Determinats of dividend policy case study , listed companies on hochiminh stock exchange

---o0o---KHƯU CẨM TÚ DETERMINANTS OF DIVIDEND POLICY CASE STUDY: LISTED COMPANIES ON HOCHIMINH STOCK EXCHANGE MASTER THESIS Ho Chi Minh City - 2011... ---o0o---KHƯU CẨM TÚ DETERMINANTS O

-o0o -KHƯU CẨM TÚ

DETERMINANTS OF DIVIDEND POLICY CASE STUDY: LISTED COMPANIES ON HOCHIMINH STOCK EXCHANGE

MASTER THESIS

Ho Chi Minh City - 2011

-o0o -KHƯU CẨM TÚ

DETERMINANTS OF DIVIDEND POLICY

CASE STUDY: LISTED COMPANIES ON HOCHIMINH STOCK EXCHANGE

MAJOR: BANKING AND FINANCE

MAJOR CODE : 60.31.12

MASTER THESIS

INSTRUCTOR : DOCTOR TRẦN PHƯƠNG NGỌC THẢO

Ho Chi Minh City - 2011

First and foremost I would like to thank my supervisor, Doctor Tran Phuong Ngoc Thao for her guidance and support.

In addition I would also like to express my gratitude to Doctor Vo Xuan Vinh for his valuable feedback on the problems of the study.

Special thanks also go to all of my lecturers at Banking and Finance Faculty, University of Economics Hochiminh City for their teaching and guidance during

my Master of Banking and Finance course.

I also wish to thank my friends in Dong A Securities, Au Viet Securities, Thang Long Securities and China Trust Bank for their great support.

Finally, my greatest thanks would go to my family including my parents, my husband and my baby who are the greatest encouragement for me to overcome all difficulties in my life.

This thesis aims to add empirical evidence to the corporate finance literature bylooking at firms’ dividend policy in the context of Hochiminh Stock Exchange(HOSE) of Vietnam as an emerging market.

The thesis consists of five chapters After an introductory chapter, the studyreviews the existing literature on the dividend policy controversy with an emphasis onthe recent empirical works The following third chapter provides the researchmethodology for the study The study applies panel data procedures, random effectslogistic regression, random effects generalized least square (GLS) regression andmultinomial logistic regression to estimate and test the determinants of dividendpolicy of l09 firms listed on HOSE While the main findings of this study which areconsistent with theory are reported in the fourth chapter, there are new major insightsthat represent the special case of emerging markets in general and Vietnam inparticular These main insights, as well as the main conclusions of the study, aresummarized in Chapter 5, including some limitations and recommendations for futureresearch

Our findings is that many of the factors that are found to be significant in thedetermination of dividend policy of the firms listed on HOSE are the same as thosefound in developed capital markets such as profitability, firm size and tangibility.Besides, we find that the listed firms on HOSE fail to pay dividend only because ofmaking loss or too small profit in business operation, not because of havinginvestment opportunities in the future We also find some evidence to support the lifecycle theory Our findings suggest that firms with greater investment, as measured bygrowth rate of sales are more likely to prefer paying stock dividends to cashdividends

Keywords: Dividend policy, Vietnam, HOSE.

