years in the United States, arguing that signaling and agency costs are extant reasons to explain those policies Both explanations arise from the two main models the financial literature has posited to explain capital structure decisions: the pecking order model and the trade-off model Although initially conceived to explain capital structure choices, both models also offer predictions on how firms decide to pay dividends to their shareholders (Fama and French, 2002) In the pecking order framework, Myers (1984) posits that asymmetric information leads managers to issue risky securities when they are overpriced As a result, investors demand a premium on new and existing shares, once new issues are announced In anticipation managers can forego profitable investments if they require additional risky capital To avoid this problem, minimizing asymmetric information costs, managers prefer to finance new projects with retained earnings, then with low risk debt, risky debt, and as a last option they issue equity The pecking order model does not explain why firms pay dividends; however, once dividends are paid, firms with less profitable assets in place, large current and expected investments, and high leverage find dividends less attractive, given the financing costs attached to the issue of new risky securities ∗ Corresponding author Tel.: +57 25552334 E-mail addresses: jbenavid@icesi.edu.co (J Benavides), lberggru@icesi.edu.co (L Berggrun), hfperafan@icesi.edu.co (H Perafan) http://dx.doi.org/10.1016/j.frl.2016.03.012 1544-6123/© 2016 Elsevier Inc All rights reserved 198 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Higher stability of income can also be associated with a lower likelihood of foregoing attractive investments or the need of issuing risky securities Thus, to lower the possibility of not taking advantage of investment opportunities when cash flow is low, firms with volatile income pay less dividends The following associations, controlling for additional interactions, are expected: (1) more profitable firms pay more dividends; (2) firms with more leverage and more investment opportunities pay less dividends; and (3) firms with more volatile income pay less dividends The other main venue in explaining capital structure decisions is the trade-off model Under this model firms make capital structure decisions weighing different and opposing forces In this setting, firms weigh bankruptcy costs and tax considerations when determining a target or optimal level of debt Firms with higher leverage, more volatile income, and larger expected investment outlays are likely to set a lower leverage level to minimize distress costs Given the fiscal benefits of interest payments, one also would expect a more intense use of debt by the most profitable firms The pecking order and trade-off models make similar predictions in terms of dividends Firms set dividends as to minimize potential bankruptcy costs (bearing in mind the differential fiscal treatment of dividends versus interest payments) Thus, firms with less volatile earnings, lower leverage, and lower expected investment opportunities are more prone to pay higher dividends Conversely, firms with unprofitable assets in place are likely to have a low dividend payout ratio Under the trade-off model, agency cost considerations can also account for leverage and dividend decisions Easterbrook (1984) analyzes the effect of a consistent dividend policy in an environment characterized by agency problems within the firm One agency cost firms face is the one related to supervising management1 , a cost which shareholders must assume since the interests of shareholders and managers are not always aligned A second agency cost refers to risk aversion by management (given its human capital investment in the firm) that prompts management to take low risk projects which in many cases may not be the most beneficial for shareholders Dividends can reduce these two agency costs since they can force companies to use financial markets more frequently and in the process expose the company to a higher degree of monitoring by investors and investment bankers that ends up reducing monitoring costs initially borne by all investors Likewise, according to Easterbrook (1984), dividends can serve to adjust the level of risk taken by management to a point more in line with shareholder’s preference (higher level) In this sense, paying a dividend increases the debt-to-equity ratio benefiting shareholders and sets free an efficient mechanism2 which results in a reduction in the firm’s agency costs Jensen (1986) points out the potential cost of agency that large free cash flow, under managerial control, could pose on the firm value Without restrictions in the use of the free cash flow, managers can waste the free cash flow in negative net present value projects Larger dividends reduce those agency costs forcing managers to take better decisions before wasting resources of the firm Iturriaga and Crisóstomo (2010) confirm the role of dividends in Brazil as a disciplining mechanism to control managers that may feel inclined to pursue value-destroying projects La Porta et al (20 0) discusses two versions of the agency theory of dividends By and large, agency theory considers dividends as a mechanism to mitigate conflicts between corporate insiders and outside shareholders A first version referred to as the “outcome model” states that dividends are a result of an effective legal protection system of shareholders In this sense one would expect a positive relationship between the level of dividends and the level of investor protection across countries The latter, since investors in more protected countries can extract more dividends from companies they invest in A second version of the agency theory of dividends (“substitute model”) considers dividends and investor protection as substitutes In this version, dividends become an instrument to strengthen the reputation of companies This reputation is important since firms may occasionally need to get funding in financial markets Under this model one would expect an indirect relationship between dividends payments and the level of investor protection across countries, since it is likely that companies in low investor protection countries care more about their reputation and as a means to protect it use dividends more intensely than companies in high investor protection countries In addition to pecking order and trade off explanations on how firms pay dividends, DeAngelo et al (2006) propose a lifecycle theory of dividends as an alternative to these two often used models They claim that young firms tend to be less prone to pay dividends since they are likely to be in a capital infusion phase, and thus most of its capital is contributed (e.g., by new shareholders), not earned On the other hand, as firms mature (and most of its capital is earned not contributed) these older firms are more inclined to pay dividends as they run out of investment opportunities DeAngelo et al (2006) find supportive empirical evidence of a lifecycle explanation of dividends because they document a positive and highly significant relationship between the earned over contributed capital ratio (proxied by retained earnings over total equity, or over total assets) and the propensity to pay dividends, even after controlling for firm size, growth, and profitability Brockman and Unlu (2011) extend the evidence of a lifecycle theory of dividends in an international study of payout policies The ratio of retained earnings to equity had a positive influence in the likelihood that a firm pays dividends implying that young firms (usually with a low ratio of retained earnings to contributed capital) tend to pay lower dividends than older firms Not only age considerations play a role in explaining dividend policy; Brockman and Unlu document that For example, audit costs to avoid manipulation of financial statements and possibly, expropriation by managers That increases the probability of using the market for capital with the consequent reduction of monitoring costs of management’s actions J Benavides et al / Finance Research Letters 17 (2016) 197–210 199 agency costs considerations (related to accounting disclosure quality) also shape dividend decisions In all, they document a U-shape