Understanding a behaviour of dividend payout policy in vietnam from various financial models

Model 2 The Partial adjustment model under an adaptive expectations hypothesis Model 3 The partial adjustment model under a rational expectations hypothesis Model 4 The earnings trend mo

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

UNDERSTANDING A BEHAVIOUR OF DIVIDEND PAYOUT POLICY

IN VIETNAM FROM VARIOUS FINANCIAL MODELS

BY

DANG HUU LOC

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, MARCH 2015

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

UNDERSTANDING A BEHAVIOUR OF DIVIDEND PAYOUT POLICY

IN VIETNAM FROM VARIOUS FINANCIAL MODELS

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

Besides my supervisor, I would like to express a deep gratitude to all the lecturers at the Vietnam – Netherlands Program Especially, I am grateful to Assoc Prof Dr Nguyễn Trọng Hoài, Dr Phạm Khánh Nam and Dr Trương Đặng Thụy who facilitate me to finish my research

I would like to thank my friends for their helps during the courses as well as in the thesis writing process

Last but not the least; I am indebted to my parents: Đặng Thanh Liêm and Nguyễn Thị Ngoan, who always love unconditionally and support me spiritually every time I need

HCMC, March 2015 Đặng Hữu Lộc

Model 2 The Partial adjustment model under an adaptive expectations hypothesis Model 3 The partial adjustment model under a rational expectations hypothesis Model 4 The earnings trend model

Speed of adjustment

ABSTRACT

This study is conducted to examine and understand a behavior of dividend payout policy at Vietnam’s listed firms for the period from 2007 to 2013 In doing so, the four well known models are adopted, known as: (i) the partial adjustment model, (ii) the partial adjustment model under an adaptive expectations hypothesis (the Waud model), (iii) the partial adjustment model under a rational expectations hypothesis, and (iv) the earnings trend model Each of the models is briefly summarized below for the convenience of the readers

The first model considers the dividend behavior as a partially adjustable process to the target dividend The current profit will mainly determine the target dividend through constant desired payout ratio This model is to provide some evidences in terms of the reluctance in changing dividend and the speed of dividend adjustment The institutional ownership variable is also embedded into the model to investigate its impact on dividend policies However, it was argued that the target dividend should be explained mainly by the long-run expected earnings instead of current earnings (Harkins and Walsh, 1971) The adaptive expectations model is employed to determine the long-run expected earnings This expectation bases on the hypothesis that human can learn from the past experience and apply for life Accordingly, the model which

is integrated by both partial adjustment (Model 1) and adaptive expectation explains better dividend policies (Lee et al., 1987) This new model is also called as the Waud model and known

as Model 2 in this study Through this model, the responsibility of managers to change dividend

as well as the relationship between the institutional shareholders and dividends are tested

Robert Lucas and Thomas Sargen, criticized that the adaptive expectations hypothesis adopted in Model 2 is unrealistic because it purely bases on the past experiences and disregards available information to managements Therefore, they propagated a hypothesis which is known

as rational expectations This hypothesis states that managers are rational to optimize their forecasts which are incorporated current values and available information into the process of forming expectations The partial adjustment (Model 1) and rational expectations to consider two dividend characteristics as the Waud model, to be known as Model 3

As the last model attempted in this study, Model 4, the earning generating process is assumed to follow a random walk with trend This assumption is consistent with the view from Fama and Babiak Accordingly, any change in the dividend payout policy will include two parts:

(i) the first part is from full adjustment of the expected change of earnings; and (ii) the second part is from partial adjustment of the remainder of earnings This model is known as Model 4 in this study

The three hypotheses have been developed and tested in this empirical study, one of its

first kind in Vietnam: (i) Firms are more reluctant to decrease the dividend than to increase the

dividend; (ii) The speed of adjustment in dividends for Vietnam market is very flexible and higher than for developed markets such as Australia, Austria, Germany, Sweden, and United Kingdom; and (iii) the absence of institutional ownership reduces significantly dividends

