Ownership and identities of the largest shareholders and dividend policy: Evidence from Vietnam

!is study investigates the relationship between the level of shareholdings and identities of the largest shareholders, and cash dividend policy. !e study is conducted with a sample of 180 "rms listed on Vietnam stock exchange markets #om 2009 to 2013. !e "xed e$ect model is employed to analyze the balanced panel data. !e results show that the higher the level of holdings by the largest shareholders, the lower the dividend payout. Moreover, companies with the State and Foreign investors as the largest shareholders have higher dividend payout ratio than companies with local investors and managers as the largest shareholders. !e study also "nds that companies tend to pay higher dividends when pro"ts decrease or growth opportunities increase.


Introduction
Privatization has been proven to be a successful approach for remarkable developments in emerging countries with deep State involvement (World Bank, 1995). Privatization may generally lead to economic improvement because of enhanced resources allocation. However, the change in ownership structure resulting from privatization may also create agency problems re ecting di erent shareholders' perspectives. ere can be a con ict of interests between managers and shareholders and/or between large and minority shareholders regarding corporate decision (Shleifer & Vishny, 1997;Maury & Pajuste, 2003;Easterbrook, 1984;Dharwadkar, George & Brandes, 2000;Maury & Pajuste, 2002;Gugler & Yutoglu, 2003;Ramli, 2010;anatawee, 2013).
is paper focuses on the impact of the largest shareholder in the company on dividend policy. A high dividend payout can be a burden for companies' bo om lines. However, a low dividend payout may not be desirable to shareholders since it is an 87 essential tool to protect shareholders from management misconducts. If dividends are not paid to shareholders, managers may channel excess fund to ine cient investments (Easterbrook, 1984). Meanwhile, dividend payout may also aggravate agency con icts (Easterbrook, 1984). Faccio et al. (2001) indicate that the agency con icts between large and small shareholders, which result from their di erent incentives and power, may lead to high dividend payouts (Ramli, 2010). In some countries, shareholding of the rst largest shareholder can be far exceeding that of the second largest shareholder of rms in emerging markets (Faccio et al., 2001;Maury & Pajuste, 2002;Gugler & Yugtolu, 2003;Harada & Nguyen, 2011;anatawee, 2013, 2014. e largest shareholder, with their dominant percentage of shareholding within the rm, has more power and incentives compared to other shareholders to represent all shareholders and monitor the self-dealing managers ( Jensen et al., 1999;Maury & Pajuste, 2003;Amidu & Abor, 2006;Al-Malkawi, 2010). Nevetheless, it has been found that the large shareholding held by the largest shareholders facilitates their opportunity to extract private bene ts or collude with other large shareholders to expropriate corporate resources (Faccio et al., 2001;Maury & Pajuste, 2002;Gugler & Yutoglu, 2003).
In Vietnam, the "Doi Moi" (Economic reform) has been carried out since 1992 to improve economic e ciency through privatizing state-owned enterprises (SOEs). e inevitable outcome of privatization is the shi from sole state ownership to various ownership categories in privatized SOEs. However, Vietnam is assumed to be in the transition towards a market-based economy and the government continues to retain the decisive roles in post-privatized companies (Truong & Heaney, 2007;Le & Chizema, 2011;Le & Buck, 2009). In the process, Vietnam market is characterized by weak corporate governance. e shareholder protection index or the disclosure index are lower than the mean value, showing no remarkable improvement (Global Competitiveness Index report, 2009;e World Bank, 2012). e privatized rms' management, retained from State owned enterprises (SOEs) is in short of competency and experience to drive companies in competitive markets (Truong & Heaney, 2007).
In this institutional context, Vietnam is a great example to raise awareness of agency problem a er privatization. is paper therefore a empts to investigate the in uence of post-privatization ownership structure on dividend payout policy in Vietnam context. A sample of 180 companies privatized and listed on the two stock exchanges of Vietnam (i.e. Ho Chi Minh City Stock Exchange and Ha Noi Stock Exchange) in the period of 2009-2013 is employed for the study.
In addition to its examination of the relationship between the shareholdings of the largest shareholder and dividend policy in Vietnam, the paper distinguishes itself from previous studies in at least one aspect. Speci cally, it examines the di erential impact of the identity of the largest shareholders on the dividend policy of the Vietnamese listed companies. Results from the study can be used as reference for the privatization progress in Vietnam. Underlying problems from the impact of the largest shareholder on 88 dividend policy will be also discussed. Results will be analyzed and discussions relevant to emerging countries with relatively similar context will then be related. e next section of the paper presents the theoretical background of the study and hypotheses development. It is followed by section three, which introduces the methodology of the study, and section four, which analyzes the data results. Section ve discusses data results and concludes the paper.