relationship between dividend payments and disclosure quality Managers in transparent environments pay high dividends because they are obliged by their shareholders, while managers in opaque settings pay high dividends to attract (hesitant) external capital suppliers This paper studies dividend payment decisions of firms in six Latin American countries in the 1995–2013 period applying the Lintner (1956) model, under the framework of Fama and French (2002) tests, that incorporate firm-specific variables in analyzing the target dividend payout decision These firm-specific variables (related to profitability, investment opportunities, volatility, and the earned-contributed capital mix) allow us to examine the dividend predictions of the pecking order and trade-off models, as well as those of the lifecycle theory of dividends Previous research Rajan and Zingales (1995), Demirgüç-Kunt and Maksimovic (1999), Booth et al (2001), Bancel and Mittoo (2004), de Jong et al (2008), and Kirch and Terra (2012) emphasize the need to account for country-specific factors when examining leverage decisions worldwide In a recent study, de Jong et al (2008) show that country-specific factors (law abidance, shareholder/creditor right protection, market/bank financial system orientation, stock/bond market development and GDP growth rate) are important determinants (both directly and indirectly) of the leverage decisions of a panel of firms from 42 countries In particular, firms in countries with more developed bond markets and higher GDP growth show higher leverage We contribute to the literature by examining the extent of how both firm- and country-specific factors (mostly related to corporate governance) shape the dividend decisions of Latin American firms within the framework of the pecking order and trade off models, and the recent lifecycle theory of dividends We find that the dividend payout rate is positively associated to profitability and negatively related to past indebtedness and investment opportunities, as predicted by the pecking order and trade-off models Governance indicators at the country level are also positively correlated with the target dividend payout Mitton (2004) finds similar evidence for a sample of countries from emerging markets Furthermore, he finds that country- and firm-level governance indicators are complements rather than substitutes in explaining the positive relationship between governance and dividends We expand Mitton’s (2004) evidence and examine the association between governance (at the country level) and the way firms adjust their dividends to shifts in earnings We find that the speed to which firms adjust their dividends to changes in earnings is lower in high governance countries in the region Thus, firms pay less volatile dividends in countries with higher governance scores Contrary to what De Angelo et al (2006) report, we find a negative impact of retained earnings on dividends, older firms in our sample not pay higher dividends; we interpret this result as a consequence of the very low use of equity markets of our firms as a financing mechanism, with managers prioritizing investment flexibility (Blau and Fuller, 2008) or being less disciplined by market forces (Easterbrook, 1984) The rest of the paper is organized as follows: Section describes the dataset, while Section discusses our econometric approach and findings on dividend policy in Latin America Section reports results from robustness checks Lastly, Section concludes the paper with a summary of our findings Data Our sample includes financial data for public firms in six Latin American countries (Argentina, Brazil, Chile, Colombia, Mexico and Peru) from 1995 to 2013 We collect firm data from consolidated financial statements and expressed in U.S dollars Our source is Economatica, a database specialized in Latin American exchanges.3 In addition, we gather information on a rule of law index from the World Bank The index is built from thirty underlying sources which report perceptions from survey respondents and experts worldwide (Kaufmann et al., 2010) The scale of the index varies from -2.5 (weak) to 2.5 (strong) rule of law abidance We perform cubic spline interpolation when there is missing data for the rule of law indicator To construct our dataset we apply several screens Utilities and banks are excluded from our sample because their dividend decisions are often influenced by regulation Only firms with at least five million dollars of average assets enter our sample We not include information on share repurchases due to the lack of data for these transactions in Economatica.4 Repurchases are infrequent in the region given the highly concentrated ownership Venezuelan firms are not included in our sample given the low number of reporting firms especially for the years 2010 and 2011 In our estimations we only consider observations with positive reported dividends To mitigate the impact of outliers, we censor observations at the 2nd and 98th percentile Pt Our main variable, dividends, is calculated indirectly using two different methods In the first method: Dt = DPP St BVP St BEt , t where dividends for fiscal year t (Dt ) are calculated multiplying the current dividend yield (dividends per share, DPSt , over share price, (Pt ) times the price to book (book value per share, BVPSt ) ratio times the book value of common equity (BEt ) The second method is straightforward:Dt = DPP St MVEt , we just multiply the dividend yield times the total market capitalizat tion (MVE or market value of equity) We report our results below using the first method Nonetheless, our results remain qualitatively similar using either of the two methods Dividend information in the Emerging Markets Database in Compustat for Latin American countries is almost nonexistent We only found repurchases information for the last five years for some of the countries of our sample 200 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Given the differences in the balance sheets of firms in the region we make some adjustments to obtain comparable figures We compute for Brazil dividends as the sum of reported dividends plus shareholder equity interest Corporations in Brazil can remunerate shareholders via interest payments to obtain tax savings (Velez-Pareja and Benavides Franco, 2011) To estimate retained earnings, we construct a series for each country We add retained earnings, other reserves type A, and legal reserves to obtain retained earnings for Argentina For Chile, we add retained earnings and other reserves type A and B while for Peru we add, in addition, unrealized results Retained earnings in Brazil are the sum of retained earnings and reserves In Colombia, we assimilate retained earnings as the addition of net income, total reserves, and accumulated income As for Mexican firms, retained earnings are set equal to net income for the year, legal reserves, and reserves created in previous years The top panel of Table includes a brief description of our main variables In Panel B of Table we observe that Chilean firms pay the highest dividends as a percentage of assets Mexican and Brazilian firms are the largest (by assets) possibly because these firms come from the two biggest economies of the region Focusing on variables related to investment opportunities ( MAVt , and P P Nt+1 ), we observe, by and large, that Chilean and Peruvian tend to show the highest ratios As t for profitability measures, Et St fluctuates from 12.29% and 21.16%, while N It At+1 ranges from 3.88% to 7.41% In terms of leverage (using both book or market figures), Brazilian firms appear the more highly indebted while Colombian firms are the less leveraged in the region Mexican and Chilean firms show the higher retained earning-to-assets ratio possibly indicating that these firms are the most mature in the region In addition, Chile scores strongly in rule of law compliance5 , while Colombia shows the weakest score D The third panel of Table shows, as expected, a positive and significant correlation (0.534) between dividends At+1 and net income N It At+1 MVt At ( EStt ) The panel also shows a positive correlation between dividends and likely to be more related to profitability than to investment opportunities Profitability correlated Conversely, investment opportunities comprise in the