Abstract ii

Contents iii

List of Figures vi

List of Tables vii

Abbreviations ix

CHAPTER 1: INTRODUCTION 1.1 Introduction 1

1.2 The usefulness of the research 2

1.3 Research objectives 2

1.4 Research methodology 3

1.5 The structure of the research 4

CHAPTER 2: LITERATURE REVIEW 2.1 Introduction 6

2.2 Dividend theories and concerned empirical studies 6

2.3 Empirical studies on the determinants of dividend policy 8

2.3.1 Profitability 8

2.3.2 Firm size 9

2.3.3 Financial leverage 10

2.3.4 Growth opportunities 10

2.3.5 Asset tangibility 11

CHAPTER 3: RESEARCH METHODOLOGY

3.1 Introduction 13

3.2 Data specifications 13

3.2.1 Research sample description 13

3.2.2 Explanatory variables 15

3.2.2.1 Variables those vary both across firms and time 15

3.2.2.1.1 Return on assets (ROA) 15

3.2.2.1.2 Firm size 16

3.2.2.1.3 Financial leverage 16

3.2.2.1.4 Growth opportunities 17

3.2.2.1.5 Asset tangibility 17

3.2.2.2 Variables that vary only across firms 18 3.2.2.2.1 Business risk (Beta) 18 3.2.2.2.2 Industry dummies 18

3.2.3 Dependent variables – Dividend policy 18

3.3 Empirical model specifications 19

3.3.1 Model 1 19

3.3.2 Model 2 20

3.3.3 Model 3 21

4.2 Descriptive statistics 23

4.3 Correlation matrix of explanatory variables 26

4.4 Results of Model 1 27

4.5 Results of Model 2 30

4.6 Results of Model 3 34

4.7 Robustness tests 38

CHAPTER 5: CONCLUSION 5.1 Introduction 41

5.2 Conclusion 41

5.3 The implications of the research 42

5.4 Limitations and recommendations for future research 44

References 46

Appendix A – Regression results of 3 models 51

Appendix B – List of 109 non-financial firms, classified industries and beta 58

Appendix C – The Dividend Time Lines of 109 firms on HOSE 2007-2009 62

Appendix D – Explanatory Variables calculated from Financial Statements 74

Figure 1.1 Research methodology 4

Figure 1.2 The structure of the research 5

Figure 3.1 Outline of chapter 3 13

Figure 3.2 Summary of research data collection 15

Table 4.1 Summary of the industry structure 24

Table 4.2 Summary of the dividend policy of 109 firms listed on HOSE 24

Table 4.3 Summary of the dividend forms of payers listed on HOSE 25

Table 4.4 Descriptive statistics of the explanatory variables 26

Table 4.5 Variance inflation factor (VIF) for the explanatory variables 27

Table 4.6 Correlation coefficients among the explanatory variables 27

Table 4.7 The reported results of Model 1 28

Table 4.8 Descriptive statistic of ROA of dividend non-payers and payers 29

Table 4.9 The reported results of Model 2 30

Table 4.10 Summary of current debts/total debts ratio 33

Table 4.11 Summary of cash dividend per share of firms in the different classified industries 34

Table 4.12 The reported results of Model 3 35

Table 4.13 Summary of revenues growth of cash dividend payers and stock dividend payers 36

Table 4.14 Summary of dividend forms of firms in Basis Material sector 37

Table 4.15 Summary of dividend forms of firms in Consumer Staples sector 38 Table 4.16 Summary of dividend forms of firms in Consumer Discretionary sector

38

HOSE Hochiminh Stock Exchange

CHAPTER 1 : INTRODUCTION

Problems concerning dividend policy have puzzled both academics and corporatemanagers for several decades Dividend may enhance the market value of thecompany but on the other hand it may mean less of available internal funds and moredependence on external sources for expansion purposes Furthermore, whiledetermining dividend payment, a prudent management strikes a balance betweenshareholder’s expectation and firm’s long term interest A deeper understanding as tothe motivation behind dividends will provide opportunity to better value stock, asmost current stock valuation models include dividends as a key element Besides, itmight held the policy makers to find the loop holes and to craft a better policy forfuture

Dividend policy decision is affected by many factors These factors may varysubstantially from country to country Most of the studies on determinants ofcorporate dividend policy have been conducted using data from developed countries

To date, to the author’s knowledge, no quantitative study has been conducted usingdata from Hochiminh Stock Exchange (HOSE) of Vietnam and corollary ofinsufficient information for academics, policy makers and investors on Vietnam stockmarket This has inspired the author to conduct this study , which is expected to guidethe dividend policy of Vietnamese companies

The remainder of this chapter provides general introduction about the usefulness

of the research, the research objectives, research methodology and the structure of theresearch

1.2 THE USEFULNESS OF THE RESEARCH

Firstly, this study provides insights and guidance for investors in choosing thekind of stocks having dividend payment level and dividend form suitable to theirpreference (details presented in chapter 4)

Secondly, this study provides a useful caution that investors should not be fooledwhen the firms say “we do not pay dividend because we have investmentopportunities” In fact, the listed firms on HOSE fail to pay dividend only because ofmaking loss or too small profit in their business operation, not because of havinginvestment opportunities in the future In other words, the probability of payingdividend of firms listed on HOSE totally depends on their business profit

Finally, we also provide academics with empirical evidence on determinants ofdividend policy of Vietnam stock market Our findings is that many of the factorssignificant in the determination of dividend policy of the firms listed on HOSE arethe same as those found in developed capital markets such as profitability, firm sizeand asset tangibility We also find some evidence to support the life cycle theory Ourfindings suggest that firms with greater investment, as measured by growth rate ofsales are more likely to prefer paying stock dividends to cash dividends On the otherhand, mix of both cash and stock dividend payers are associated with bigger firm sizethan cash dividend payers

The research is planned in the context of firms listed on Hochiminh StockExchange (HOSE) of Vietnam The main purpose of this paper is to investigate thedividend policies of the non-financial firms listed on HOSE for the period of 2007-

2009, and to explain their dividend payment behavior

Research questions are specified to solve research objectives as follows :

Q1: What factors determine the probability of paying dividend of firms listed onHOSE? What reasons make the firms not pay dividends?

Q2: What factors determine cash dividend per share of firms listed on HOSE?

Q3: What factors determine the probability of choosing the different forms ofdividend payment (stock dividend, or cash dividend or mix of cash and stockdividend)?