eories and hypotheses
e impact of the largest shareholders on dividend policy will be investigated in two aspects: their shareholdings within the rms and their identities.

Shareholdings and dividend policy
e shareholdings of the largest shareholder are assumed to have a signi cant impact on corporate policies, particularly the dividend decision. However, the in uence varies across countries, especially the emerging ones with weak corporate governance context.
Classical agency perspective emphasizes the con ict of interests between managers and shareholders. Easterbrook (1984) proposes that dividend payout can play a monitoring function, reducing cash ows available to managers who can manipulate available resources and pursue negative return investments (Al-Najjar & Hussainey, 2011). Because of dispersed ownership, there is high probability for free-rider problem to occur. Monitoring of managers is more challenging in the presence of a free-rider problem (Shleifer & Vishny, 1986). Small shareholders may choose not to supervise the management since they expect the others will do. In contrast, large shareholders have a be er chance than minor shareholders to establish nancial discipline on managers. Minimum resources of rms will then be invested in low return projects (Easterbrook, 1984;Claessens & Djankov, 1999;Maury & Pajuste, 2002;Ramli, 2010;Harada & Nguyen, 2011). Moreover, because of bearing higher cost for monitoring the management than small shareholders, large shareholders have more incentives to require higher dividend payment to compensate for such cost (Easterbrook, 1984;Maury & Pajuste, 2003;Ramli, 2010;Harada & Nguyen, 2011). e positive relationship between shareholdings of large shareholders and dividend payout is found in both Ramli (2010) for Malaysia and anatawee (2013) for ailand.
However, in a weak corporate governance context, Vishny (1997), La Porta et al. (1999), Faccio et al. (2001), and Ramli (2010) argue that not the agency problem between managers and shareholders but the con ict of interests between large and small shareholders may be dominant in these markets. According to Shleifer and Vishny (1997) and Claessens and Djankov (1999), for large shareholders, the bene ts from in uencing management to make favorable decisions for their own sakes may outweigh the bene ts of representing other shareholders in the monitoring of managers. us, large shareholders may have the tendency to act for their own bene ts at the 89 expense of other investors. Empirically, the negative relationship between high level of shareholdings and dividend payout is documented (e.g., Maury & Pajuste, 2002;Gugler & Yutoglu, 2003;Harada & Nguyen, 2006;Bena & Hanousek, 2008). Practically, they may force managers to use free cash ows to make investment decisions with the companies they own even though returns on these investments are not desirable.
According to Nguyen (2008), in the case of Vietnam, the external monitoring system is considered underdeveloped, and the corporate governance is generally weak in terms of several categories such as minority shareholder protection and disclosure requirement. e largest shareholder has superior shareholdings compared to other shareholders, and minority shareholders are not su ciently protected (IFC reports on governance of 2012, 2013). ese characteristics of the market tend to create the agency problem where large shareholders expropriate minority shareholders. Based on widely documented tunneling behavior of the largest shareholder in the institutional context of weak corporate governance, the ownership held by the largest shareholder is speculated to have a negative in uence on dividend payout policy as the result of the low minority shareholder protection (La porta et al., 2000). e rst hypothesis is accordingly established: H 1 : ere is negative relationship between percentage of shareholdings of the largest shareholder and the dividend payout ratio.