ratio of P P Nt+1 At t+1 suggesting that this variable is and dividends are also positively and dividends are negatively correlated al- though the correlation is not statistically significant, which contrast with those include in the ratio At+1 At+1 that has a positive statistically significant correlation with dividends Leverage is negatively related with dividends All the previous associations are very much in line with predictions of the pecking order and trade off models The lifecycle theory of dividends suggests a positive association between dividends and retained earnings (or RAEt ); however, this correlation is negative and significant t for our sample (−0.37), when discussing the results from Table we will elaborate on this issue Interestingly, dividend payments and the rule of law indicator are positively and significantly correlated lending univariate support to the outcome model of the agency theory of dividends (La Porta et al., 20 0) In Table 1, Panel D we note that our unbalanced panel comprises 666 firms and 3798 firm-year observations Brazilian and Chilean firms constitute more than 50% of the sample The average number of observations per firm is 5.70 The bottom panel (E) of Table reports the evolution in time of one of our main independent variables The ratio of net income to assets appears more stable (measured by the coefficient of variation) in Chile and Peru Conversely, and Mexican and Argentinean show the most volatile profits The last four years have witnessed an increase in profits for all the countries in the sample, in line with the economic recovery of the region Econometric approach and results Lintner (1956) in an influential survey on dividend policy, still being applied to explain firm’s payouts Andres et al., (2015), found that firms pay considerable attention to the existing rate of dividend payments when determining the upcoming dividend Furthermore, changes in dividends were strongly affected by variations in current earnings Most of the firms in their sample intended to keep a roughly constant target dividend payout (TP) ratio (with 50% being the most common target) In all, Lintner (1956) posits that firms have a long term target payout ratio that affects target dividends in the following way: T Dt+1 = T P ∗ N It (1) In Eq (1), T Dt+1 is target dividend measured in year t + 1, and NIt is the net income that backs the observed dividends Adjustment costs produce just a partial movement to the target in year t + 1, thus the change ( ) in dividends is the result of the difference between the target dividend and the actual dividend times the speed of adjustment (SOA): Dt+1 = SOA(T Dt+1 − Dt ) + εt+1 (2) In order to estimate Eq (2), Lintner (1956) replaced TDt + to obtain an empirical counterpart of his model: Dt+1 = α1 N It + α2 Dt + εt+1 The speed of adjustment is SOA = −α2 and the target payout is T P = (3) α1 SOA The rule-of-law indicators remain (not shown) relatively stable or improve throughout the years, except for Argentina which suffered a sudden decline in the rule of law index in the 20 0–20 02 period J Benavides et al / Finance Research Letters 17 (2016) 197–210 201 Table Selected sample statistics Panel A Description of variables Variable Definition D/A Dt + /At + Ln(A) A t + /At + NIt /At + MV MV/A E/S PPNt + /At L/A L/MV Dt /At + RE/A GI Dividends over assets Change in dividends over assets Natural logarithm of assets (in millions of US dollars) Proportional change of assets Net income over assets of next fiscal year Market value = market equity + liabilities (in millions of US dollars) Market value over assets Earnings before interest and taxes over sales Change in net plant and equipment over assets of past fiscal year Liabilities over assets Liabilities over market value Dividends over assets of next fiscal year Retained earnings over assets Governance index: rule of law (World Bank governance indicators) Panel B Mean values of selected variables per country (1995–2013) Dt + / At + (%) Dt + / At + (%) At ($) At + / At + (%) NIt / At + (%) MVt / At Et /St (%) PPNt + /At (%) Lt + /At + (%) Lt + /MVt + (%) Dt / At + (%) REt /At (%) GIt + Argentina Brazil Chile Colombia Mexico Peru Total 3.48 0.44 $1,571 4.84 6.70 1.22 18.26 1.91 44.94 40.71 3.04 18.51 −0.41 2.70 0.35 $2,004 5.68 5.46 1.21 14.17 2.70 53.94 50.47 2.35 13.17 −0.31 3.56 0.26 $1,258 7.21 5.59 1.36 15.04 4.37 47.91 39.82 3.30 21.22 1.23 2.08 0.06 $1,787 8.46 3.88 0.97 12.29 3.56 34.21 36.51 2.02 12.89 −0.69 2.53 0.16 $3,284 6.08 6.50 1.58 14.36 3.03 46.40 32.53 2.37 27.10 −0.54 3.27 0.34 $640 8.08 7.41 1.37 21.16 4.52 42.09 37.17 2.94 14.92 −0.67 3.04 0.31 $1,682 6.37 5.80 1.28 15.38 3.33 49.52 44.12 2.73 16.78 0.07 Panel C Correlation matrix (1995–2013) Dt+1 / At+1 Dt + /At + At At + /At + NIt /At + MVt /At Et /St PPNt + /At Lt + /At + Lt + /MVt + Dt /At + REt /At GIt 0.457∗∗∗ −0.038∗∗ −0.050∗∗∗ 0.534∗∗∗ 0.353∗∗∗ 0.277∗∗∗ −0.009 −0.169∗∗∗ −0.417∗∗∗ 0.795∗∗∗ 0.081 0.107 At 0.002 0.060∗∗∗ 0.126 0.107 0.079 0.051∗∗∗ −0.038∗∗ −0.146∗∗∗ −0.176∗∗∗ −0.012 −0.018 At+1 / At+1 NIt / At+1 MVt / At Et / St PPNt+1 / Lt+1 / At At+1 Lt+1 / MVt+1 Dt / At+1 REt / At GIt 0.048∗∗∗ −0.094 0.107 0.138∗∗∗ 0.044∗∗∗ 0.226∗∗∗ 0.049∗∗∗ −0.043∗∗∗ −0.047∗∗∗ −0.075 −0.050∗∗∗ 0.173∗∗∗ 0.051∗∗∗ 0.670∗∗∗ 0.056∗∗∗ −0.171∗∗∗ −0.097 0.037∗∗ −0.009 0.392∗∗∗ 0.372∗∗∗ 0.201∗∗∗ −0.019 0.140∗∗∗ 0.102 −0.271∗∗∗ 0.097 −0.088 0.052∗∗∗ −0.455∗∗∗ −0.471∗∗∗ −0.234∗∗∗ −0.123 0.634∗∗∗ 0.505∗∗∗ 0.317∗∗∗ 0.252∗∗∗ −0.045∗∗∗ −0.161∗∗∗ −0.362∗∗∗ ∗∗∗ ∗ 0.345 0.125 0.014 0.027 −0.391∗∗∗ −0.370∗∗∗ 0.097 −0.053∗∗∗ 0.074 −0.032∗∗ 0.033∗∗ −0.014 −0.092 0.131 0.202∗∗∗ Panel D Firms per country (1995–2013) Argentina Brazil Chile Colombia Mexico Peru Total Firms Observations (firm-year) Average (Obs/Firms) 60 319 141 19 60 67 666 293 1,758 1,099 87 192 369 3,798 4.88 5.51 7.79 4.58 3.20 5.51 5.70 Panel E Average NIt /At + per country and year Argentina Brazil Chile Colombia Mexico Peru Total 1995 1996 1997 1998 1999 20 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 7.39% 4.33% 6.23% 2.75% 3.20% 9.07% 5.44% 5.94% 5.01% 5.81% 4.32% 9.40% 7.53% 5.68% 5.75% 5.04% 6.18% 4.87% 7.67% 9.37% 5.99% 6.14% 5.31% 6.31% 4.92% 4.92% 6.64% 5.82% 4.23% 4.25% 5.65% 2.66% 7.28% 7.77% 4.99% 4.74% 6.07% 5.15% 5.26% 6.00% 7.21% 5.77% 8.55% 6.67% 5.21% 5.32% 6.23% 6.42% 6.14% 2.24% 4.06% 4.24% 4.66% 7.92% 5.92% 4.29% 6.44% 5.89% 5.13% 4.62% 9.85% 6.75% 5.72% 7.51% 6.60% 5.87% 2.53% 4.70% 7.43% 6.38% 8.85% 6.13% 6.07% 2.22% 6.84% 6.85% 6.31% 8.43% 4.98% 5.97% 2.18% 7.03% 7.40% 5.95% 7.40% 7.21% 6.36% 2.87% 8.39% 8.14% 7.16% 5.95% 4.41% 4.66% 1.59% 5.03% 7.23% 4.88% 5.62% 4.87% 5.43% 2.19% 6.03% 7.86% 5.53% 8.50% 6.48% 6.46% 5.28% 7.47% 8.23% 6.91% 8.85% 5.47% 4.53% 3.93% 5.82% 7.55% 5.68% 6.15% 4.87% 4.79% 3.19% 4.94% 5.53% 5.01% (continued on next page) 202 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Table Continued Argentina Brazil Chile Colombia Mexico Peru Total Avg (%) σ (%) σ /μ (%) 6.59 5.43 5.56 3.63 6.60 7.38 5.76 1.78 0.94 0.69 1.28 1.73 0.98 0.71 27.04 17.30 12.45 35.39 26.16 13.26 12.26 Note: The table describes the variables we use (Panel A) from Economatica and World Bank Governance Indicators Panel B reports average values of our main variables per country, all variables are ratios except Assets (At ) which is measured in millions of US dollars Panel C reports correlations among variables, where ∗∗∗, ∗∗ and ∗ denote significance of the pairwise correlation (Ho: ρ = 0) at the 0.01, 0.05 and 0.10 levels The next panel (Panel D) shows the number of firms and observations per country It over the sample years per country, the last three columns present the average, the standard deviation and Panel E reports ANt+1 the coefficient of variation, respectively 3.1 Target payout and speed of adjustment Following Fama and French (2002) we examine the target payout and the speed of adjustment taking into account the effect of the driving variables of dividend decisions according to the pecking order and trade off models as well as the lifeA cycle theory of dividends We proxy profitability with ESt and MAVt ; investment opportunities with At+1 and PAPNt Leverage L t L is measured using accounting ( At+1 ) or market ( MVt+1 ) figures t+1 t+1 t REt At t t is our proxy for the earned over contributed capital ratio which is the main driving variable of dividend