Q4: Do firms in different classified industries have different dividend policies?

Using multinomial logistic regression to test the determinants of the

probability of choosing the different forms of dividend payment (stock

dividend, or cash dividend or mix of cash and stock dividend)

Stata software version 11 is used as an data analysis tool to implement this

research

with theory?

Yes

Conclusionand feedback

1.5 THE STRUCTURE OF THE RESEARCH

The organization of the paper is as follows The next section reviews someimportant previous studies abroad The third section describes the data andmethodology The fourth section presents the results and the last section contains themain conclusions of the study

The structure of the research is summarized in figure 1.2 as follows :

Chapter 1 Introduction

CHAPTER 2 : LITERATURE REVIEW

2.1 INTRODUCTION

This chapter reviews relevant theories and models of dividend problem in theliterature The purpose of this review is to provide the background for the researchhypotheses as well as the research models which will help to test our mentionedresearch objectives in Vietnam stock market as an emerging market

2.2 DIVIDEND THEORIES AND CONCERNED EMPIRICAL STUDIES

The bird-in-hand theory, a pre-Miller-Modigliani theory, postulates that investorsprefer dividends from a stock to potential capital gains because of the inherentuncertainty of the latter In other words, investors prefer the certainty of dividendpayments to the possibility of substantially higher future capital gains Therefore, ahigh payout ratio will reduce the required rate of return (cost of capital), and henceincrease the value of firm (Gordon, 1959)

However, Miller and Modigliani (1961) demonstrate that under certainassumptions such as rational investors and a perfect capital market, the market value

of firm is independent of its dividend policy In actual market practices, nonetheless,these assumptions do not hold and dividend policy does seem to matter Over theyears, dozens of theories have attempted to explain the dividends phenomenonwithout reaching any consensus Many of the theories regard dividends as asignaling device to mitigate information asymmetry problems Others view agents asrational and dividends either serve as an efficient way to resolve agency problems, or

to reduce transaction costs to shareholder in managing their funds

According to signaling models (Bhattacharya, 1979; John and Williams, 1985;and Miller and Rock, 1985) dividends contain the information about the current orfuture level of earnings and therefore can be used as a signaling device to influencethe share price Only good-quality firms can send signals to the market through

dividends and poor-quality firms cannot mimic these because of the dissipativesignaling costs (for example, transaction costs of external financing, tax penalties ondividends, or distortion of investment decisions) An announcement of dividendincrease is considered as good news and accordingly the share price reacts favorably,and vice versa Support for the signaling hypothesis can be found, for example, inNissim and Ziv (2001), and Bali (2003).

A different explanation for dividend policy is based on the agency cost (La Porta

et al., 2000) The agency cost theory predicts that dividend payments can reduce theproblems associated with separation of ownership and management control.Dividends serve as a mechanism to reduce free cash flow under management control,and thus help to mitigate the agency problems Reducing funds under managementdiscretion may result in forcing them into the capital markets more frequently, thusputting them under the scrutiny of capital suppliers (Rozeff,1982) Many researchershave offered empirical support for the agency-theory explanations of why firms paydividends such as Holder et al (1998), among others They document that the firmswith higher levels of free cash flow have higher agency costs and need higherdividend payout ratios to reduce those agency costs Others including Yoon andStarks (1995), and Lie (2000) provide little support or reject the hypothesis

Another explanation relates dividend policy to the transaction costs Due topaying dividends, the firm has to raise external financing in order to meet investmentneeds Transaction cost include flotation costs to the firm of raising additionalexternal finance such as underwriter fees, administration costs, management time andlegal expenses This theory indicates that firms incurring large transaction costs, will

be required to reduce dividend payouts to avoid the costs of external financing (Alli

et al.,1993; Holder et al., 1998)

Another vein of the literature ties dividend payout to firms’ life cycle Inparticular, numerous papers observe that firms that pay dividends tend to be more

mature and less volatile According to Grullon et al (2002), firms that exhaust theirinvestment opportunities increase their dividends, and thus dividends indicate firmmaturity rather than signaling future profitability Several papers highlight the linkbetween dividends and idiosyncratic risk Fink, Fink, Grullon, and Weston (2004)document that dividend-paying firms have lower idiosyncratic volatility Bradley,Capozza, and Seguin (1998) and Chay and Suh (2008) explain the link betweendividends and volatility in selection : Only firms with low cash-flow uncertainty feelcomfortable with committing to paying dividends, an attitude consistent with theconservative managerial views in Lintner (1956) and Brav et al (2005) Hoberg andPrabhaha (2009) determine that the disappearance of dividends (Fama and French,2001) is associated with an increase in idiosyncratic risk Supporting the view that thedecline in idiosyncratic risk is related to firm maturity, studies find that idiosyncratic

is negatively correlated with the firm governance index ( Ferreira and Laux, 2007)and firm age ( Fink et al., 2004) DeAngelo, DeAngelo, and Stulz (2006) also findsupporting evidence for the life-cycle theory : Firms are more likely to pay outdividends when their equity is earned through operations, rather than contributed byinvestors