Key findings in this empirical study reveal that three above hypotheses are plausible in the

case of Vietnam First, the empirical results reveal that managers are more afraid of cutting dividends than raising dividends Second, the listed firms in Vietnam are very flexible to change the dividend policies Third, the absence of institutional shareholders in the firms will significantly decrease the level of dividend payouts Fourth, the findings also confirm that the

adaptive expectations hypothesis is more appropriate than the rational expectations hypothesis in explaining the dividend behavior for the emerging markets, in particular for Vietnam, regardless

of the presence or absence of the institutional ownership in a firm

Contents

CHAPTER 1: 1

INTRODUCTION 1

1.1 Problem statement 1

1.2 Research objectives 2

1.3 The structure of study 3

CHAPTER 2 4

LITERATURE REVIEW 4

2.1 Dividend and characteristics 4

2.2 Dividend theory 7

2.3 Institutional ownership 9

2.4 The relationship between dividend policy and institutional ownership 9

2.4.1 Taxation 10

2.4.2 Agency theory 11

2.4.3 Signaling 12

2.4.4 Summary of empirical evidence 14

CHAPTER 3 15

RESEARCH METHODOLOGY 15

3.1 The partial adjustment model 15

3.2 The Partial adjustment model under an adaptive expectations hypothesis (The Waud model) 17

3.3 The partial adjustment model under a rational expectations hypothesis 19

3.4 The earnings trend model (ETM) 21

3.5 Hypotheses development 22

CHAPTER 4 24

SAMPLE, VARIABLES AND ECONOMETRIC ANALYSES 24

4.1 Sample selection 24

4.2 Variables 24

4.2.1 Dividend per share 25

4.2.2 Earnings 25

4.2.3 Institutional ownership 25

4.3 Econometric Analyses 25

CHAPTER 5: 30

DATA DESCRIPTIONS AND RESULTS 30

5.1 Data descriptions 30

5.2 Results 33

CHAPTER 6: 37

CONCLUSIONS AND IMPLICATIONS 37

6.1 A brief summary of the four models adopted 37

6.2 Conclusions 38

6.3 Implications 39

6.4 Limitations 39

APPENDIX 46

CHAPTER 1:

INTRODUCTION 1.1 Problem statement

The first stock market in Vietnam was established in Ho Chi Minh City in 2000 Initially, there were only two companies listed: Refrigeration Electrical Engineering Joint Stock Corporation (REE) and Saigon Cable and Telecommunication Material Joint Stock Company (SAM), with a small market capitalization of 270 billion VND It has been quiet in the Vietnam Stock market for a long time However, the stock market was really booming in 2006 on all three trading floors: Ho Chi Minh City Stock Exchange (HOSE), Ha Noi Stock Exchange (HNX) and the Over-The-Counter market (OTC) So far, there are 760 listed firms with the market capitalization of 52 billion USD In fact, the capitalization of the Vietnam stock market reaches 32% GDP of Vietnam in 2014 Listed firms play a more and more important role in the Vietnamese economy As such, it is important to understand key characteristics of listed companies in Vietnam, in which decisions to pay dividend have attracted attention from academics and practitioners

From an interdisciplinary perspective, dividend policy has long been captivating economists as the major puzzle of corporate finance, so a great deal of effort has been spent on unraveling this subject A well-known theorem introduced by Modigliani and Miller (1961) argued that dividend policy is irrelevant to value of a firm This view is considered under the assumptions of a perfect market However, practice always deviates from the theory due to the existence of market imperfections such as taxes, transaction costs, agency problem, and information asymmetry To assess the influence of dividends in Germany, Amihud & Murgia (1997) considered a sample of 200 German firms to conclude that dividend change play a significant role in future prospects of firms The reason is that a change in payout may provide information about management’s confidence in the future and so affect the stock price (Breadley, Myers & Allen, 2011, p.397) In the case of the Austrian firms, dividend also has a negative correlation with investment (Gugler, 2003) Although there are still many controversies surrounding this issue, no one can deny that dividend policy has been considered as one of the most crucial decisions in corporate financial management

Dividend payout policy is even more important in the emerging market and Vietnam market is one of them Due to asymmetric information, changes in dividend were considered as a signal about the company’s prospects in the future (Short, 2002) Especially, it influences substantially shareholders who often plan and expect stable future cash flows for retirees, pension funds and insurance companies In fact, on January 20th 2014, Decision No4/2014 was promulgated to approve the establishment of Voluntary Pension Fund This decision will pave the way for vibrant fund market in near future In relation to asset valuation, multiple models forecasting stock price in the long run is based on dividend The models present a necessity to understand why companies practice and change dividend policies Moreover, the highly profitable companies that do not pay dividends usually get the backlash from shareholders Therefore, it is necessary to understand dividend behavior for the Vietnam market in this study

The development of stock market has also associated with the development of the institutional investors Institutional investors play an important role not only in corporate control but also in creating liquidity for the market Smith (1996) argued that the institutional shareholders are the resource of monitoring management and lead to changes of governance structures and performance were targeted At the end of 2013, the institutions invested approximately 5 billion USD into the Vietnam stock market Moreover, 86% listed firms in 2013 was in existences of institutions owning 5% or more of equity in a company Thus, it is stated that institutional investors affect significantly the policies of firms, including dividend policies, but there are not any researches to address how institutional shareholders influence on this policy

in the case of Vietnam For this reason, this study discusses the relationship between dividends and institutional ownership from the period of 2007-2013