decisions according to the lifecycle theory of dividends We use the natural logarithm of size as a proxy for volatility (as Fama and French (2002)) Larger firms are more likely to have low volatility given their more stable cash flows and earnings We model how TP and SOA are affected by firm specific variables Thus we follow Fama and French (2002), who argued that a more robust approach to test theories about dividends determinants is that firm specific factors affect directly the magnitude of TP and SOA We also include country effects in our estimation to assess the effect of unobserved heterogeneity among countries on dividend policies Additionally, there is a substantial body of research (see Section 1) that stresses the need to account for country specific factors in modeling leverage decisions We expand this approach to dividend decisions by estimating a panel model with country fixed effects Our model derived from Eq (1) allows TP to vary using firm specific factors: Dt+1 = a0 + d ci + At+1 + εt+1 a1 + a2 MVt Et + a3 + a4 At St P P Nt REt + a5 ln (At ) + a6 Levt+1 + a7 At At Opposite Hypotenuse N It At+1 (4) The dummy variables dci account for differential intercepts for each country, allowing the estimation of differential target payouts We initially estimate Eq (4) using a panel regression with country dummies and clustering errors at the year level The relation between dividends and the exogenous variables is modeled in five ways The first model (a.) includes country dummies, but assumes no interaction between ANIt and the proxies used for profitability, investment opportunities, volatility, t+1 leverage and the earned/contributed capital (i.e., a2 = a3=…a7 = 0) The second approach (b.) allows for the interaction of N It only with our proxies of profitability, investment opportunities, and volatility The following two specifications (c and A t+1 d.) expand specification b and include leverage effects in terms of book or market leverage The last specification (e.) resembles specification d and includes lifecycle effects Argentina is use as the omitted country in deriving our country dummies At the bottom of Panel A of Table we report the number of firms and yearly observations used under each specification Wald tests’ results strongly support the inclusion of country dummies (which are mostly negative) We can decisively reject the null that the joint value of the country dummies is equal to zero Moreover, the fit of the model (that never falls below 0.26) increases slightly as we move from specification a to specification e Our evidence of a positive relationship between MAVt and dividend is similar to that of Fama and French (2002) for a U.S t sample and is in accordance with the signaling hypothesis (Bozos et al., 2011) However, the positive sign of the interaction N It MVt variable A ∗ A is unexpected since under the pecking order and trade-off models firms with high investment prospects t+1 t are expected to pay lower dividends Perhaps this positive sign can be rationalized under the premise that MAVt can also be t considered as a proxy the current profitability of firms Et N It The positive coefficients of the interaction variables related to S and A concur with the pecking order and trade off t t+1 models that assume that the most profitable firms are more prone to pay higher dividends Under the trade-off model these higher dividends are explained as a means to counter agency problems prompted by excess cash flows In the pecking order model, these higher dividends are explained by the use of more profitable assets that allow firms to maintain a low risk J Benavides et al / Finance Research Letters 17 (2016) 197–210 203 Table Determinants of dividend payout ratio Panel A Estimates with country effects Specification: a No interaction Intercept NIt / At + d-Brazil d-Chile d-Colombia d- Mexico d-Peru NIt /At + ∗ MVt /At NIt /At + ∗ Et /St NIt /At + ∗ PPNt + /At NIt /At + ∗ ln(at ) NIt /At + ∗ Lt + /At + NIt /At + ∗ Lt + /MVt + NIt /At + ∗ REt /At Wald Adjusted R2 N Firms 0.012∗∗∗ 0.335∗∗∗ −0.003 0.005∗∗∗ 0.0 0 −0.008∗∗∗ −0.005∗∗ 160.6∗∗∗ 0.268 6037 836 b No leverage c Book leverage 0.014∗∗∗ 0.014∗∗∗ 0.122∗∗ −0.003 0.004∗∗ −0.003∗ −0.009∗∗ −0.006∗∗∗ 0.107∗∗∗ 0.336∗∗∗ −0.174∗∗ −0.002 0.143∗∗∗ −0.002 0.004∗∗∗ −0.004∗ −0.009∗∗ −0.006∗∗∗ 0.108∗∗∗ 0.338∗∗∗ −0.182∗∗ −0.002 −0.034 114.4∗∗∗ 0.322 4962 770 134.5∗∗∗ 0.322 4912 769 d Market leverage e Retained earnings 0.014∗∗∗ 0.015∗∗∗ 0.394∗∗∗ −0.003 0.004∗∗ −0.006∗∗ −0.010∗∗∗ −0.007∗∗∗ 0.063∗∗∗ 0.260∗∗∗ −0.208∗∗ 0.005 −0.430∗∗∗ −0.588∗∗∗ −0.397∗∗∗ 111.0∗∗∗ 0.366 4838 767 0.274∗∗∗ −0.002 0.004∗∗ −0.005∗∗ −0.011∗∗∗ −0.006∗∗∗ 0.074∗∗∗ 0.218∗∗∗ −0.221∗∗∗ 0.004 140.0∗∗∗ 0.348 4911 770 Panel B Target payout Argentina Brazil Chile Colombia Mexico Peru a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.555 0.509 0.641 0.549 0.407 0.467 0.533 0.484 0.604 0.477 0.380 0.431 0.528 0.484 0.603 0.465 0.376 0.426 0.509 0.478 0.573 0.427 0.314 0.405 0.524 0.477 0.588 0.425 0.341 0.401 Panel C Estimates with governance indicators Specification: a No interaction Intercept NIt / At + NIt ∗ GIt NIt /At + ∗ MVt /At NIt /At + ∗ Et /St NIt /At + ∗ PPNt + /At NIt /At + ∗ ln(at ) NIt /At + ∗ Lt + /At + NIt /At + ∗ Lt + /MVt + NIt /At + ∗ REt /At Adjusted R2 N Firms 0.011∗∗∗ b No leverage 0.012∗∗∗ 0.342∗∗∗ 0.110∗∗∗ 0.012∗∗∗ 0.104∗∗ 0.112∗∗∗ 0.095∗∗∗ 0.354∗∗∗ −0.175∗∗ 0.002 0.285 6037 836 c Book leverage 0.343 4962 770 d Market leverage e Retained earnings 0.012∗∗∗ 0.013∗∗∗ 0.368∗∗∗ 0.116∗∗∗ 0.053∗∗∗ 0.297∗∗∗ −0.204∗∗ 0.009∗∗ −0.369∗∗∗ −0.551∗∗∗ −0.449∗∗∗ 0.385 4838 767 0.126∗∗∗ 0.115∗∗∗ 0.095∗∗∗ 0.369∗∗∗ −0.186∗∗ 0.0 0 −0.012 0.241∗∗∗ 0.104∗∗∗ 0.066∗∗∗ 0.253∗∗∗ −0.220∗∗∗ 0.006 0.343 4912 769 0.363 4911 770 Note: The data is from public Latin-American firms in six countries and it covers seventeen years (1995–2013) The dependent t+1 variable is DAt+1 , dividends for fiscal year t + divided by assets in year t + Panel A reports the results of panel regressions; all regressions include country dummies; dci is the dummy for country i Specification a does not include an interaction It It The next specification, includes interaction terms with ANt+1 for MAVt t , EStt , PPANt t+1 , and ln (At ) Specifications c term with ANt+1 vt+1 N It t+1 with LeAvt+1 and Le At+1 MVt+1 N It REt Panel B presents the and At+1 At a2 Mn MAVt t + a3 Mn EStt + a4 Mn PAPt Nt + and d., expand specification b interacting d with an interaction term of estimated as a0 +d ci NI Mn ( A t ) + a1 + ( ) ( ) ( ) respectively The last specification augments specification implied target payout per country The target payout is a5 Mn(ln(At ) ) + a6 Mn(Levt+1 ) + a7 Mn( RAEtt ), where Mn(.) is t+1 the sample mean of a variable, and Levt+1 is either book leverage or market leverage in t + Panel C shows regression results It replacing country dummies with a governance indicator variable, interacted with ANt+1 We estimate coefficients’ significance based on standard errors clustering by time R2 is the adjusted R2 , and N is the number of observations of each model ∗∗∗, ∗∗, and ∗ denote significance at the 0.01, 0.05, and 0.10 levels Wald test proves the null hypothesis that the joint value of the country dummies is equal to zero debt capacity to finance investment Furthermore, higher profits, ceteris paribus, reduce the financing (or funds flow) deficit (Shyam-Sunder and Myers, 1999) allowing firms to pay higher dividends The change in net plant and equipment carries the expected negative sign in line with predictions of the pecking order and trade off models In addition, the slopes for our leverage proxies show an expected and significant negative sign In the pecking order model where firms balance current and future financing costs this negative relation is natural since if more levered firms pay a higher fraction of their earnings in dividends this would increase the probability of using higher cost 204 J Benavides et al / Finance Research Letters 17 (2016) 197–210 financing In the trade-off model dividends and leverage are considered as substitutes to mitigate agency problems Thus it is sensible for more indebted firms to control their dividends payments ln (At ), our risk proxy, shows a positive slope in our most complete specifications (d and e.) This result does support the pecking order and trade- off models that