2.3 EMPIRICAL STUDIES ON THE DETERMINANTS OF DIVIDEND

POLICY

This section reviews empirical studies concerning the effects of individualvariables on dividend policy These empirical studies will provide background for theauthor to collect research data and to establish the research hypothesis andmethodology The considered individual variables include profitability, firm size,financial leverage, growth opportunities, assets tangibility and business risk

2.3.1 Profitability

The financial literature documents that a firm’s profitability is a significant andpositive explanatory variable of dividend policy (Jensen et al., 1992; Han et al., 1999;Fama and French, 2001; Al-Malkawi, 2007) However, there is a significant

difference between dividend policies in developed and developing countries Wang et

al (2002) compare the dividend policy of Chinese and UK listed companies, and findthat the former tends to vote for a higher dividend payout ratio, than the latter.Moreover, UK companies have a clear dividend policy in which annual dividendincreases and most of companies pay a cash dividend In contrast, Chinese companieshave unstable dividend payments and their dividend ratios are heavily based on firmearnings in the same year, not on any other factor The latter finding is consistent withthat of Adaoglu (2000), who states that the main determinant in the amount of cashdividends in the Istabul Stock Exchange is earnings for the same year

2.3.2 Firm size

Jensen et al (1992) and Redding(1997) indicate that large firms distribute ahigher amount of their net profits as cash dividends, than small firms do Otherstudies relate the positive association between dividends and firm size to transactioncosts For example, Holder et al (1998) reveal that larger firms have better access tocapital markets and find it easier to raise funds at lower costs, allowing them to payhigher dividends to shareholders This demonstrates a positive association betweendividend payouts and firm size Furthermore, Sawicki (2005) illustrates that dividendpayouts can help to indirectly monitor the performance of managers in large firms.That is, in large firms, information asymmetry increases due to ownership dispersion,decreasing the shareholder’s ability to monitor the internal and external activities ofthe firm, resulting in the inefficient control by management Paying large dividendscan be a solution for such a problem because large dividends lead to an increase inthe need for external financing, and the need for external financing leads to anincrease in the monitoring of large firms, because of the existence of creditors

Several studies have tested the impact of firm size on the dividend-agencyrelationship Lloyd et al (1985) are among the first to modify Rozeff’s model by

adding ‘firm size’ as an additional variable They find it as an important explanatoryvariable, as large companies are more likely to increase their dividend payouts todecrease agency costs.

2.3.3 Financial leverage

A growing number of studies have found that the level of financial leveragenegatively affects dividend policy (Gugler and Yurtoglu, 2003; Al-Malkawi, 2007).Their studies infer that highly levered firms look forward to maintaining their internalcash flow to fulfill duties, instead of distributing available cash to shareholders andprotect their creditors

Furthermore, Mollah et al (2001) examine an emerging market and find a directrelationship between financial leverage and debt-burden level that increasestransaction costs Thus, firms with high leverage ratios have high transaction costs inraising external finance, and therefore are in a weak position to pay high dividends

fast-Moreover, La Porta et al (2000) investigate countries with high legal protectionand reveal that fast-growth firms pay lower dividends, as the shareholders are legallyprotected, allowing them to wait to receive their dividends when the investmentopportunities are good On the other hand, in countries with low legal protection forshareholders, firms keep the dividend payout high, to develop and maintain a strongreputation, even when they have better investment opportunities

2.3.5 Assets Tangibility

The previous literature assumes that there is a relationship between the firm’s

assets structure and firm dividend policy “Firm with more tangible assets have

greater tax benefits without relying on debt, and therefore might be more inclined to use dividend policy to influence information asymmetry and agency costs” (Koch and

Shenoy, 1999) On the contrary, it is argued that asset tangibility has an inverserelationship with dividend policy, especially in developing markets Aivazian et al

(2003) state “when the assets are more tangible, fewer short-term assets are

available for bank to lend against This imposes financial constrains on firm operating in more primitive financial systems, where, the main source of debt is short-term bank financing”.