This study is conducted to meet the following two research objectives

 First, this study aims at examining and quantifying two major characteristics of

dividends in Vietnam using the Partial Adjustment model

 Second, the hypothesis that whether the absence of institutional ownership reduces

the dividend is examined Four models are applied to test this hypothesis All data are collected from listed firms of the Vietnam stock exchange for the period of 2007-

2013

1.3 The structure of study

There are six chapters in this study First, the introduction chapter presents the problem statement as well as the research objectives The second chapter presents the overall literature relevant to the issues mentioned in the introduction Chapter 3 demonstrates clearly why four models are employed to test the hypotheses in this study Measurements of all variables and econometric analyses are proposed in the fourth chapter Based on the discussion in the previous chapters, Chapter 5 presents the findings drawn from the regression results of the four models Finally, conclusions, implications and limitations are presented in Chapter 6

CHAPTER 2 LITERATURE REVIEW

This chapter will introduce a concept and common characteristics of dividends Then, views on the dividend policies can be classified into two distinct theories: (i) dividend irrelevance theory and (ii) dividend relevance theory Afterwards, the definition of institutional ownership as well as its effects on dividend policies is examined This relationship will be clarified through: taxation, agency theory and signaling At the end, overall impacts of institutional ownership on dividends will be summarized to provide final conclusions

2.1 Dividend and characteristics

According to Frankfurter et al (2003), dividend is a portion of earnings which a firm distributes to shareholders in proportion to their ownership Companies usually pay dividend annually or quarterly in the form of cash This payout results in a reduction in cash items and retained earnings on balance sheet of a firm Therefore, dividend policy refers to the financial policy, in which the management decides size and pattern of dividend to distribute among the shareholders (Lease et al., 2000)

By surveying the opinions of managers, Lintner (1956), Baker and Powell (1999), Brav et

al (2005) provided evidences to determine the characteristics of dividend policy in which the theoretical and empirical researches are not sufficient to explain Although there is the amount of debate around dividend problem, the financial economists almost agree with the most following common properties of dividends:

Figure 2.1: The result of Brav et al.’s survey in 2005

1) First, firms are more reluctant to decrease the dividend than to increase the dividend A

reason under the information asymmetry hypothesis states that outside investors cannot access to more information of the company as managers, so dividends are considered as a good indicators to reveal the firm’s future prospects (Bhattacharya, 1979; John and Williams, 1985) Accurately, a higher dividend announcement predicts higher current or future earnings Many researches show that the return which exceeds expectations usually follows a positive change of dividends (Pettit, 1972; Asquith and Mullins, 1983) Hence, a decision of cutting dividends may be a bad news for investors, so it can cause falls in stock price For this reason, decreasing dividends is more costly than increasing dividends 2) Second, managers restrict to make the decision of changing in dividend that can be

possibly reversed within a short time, so most of companies adjust partially to the target dividends within each year Changing gradually dividends is considered as a buffer against the uncertainty of the future earnings which affect negatively to dividend policy In other words, firms usually smooth their dividend However, the extent of dividend smoothing is different between developing countries and developed countries A popular approach to

measure the dividend smoothing is an application of the partial adjustment model to estimate the speed of adjustment (SOA) A research of Javakhadze et al (2014) in the period of 1999 to 2011 shows that SOA in developed countries is less than in developing countries In detail, the averaged SOAs of developed countries such as Australia, Austria, Germany, Sweden, United Kingdom equal 0.4557, 0.3578, 0.3306, 0.4962, 0.2303, respectively Meanwhile, its values in developing countries including China, Nigeria, Pakistan, and South Africa, in turn, are 0.6030, 0.5332, 0.6273, and 0.6740 According to Glen et al (1995), the volatility of dividends is less concerned in emerging markets than developed markets In other words, the signaling in dividend is less important in developing countries Hence, the SOAs in emerging markets can be higher than in developed markets