Does identity of the largest shareholders ma er?
Owners may di er in operating targets, motivation, risk preference, capability and control of nancial resources as well as managerial expertise (Maury & Pajuste, 2003;Gugler & Yugtolu, 2003;Ramli, 2010). erefore, it is speculated that di erent identities of the largest shareholders will not have the same impact on dividend policy (Lace et al., 2013).

The State and dividend policy
According to Bradford (2013), privately owned companies have been limited in accessing external capital resources. us, they have to rely on internal sources for investment and hence apply a low dividend payout policy. Meanwhile, the state shareholder 90 can easily access external sources. Companies with the State as a large shareholder can have support from government to obtain external sources such as favorable loan terms (Le & Chizema, 2012;Le & O'Brien, 2010). erefore, companies with high levels of state ownership tend to pay higher levels of cash dividend. In addition, according to the signaling hypothesis, the state, which has a tendency to play a pivotal role in strategic sectors which are important in economy, desires to strengthen its position, signaling a good image. erefore, companies with large state ownership may pay higher cash dividend to signal their positive performance (Bradford, 2013;Sulong & Nor, 2008;Wang et al., 2011).  also nd that dividend payouts increase when government ownership increases in China market. However, as transfer of State shares can only be realized by the government's approval, the tactic of paying high dividend may facilitate the state in transferring a portion of non-tradable shares to other shareholders. is suggests that dividend may be manipulated to serve the purpose of the large shareholders, i.e. the State, instead of protecting shareholders. According to Sun et al. (2005), rms with high levels of state ownership have tendency to take disproportional pro t to compensate for the support they o er to companies (Xu & Wang, 1999;Nguyen, 2008;Wang et al., 2011;and Bai et al., 2013). In general, the State shareholder is assumed to prefer high level of dividend payments.

Managerial shareholdings and dividend policy
Managers tend to pay low level of dividend and retain high level of earnings in order to grasp investment opportunities at their convenience (Roze , 1982;Alli et al., 1993;Chay & Suh, 2009;Chen & Dhiensiri, 2009). From a conventional agency perspective, holding a position in management, large shareholders can have more opportunities to be er supervise and alleviate the management discretion ( Jensen & Meckling, 1976;Short et al., 2002;Chen & Dhiensiri, 2009). Hence, the accountability of investment decisions is improved ( Jensen & Meckling, 1976;Ang, Cole & Lin, 2000;Short et al., 2002;Chen & Dhiensiri, 2009). In this context, paying high dividend as a device to monitor managers and increase accountability of their actions is not considered as an e ective practice (Chen & Steiner, 1999;Fenn & Liang, 2001;Al-Malkawi, 2005). However, aligning the interests between principals and agents in the condition of weak corporate governance may be a challenge because of the entrenchment problem. Speci cally, Jensen (1983) suggests that managerial entrenchment is considered as one of the costliest manifestations of agency problem (White 1996;Fenn & Liang, 2001;Maury & Pajuste, 2002). As the amount of managerial stockholding increases to a certain level, managers start seeking for their personal utility through non-value-maximizing behaviors such as high salary, empire building and so forth (Maury & Pajuste, 2002;Lins, 2003;Miguel, Pindado & Torre, 2004;Bunkanwanicha et al., 2008). In general, companies with the managers as the largest shareholders may pursue lower levels of cash dividend than companies with other type of the largest shareholders.  Sulong and Nor (2008) found that foreign investors in Malaysia prefer paying low cash dividends. ey argue that foreign investors, employing be er monitoring disciplines in their companies in emerging markets, do not require high cash dividend payment to reduce agency con ict. Moreover, due to the costs of transferring dividends overseas which may be taxed in their home countries, foreign investors may prefer low dividend payments (Sulong & Nor, 2008;Chai, 2010;Ullah et al., 2012;Abdullah et al., 2012). However, in a weak corporate governance context, foreign owners, who are highly disadvantageous in terms of information on rm performance and market and legal changes, may desire more for dividend payments. In other words, foreign investors as large shareholders may require a higher cash dividend payment compared with local investors. Cook and Jeon (2006), Baba (2009), Warrad et al. (2012 and anatawee (2013) also suggest that foreign shareholdings are associated with a higher dividend payout than domestic shareholdings. Overall, in the weak governance context of Vietnam, it is expected that foreign largest shareholders may prefer higher payouts than local largest shareholders. e second hypothesis is accordingly established:

Foreign investor and dividend policy
H 2 : ere is di erential impact of identities of the largest shareholders on dividend payout ratio.