hypothesize that firms with more volatile cash flow tend to be less levered as well as more conservative in their dividend policies Nonetheless, the coefficient for our risk proxy is not statistically significant The coefficient on the earned to contributed capital variable is negative The indirect relationship between dividends and the earned-contributed capital mix does not support the lifecycle theory of dividends which predicts a positive relationship Possibly, older firms in the region abstain from paying more generous dividends since they not recur to equity markets for regular financing needs, as their counterparts in more developed markets do, being less disciplined by the market (Easterbrook, 1984) and prioritizing investment flexibility (Blau and Fuller, 2008) over shareholders satisfaction Given the high ownership concentration of the listed firms in our sample (Benavides, 2005), is not surprising the reluctance to use equity markets for financing needs, as controlling shareholders seek to maintain their grip on the firm In Panel B of Table 2, the implied target payout is calculated for each country in the sample using the same five specifications of the top panel of Table The target payout is calculated as: TP = ( a0 + d ci ) + a3 M n Et St + a5 Mn ln (At ) + a6 Mn(Levt+1 ) + a7 Mn REt At Mn N It At+1 + a1 + a2 M n MVt At + a4 M n P P Nt+1 At (5) Focusing on the last two columns of Panel B we see that target payouts fluctuate widely from 0.314 to 0.588 In the last column, Chile shows the highest payout, followed by Argentina, Brazil, Colombia, Peru, and Mexico Interestingly, Chile which according to Panel B of Table has the highest rule of law indicator, also presents the most generous target payout On the other hand, firms in Peru that face an environment of weak rule of law abidance, showed one of the lowest target payouts In fact, the correlation between the country orderings by target payout and by rule of law compliance (Chile, Brazil, Argentina, Mexico, Peru, and Colombia) is strong (higher than 0.6) The positive association between the target payout and rule of law abidance gives support to the “outcome model” of the agency theory of dividends Our results for Latin America mirror those of La Porta et al (20 0) who in a cross sectional analysis of more than 40 0 firms in 33 countries for 1994 found evidence favoring the “outcome model” La Porta et al (20 0) document higher dividend payout rates in countries with better investor protection (generally Anglo-Saxon or “common law” countries in contrast to civil law countries that often have lower levels of investor protection) Likewise, Bebczuk (2007) finds that improvements in governance at the firm level (e.g., in transparency and disclosure) are usually accompanied by a higher dividend payout rate in Argentina In the bottom panel of Table we examine the positive association between TP and the rule of law variable in more detail Here we use the same model specifications as those of Panel A of Table but with two modifications We exclude the country dummies and we expand the model with a new interaction ( ANIt and GIt+1 ) to capture the effect of rule of law t+1 compliance at the country level and TP Panel C of Table shows that the magnitude, sign, and significance of our variables to proxy profitability, investment opportunities, risk, and lifecycle effects resemble those of the second panel of Table Importantly, our regression results show that the coefficient of the interaction of the rule of law variable (and ANIt ) is positive and highly significant This t+1 finding gives further credence to the idea that investors in better investor protection countries are more likely to benefit from higher dividends 3.2 Variations in investments and dividends This section examines how firms alter their dividends to accommodate variations in their investment outlays Based on A the Lintner (1956) model, and including the variable A t+1 to account for contemporaneous investment one could estimate t+1 the following dynamic model: Dt+1 N It Dt At+1 = a0 + α1 + b1 + c1 + εt+1 At+1 At+1 At+1 At+1 (6) Nonetheless, Fama and French (2002) argue that the model of Eq (6) is misspecified since TP and SOA are likely to vary across firms Thus, Eq (6) is modified to take into account both firm characteristics as well as country effects which likely affect dividend policy The resulting model follows: Firms would then have to use either debt at a higher interest rate or equity financing J Benavides et al / Finance Research Letters 17 (2016) 197–210 Dt+1 = a0 + At+1 a1 + a2 + b1 + b2 + c1 MVt Et + a3 + a4 At St MVt Et + b3 + b4 At St P P Nt REt + a5 ln (At ) + a6 Levt+1 + a7 At At 205 N It At+1 P P Nt REt Dt + b5 ln (At ) + b6 Levt+1 + b7 + d ci At At At+1 At+1 + εt+1 At+1 (7) Eq (7) implies a speed of adjustment specific for each country, equal to (where Mn stands for mean): SOA = − b1 + b2 Mn MVt At + b3 Mn Et St + b5 Mn(ln (At ) ) + b6 Mn (Levt+1 ) + b7 Mn The country dummy is interacted with Dt At+1 P P Nt At + b4 Mn REt At + d ci (8) because the SOA is likely to be affected by country characteristics (Adaoglu (20 0) and Andres et al (2009)) Given that TP is defined as the ratio of the coefficients accompanying N It At+1 evaluated at the mean values of the independent variables over SOA, then TP is also modified by the country dummies Instead of a panel regression, each year we estimate cross sectional regressions and average the estimates through the years (Fama and MacBeth, 1973) In Panel A of Table we estimate the model of Eq (7) using a pooled panel regression and clustering errors by year Table shows our results for Eq (7), the reported coefficients are the result of the interaction terms evaluated at the mean values of the independent variables that account for firm characteristics We employ five different specifications similar to those described in Section 3.1 for Table Specification is the simplest and does not consider interactions for firm characteristics, just for the country dummies Specification expands specification and includes interactions except with leverage In the following two specifications, we additionally include an interaction term for leverage, the first one using book leverage and the second one market leverage Specification (or the full model of Eq (7)) repeats specification plus an additional interaction term for retained earnings The positive and significant coefficient of ANIt supports the idea that dividend changes are influenced by firms’ profits t+1 The coefficients to gauge how dividends change after investments outlays is negative and fluctuates from −0.003 to −0.006 The finding of a negative coefficient supports the idea that firms cut back on dividends when investment requirements grow Nonetheless, and similar to previous studies for the U.S Myers (1984), Shyam-Sunder and Myers (1999) and Fama and French (2002), the magnitude of the coefficient is economically small since the change in dividends absorbs roughly just 0.6% of the change in assets Furthermore, in the top panel of Table we can see that the interacted country dummies are mostly negative but only significant for Mexico In Panel B of Table we report the speed of adjustment of dividends to changes in net income controlling for past dividends and concurrent investment needs SOA (focusing on the last column of the table) fluctuates from 0.34 to 0.72 Interestingly, our speed of adjustment estimates for Latin American tend to surpass those reported (that range from 0.28 to 0.33) for the U.S by Fama and French (2002) In a related paper, Chemmanur et al (2010) compare dividend policies in the U.S and Hong Kong Firms in Hong Kong appear to have more flexible dividend policies since they smooth dividends to a lower extent (i.e., they have a higher SOA) than in the U.S., perhaps as ownership is more closely aligned (alleviating manager-shareholders conflicts) In addition, Hong Kong managers seem less concerned with the informational content of dividend changes (e.g., reflected in abnormal returns around dividend increases and cuts) and consequently are inclined to alter dividends more swiftly than U.S managers Both factors that favor a higher speed of adjustment in Hong Kong (i.e., a close alignment between managers and shareholders and a disregard of market signals related