2.3.6 Business risk

It has been argued that high-risk firms tend to have a higher volatility in theircash flows, than low-risk firms Consequently, the external financing requirement ofsuch firms will increase, driving them to reduce the dividend payout to avoid costlyexternal financing (Chen and Steiner, 1999) Jensen et al (1992) contend that greatersystematic risk increased the uncertainty of the direct relationship between currentand expected future profits Hence, firms avoid commitment to pay large dividends,

as the uncertainty about earnings increases Moh’d et al (1995) also report an inverserelationship between the dividend ratio and intrinsic business risk, proxied by theshare’s beta They indicate that firms with unstable earnings pay lower dividends, in

an attempt to keep the dividend payout stable and to avoid the high cost of externalfinancing In contrast, Mollah (2001) finds that firms listed on the Dhaka StockExchange pay a large dividend, even though the beta for their stock is high He thenargues that in an emerging stock exchange, dividend may not be the most appropriatetool to convey correct information about transaction costs to the market

2.4 CONCLUSION

Dividend policy in emerging markets is often different in its nature,characteristics, and efficiency, from that of developed markets The above review ofthe literature reveals several different explanations for the factors which effect ondividend policy These factors vary substantially from country to country That is one

of the reasons why we want a study for Vietnam stock market because we do notknow yet which theory applies to Vietnam firms and which empirical studies isconsistent with Vietnam stock market

CHAPTER 3 : METHODOLOGY

3.1 INTRODUCTION

This chapter provides the details of research data collection and applied researchhypothesis and methodology to test research objectives The outline of this chapter issummarized in figure 3.1 as follows :

Figure 3.1 Outline of chapter 3

Research SampleDescription

Explanatory Variablesand Hypothesis

3.3 Empirical Model Specifications

3.2 DATA SPECIFICATIONS

3.2.1 RESEARCH SAMPLE DESCRIPTION

For the study, the sample of 109 firms of Hochiminh Stock Exchange has been taken into the consideration We select 109 firms based on the following criteria :

1 Listed on Hochiminh Stock Exchange during the period 2007-2009

2 Be non-financial firms In other words, funds, securities companies, banksand other financial firms are excluded from the sample Following previousresearch (Fama and French (2001), DeAngelo et al (2006)), we excludefinancial firms because these firms operate in a highly regulated environment

3 Must have non-missing values on dividends and the financial database

4 Trading sections of the firms are not be interrupted because of their violation

of the regulations of Hochiminh Stock Exchange

The present study includes both dividend-paying as well as non-dividend-payingfirms Because if value-maximizing firms choose not to pay dividends, a sample thatcontains only dividend paying firms will be subject to a selection bias Kim andMaddala (1992) demonstrate that it is important to allow for zero observations ondividends in the estimation of models of dividend behavior Likewise, Deshmukh

(2003) states “ If firms find it optimal to not pay dividends, then their exclusion from

any empirical analysis may create a selection bias in the sample, resulting in biased and inconsistent estimates of the underlying parmeters” The data has been collected

from the web of Hochiminh Stock Exchange for the period 2007 to 2009, specified asfollows :

The explanatory variables of the study including return on assets (ROA), firm size, financial leverage, asset tangibility, growth opportunities have been calculated from the Audited Annual Financial Statements of 109 firms for the period of 2006 to 2009

The last explanatory variable – business risk (beta) has been calculated fromhistory of adjusted stock price of these 109 firms and history of VNindex.The dependent variable – dividend per share has been collected from the dividend time line of 109 firms for the period 2007 to 2009 For firms paying more than once per year, we add all times of the dividend payment to calculatedividend per share for the whole year (see more in Appendix B)

The classification of industries has been adjusted with reference of the

“Market Review” report of Vietnam International Securities Company

(http://vise.com.vn/)

Panel structure is by year

Figure 3.2 Summary of Research Data Collection

Collecting Collecting The history Classifying

The Audited of adjusted

Resolution of Dividend ROA; Firm Business risk

Growth

Independent Explanatory Variables Industry

3.2.2 EXPLANATORY VARIABLES AND HYPOTHESIS

3.2.2.1 Variables those vary both across firms and time

3.2.2.1.1 Return on Assets (ROA)

Based on the above mentioned literature review, profitability of the firm isundoubtedly expected to have the largest influence on dividend payment decision.Loss making and low profit margin firms are more likely to omit dividend whereaspoor quality firms cannot afford to match dividend payments because they face hightransaction cost when the cash flows do not materialize As a proxy, this studymeasures firm profitability by the return on assets (ROA) (Kamat and Kamat, 2009).The formula for return on assets as follows:

Net profitROA =

Total assets

The hypothesis is formulated to test the ROA as follows : the dividend payout ispositively associated with a firm’s current profitability.