Research design and sample size
Data were collected from 2009 to 2013 and organized into a balanced panel. e nonprobability sampling method is applied. e sample is comprised of companies listed on two well-recognized stock exchanges in Vietnam, which are Ho Chi Minh City Stock Exchange (HOSE) and Hanoi Stock Exchange (HNX). e rms were all SOEs before listing, i.e. listed rms which were not SOEs before being listed are excluded. In addition, the nancial sector including banks, real estate, securities and insurance companies is excluded from the data due to its distinguished characteristics of corporate structures and revenue models. Companies with insu cient data in the study period are also excluded as the unbalanced panel may introduce the noise of unit heterogen-92 ity. Converting unbalanced panel data into a balanced panel data may result in a biased sample if the missing data is not random (Cameron & Trivedi, 2005). Finally, a sample of 180 rms is chosen for the research.

Model speci cation
Two multiple regression models are constructed to test the two hypotheses of the study as follows: (1) Model (1) is constructed to examine the relationship between shareholdings of the largest shareholder and dividend policy. (2) Model (2) is constructed to account for the impact of di erent types of the largest shareholder (the identity of the largest shareholders) including the State, managers, foreign investors and local investors on dividend payout policy.

Dependent variable
Dependent variable is Dividend payout ratio (DPR) measured as the ratio of cash dividends per share divided by earning per share. DPR is considered as a more appropriate indicator of dividend policy than dividend yield or dividend per share since payout ratios and ploughed back ratios are taken into consideration (Roeze , 1982). Moreover, cash dividend is used because it directly a ects the equity and cash holding of a company.

Independent variables/ Ownership structure variables
Ownership structure refers to two dimensions: Shareholding of the largest shareholder (SHARE) is calculated as the percentage of ownership of the shareholders who directly own the highest volume of shares within the company.
Ownership identity dummies: Binary variables are used to capture the identity of the largest owner (Ramli, 2010;anatawee, 2013;anatawee, 2014). In the particular context of Vietnam, this study classi es the identities of the largest shareholders into 4 categories: (1)

93
(4) local investor (LOC) 2 . Companies in which the largest shareholders are local investors are classi ed as a base category.

Control variables
Pro tability (PROF) is measured as net income divided by total asset. According to the Pecking order theory, companies with low pro tability may pay low dividends retaining high levels of earning for investment since issuing debt or equity for investment is expensive (Litner, 1986;Jensen et al., 1986;Fama & French, 2000).
Firm size (SIZE) is proxied by (logarithm of) total assets as in Chay and Suh (2009) and Chen and Dhiensiri (2009) since large rms may depend less on internal funds for future investment because they have easier access to external debt with their reputations (Holder et al., 1998).
Leverage (LEV) is de ned as total debt to total assets. LEV is controlled to account for the impact of debt on dividend payouts because companies with more nancial obligations imposed by debt nancing practices may not have su cient fund to pay high cash dividends ( Jensen, 1986). High leveraged companies are more likely to pay low dividends to avoid using external debts with unfavorable loan terms (Roze , 1982;Gugler & Yugtolu, 2003).
Firm growth (GROW) is the percentage of change in a rm's sales. GROW should be controlled since rms with growth opportunities require more capital for investing purposes (Roze , 1982;Al-Malkawi, 2010;Chen & Dhiensiri, 2009). To avoid transaction costs due to external nancing, rms may keep high retention levels, i.e. pay less dividend to reduce reliance on debt (Myers & Majluf, 1984). us, the negative relationship between growth opportunities and payout is expected.