to dividend increases or omissions) are likely to apply as well in Latin America We hypothesize that these factors play a role in understanding our finding of a higher SOA in the region Another possible explanation for our finding of a more volatile dividend policy in Latin America than in the U.S may be related to the more prevalent use of private debt in Latin America versus public debt (which is likely to be more pervasive in the U.S.) Aivazian et al (2006) find that firms in the U.S that rely on private debt are more prone to follow a purely residual dividend policy as opposed to firms with rated (public) debt Companies that depend on public debt tend to smooth dividends (as in Lintner (1956) perhaps to ameliorate signaling and agency problems brought about by their reliance in bond financing which is a funding source more susceptible to informational asymmetries than bank financing Thus, the fact that firms in Latin America are likely to rely on private debt allows them to carry a more flexible (and volatile) dividend policy in which market signals play a lesser role in determining the rate of change in dividends Furthermore, we find in Panel B of Table that firms in countries with low scores on rule of law obedience (Colombia, Mexico, and Peru) are more likely to have higher speed adjustments than firms located in countries with higher relative scores on rule of law compliance (Argentina, Brazil, and Chile) In the literature there is some support for the negative relation between how firms alter their dividends after changes in profits and the environment in which a firm operates For example, Andres et al (2009) claim that dividend volatility is less of a concern for firms in emerging markets (when compared to firms in developed markets) Further, Adaoglu (20 0) finds 206 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Table Lintner model with dynamic adjustment according to Eqs (5) and (6) Panel A Estimates with country effects Specification: a No interaction Intercept NIt / At + At + / At + Dt /At + Dt /At + ∗ d-Brazil Dt /At + ∗ d-Chile Dt /At + ∗ d-Colombia Dt /At + ∗ d- Mexico Dt /At + ∗ d-Peru R2 N Firms b No leverage c Book leverage 0.003∗∗∗ 0.001 0.113∗∗∗ −0.003 −0.234∗∗∗ −0.006 0.088 −0.015 −0.051 0.029 0.163 5,227 798 d Market leverage 0.004∗∗∗ 0.113∗∗∗ −0.006∗∗ −0.339∗∗∗ 0.037 0.049 −0.065 −0.164 −0.035 0.251 4,625 762 0.004∗∗∗ 0.106∗∗∗ −0.005∗∗ −0.334∗∗∗ 0.041 0.045 −0.094 −0.206 −0.063 0.263 4,579 760 e Retained earnings 0.004∗∗∗ 0.121∗∗∗ −0.005∗∗ −0.385∗∗∗ 0.020 0.042 −0.077 −0.334∗∗ −0.092 0.292 4,500 752 0.105∗∗∗ −0.005∗∗ −0.382∗∗∗ 0.049 0.040 −0.073 −0.240∗ −0.043 0.270 4,566 756 Panel B Speed of adjustment Argentina Brazil Chile Colombia Mexico Peru a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.23 0.24 0.15 0.25 0.28 0.20 0.34 0.30 0.29 0.40 0.50 0.37 0.33 0.29 0.29 0.43 0.54 0.40 0.38 0.33 0.34 0.46 0.62 0.42 0.38 0.37 0.34 0.46 0.72 0.48 a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.48 0.47 0.78 0.45 0.40 0.55 0.33 0.37 0.39 0.28 0.22 0.30 0.32 0.36 0.37 0.25 0.20 0.27 0.27 0.31 0.31 0.23 0.17 0.25 0.31 0.33 0.35 0.26 0.17 0.25 Panel C Target payout Argentina Brazil Chile Colombia Mexico Peru Panel D Governance indicators Specification: a No interaction Intercept NIt / At + At + / At + Dt /At + Dt /At + ∗ GIt R2 N Firms b No leverage 0.003∗∗∗ 0.0 0 0.111∗∗∗ −0.001 −0.194∗∗∗ 0.048∗∗ 0.125 5,227 798 0.111∗∗∗ −0.004∗ −0.308∗∗∗ 0.026 0.221 4,625 762 c Book leverage 0.003∗∗∗ 0.108∗∗∗ −0.004∗ −0.313∗∗∗ 0.032∗ 0.233 4,579 760 d Market leverage 0.004∗∗∗ e Retained earnings 0.004∗∗∗ 0.124∗∗∗ −0.004∗ −0.377∗∗∗ 0.049∗∗∗ 0.260 4,500 752 0.106∗∗∗ −0.004∗ −0.355∗∗∗ 0.025 0.242 4,566 756 Note: The data is from public Latin-American firms in six countries and it covers seventeen years (1995–2013) The dependent t+1 , the change in dividends for fiscal year t + versus year t divided by assets in year t + Panel A reports variable is ADt+1 the results of pooled panel regressions; all specifications include country dummies interacting with for country i The slope on N It At+1 is the average across years of a1 + a2 Mn( MAVt t ) + a3 Mn( EStt ) + a4 Mn( Dt+1 ,; At+1 P P Nt At dci is the dummy ) + a5 Mn(ln(At ) ) + a6 Mn(Levt+1 ) + a7 Mn( RAEtt ), where Mn(.) is the sample mean of a variable, are the regression coefficients from Eq (6) and Levt+1 is either book leverage or market leverage in t + Meanwhile, the slope on b1 + b2 Mn( MVt At ) + b3 Mn( ) + b4 Mn( Et St P P Nt At ) + b5 Mn(ln(At ) ) + b6 Mn(Levt+1 ) + b7 Mn( REt At Dt At+1 is the average across years of ) where bi are the regression coefficients from Eq (6) Panel B presents the speed of adjustment per country, which is the negative of the sum of the slope on the dci The implied target payout in Panel C is the slope on N It At+1 Dt At+1 and divided by the speed of adjustment Panel D shows regression results replacing country dummies with a governance indicator variable, interacted with Dt At+1 We estimate coefficients’ signifi- cance based on standard errors clustering by time R2 is the adjusted R2 , and N is the number of observations of each model ∗∗∗, ∗∗, and ∗ denote significance at the 0.01, 0.05, and 0.10 levels that firms in Turkey follow a pure residual policy (i.e., he reports a SOA of 1.0) before and after a change in regulation that took place in 1995 that exempted firms from paying a minimum mandatory dividend Even though the regulation shift should have prompted firms listed on the Istanbul Stock Exchange to adopt more flexible dividend policies these firms stuck to an unstable dividend policy Consequently, smoothing (omitting) dividends appears less (more) prevalent in emerging markets which often score poorly on rule of law compliance In all, we find that firms in the region adjust dividends, more swiftly than their U.S counterparts, to positive and negative shocks in earnings A higher alignment between management and shareholders in an environment of high concentration of J Benavides et al / Finance Research Letters 17 (2016) 197–210 207 ownership as well as a lower concern by Latin American managers of the impact of dividend policy changes on firm value could explain our results In addition, we expand the international evidence on the positive relationship between SOA and the extent of rule of law compliance, in line with previous studies Target payouts reported in Panel C of Table are roughly consistent (but somewhat smaller) than those reported in Panel B of Table Yet again, firms in countries with higher rule of law scores reward their investors with richer dividends The bottom panel of Table reports an alternative model of Eq (7) The country dummies are replaced by the rule of law index Confirming our previous findings the coefficient of the governance index is positive (indeed, in 13 out of 18 years the coefficient is positive) and significant, implying that countries with a higher governance score tend to have a lower SOA Robustness checks In this section we discuss several robustness tests The main focus of the tests is to examine whether changes in the estimation period or in our regression specifications have any impact on our conclusions Using similar specifications to those in Table 3, Table reports our estimates of the speed of adjustment, the target payout, and the impact of rule of law compliance on dividend payments for two sub-periods The first period extends from 1995 to 2003, and the second covers from 2004 to 2013 We examine the stability of our coefficients or estimates of interest before and after the crisis in Argentina that was triggered by an abrupt devaluation of the Argentinean peso on December of 2001 In the first four panels of Table we observe several interesting patterns In the second period all countries raised the target payout with respect to the first except for Argentina (possibly as a consequence of the negative effects of the crisis on Argentinean firms’ profitability) For example, under specification e Colombian firms increased the target payout from 0.24 to 0.45 (for an 88% increase) Similar to our findings in Table 3, we document a positive relationship between the standing of a country in terms of the level of law enforcement and the target payout For example, in 2002–2013, firms in Chile, Brazil, and Argentina showed a higher target payout that firms located