An empirical research of Adaoglu (2000) from the Istanbul Stock Exchange argued that stable dividend policies are not adopted in listed firms The reasons’ Chemmanur et al (2010) suggest that in developing countries, much of listed firms are still small, have greater growth opportunities and are financially constrained; therefore, they have a higher cost to pay dividend than to retain earnings Simultaneously, the dividend policies are also influenced by volatilities in Vietnam economy in this period, including both of external effects and internal effects First, financial crisis in 2007-08 and European debt crisis in 2010-12 affect seriously the Vietnam export and reduce the economic growth Second, high inflations come in 2007, 2008, 2010 and 2011, combining the tight monetary policies

in 2011 and 2012 Hence, many companies concentrated to overcome the difficulties rather than to maintain a stable dividend policies The role of dividends which acts as signaling device is omitted Managements adjust simply dividends in proportion to profits Generally, the firms are predicted to adjust quickly their current dividend to the target dividends for the case of Vietnam in period from 2007 to 2013 In this study, I will also reinvestigate that whether SOA in Vietnam is more flexible and higher than other developed countries as mentioned above

Developed countries Number of firms Mean Median

on the topic can be classified into two distinct theories: (i) dividend irrelevance theory and (ii) dividend relevance theory

In 1961, the publication (hereafter MM theory) of two noble laureates, Miller and Modigiliani, laid the foundation of the irrelevance theory Under the assumption of perfect capital market, they argued that the market value is only maximized by an optimal investment policy, not by dividend policy Specifically, it is assumed that there are two identical firms except for financial structure, the first financed totally by equity, the second financed by the mixture of equity and debt MM theory inferred that the values of the two firms are equal in spite

of the difference in financial policies From the investor’s perspective, the way which firms distribute profits does not affect to the earnings maximization of shareholders For example,

when a firm pays out all profits to shareholders as dividends, the stock price will be decreased equally by the amount of a dividend per share on the ex-dividend date As a supplement, Brennan (1971) suggested the rejection of MM theory is synonymous with the rejection of symmetric market rationality as well as independence of irrelevance information To support empirical evidence, Black and Scholes (1974) tested the effect of dividend yield on stock prices, based on the data of listed firms on the New York Stock Exchange (NYSE) They concluded that changes in firm value are independent of changes in dividend payout “The intuitive argument in favor of the hypothesis is that if management could increase the market value of the firm's stock

by changing dividend policy, why has it not done so already” (Glen et al., 1995, p.3) Subsequently, there are more empirical evidences in support of MM theory (Miller & Scholes, 1978; Hess, 1981; Miller, 1986; Bernstein, 1996)

On the other hand, firms which do not pay out dividends receive negative reactions from shareholders In fact, the dividends are not always reflected completely into the stock price When the market is depressed or appreciated, the stock price will be undervalued or overvalued, respectively Simultaneously, long-run investors usually require dividends as regular expenses Therefore, the regardless of taxes and the unrealistic assumption of efficient capital markets are controversial in an aspect of applying the rules in the facts By embodying in taxes, signaling information, agency cost and human behavior, the economists have proved the opposite Some empirical studies reveal that an increasing dividend payment will result in an increase of firm value (Gordon& Shapiro, 1956; Lintner, 1962; Walter, 1963) Theoretical background of this argument is the intensity of preference of dividend today over possible shares or dividend tomorrow In other words, most of people prefer “a bird in the hand” to “two birds in the bush” Moreover, a higher dividend conveys a credible signal of good future prospects, which appeal to investors willing to pay higher stock price

Another issue is very noticeable that the existence of tax and differences in tax rate may distort the dividend decisions The tax effect hypothesis stated that if dividend tax is higher than capital gains tax, management will respond by decreasing the dividend payout to ensure that a shareholder wealth is maximized In this case, decreasing dividend payout may increase the value of a firm Brennan (1970) constructed a model to illustrate this theory This model was subsequently extended by Litzenberger and Ramaswamy in 1979 Finally, under agency theory, paying higher dividend leads to lower agency cost, so company performances will be improved

In summary, although the assumptions in MM theory are difficult to be met in the real world, it is still referred and studied in many universities It is crucial since this theorem mentioned determinants of optimal capital structure and their effects on this structure Many empirical studies have been implemented in this topic However, in an emerging market like Vietnam, lacking empirical studies of this subject require researchers to pay more attention

2.3 Institutional ownership

In 1976, Jensen and Meckling defined the ownership structure as the capital allocation of a company among debt, inside equity held by the management and outside equity There are also many different definitions formulated in later studies The most widely used definition in relation to the institutional ownership is that “the ownership structure is defined by the distribution of equity with regard to votes and capital but also by the identity of the equity owners” (Sivathaasan, 2013, p.1) To facilitate researches, many economists and institutions proposed different classifications of ownership structure