Methodology
Firstly, in order to test the two hypotheses of the study, regression analysis is conducted in Pooled OLS, REM and FEM.
Secondly, the speci cation tests, including F-statistic test and Breusch and Pagan Lagrange multiplier test and Hausman test are used to determine appropriate models. en, the Breusch-Pagan test and Wooldridge test for autocorrelation in panel data are conducted to check for heteroskedasticity and autocorrelation problems. If panel data has heteroskedasticity and/or autocorrelation problem, one common practice to cure the problems is to use cluster-robust standard errors. Clustering at the crosssectional panel level will produce not only consistent but also more e cient estimates 2 According to Article 2, Chapter 1, Decision 121/2008/QD-BTC, foreign investors are "individuals with foreign nationality who reside overseas or in Vietnam, including people of Vietnamese origin with foreign nationality; organizations established and operating pursuant to foreign law and their branches including branches operating in Vietnam; organizations established and operating pursuant to the law of Vietnam with 100% foreign capital contribution, and their branches; investment funds established and operating pursuant to foreign law and investment funds established and operating pursuant to the law of Vietnam with 100% foreign capital contribution. " of standard error (Arellano, 1987;Baltagi, 2001;Wooldridge, 2010). However, even robust standard errors will be biased downward if residuals are correlated across sections. In this case, Driscoll and Kraay (1998) standard errors can be a solution to both cross-sectional and time dependence form of residuals. In order to choose which approach to employ for mitigating the impact of heteroskedasticity and autocorrelation problem, Pesaran cross-sectional dependence is performed. e null hypothesis is that errors are not correlated across entities. If the null hypothesis is rejected, Driscoll and Kray standard errors will be applied. Otherwise, cross-sectional cluster robust standard errors will be applied to chosen models.

Descriptive statistics
e descriptive analysis presented in Table 1 provides an overview of variables employed in the two models. e high gap between the maximum and the minimum DPRs re ects the wild uctuations in the dividend payment practices of the samples. e mean value of DPR indicates that, on average, these companies use 50% of their earnings to distribute cash dividends to shareholders. e mean value is approximate to that of companies in ailand (47%) ( anatawee, 2013), while signi cantly higher than the values of the companies in China (16.81%) ( anatawee, 2014), Japan (33%) (Harada & Nguyen, 2011), Malaysia (22%) (Ramli, 2010) and Canada (32.8%) (Adjaoud & Ben-Amar, 2010). It is also notable that the maximum of the dividend payout ratio is 1.23, which means that there is at least a rm that pays dividend exceeding its earning. Table 2a shows that the mean value of shareholdings of all largest shareholders is 39%. e declining number of companies with the State as the largest shareholder over years re ects the reforming e ort of the government to reduce their shares in privatized companies. However, across the panel sample, the State remains the largest shareholder, con rming the dominance of state ownership in the Vietnamese privatization. Table 2b shows dividend payout ratios by companies with speci c category of the largest shareholder. From the summary, it is worth noting that the company paying dividend in excess of its earning has the State as the largest shareholder. Moreover, companies with the State as the largest shareholder also have the highest average dividend payout ratio. We project that the companies with the State as the largest shareholder will pay higher dividend than other companies. e result will be con rmed in the regression model.

Multicollinearity test
In Table 3, there are only two comparatively high correlations. First, there is a positive relationship between leverage and size as discussed in the descriptive statistics. Secondly, the relationship between leverage and pro tability suggests that if companies largely depend on debt, there may be an improvement in the bo om line. However, overall, the

Testing Model (1) -the relationship between shareholdings of the largest shareholder and dividend payout ratio
As in Table 4, diagnostic tests indicate that FEM is an appropriate model. Time xed e ects are also not required in the model. e data has both a heteroskedasticity and autocorrelation problem. However, there is no cross-sectional dependence of residuals, cluster robust standard error will produce standard errors that are robust to heteroskedasticity and within panel serial correlation (Arellano, 1987). erefore, the e ect of   the identity of the largest shareholder on dividend payout ratio is analyzed by entity xed e ect with cluster robust standard errors. Shareholdings of the largest shareholder have been found to have a signi cant negative relationship with dividend payout ratio at a signi cance level of 1%. Hypothesis (H 1 ) is therefore supported.