in countries with lower rule of law scores (Colombia, Peru, and Mexico) Focusing on the speed of adjustment across sub-periods, we document a general decrease in SOA for all countries except Argentina It seems that most firms in the region pursued a less volatile dividend policy on the second half of the sample On the other hand, Argentinean firms pursued a slightly more erratic dividend policy after the crisis Furthermore, this higher dividend volatility coincided with a significant drop in 2002 in Argentina’s score on rule of law abidance (the score remained significantly lower, with respect to pre-crisis levels, until the end of the sample) Our findings for Argentina reinforce the idea of a negative relationship between the level of law compliance and the velocity in which firms alter their dividends as a response to shocks in profits In Panel E and Panel F of Table we replace again the countries dummies for the rule of law indicator In our analysis we will focus on the coefficient on ADt ∗ GIt+1 In the first period the coefficient is positive, however, in none of the specit+1 fications, the coefficient is statistically significant During the second period (2002–2013) the coefficients on rule of law are positive and higher than those for the first period Importantly, the coefficient on the interaction variable to account for rule of law effect is significant In all, the effect of rule of law on dividend policy (and in particular on the speed of adjustment) appears to show more strongly in the second half of our sample We conduct additional robustness checks not reported in tables We include industry dummies in Eq (7) (Table 2) with similar qualitative results Furthermore, we proxy country governance in Table with a voice and accountability (VA) indicator instead of a rule of law index VA tries to capture the degree of freedom of expression within a country In short, our main findings of a positive (negative) effect of country governance on the target payout (speed of adjustment) remain unaltered In a world with frictions (e.g., frictions related to trading costs and microstructure effects) dividends are possibly a less costly option for shareholders (than selling their stocks) to obtain cash from their investments in the firm Banerjee et al (2007) find support for the notion of a substitution role of liquidity and dividends since that less liquid firms are more likely to pay dividends than more liquid firms In this untabulated robustness check, we examine whether a firm’s liquidity affects its dividend decisions To this end, we use Amihud’s (2002) illiquidity measure (the yearly average of the ratio of the absolute daily return over daily dollar volume) to proxy for liquidity effects When included in Tables and 3, we observe no statistically significant effect of illiquidity on the target payout and the speed of adjustment (the magnitude and significance of the remaining variables continues to be qualitatively similar) Perhaps dividend policies of Latin American firms are more reactive to shocks in aggregate or overall liquidity rather than to changes in the liquidity of the firm’s shares Examining this last assertion, nevertheless, is beyond the scope of this paper Concluding remarks We contribute to the literature by analyzing dividend payment decisions of firms in six Latin American countries from 1995 to 2013 Based upon the classic Lintner model (1956), and examining the predictions of three guiding theories (pecking order model, trade-off model and the lifecycle theory of dividends), our analysis highlights the importance of including both firm and country specific factors when analyzing how firms adjust their dividends after changes in earnings 208 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Table Robustness checks Panel A Speed of Adjustment (1995–2003) Argentina Brazil Chile Colombia Mexico Peru a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.20 0.39 0.24 0.35 0.33 0.18 0.30 0.45 0.41 0.39 0.52 0.42 0.29 0.45 0.41 0.42 0.65 0.47 0.33 0.48 0.46 0.47 0.71 0.49 0.32 0.51 0.45 0.51 0.79 0.58 Panel B Speed of adjustment (2004–2013) Argentina Brazil Chile Colombia Mexico Peru a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.32 0.10 0.08 0.05 0.27 0.25 0.39 0.15 0.17 0.19 0.48 0.35 0.42 0.14 0.18 0.16 0.48 0.36 0.44 0.18 0.23 0.30 0.55 0.36 0.45 0.20 0.23 0.26 0.52 0.38 a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.44 0.23 0.37 0.25 0.27 0.48 0.35 0.23 0.26 0.27 0.20 0.25 0.36 0.23 0.25 0.25 0.16 0.22 0.31 0.21 0.22 0.21 0.14 0.21 0.38 0.24 0.27 0.24 0.15 0.21 a No interaction b No leverage c Book leverage d Market leverage e Retained earnings 0.40 1.31 1.52 2.79 0.48 0.52 0.29 0.77 0.66 0.60 0.24 0.32 0.28 0.86 0.64 0.72 0.24 0.32 0.24 0.60 0.47 0.35 0.19 0.30 0.26 0.58 0.52 0.45 0.22 0.31 Panel C Target payout (1995–2003) Argentina Brazil Chile Colombia Mexico Peru Panel D Target payout (2004–2013) Argentina Brazil Chile Colombia Mexico Peru Panel E Estimates with governance indicators (1995–2003) Specification: a No interaction b No leverage c Book leverage d Market leverage e Retained earnings Intercept NIt / At + At + / At + Dt /At + Dt /At + ∗ GIt R2 N Firms 0.001∗∗ 0.085∗∗∗ 0.001 −0.280∗∗∗ 0.037 0.168 2,155 567 0.003∗∗∗ 0.094∗∗ −0.001 −0.384∗∗∗ −0.003 0.249 1,829 499 0.003∗∗∗ 0.094∗∗∗ −0.001 −0.406∗∗∗ 0.007 0.267 1,806 496 0.004∗∗∗ 0.094∗∗ −0.002 −0.446∗∗∗ 0.0 0 0.284 1,805 494 0.003∗∗∗ 0.113∗ −0.001 −0.462∗∗∗ 0.033 0.308 1,797 493 c Book leverage d Market leverage e Retained earnings Panel F Estimates with governance indicators (2002–2013) Specification: Intercept NIt / At + At + / At + Dt /At + Dt /At + ∗ GIt R2 N Firms a No interaction 0.0 0 0.123∗∗∗ 0.0 0 −0.135∗∗∗ 0.057∗∗ 0.094 2,819 588 b No leverage 0.003∗∗ 0.112∗∗∗ −0.005 −0.218∗∗∗ 0.057∗ 0.167 2,587 570 0.003∗∗ 0.118∗∗∗ −0.006 −0.220∗∗∗ 0.054∗ 0.179 2,564 565 0.004∗∗∗ 0.003∗∗ 0.123∗∗∗ −0.004 −0.271∗∗∗ 0.070∗∗ 0.207 2,502 558 0.108∗∗∗ −0.004 −0.249∗∗∗ 0.054∗ 0.191 2,556 566 Note: The data is from public Latin-American firms in six countries and it covers seventeen years (1995–2013) The depent+1 „ the change in dividends for fiscal year t + versus year t divided by assets in year t + We split the dent variable is ADt+1 dataset in two periods: period a goes from 1995 to 2003 and period b goes from 2004 to 2013 Panel A to Panel D present t Panel A and Panel SOA and TP from pooled panel regressions of Eq (6) that include country dummies interacted with ADt+1 B present the speed of adjustment per country, which is the negative of the sum of the slope on Dt At+1 and the dci , for pe- riods a and b respectively The implied target payout in Panel C (period a.) and Panel D (period b.) is the slope on vided by the speed of adjustment The slope on N It At+1 is the average across years of a1 + a2 Mn( MAVt t ) + a3 Mn( EStt ) + a4 Mn( N It At+1 P P Nt At di- )+ a5 Mn(ln(At ) ) + a6 Mn(Levt+1 ) + a7 Mn( RAEtt ), where Mn(.) is the sample mean of a variable, are the regression coefficients from Eq (6) and Levt+1 is either book leverage or market leverage in t + Meanwhile, the slope on Dt At+1 is the average across years of b1 + b2 Mn( MAVt t ) + b3 Mn( EStt ) + b4 Mn( PAPt Nt ) + b5 Mn(ln(At ) ) + b6 Mn(Levt+1 ) + b7 Mn( RAEtt ), where bi are the regression coefficients from Eq (6) Panel D and Panel E shows regression results replacing country dummies with a governance indicator t for periods a and b respectively We estimate coefficients’ significance based on standard errors variable, interacted with ADt+1 clustering by time R2 is the adjusted R2 , and N is the number of observations of each model ∗∗∗, ∗∗, and ∗ denote significance at the 0.01, 0.05, and 0.10 levels J Benavides et al / Finance Research Letters 17 (2016) 197–210 209 We find broad support for the common empirical predictions of the pecking order and trade-off models More profitable firms tend to pay a higher relative (e.g., with respect to assets) dividend while more indebted firms or firms with higher investment needs are more likely to pay lower dividends We not find a significant effect of volatility (proxied by firm size) on the dividend payout ratio Contrary to predictions of the lifecycle theory of dividends, the effect of the earned to contributed capital in the dividend payout is negative We hypothesize that for the listed firms in the