According to Morten Balling (1997), ownership structure is classified into seven sectors, including: (1) central bank and other supervisory authority, (2) the government; (3) banks and other financial institutions; (4) institutions such as pension fund; (5) individuals; (6) non-financial companies; and (7) foreign holdings However, in this study, we only concentrate to the institutional ownership which reflects the organizations, companies, funds holding 5% or more

of equity in a company The benchmark 5% of equity owner was used in accordance with the regulations of Vietnam government on information disclosure of listed companies It is argued that due to the large percentage they hold in ownership structure, institutions are more capable of affecting to the company’s decisions, including dividend policy, rather than individual investors The procedure formulating the relationship between dividend policy and institutional ownership

is described as follows

2.4 The relationship between dividend policy and institutional ownership

Dividend policy and ownership structure of a firm have been considered as one of the most crucial and debatable relationship in finance for a number of years The relationship between them was recognized from 1964 by Williamson Afterward, many researchers (Leland and Pyle,

1977; Jensen, 1986) have been attempted to establish literatures on this relationship Short et al (2002) argued that institutional ownership creates the incentive to increase the dividends, and it

is the first time that this relationship is investigated in Vietnam In particular, dividend policy is influenced by institutional ownership through three mechanisms, including: taxation, agency theory and signaling These mechanisms will clarify why the institutional ownership is the cause

to maintain the high dividend Each of these mechanisms is discussed in turn below

2.4.1 Taxation

Tax system can distort behaviors of investors when the tax rates on dividend are different from the tax rates on capital gain In the United State of America, the government has regulated that the tax imposed on dividend is separate with income tax of corporations on their profits This means that in dividend paying companies, the shareholders was taxed twice, the first in term

of corporation tax on firm’s profits and the second in term of income tax on dividend receipt As

a result, taxpayers are favorable of the incomes retaining policies rather than the high dividend paying policies and tax-exempt shareholders is neutral in respect of dividend policies This leads the questions “why do firms pay dividend” in Black’s study (1976) In contrast, the tax system in Vietnam is very different from the U.S Circular No.128/2003/TT-BTC regulated that the tax rates of capital gain and dividend are 28% and 0%, respectively Thereafter, Vietnam’s corporate income tax law introduced on June 2008 set up the tax rate 25% on capital gains and the tax rate 0% on dividend receipts Therefore, if institutions are large shareholders in companies, to maximize their wealth, they will have a propensity of putting pressure on managements to pay higher dividends

Another issue to be addressed is “the need of institutional shareholders for funds on an ongoing basis” (Short et al., 2002, p.108) More precisely, institutions rely not only on capitals gains but also on dividend to fund their activities such as paying insurance, funding pension Furthermore, in developed countries such as the US and the UK, pension funds which manage trillions of dollars to invest in many stocks, is always the most influential on the market These institutions usually pay pensions from investment income rather than from new contributions Institutions usually are long-run investors, so they need dividends to pay regular expenses Thus, they usually require a stability of dividends to match their liabilities as planned These

requirements may force companies to pay higher dividends than the managers intended, especially in circumstances of recession and low income In spite of lack of pension fund in Vietnam now, this aspect need pay more attention Since on January 20th 2014, Decision No4/2014 was promulgated to approve the establishment of Voluntary Pension Fund, the discussion above will be the foundation of researches in the future

In short, on the ground of this consideration, it is feasible to suggest that there are a positive relationship between institutional shareholdings and dividend payout In other words, firm with an absence of institutional ownership is lower dividend than others In future, the need

of maintaining the cash flow of new pension funds in Vietnam will consolidate this hypothesis stronger unless there are changes in financial policies

2.4.2 Agency theory

In corporate finance, the agency problem arises when there are interest conflicts between managers of a firm and shareholders of the firm (Jensen and Meckling, 1976) Managers are employed to administer firms in accordance with maximizing shareholder wealth Nevertheless, managers sometimes act in their own best interests that differ from the best interests of the shareholders This leads a necessary monitoring to alleviate losses that managers can abuse from their positions

Shleifer and Vishny (1986) argued that large shareholders as institutions have more incentive than small ownership to monitor the management In addition, they also have more voting power and ability to implement the monitoring effectively However, the free rider problem may eliminate the incentive of institutions to offer a direct supervision because they must bear expenses alone Accordingly to Maug (1998), this problem will be not serious in more liquid market, since investors can be compensated for the monitoring cost as a consequence of informed trading Nevertheless, an emerging market like Vietnam is less liquid, and an event that institutional shareholders sale a large portion of their equity depresses such stock prices Hence, Easterbrook (1984) implied that a method to solve this problem is an increase in dividend payout

to suppress excessive retention of cash flow Assuming that the firms continue their projects as planned, they must borrow from the capital market to meet the demand To obtain loan approval, they are subject to the monitoring of external financial institutions, for this reason, agency costs will be reduced