Testing Model (2) -the effect of identity of the largest shareholder on dividend payout ratio
Similar to model (1), the e ect of the identity of the largest shareholder on dividend payout ratio is also analyzed by panel xed e ect with cluster robust standard errors (Table 5). It is important to note that the coe cients of dummy variables refer to di erence in average levels of dividend payment pursued by di erent types of the largest owners, holding other variables constant. Since the State and Foreign variables are signi cantly di erent from reference group with a 5% level of signi cance, it can be concluded that on average, companies with the State or foreign investors as the largest shareholders pay higher level of payout than base companies. Meanwhile, companies with the manager as the largest shareholder, as speculated, pay lower average level dividend than the reference group. However, the coe cient is not signi cant (only at 13.7% level of signi cance).

Discussions and Conclusions
Privatization is the important part of economic restructuring programs in most transition economies (Dharwadkar et al., 2000;Megginson & Ne er, 2001). Generally, privatization is associated with the alteration in ownership structure. While changes in ownership structure are believed to improve companies' pro tability and e ciency (Megginson & Ne er, 2001;Truong & Heaney, 2007;Le & Buck, 2009), in Vietnam and other transition markets, it may induce corporate governance issues (Truong & , 2007;La Porta et al., 1999;Le & Buck, 2009). is study examines the in uence of the largest shareholder on dividend policy, an important nancing decision that can a ect shareholders' wealth and protect them against misconducts by management.
Results support the rst hypothesis that the more shareholdings held by the largest shareholder, regardless of their identity, the lower the dividend payout ratio. In general, this can be explained by the fact that in the particular corporate governance context of Vietnam, where information asymmetry is one of the most striking problems, minority shareholders may not be promptly and su ciently provided with information on rm performance. As a consequence, they may be appropriated by the large shareholders since minority shareholders may have failed to ask for a higher dividend payment. is would reveal a common practice in emerging markets (Dharwadkar et al., 2000) that the higher the shareholding of the largest shareholders, the more opportunities and incentives for this type of shareholders to expropriate others, the lower is the company payout. Speci cally, the largest shareholders are more likely to grasp the bene ts supposed to be shared by other shareholders. Other shareholders are not protected if earnings are used in manners that are not bene cial to all shareholders. e second hypothesis incorporates the identities of the largest shareholders to determine whether they have any e ect on dividend payouts. e signi cant coe cients of variables indicate that companies with di erent identities of their largest shareholders do not pay similar levels of cash dividends, suggesting that some types of the largest shareholder with di erent characteristics in terms of motivations, risk a itudes and capabilities may have more or less incentive to force managers to pay dividends.
In particular, the positive and signi cant coe cient of the State variable reveals that privatized companies with the State as the largest shareholder will pay higher dividend than a base company (company with the largest shareholders as local investors). In the speci c context of Vietnam, one possible explanation is that the government has eliminated many tax barriers a er becoming a member of the World Trade Organization in 2005. As a result, that might lead to the decline of the tax revenues for the government.
us, the State may have incentive to demand for high payouts to make up for the decrease of tax revenues. Companies having the State as the largest shareholder may not have to rely much on internal resources for investment purposes because as the largest shareholder and the regulator, the State can either implicitly or explicitly back up companies to borrow at favorable loan terms (Ngoc & Mohnen, 2005;Truong & Heaney, 2007). As a consequence, they may not be constrained to pay out. In addition, the companies may also choose to pay high dividend to a ract other shareholders in order to accelerate equitization in Vietnam. In contrast, unlike the State shareholder, local investors cannot be assured of accessing other sources of nancing when their companies need. ey therefore wish to retain funds within the companies for investment purposes, especially in the other companies they own or invest. In other words, they may not be willing to pay out as high as companies with the State as the largest shareholders. 100 However, if the State as the largest shareholder abuses their power to in uence management, other shareholders, especially strategic investors who can improve operational e ciency may be reluctant to participate in the management of companies. Future research can further investigate this implication.