region paying a generous dividend is not in their agenda since they have not yet exhausted their investment opportunities, their high ownership concentration and, consequently, the very low frequency of new financing through the equity markets Furthermore, we find differential target payouts and speeds of adjustments per country Importantly, we find that firms in countries with a higher rule of law compliance are more likely to pay a higher rate of dividends This positive association between the target payout and rule of law abidance supports the “outcome model” of the agency theory of dividends (La Porta et al., 20 0) It appears that investors in more relatively law abiding countries (Argentina, Brazil, and Chile) are able to extract higher dividends than those investors in countries where the rule of law is weaker (Colombia, Mexico, and Peru) In terms of the speed of adjustment we document an indirect relationship between SOA and rule of law indices In all, it appears that firms in low rule of law countries are more prone to conduct a more erratic dividend policy than firms in high rule of law countries We thus extend previous evidence (Adaoglu (20 0) and Andres et al (2009)) that suggests a close link between how quickly firm adjust their dividends to changes in earnings, and country characteristics in which a firm is located A series of robustness checks gives further credence to our results We split the sample in two periods to isolate the effect of the Argentinian crisis (in the end of 2001) It seems that the effect of rule of law on the dividend payout is stronger and more significant in the second part of our sample Nonetheless, our positive (negative) relationship between the rule of law country scores and the target payout (speed of adjustment) holds both for the pre- and post-crisis periods Different variations in how we proxy for rule of law and the use of alternative estimation techniques directed us to the same conclusions We leave for future research the effect on dividend policy of the ownership concentration and firm type (whether the firm is owned by a family, it is a widely held corporation, or a state owned firm) Related research has studied the effect of ownership concentration on dividends in a country level (Lefort and Walker, 2005) Nonetheless, research in cross country differences and the effect of ownership type in dividend policies is scant and can shed light on how finance and governance theories interact for emerging markets Repurchases are becoming popular among U.S managers as a way to distribute excess cash (Brav et al., 2005; Brawn and Sevic, 2015) Examining whether dividend payments and repurchases are concurrent or not (Fama and French, 2001), and repurchase determinants in Latin America is also in our agenda to gain a more comprehensive understanding of payout policy References Adaoglu, C., 20 0 Instability in the dividend policy of the Istanbul Stock Exchange (ISE) corporations: evidence from an emerging market Emerg Mark Rev 1, 252–270 Aivazian, V.A., Booth, L., Cleary, S., 2006 Dividend Smoothing and Debt Ratings J Financial Quant Anal 41, 439–453 Amihud, Y., 2002 Illiquidity and stock returns: cross-section and time-series effects J Financial Mark 5, 31–56 Andres, C., Doumet, M., Fernau, E., Theissen, E., 2015 The Lintner model revisited: dividends versus total payouts J Bank Finance 55, 56–69 Andres, C., Betzer, A., Goergen, M., Renneboog, L., 2009 Dividend policy of German firms: a panel data analysis of partial adjustment models J Empir Finance 16, 175–187 Banerjee, S., Gatchev, V.A., Spindt, P.A., 2007 Stock market liquidity and firm dividend policy J Financial Quant Anal 42, 369–397 Bancel, F., Mittoo, U.R., 2004 Cross-country determinants of capital structure choice: a survey of European firms Financial Manag 33, 103–132 Bebczuk, R.N., 2007, Access to credit in Argentina (United Nations, CEPAL) Benavides, J., 2005 Ownership Concentration and Accounting Performance: The Latin-American Case Blau, B., Fuller, K.P., 2008 Flexibility and dividends J Corp Finance 14, 133–152 Booth, L., Aivazian, V., Demirguc-Kunt, A., Maksimovic, V., 2001 Capital structures in developing countries J Finance 56, 87–130 Bozos, K., Nikolopoulos, K., Ramgandhi, G., 2011 Dividend signaling under economic adversity: evidence from the London Stock Exchange Int Rev Financial Anal 20, 364–374 Brav, A., Graham, J.R., Harvey, C.R., Michaely, R., 2005 Payout policy in the 21st century J Financial Econ 77, 483–527 Brawn, D., Sevic, A., 2015 Net payout return: An alternative to the traditional returns approach based on dividends and share repurchases Finance Res Lett 13, 66–73 Brockman, P., Unlu, E., 2011 Earned/contributed capital, dividend policy, and disclosure quality: an international study J Bank Finance 35, 1610–1625 Chemmanur, T.J., He, J., Hu, G., Liu, H., 2010 Is dividend smoothing universal?: New insights from a comparative study of dividend policies in Hong Kong and the U.S J Corp Finance 16, 413–430 de Jong, A., Kabir, R., Nguyen, T.T., 2008 Capital structure around the world: the roles of firm- and country-specific determinants J Bank Finance 32, 1954–1969 DeAngelo, H., DeAngelo, L., Stulz, R.M., 2006 Dividend policy and the earned/contributed capital mix: a test of the life-cycle theory J Financial Econ 81, 227–254 Demirgüç-Kunt, A., Maksimovic, V., 1999 Institutions, financial markets, and firm debt maturity J Financial Econ 54, 295–336 Easterbrook, F.H., 1984 Two agency-cost explanations of dividends Am Econ Rev 74, 650 Fama, E.F., French, K.R., 2001 Disappearing dividends: changing firm characteristics or lower propensity to pay? J Financial Econ 60, 3–43 Fama, E.F., French, K.R., 2002 Testing trade-off and pecking order predictions about dividends and debt Rev Financial Stud 15, 1–33 Fama, E.F., MacBeth, J.D., 1973 Risk, return, and equilibrium: empirical tests J Political Econ 607–636 Floyd, E., Li, N., Skinner, D.J., 2015 Payout policy through the financial crisis: the growth of repurchases and the resilience of dividends J Financial Econ 118, 299–316 Iturriaga, F.J.L., Crisóstomo, V.L., 2010 Do leverage, dividend payout, and ownership concentration influence firms’ value creation? an analysis of Brazilian firms Emerg Mark Finance Trade 46, 80–94 210 J Benavides et al / Finance Research Letters 17 (2016) 197–210 Jensen, M.C., 1986 Agency costs of free cash flow, corporate finance, and takeovers Am Econ Rev 76, 323 Kaufmann, D., Kraay, A., Mastruzzi, M., 2010 Response to ’what the worldwide governance indicators measure?’ Eur J Dev Res 22, 55–58 Kirch, G., Terra, P.R.S., 2012 Determinants of corporate debt maturity in South America: institutional quality and financial development matter? J Corp Finance 18, 980–993 La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R.W., 20 0 Agency problems and dividend policies around the world J Finance 55, 1–33 Lefort, F., Walker, E., 2005 The effect of corporate governance practices on company market valuation and payout policy in Chile SSRN, 676710 Lintner, J., 1956 Distribution of incomes of corporations among dividends, retained earnings, and taxes Am Econ Rev 46, 97–113 Mitton, T., 2004 Corporate governance and dividend policy in emerging markets Emerg Mark Rev 5, 409–426 Myers, S.C., 1984 The capital structure puzzle J Finance 39, 575–592 Rajan, R.G., Zingales, L., 1995 what we know about capital structure? some evidence from international data J Finance 50, 1421–1460 Shyam-Sunder, L., Myers, S.C., 1999 Testing static tradeoff against pecking order models of capital structure J Financial Econ 51, 219–244 Velez-Pareja, I., Benavides Franco, J., 2011 Cost of capital when dividends are deductible Braz Rev Finance 9, 309–334  ... shape the dividend decisions of Latin American firms within the framework of the pecking order and trade off models, and the recent lifecycle theory of dividends We find that the dividend payout rate... more dividends from companies they invest in A second version of the agency theory of dividends (“substitute model”) considers dividends and investor protection as substitutes In this version, dividends... specialized in Latin American exchanges.3 In addition, we gather information on a rule of law index from the World Bank The index is built from thirty underlying sources which report perceptions from survey