Rozeff (1982) constructed a model in which two market imperfections, agency costs and transaction costs, were combined to illustrate the dividend payout ratios, and known as the cost minimization model The implemented empirical study based on the data of 1000 firms in 64 industries over the period 1974-1980; and agency costs and transaction costs were represented by five variables in which two proxies of agency costs are significantly negative with dividend ratios In other words, this research supported strongly the hypothesis that increased dividends

lower agency costs Motivating for this view, Lloyd et al., (1985) replicated and expanded the

Rozeff’s cost minimization model by adding firm size as the important variable The result strengthened the Rozeff’s hypothesis in introducing agency costs as a major determinant of dividend payout ratios Subsequently, this result also was reaffirmed in many studies such as Schooley and Barney (1994), Moh’d et al (1995)

Moreover, excessive retention of cash flow may be diverted by managers for wasteful expenditures or unprofitable project More precisely, a scarcity of resources promotes managers

to think deeply in spending money Therefore, La Porta et al (2000) suggested that increasing dividend is a good device to protect investors in unprotected legal environments, avoiding the loss of assets to shareholders

In summary, instead of self-monitoring, outside institutional shareholders usually have a propensity to force managers paying higher dividend, and reduced agency cost is the major incentive By the way, it is also argued that there is a positive correlation between the higher dividend and presence of institutional under the view of the agency theory

2.4.3 Signaling

Signaling theory was firstly addressed in Lintner’s study (1956) and was supported later

by Miller and Modigiliani (1961) They argued that managers usually adopt a stabilization of dividend policies with long-run payout ratios, so the changes in dividend are perceived as the changes in the views of managers about future earnings of firms Precisely, Miller and Rock (1985) concluded that a higher dividend is a good indicator in relation to the company’s better prospects Zeckhauser and Pound (1990) stated that both dividends and institutional shareholders are good proxies for signaling devices Therefore, the effect of institutional ownership can crowd out the effect of dividends as a credible signal to investors Under this hypothesis, the

relationship between dividend and institutional ownership is negative However, Short et al (2002) stated that it is not clearly to explain how stock purchase plans of institutions can convey

a signal of the firm’s future earnings to the market They proposed two possibilities First, some investors expect that the presence of institutions can reduce the agency cost due to their monitoring activities However, as mentioned above, since the free rider problem arises, institutions will adopt higher dividend policy instead of providing freely direct monitoring Second, an argument is that institutional investors are more professional than retail investors in the ability to evaluate the company outlook Thus the firm which is invested by large institutions

is guaranteed implicitly for a good profitability in the future Although this hypothesis holds some attractions, they suggested that there is not much strong evidence to support this scenario which portfolio of institutions is more profitable than individuals

Fama and French (2001) pointed out that given firm characteristics, the proportion of listed firms paying cash dividend is in a fall propensity from 1978 To explain the “disappearing dividend phenomena”, Amihud and Li (2006) stated that the decline in the information content

of dividends causes the decreasing of paying dividends which have been used as a credible signal

to market The reason is the existence of institutional investors which are more informed than retail investors The informational asymmetry problems between institutional investors and retail investors are derived from two reasons First, given amount of information, institutions gain more profit, so they have more incentives to gather information Second, their proficiency can help them reduce marginal cost in collecting and processing information It is obvious that they also have more financial powers which are manipulated to acquire information “Since the more informed institutional investors use their information in trading stocks, by the time a dividend change is announced, part of the information that it conveys about the firm’s value is already incorporated in the stock price” ( Amihud and Li, 2006, p.646) In other words, the increasing of holdings by institutions is the reason for the decline in the information content of dividend, leading lower dividend payout

However, all studies were implemented in developed countries, these references to appropriate with an emerging market is still an open question In Vietnam, the magnitude of the market is still small, less liquidity and not transparent, so large institutions can abuse their market power to manipulate stocks, not basing on the evaluation of the future prospect of the firms Accurately, they can convey a wrong signal to the market Therefore, the opinion that the

role of institutional shareholders is likely a good signaling is more controversial in the case of Vietnam