With regard to rms with the largest shareholders as foreign investors, it was found that, on average, the rms pay higher cash dividend payout than a base company. is can be explained by the fact that because of disadvantages in terms of geographic distance, and higher information uncertainty compared to the managers, foreign investors as rms' largest shareholders may require higher cash payouts to monitor the management (Cook & Jeon, 2006;Baba, 2009;Warrad et al., 2012;anatawee, 2013). is practice implies that foreign investors may embrace a risk-averse a itude towards the weak corporate governance in Vietnam.
Holding other variables constant, companies with the manager as the largest shareholder pay lower ratio than a base company, while the coe cient is not strongly signi cant. However, compared to those with the State or foreign investors as the largest shareholders, companies with the largest shareholders in management pursue lower level of cash dividend. As the manager, the largest shareholder may have more information about investment opportunities and business issues than the State and foreign investors. erefore, they may prefer to retain higher portion of earnings to exploit the resources at their convenience. Meanwhile, companies with the largest shareholder in management board do not pay lower level of payout than companies with local investor as the largest shareholder. One possible explanation is that the largest shareholders who hold management position may share some advantages with local investors in understanding the legal instability and investment opportunities of the local market. erefore, the largest shareholders are not signi cantly di erent from local largest shareholders. Most importantly, they are more constrained than local largest shareholders in accessing external debts. us, they may desire to pay lower dividend so that they can tunnel the resources into the other companies. e relationship between growth opportunities or pro tability and dividend payout ratio is signi cantly positive. is result is opposed to the Pecking order theory (Roze , 1982;Lloyd et al., 1985;Jensen et al., 1992) and signaling theory. Given weak corporate governance context with severe asymmetric information, shareholders may desire to be protected by high dividend payouts and consider dividend as an important indicator to evaluate companies for investment decisions. With a strong incentive to maintain reputation and a ract potential investors, companies remain paying high dividend and rely on external sources for growth opportunities (Harada & Nguyen, 2011). It is true that shareholders may gain bene t from high dividend. However, this may turn out to be a problem for the future prospect of companies. Speci cally, companies may have to depend on debt to nance investment opportunities. erefore, internal sources for future prospects may be employed to pay current dividend. ese practices are considered unfavorable to shareholder's wealth in the long run (Harada & Nguyen, 2011).
In general, the results do provide quantitative evidence on the impact of the largest shareholder on dividend policy. e negative relationship between shareholdings of the largest shareholders and dividend payout ratio may reveal the probability of expropriating behaviors of large shareholders. From the economic viewpoint, it will be detrimental to company if available dollar for investing activities is allocated to low return investment facilitated by the expropriating behaviors of the largest shareholder. e ndings of the study also reveal di erences in average levels of dividend payout observed from di erent types of the largest shareholders. Among them, companies with the State or foreign investor being the largest shareholder may pursue a higher level of payout than companies with local investors or managers as the largest shareholders. is implies that certain types of the largest shareholders may have more preference towards high earnings retention when employed in manners that do not bene t others. Moreover, while some largest shareholders may prefer higher dividend than others, it does not mean that shareholders may be bene ted. It may indicate that the largest shareholders serve their own purposes, as in the case of the State as the largest shareholder. Or, the largest shareholders may concern the uncertainty of the company and hence require higher dividend as in the case of the foreign investor as the largest shareholder. In addition, dividends may be employed to a ract investors at the risk of indebtedness as suggested by the relationship between the pro tability/ growth opportunities and dividend payout ratio.
is study has its own limitations. Firstly, the study has drawn conclusions based on the reliability of data resource and data suppliers. However, using secondary data has limitations under severe transparency problem in Vietnam. Secondly, some variables such as free cash ow, company's risk and age may be controlled for to provide with more profound understanding of dividend behavior of companies. e generalization of the results will be more compelling if the institutional elements of the researched countries are taken into consideration. irdly, the complex cross-shareholdings may con ne the ability of the study to understand precise in uences of each type of the largest shareholders. In addition, the study may not precisely reveal the exact expropriating behaviors of the largest shareholders. is question is open to future investigation. e paper may be considered as an initial a empt to open an interesting discussion on how di erent types of ownership in uence dividend paying behavior of companies.