3.2.2.1.2 Firm size

Based on the above discussion of the literature review, the firm size variable isexpected to have a positive relationship with dividend payouts A large firm typicallyhas better access to capital markets and finds it easier to raise funds with lower coastand fewer constraints compared to a small firm This suggests that the dependence oninternal funding decreases as firm size increases Therefore, ceteris paribus, largefirms are more likely to afford paying higher dividends to shareholders This studyuses the natural logarithm of total assets as a proxy of firm size (Gill et al., 2009).The formula for firm size as follows :

Firm size = Ln(total assets)The hypothesis in regard to firm size is formulated as : the dividend payout ispositively associated with firm size

3.2.2.1.3 Financial leverage

According to the literature review, an inverse relationship between financialleverage ratio and dividends is expected When a firm acquires debt financing itcommits itself or fixes financial charges embodied in interest payments and theprincipal amount, and failures to meet these obligations may lead the firm intoliquidation The risk associated with high degrees of financial leverage may thereforeresult in low dividend payments because, ceteris paribus, firms need to maintain theirinternal cash flow to pay their obligations rather than distributing the cash toshareholders As a proxy, this study measures financial leverage by debt ratio(Aivazian et al., 2003; Achy, 2009) The formula for debt ratio as follows:

Total DebtDebt ratio =

Total AssetsThe hypothesis is formulated for further investigation as : the dividend payout isnegatively associated with financial leverage

3.2.2.1.4 Growth Opportunities

A review of literature reveals several explanations for the relationship betweengrowth opportunities and dividend policy One explanation is that a firm tends to useinternal funding sources to finance investment projects if it has large growthopportunities and large investment projects Such a firm chooses to cut, or pay lowerdividends, to reduce its dependence on costly external financing On the other hand,firms with slow growth and fewer investment opportunities pay higher dividends toprevent managers from over-investing company cash This study uses sales/revenuesgrowth rate as a proxy variable for growth opportunities (Holder et al., 1998; Chen etal., 1999; Saxsena, 1999; Manos, 2002) The formula for sales/revenues growth rate

is negatively associated with asset tangibility

3.2.2.2 Variables that vary only across firms

3.2.2.2.1 Business Risk (Beta-β)

This study uses beta as a common proxy for firm business risk, which represents

a firm’s operating and financial risk (Holder et al., 1998; Chen et al., 1999; Saxsena,1999; Manos, 2002) The formula for beta as follows:

Cov(Ri, Rm)Beta =

Var (Rm)Where : Ri: stock return; Rm: market return (VNindex)

Based on the previous discussion, the following hypothesis is formulated : thedividend payout is negatively associated with systematic risk

3.2.2.2.2 Industry Dummies

Many characteristics of the firms may be reasonable similar within the industrygroups, but cannot be captured easily Because firms in the same industry also followsome different characteristics or procedures For these reasons the industryclassification of firms included in our specification results from the reference of the

“Market Review” report of Vietnam International Securities Company(http://vise.com.vn/) Eleven industry classifications used here are Basic Materials,Consumer Staples, Consumer Discretionary, Oil/Gas, Real Estate, Industrials,Information Technology, Utilities, Transportation, Constructions and Materials, andMulti-scope Business and Group Related to this prediction is the observationreported in Pandey (2001), that a firm’s industrial classification is an importantdeterminant of dividends

3.2.3 DEPENDENT VARIABLE – DIVIDEND POLICY

The Dividend Time Lines of listed firms on HOSE have been announced from 1

to 3 times each year Their contents include dividend declaration date, ex-dividenddate, record date, dividend payment date, the level of dividend payment as apercentage of par value per share and the forms of dividend payment

In different models/regressions we use different dependent variables capturing different aspects of a firm’s dividend policy including :

The probability of paying dividend (pay dividend or not to pay dividend) Cash dividend per share

The probability of choosing the paid dividend forms ( stock dividends, or cashdividends or mix of cash and stock of dividends.)

3.3 EMPIRICAL MODEL SPECIFICATIONS

3.3.1 MODEL 1 : This model tests the determinants of probability of paying

dividend (including cash dividend, stock dividend and both of cash and stockdividend) of firms listed on HOSE

3.3.1.1 Hypotheses :

H1: Profitability is positively associated with the probability of dividend paymentH2: Firm size is positively associated with the probability of dividend paymentH3: Financial leverage is negatively associated with the probability of dividendpayment

H4: Asset tangibility is negatively associated with the probability of dividendpayment

H5: Growth opportunities is negatively associated with the probability of dividendpayment

H6: Business risk is negatively associated with the probability of dividend payment.H7: Classified industries influence the probability of dividend payment

3.3.1.2 Model :

Random effects logistic regression model is used to estimate the probability ofdividend payment occurring We code the probability of dividend payment as follows:

1 = the company pays dividend (cash dividend, or stock dividend or both of cash dividend and stock dividend) in that particular year

0 = the company does not pay dividend at all

The static panel data model as :

Log(odds) = Log (p/(1-p)) = α + β 1 ROA it + β 2 SIZE it + β 3 DEBT it + β 4 TANG it +

u it is composite residual comprised of a firm-specific component.

3.3.2 MODEL 2 : This model tests the determinants of cash dividend per share of

firms listed on HOSE

3.3.2.1 Hypotheses

H1: Cash dividend per share is positively associated with profitability

H2: Cash dividend per share is positively associated with firm size

H3: Cash dividend per share is negatively associated with financial leverage

H4: Cash dividend per share is negatively associated with asset tangibility

H5: Cash dividend per share is negatively associated with growth opportunitiesH6: Cash dividend per share is negatively associated with business risk

H7: Cash dividend per share is positively/negatively associated with classifiedindustry

3.3.2.2 Model :

Random effects generalized least square (GLS) regression model is used toestimate the determinants of cash dividend per share The static panel data asfollows :

Yit = α + β 1 ROA it + β 2 SIZE it + β 3 DEBT it + β 4 TANG it + β 5 RISK i + β 6 GROWTH it +

Yit refers to cash dividend per share for firm i in year t, the dependent variable

is calculated as percentage on face value of per stock

The parameter to be estimated are α and the seventeen β that capture the impact of profitability, firm size, financial leverage, tangibility, business risk, growth opportunities and eleven classified industries on cash dividend per share

u it is residual error for firm i in year t

3.3.3 MODEL 3 : This model tests the determinants of the probability of choosing

form of dividend payment (cash dividend or stock dividend or both of stock and cashdividend occurring)

3.3.3.1 Hypotheses

H1: Profitability influence the probability of choosing forms of dividend paymentH2: Firm size influence the probability of choosing forms of dividend paymentH3: Financial leverage influence the probability of choosing forms of dividendpayment

H4: Asset tangibility influence the probability of choosing forms of dividend payment

H5: Growth opportunities influence the probability of choosing forms of dividend payment

H6: Business risk influence the probability of choosing forms of dividend payment.H7: Classified industries influence the probability of choosing forms of dividend payment

3.3.3.2 Model :

We code the forms of dividend payment as follows :

1 = the company has a policy to pay cash dividends

2 = the company has a policy to pay stock dividends

3 = the company has a policy to pay mix of cash and stock dividends

Multinomial logistic model is used as follows :

Log(P2/P1) = α + β 1 ROA it + β 2 SIZE it + β 3 DEBT it + β 4 TANG it + β 5 RISK i +

P1 is the probability of choosing cash dividend payment, P2 is the probability

of choosing stock dividend payment, P3 is the probability of choosing both of stock and cash dividend payment

The parameter to be estimated are α and the seventeen β that capture the impact of profitability, firm size, financial leverage, tangibility, business risk, growth opportunities and twelve classified industry on the probability of choosing forms of dividend payment

u it is residual error for firm i in year t

CHAPTER 4 : DATA ANALYSIS AND FINDINGS

4.1 INTRODUCTION

This chapter presents the analysis of results from the study We use descriptivestatistics to explore the features of explanatory variables and correlation matrix topresent the relationship between explanatory variables Furthermore, we use thedifferent forms of regression analysis to explore the determinants of dividend policy

of the firms listed on HOSE

4.2 DESCRIPTIVE STATISTICS

Among 109 selected companies in the study, there are 19 companies in BasisMaterials sector, 24 companies in Consumer Staples sector, 11 companies inConsumer Discretionary sector, 8 companies in Oil/Gas sector, 4 companies in RealEstate sector, 10 companies in Industrials sector, 1 company in InformationTechnology sector, 6 companies in Utilities sector, 7 companies in Transportationsector, 17 companies in Constructions and Materials sector and 2 companies in Multi-Scope Business and Group sector The percentage of Consumer Staples sector is22.02 percent, highest in the industry structure The percentage of InformationTechnology sector is 0.92 percent, lowest in the industry structure (see more in Table4.1)

Table 4.1: Summary of the industry structure

in Table 4.2)

Table 4.2: Summary of the dividend policy of 109 firms listed on HOSE

companies (90.00%) in 2008 and 75 companies out of 98 companies (76.53%) in

2009 among the dividend payers (see more in Table 4.3)

Table 4.3: Summary of the dividend forms of payers listed on HOSE

Furthermore, as noted in Table 4.3 :

1 The number of stock dividend payers sharply decrease from 14 companies

in 2007 to 4 companies in 2008

2 The number of both of cash and stock dividend payers also sharply fall

from 17 companies in 2007 to 5 companies in 2008

3 The number of cash dividend payers increase from 76 companies in 2007

to 81 companies in 2008

These are also the other results from the global economic crisis in 2008 Inparticular, Vietnam’s stock index take a deep plunge in March 2007 The highestpoint of Vnindex is nearly 1200 (March 2007) and falls to 370 points (October 2008).Due to the plunge of stock price, paying stock dividend not only does not attract theinvestors but also provides more burden for them because of the illiquidity of stock inHochiminh Stock Exchange Besides, stock dividend payment will make the number

of outstanding stocks increasing With the unpredicted difficulties of 2008’s globalcrisis as well as the uncertainty of attaining the planned profit announced in AnnualGeneral Meeting, the more the outstanding stock exist, the more diluted earnings pershare (EPS) is

The descriptive statistics of explanatory variables from the period of 2007 to

2009 contain a sample of 109 non-financial listed firms of Hochiminh Stock

Exchange and show the average indicators of variables computed from the financialstatements From the descriptive statistics we find that mean level (Average) return

on asset (ROA) is 9.17% in our full sample The mean value of leverage is 44.63%which shows the firms in our sample use the debts to finance their assets The averagegrowth rate in sales of listed non-financial firms is 29.55% The mean of the firm size

is 27.25 which shows that the listed firms of HOSE do not invest more in their asset.The mean of the asset tangibility is 30.28% which shows that fixed assets is not muchinvested in the asset structure of firms The explanatory variable of beta shows thebusiness risk with the mean average of 0.944 (see more in Table 4.4)

Table 4.4 : Descriptive statistics of the variables used in the study for the

non-financial firms listed on HOSE for the period of 2007 to 2009

Profitability (ROA) 0.091710 0.146615 -1.859 0.525

Firm size ( SIZE) 27.25962 1.089692 25.207 30.292

Firm leverage (DEBT) 0.446355 0.214124 0.032 0.989

Tangibility (TANG) 0.302865 0.199051 0.007 0.939

Growth rate (GROW) 0.295523 0.641780 -0.671 7.255

Business risk (BETA) 0.944503 0.179570 0.455 1.3076

4.3 CORRELATION MATRIX OF EXPLANATORY VARIABLES

Because multicollinearity between explanatory variables may result in the wrongsigns or implausible magnitudes in the estimated model coefficients, and the bias ofthe standard errors of the coefficients To avoid this problem, the Variance InflationFactor (VIF) test is used The results of this test are presented in Table 4.5 The meanVIF is 1.15, which is much lower than the “rule of thumb” of 10 The VIF forindividual variables is also very low This indicates that the explanatory variablesincluded in the model are not substantially correlated with each other (see more inTable 4.5)

Table 4.5: Variance Inflation Factor (VIF) for the explanatory variables

Business risk (BETA) 1.25 0.8027

Profitability (ROA) 1.12 0.8907Firm leverage (DEBT) 1.12 0.8924

To further test whether the explanatory variables are correlated, a pair-wisecorrelation matrix among the explanatory variables is estimated The results areillustrated in Table 4.6, where it can be seen that the correlation coefficient are low(all < 0.330), suggesting that there is no multicollinearity problem among thesevariables

Table 4.6 : Correlation coefficients among the explanatory variables

Profitability (ROA) 1.0000

Firm size (SIZE) -0.0036 1.0000

Firm leverage (DEBT) -0.2593 0.1647 1.0000

Tangibility (TANG) 0.1523 0.1524 0.1123 1.0000

Growth rate (GROW) 0.0840 0.1480 0.1030 -0.0563 1.0000

Business risk (BETA) 0.1556 0.3227 0.0022 -0.2294 0.0565 1.0000

4.4 RESULTS OF MODEL 1

Model 1 aims to shed the light on the factors that may affect the probability ofpaying dividend Since the dataset is a cross-sectional time-series dataset, the randomeffects logistic regression (LOGIT) is used instead of the standard logistic regression

Table 4.7 : The reported results of Model 1

Explanatory Variables Coef t-statistics Coef t-statistics

***,**,* significant at 1%, 5% and 10% respectively

Prob > chi2 of Model A means that the model overall has explanation power at 1% significant level

Prob > chi2 of Model B means that the model overall has explanation power at 5% significant level

As can be seen from Model A of Table 4.7, the statistically significant variable atthe 99% confidence level is return on assets (ROA) The insignificant variables arefirm size, financial leverage, asset tangibility, growth opportunities and business risk.Since the variables of firm size, financial leverage, asset tangibility, growthopportunities and business risk are not significant, the hypotheses H2, H3, H4, H5and H6 of Model 1 cannot be supported by the data from the 109 non-financial

firms considered in this study The estimated results from Model A present thatprofitability is positively associated with probability of paying dividend This isconsistent with the hypothesis 1 (H1) of the Model 1 One explanation is thatVietnamese companies have unstable dividend payments and their dividend paymentsare heavily based on firm earnings for the same year, not on any other factor Thisresult supports all of the historical researches of the authors mentioned in literaturereview, especially the research of Wang et al.(2002) on Chinese market.

Indeed, dividend payers are associated with higher profitability, illustrated inTable 4.8 below, the median return on assets (ROA) is 11.08% for dividend payersand -8.47% for non-payers This presents that most of listed firms on HOSE do notpay dividend because of making loss in their business operation, not because ofhaving many investment opportunities in the future

Table 4.8: Descriptive Statistic of ROA of dividend non-payers and payers