2.4.4 Summary of empirical evidence

Overall, the effect of institutions in dividend decisions is formulated through three ways The sum of tax, agency cost and signaling considerations determines the property of this relationship (negative, positive or not significant) Under the view of tax effect, to maximize the wealth and to match the liabilities are the clear incentives of the institutions to demand a high dividend Under the view of agency theory, paying out dividend is a good approach not only to reduce agency cost but also to avoid the free rider problem Moreover, a high dividend policy restricts the excess of free cash flow to be used for wasteful expenditure In contrast, the relationship between institutional ownership and dividend under signaling hypothesis is ambiguous, and there are not strong evidences in emerging markets like Vietnam Therefore, a positive association between the high dividend and institutional ownership is predicted, and this

is the first times this hypothesis is investigated in the case of Vietnam To support to this hypothesis, in the case of Canada, Eckbo and Verma (1994) proved that dividend is proportional

to the voting power of institutional ownership and inversely proportional to the voting power of managerial ownership Short et al (2002) revealed the strong evidence that institutional ownership affect significantly to payout ratios from the sample of the U.K listed firms

CHAPTER 3 RESEARCH METHODOLOGY

Every market has different characteristics, so many models were employed to test the hypotheses more accuracy There are four models adopted in my study, including: (i) the Partial Adjustment model, (ii) the Partial adjustment model under an adaptive expectations hypothesis(the Waud model), (iii) The partial adjustment model under a rational expectations hypothesis, and (iv) the Earnings Trend Model To test the hypothesis, a dummy variable which describes the absence of the institutional ownership is added in four models If the absence of an institution owning 5% or more of equity appears, it will be equal 1 and 0 otherwise Subsequently, the hypothesis development and expected sign table will be represented at the end

of this chapter

3.1 The partial adjustment model

Lintner (1956) proposed firstly a dividend model, basing on the positive dividend-earnings relationship From available information of 600 U.S listed, he selected carefully 28 companies to interview for detailed investigation Although dividend policy is significantly different across companies, some common patterns were withdrawn from his result These are clarified as follows:

- Managers almost believe that the target payout ratio should be remained with a little deviation in long-run

- In making dividend decision, managers pay attention to the change in the existing payout, not to the amount of dividend level

- Earnings are the major factor to dominate the dividend policy better than other factors

- Managers restrict to make the decision of changing in dividend that can be possibly reversed within a short time Therefore, most of companies adjust partially to the target dividends within each year, and the speed of adjustment should be done a certain way Based on these characteristics, Lintner (1956) established a dividend behavior model, named partial adjustment model For a given year t with firm i, there is a desired payout ratio r between the target dividends, D∗

ti, and the profit Eti:

D∗ti = rEti (1) Firms adjust partially annual dividends to the target dividends, so a new equation is as below:

Dti- D(t−1)i= a + b (D∗ti-D(t−1)i) + uti (2) where a = a constant, and b is a coefficient representing the speed of adjustment of the actual dividends to the target dividends in a year The reduction of net profits leads to decrease the target dividends and put pressure on cutting current dividend However, the current dividends can be not reduced if the constant in the function (2) is positive If the constant is negative, the current dividends can be cut even earnings increasing Therefore, the intercept reveals the reluctant of the firm to decrease or increase the dividend According to Lintner, the firms adjust partially dividends year by year, so the speed is subject to 0<b≤ 1 Substituting (1) into (2), the reduced form becomes:

Dti- D(t−1)i= a + brEti - bD(t−1)i + uti (3) Brittain (1966), Fama and Babiak (1968) indicated that equation (3) explains well in predictions of dividend with small mean squared error Furthermore, all variables in this equation are significant, and the R2 does not increase significantly when financial variables are embedded Similarly, other kinds of economic behaviors, especially supply response and inventory investment, are characterized by the partial adjustment process, so this model was modified to employ in numerous empirical studies (Nerlove, 1958; Hill, 1971; Burrows and Godfrey, 1973)

As discussed in the literature review, institutional ownership have significant effects on making dividend decisions, and this is reflected in my model by transforming the desired payout ratio ri

Substituting (4) into (2), the model becomes:

Dti- D(t−1)i = a + brEti + briEtiIns - bD(t−1)i+ uti (5)

According to the theory, the coefficient (b𝑟𝑖) is predicted to be negative If constant term is positive, it implies that managers are more reluctant to decrease than to increase dividend

3.2 The Partial adjustment model under an adaptive expectations hypothesis (The Waud model)

Although the model proposed by Lintner is very useful to determine the dividend of firms, the formation of the target dividends is unlikely to be sufficiently described Harkins and Walsh (1971) suggested that the long-run expected earnings are more important than current earnings in dividend decision by most managers By the way, the expected dividends D∗ti are related to the long-run expected earnings Eti∗, instead of current earnings Eti, by a ratio r

Through the long-run expected earnings, the target dividends are determined not only by current dividend but also by other factors such as past experiments or relevant information in outlining expectations In other words, managers can rely on the reaction of shareholders in the past or permanent prospect of the firm to decide further increase or decrease A method to model the expectation is popularized by Cagan (1956) and Friedman (1957), in which “economic agents will adapt their expectations in the light of past experience and that in particular they will learn from their mistakes” (Shaw, 1984, p.25) According to this hypothesis, the equation of expectations is formed:

Eti∗- E(t−1)i∗ = c (Eti–E(t−1)i∗ ) (7) where c is the adjustment speed of expectations, such that 0≤ c ≤1

Equation (7) is called as adaptive expectation or error learning hypothesis More detailed, modified expectations in each period are proportional to a fraction c of the difference between the current earnings and previous expected earnings If c=1, Eti∗ = Eti, implying that current earnings are perceived totally to transfer to long-run expected earnings On the contrary, if c=0,

Eti∗=E(t−1)i∗ , managers believe that long-run expectations are not influenced by the current earnings It provided a simple way to model an expectation basing on the belief that human usually is dominated by habit persistence and study from their experiences By doing repeated algebra, the equation (7) is transformed as:

Eti∗ = cEti + c(1-c)E(t−1)i + c(1 − 𝑐)2E(t−2)i + c(1 − 𝑐)3E(t−3)i +… (8) The above equation reinforces the adaptive expectation in meaning that more distant events are less influential than more recent events It also represents that the expectations are accumulated data from previous years In 1968, Roger N Waud combined two models, the

adaptive expectation model and the partial adjustment model, to generate the Waud model Similarly Lintner model, company adjusts partially actual dividend to the expected dividend, thus:

𝑏 (D(t−1)i- (1-b)D(t−2)i - a - u(t−1)i) (11)

Substituting (6) and (11) into (7), then we receive:

D∗ti= crEti + (1−𝑐)

𝑏 (D(t−1)i- (1-b) D(t−2)i - a - u(t−1)i ) (12)

Combining (12) and (9), we have a reduced model:

Dti- D(t−1)i = ac + bcrEti + (1-b-c)D(t−1)i + (1-b)(1-c) D(t−2)i + 𝑣𝑡𝑖 (13)

where 𝑣𝑡𝑖 = uti- (1-c) u(t−1)i or 𝑣𝑡𝑖 = (1-c) 𝑣𝑡𝑖+ uti + (1 − c)2u(t−2)i

If we assume that uti~ N(0, 𝜎𝑢2) and is independent, the statistic property of 𝑣𝑡𝑖is serially correlated It is noticeable due to trigger the inefficient estimation in OLS The solution for this problem will be exhibited in the later chapter Lee et al.’s (1987) empirical research indicated that the model which is integrated by partial adjustment and adaptive expectations explains better dividend policies Due to significant effects of institutional ownership on the desired ratio r, the equation (6) becomes:

D∗

By the reason of ownership structure, the equation (13) now becomes:

Dti- D(t−1)i = ad + cdrEti + cdriEtiIns + (1-d-c) D(t−1)i - (1-d)(1-c)D(t−2)i - vti(15)

As previously presented, the coefficient (cd𝑟𝑖) is predicted to be negative, whereas the intercept is predicted to be positive

3.3 The partial adjustment model under a rational expectations hypothesis

Although the adaptive expectations model has employed popularly in empirical studies, some leading economists, especially Robert Lucas and Thomas Sargent, criticized that it is unrealistic because it purely bases on the past experiences and disregards available information

to managements Therefore, they propagated a hypothesis which is in response to perceived mistakes in adaptive expectations is known as rational expectations In other words, this hypothesis stated that managers are rational to optimize their forecasts which are incorporated current values and available information in the process of forming expectations Suppose that the long-run expected earnings is given by:

Eti∗ = α { ∑∞ βj

where:

α: the rate of return of firm value on the market

β: the discount rate, 0< β <1

e𝑡: the expectation operator in the sense of the existing information set I(t) available to forecast at period t

A equivalent expression is et(Et) = e(Et| I(t))= Et and et(Et+j) = e(Et+j| I(t)) At time t,∑∞j=0βjet(Et+j)} stands for the expected market value of firm which is calculated by all discounted future earnings

This identification is recognized broadly in many studies about rational expectations, permanent income and efficient-market hypothesis, for example Hall (1978), Bernanke (1984), and Nakamura (1985) According to Ball and Watts (1972), Gonedes (1973), Watts and Leftwich (1977), the earnings generating process is a random walk with drift: