DO CHANGES IN DIVIDEND POLICY SIGNAL THE FUTURE OR THE PAST? LIM WEI LING FACULTY OF BUSINESS AND ACCOUNTANCY UNIVERSITY OF MALAYA

Transcription

1 DO CHANGES IN DIVIDEND POLICY SIGNAL THE FUTURE OR THE PAST? LIM WEI LING FACULTY OF BUSINESS AND ACCOUNTANCY UNIVERSITY OF MALAYA NOVEMBER 2008

2 Do Changes in Dividend Policy Signal the Future or the Past? Lim Wei Ling Bachelor of Business Administration Universiti Kebangsaan Malaysia 2001 Submitted to the Graduate School of Business Faculty of Business and Accountancy University of Malaya, in partial fulfilment of the requirements for the Degree of Master of Business Administration November 2008

3 ABSTRACT This research study investigates whether Malaysian Main Board companies pursue changes in dividends to convey information on the future profitability of the companies by studying on a total of 2,679 firm-year observations from year 1998 to The Ordinary Least Square regression results show that Main Board listed companies do not use dividends as a signalling tool to convey information on the future prospect of the companies. In fact, positive and significant relationship is found to be stronger between changes in dividends in year T=0 with concurrent changes in earnings in year T=0, which is consistent with the previous findings by Benartzi, Michaely and Thaler (1997) and Nissim and Ziv (2001). Further, regression analysis on sub-samples of firm-year observations categorized by the period of stable dividends before the dividend change events (2 years, 3 years and 4 years) shows no relationship exists between the stability of dividends before dividend change events with the extent of dividend signalling. No relationship is found between size of dividend change, size of dividend yield with the extent of dividend signalling. Multiple regression by incorporating industry dummies in the regression equation shows that no difference in changes in earnings between the 3 major sectors (Industrial, Trading/Service and Consumer) when there are changes in dividends, indicating that industry effect does not have any influence on the extent of dividend signalling of the companies. ii

4 ACKNOWLEDGEMENTS I would like to express my gratitude to my supervisor of my research project, Ybhg. Professor Dato' Dr. Mansor for his patience, support and guidance given on my research area on dividend signalling. Despite the difficulties encountered in my research due to meticulous works involved and constraints in getting complete financial database needed in this study, I am glad to be able to complete my research on time under the intellectual guidance and share of research experiences by Professor Dato Dr. Mansor. Million of thanks to my beloved family for their unconditional supports and understanding despite of my less time with them due to my busy schedule for both work and research project. Appreciation is also dedicated to my superior and colleagues at work for their support and understanding during my entire MBA course period. Special thanks to Mr CG Teh, Mr KL Ooi and Ms Shirley Lo who have provided their professional insights on dividend signalling in Malaysia. Being top management involved in the dividend decision of three Main Board listed companies, their feedbacks are really useful in explaining the reasons of dividend signalling not applicable in Malaysia. Last but not least, I would also like to express my appreciation to my fellow MBA coursemates who have accompanied me for the whole MBA course with joy and unforgettable memories. Special thanks to Mr Cheong Kok Loong who has provided his assistance and support to me during my journey of completing this research study. iii

5 TABLE OF CONTENTS ABSTRACT. ii ACKNOWLEDGEMENTS iii LIST OF SYMBOLS AND ABBREVIATIONS.. viii CHAPTER 1: INTRODUCTION PURPOSE AND SIGNIFICANCE OF THE STUDY OBJECTIVE AND SCOPE OF THE STUDY LIMITATION OF THE STUDY ORGANISATION OF THE STUDY... 5 CHAPTER 2: LITERATURE REVIEW WHAT IS DIVIDEND POLICY THE DIVIDEND SIGNALLING THEORY Previous literatures that support earnings and past dividends determine current dividends Previous literatures that support changes in dividends do not provide signals to the market Previous literatures that support changes in dividends provide signals to the market 20 CHAPTER 3: RESEARCH METHODOLOGY DEVELOPMENT OF THE HYPOTHESES SELECTIONS OF MEASURES Measure of unexpected earnings Measure of changes in dividends SAMPLING DESIGN DATA COLLECTION PROCEDURE DATA ANALYSIS TECHNIQUES Data Filtering Assumptions adopted in the regression analysis Analysis on the relationship between changes in dividends in year 0 with changes in earnings in the concurrent year and subsequent 5 years Analysis on the extent of dividend signalling with the influence of industry effect using multiple regression.. 36 CHAPTER 4: RESEARCH RESULTS SUMMARY STATISTICS Overall Dividend Payment Trend of Main Board Listed Companies in Bursa Malaysia Descriptive summary on the selected samples ANALYSIS OF MEASURES Analysis of Regression Result on All Firm-Year Observations Analysis of Regression Result on Each Individual Year Observations From Year 1998 to Analysis of Regression Result on Dividend Change Events During the Financial Crisis Period ( ) and Post- Financial Crisis Period ( ) Analysis of Regression Result on Dividend Change Events Occurred After Stable DPS for Consecutive 2 years, 3 years and 4 years iv

6 4.2.5 Analysis of Regression Result on the Dividend Change Events Categorized by Size of Dividend Change Analysis of Regression Result on Dividend Change Events Categorized by Size of Dividend Yield Analysis of Regression Result on All Firm-Year Observations by Incorporating Industry Effect SUMMARY OF RESEARCH RESULTS 97 CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS CONCLUSION SUGGESTIONS FOR FUTURE RESEARCH REFERENCES 119 APPENDICES v

7 LIST OF FIGURES Figure 4.1 Composition of dividend paying and non-dividend paying companies in the Main Board of Bursa Malaysia ( ) 37 Figure 4.2 Dividend paying companies in the Main Board of Bursa Malaysia (by sector), Figure 4.3 Type of dividend changes for dividend paying companies ( ).. 41 Figure 4.4 Average DPS of Main Board Companies, categorized by sector ( ) Figure 4.5 Average DPR of listed companies between Figure 4.6 Average Dividend Payout Ratio of Main Board Companies categorized by sector ( ) 50 Figure 4.7 Average Dividend Yield (%): Figure 4.8 Average Dividend Yield of Main Board Companies categorized by sector ( ) 53 Figure 4.9 DPS, EPS and DPR of Construction Sector ( ). 55 Figure 4.10 DPS, EPS and DPR of Consumer Sector ( ). 56 Figure 4.11 DPS, EPS and DPR of Industrial Sector ( ).. 57 Figure 4.12 DPS, EPS and DPR of Plantation Sector ( ). 58 Figure 4.13 DPS, EPS and DPR of Properties Sector ( ). 59 Figure 4.14 DPS, EPS and DPR of Technology Sector ( ).. 60 Figure 4.15 DPS, EPS and DPR of Trading / Services Sector ( ) 61 LIST OF TABLES Table 3.1 List of Regression Analysis Table 4.1 No. of dividend paying companies from year 2002 to Table 4.2 Percentage of dividend paying companies by sector (Year ).. 39 Table 4.3 Type of dividend changes for dividend paying companies ( ).. 41 Table 4.4 No. of sample companies by sector (1998 to 2007).. 54 Table 4.5 Regression result on dividend change events for all firm-year observations ( ).. 68 Table 4.6 Regression result on dividend increase events for all firm-year observations ( ).. 69 Table 4.7 Regression result on dividend decrease events for all firm-year observations ( ). 69 Table 4.8 Regression result on all firm-year observations, categorized by type of dividend change (1998). 71 Table 4.9 Regression result on all firm-year observations, categorized by type of dividend change (1999). 72 Table 4.10 Regression result on all firm-year observations, categorized by type of dividend change (2000). 73 Table 4.11 Regression result on all firm-year observations, categorized by type of dividend change (2001). 74 Table 4.12 Regression result on all firm-year observations, categorized by type of dividend change (2002). 75 vi

8 Table 4.13 Regression result on all firm-year observations, categorized by type of dividend change (2003). 76 Table 4.14 Regression result on all firm-year observations, categorized by type of dividend change (2004). 77 Table 4.15 Regression result on all firm-year observations, categorized by type of dividend change (2005). 78 Table 4.16 Regression result on all firm-year observations, categorized by type of dividend change (2006). 79 Table 4.17 Regression result on all firm-year observations, categorized by type of dividend change (2007). 80 Table 4.18 Regression result on all firm-year observations categorized by type of dividend change during the financial crisis period ( ).. 83 Table 4.19 Regression result on all firm-year observations categorized by type of dividend change during the post- financial crisis period ( ).. 84 Table 4.20 Regression result on dividend change events that occurred after stable dividends for 2 years, 3 years and 4 years.. 87 Table 4.21 Regression result on dividend increase events that occurred after stable dividends for 2 years, 3 years and 4 years.. 87 Table 4.22 Regression result on dividend decrease events that occurred after stable dividends for 2 years, 3 years and 4 years.. 87 Table 4.23 Summary on the regression result for dividend change events categorized by different sizes of dividend change 88 Table 4.24 Summary on the regression result for dividend change events categorised by different sizes of dividend yield 90 Table 4.25 Multiple regression on all firm-year observations controlled by industry effect: Industrial sector as base.. 93 Table 4.26 Multiple regression on all firm-year observations controlled by industry effect: Trading / Services sector as base 93 Table 4.27 Multiple regression on all firm-year observations controlled by industry effect: Consumer sector as base.. 94 Table 4.28 Multiple regression on dividend increase and dividend decrease sub-samples controlled by industry effect: Industrial sector as base 96 Table 4.29 Multiple regression on dividend increase and dividend decrease sub-samples controlled by industry effect: Trading / Services sector as base. 96 Table 4.30 Multiple regression on dividend increase and dividend decrease sub-samples controlled by industry effect: Consumer sector as base 97 Table 4.31 Regression result for all firm-year observations from year 1998 to Table 4.32 Regression result for all dividend increase observations from year 1998 to Table 4.33 Regression result for all dividend decrease observations from year 1998 to Table 4.34 Multiple regression result for industry dummies vii

9 LIST OF SYMBOLS AND ABBREVIATIONS The symbols and abbreviations used in this research paper and the definition of each are illustrated below: UE i,t : Unexpected earnings of firm i in year t E i,t : Earnings of firm i in year t MV i,0 : Market value of equity of firm i on the 1 st trading day of the announcement year EPS i,t : Earnings per share of firm i in year t P i,0 : Share price of firm i at the beginning of dividend change year 0. EPS i,t : Changes in earnings per share of firm i in year t Div i,t : Changes in dividend per share of firm i in year 0 D i,0 : Dividend per share of firm i in year t REITS : Real Estate Investment Trusts PN4 : Practice Note 4, as per Bursa Malaysia s published practice notes for listed companies (repealed and deleted with effect from 3 January 2005). Listed companies categorized under PN4 are companies with financial condition that does not justify continued trading and/or listing. PN17 : Practice Note 17, as per Bursa Malaysia s published practice notes for listed companies. Listed companies categorized under PN17 are companies with financial condition and level of operations on a consolidated basis do not warrant continued trading and/or listing as follows:- (a) the shareholders equity on a consolidated basis is equal to or less than 25% of the issued and paid-up capital of the company and such shareholders equity is less than the minimum issued and paid-up capital as required under Bursa Malaysia s Listing Requirements; (b) receivers and/or managers have been appointed over the asset of the company, its subsidiary or associated company which asset accounts for at least 50% of the total assets employed of the company on a consolidated basis; viii

10 (c) a winding up of the company s subsidiary or associated company which accounts for at least 50% of the total assets employed of the company on a consolidated basis; (d) the auditors have expressed an adverse or disclaimer opinion in the company s latest audited accounts; (e) the auditors have expressed a modified opinion with emphasis on the company s going concern in the company s latest audited accounts and the shareholders equity on a consolidated basis is equal to or less than 50% of the issued and paid-up capital of the company; (f) a default in payment by the company, its major subsidiary or major associated company and the company is unable to provide a solvency declaration to Bursa Malaysia. (g) the company has suspended or ceased:- (i) all of its business or its major business; or (ii) its entire or major operations, for reasons of:- (aa) the cancellation, loss or non-renewal of a licence, concession or other rights necessary to conduct its business activities; (bb) the disposal of the company's business or major business; or (cc) a court order or judgment obtained against the company prohibiting the company from conducting its major operations on grounds of infringement of copyright of products etc; or (h) the company has an insignificant business or operations IPC : Infrastructure Project Companies α 0 α 1, α 2, α 3. and any other subsequent alpha values : Constant term in the regression equation, represents the intercept of the regression line on axis Y (dependent variables) when the value of X (independent variable) is equal to zero : Beta coefficient for the independent variable and any other subsequent independent variables in the regression equation, measured on the effect of changes in independent variables on the dependent variables T : Error term in the regression DPS : Dividend per share EPS : Earnings per share DPR : Dividend payout ratio which shows how much dividend is payout from every Ringgit Malaysia of EPS. The formula for dividend payout ratio is DPS / EPS ix

11 DY : Dividend yield which measures how much cash flow an investor can get for each Ringgit Malaysia invested in the share of a company. The formula for dividend yield is DPS / Price per share. x

12 CHAPTER 1: INTRODUCTION 1.1 PURPOSE AND SIGNIFICANCE OF THE STUDY Dividend policy has become one of the major areas of research amongst the finance scholars since 1950s. It is obvious from the research results of these scholars that dividend decision is the most elusive and controversial in financial decision making, hence remained unsolved with puzzles (Black, 1976). The dividend decision of a company involves retaining a proportion of net earnings for investment needs in the future while distributing the rest as dividend to shareholders. A good dividend policy not only attracts investors and facilitates fund raising from the stock market; it also caters for the future investment needs of the company. The association between dividend decision, earnings and future investment needs therefore makes dividend announcement a source of information to the investors in accessing the future prospects of a company. In other words, dividend signals information to investors. Majority of the studies on dividend signalling of corporations in developed markets e.g. United States and Europe concluded with mixed results with some level of controversial in the theory of dividend signalling. Researchers who support the theory of dividend signalling claimed that the signalling effect from dividend announcements help to overcome informational asymmetries between the management of a company and investors who are less informed about the financial prospects of the company. Investors may view an increase in dividend payout as a signal that the company has sufficient future cash flows to meet its debt and dividend payment in the future. Such positive signal will stimulate positive reaction from investors and further cause an increase in the share price. (Handjinicolaou and Kalay, 1984). However, there are 1

13 also findings that showed dividend signalling does not exist, especially in an efficient market. The research findings in the developed market may not applicable in Malaysia in view of different firm characteristics such as ownership structure (Mancinelli and Ozkan, 2006), investment decision of firms (Fama, 1974) as well as the industry and market `characteristics differentiated by industry classification (Baker, 1998). Therefore, this research study is to examine whether Malaysian Main Board listed companies use dividend changes to signal their future financial prospects, i.e. earnings of the companies to the investors. 1.2 OBJECTIVE AND SCOPE OF THE STUDY The objective of this research is to test whether dividend signalling applies in Malaysia, given the fact that most previous studies on dividend signalling were conducted in developed countries. Given the mixed results on dividend signalling in developed countries, the study on dividend signalling in the context of Malaysia is interesting due to different market structure, legal enforcement and ownership structure in the Malaysian stock market. Hence, the objectives of this study are generally as follows:- (1) To study whether Main Board companies in Bursa Malaysia use changes in dividend to convey earnings prospect of the companies (2) To examine the effect of dividend signalling in the subsequent 5 years following the changes in dividend, should dividend signalling is proven to exist amongst the Main Board companies in Bursa Malaysia. In other words, 2

14 the research will examine whether the signalling effect becomes stronger (weaker) in subsequent years following the dividend change events. (3) To examine whether the dividend signalling effect becomes stronger when companies change their dividend policy after some period of stable dividend. (4) The scope of study is further extended to examine whether the larger the change in dividend, the stronger the signalling effect based on the behavioural norms of investors who are more concerned with larger changes as compared with smaller and insignificant changes in dividends, (5) To examine the clientele effect by creating the linkage between dividend yield and dividend signalling. Companies with higher dividend yield will place more emphasis on dividend in their share valuations as compared with companies with low dividend yield. (6) To examine the industry effect or peer group effect in dividend signalling hypothesis. Under such industry or peer group effect, a firm will adjust their dividend decisions to conform with the industry dividend practices (Baker and Powell, 2000; Baker, Veit, and Powell, 2001). This scope of this research paper is generally to examine whether managers use changes in dividends to signal future prospects to the investing public i.e. increase (decrease) in earnings in the subsequent years following changes in dividends to the investing public. The relationship between changes dividends and changes earnings is tested by focusing the relationship between dividend changes in year 0 with future earnings changes in the concurrent (year 0) and subsequent 5 years following the dividend changes (year 1 to year 5) of the companies listed on the Main Board of Bursa Malaysia. 3

15 1.3 LIMITATION OF THE STUDY In studying the information content of dividend, changes in future earnings (following changes in dividends), in some extent may take into account the effect of changes in earnings in the current year. In Malaysian stock market, the announcements of dividends and earnings are make simultaneously via the release of annual reports to the public investors. As highlighted by Aharory and Swary (1980), the major difficulty in studying the information content of dividend is the synchronization of dividend and earnings announcement. The study may not cover adequate width of studies, i.e. the size and pool of the sample is smaller and may not sufficient for the testing of dividend signalling hypothesis as compared with the previous studies conducted in more established stock exchange. The smaller sample with fewer number of dividend paying companies in each sector is due to the fact that Malaysian stock market is considered as a developing stock market with smaller number of listed companies. Furthermore, the number of Main Board listed companies is further eliminate in the filtering process as (1) some companies did not pay dividends consistently throughout the years; and (2) companies with incomplete financial information on dividends and earnings are further eliminated from the study. The Ordinary Least Square Regression (OLS Regression) adopted in this study has its limitations in terms of the violation of a host of auxiliary assumptions. For instance, the error terms in the regression equation might be (1) correlated and may not concern the effect of outliers or (2) obtained poorly behaved error terms on estimates. Moreover, the outcome of the OLS Regression might be skewed i.e. when the 4

16 unexpected earnings distribution has fat-tails and is heavily right skewed, the estimates generated by the OLS Regression will be driven by the data in the tails of the unexpected earnings distribution. Signalling theory has limitations in the form of monotonous restriction (Bernhardt, Douglas and Robertson, 2005) by averaging across the changes in dividend signals, but in fact there are many factors which can influence the result. The problem arises when there are larger reductions in dividend signal (Bernhardt, et. al, 2005). 1.4 ORGANISATION OF THE STUDY The study on the dividend signalling of Main Board listed companies in Bursa Malaysia is presented in the subsequent sections. Chapter 2 summarizes the result of the studies done by the past scholars and researchers on dividend theories and dividend signalling. Chapter 3 explains the methodologies adopted in this research study which include the hypotheses being tested, measurements and formulas of the variables in the study, sampling design, data collection procedures and data analysis techniques. Chapter 4 presents the descriptive summaries of the sample and the results of the analysis. Chapter 5 concludes the research findings and suggests for the possible area for future research. Chapter 6 lists down all the references in relation to this research study. 5

17 CHAPTER 2: LITERATURE REVIEW 2.1 WHAT IS DIVIDEND POLICY Dividend can be defined as distribution or payment in either cash or shares to the shareholders of the company out of the firms earnings (Ross, Westerfield and Jordan, 2003; Investopedia online). The decision on the amount of net earnings to be paid out as dividend to the shareholders involves several factors to be considered such as the firm s current earnings, future investment needs, cash flow position, shareholders preferences (or composition of shareholders), market sentiment as well as dividend decisions of other companies within the same industry. Due to the complexities involved in dividend decision, such decision is normally determined by the top management of the company such as chief financial officers, treasurers and board of directors. Dividend policy is a payout matter considered by firms in relation to when, how much of the net earnings to be payout as dividends and in what forms the dividends to be paid. In simpler context, dividend policy relates to the time pattern of dividend payout (Ross et al, 2003) and is determined by the changes in earnings, after taking into consideration of investment decisions (Lumby and Jones, 1981). We refer such condition as residual dividend policy. High-growth firms will not pay out large portion of their earnings as dividends, in view of the requirement to sustain higher than average growth (Investopedia online). In fact, firms that increase dividends are found to be larger and more profitable than firms that cut or maintain their dividends (Grullon, Michaely, Benartzi and Thaler, 2005). Although dividend increase is more frequent than dividend decrease, dividend increase is smaller in magnitude than 6

18 dividend decrease (DeAngelo and DeAngelo, 1990; Nissim and Ziv, 2001). There are some studies conducted on the pattern of dividend payout in other countries, i.e. companies in the United States distribute large part of their earnings as dividends and try to maintain a stable dividend policy. The establishment of dividend in these companies is in accordance with the level of current earnings as well as dividends in previous years (Lintner, 1956). However, dividend policy in emerging markets had showed some level of differences with developed markets in which the former pursued less stable dividend policy with lower payout ratio (Glen, Karmokolias, Miller and Shah, 1995). But there are also cases in which Asian companies in India and Singapore pursue a stable dividend policy (Pandey and Bhat, 1994; Ariff and Johnson, 1994). Many studies were conducted on the dividend policy in Malaysia. Studies in the 1990 s showed that the dividend behavior of Malaysian companies was stable and confirmed the applicability of the Lintner model in Malaysia (Isa, 1992; Annuar and Shamser, 1993; Gupta and Lok, 1995; Kester and Isa, 1996). However, subsequent studies showed that dividend behavior of Malaysin companies was sensitive to dividend changes (pursued less stable dividend policy) but do not immediately omit dividends when earnings decreased (Pandey, 2001). Pandey (2001) s result was consistent with study done by DeAngelo and DeAngelo (1990), Michaely, Thaler and Womack (1995), Nissim and Ziv (2001) and Grullon et al (2005) which showed that dividend cuts are less common than dividend increase and more extreme in magnitude. Following Pandey s studies, Al-Twaijry (2007) tested the relationship between DPS and EPS of listed companies in the Bursa Malaysia and concluded that companies 7

19 follow dividend policies which are not strongly attached to current earnings and negative (but insignificant) relationship was found between dividend payout ratio and future earnings. The following section elaborates on dividend related theories, namely the (1) dividend irrelevance theory, (2) agency theory, (3) information asymmetry, (4) Bird-in-Hand fallacy and (5) clientele effect. (a) Dividend Irrelevance Theory According to Miller and Modigliani (1961), the pattern of dividend is irrelevant and therefore the value and the investment decisions of a firm are independent from dividend policy in a (1) perfect and efficient market and (2) a world without taxes and transaction costs. Under the dividend irrelevance theory, a firm is free to determine any dividend policy in a free of tax environment without affecting the stream of cash flows or value of the firm in the following manner:- (i) Pay dividend in excess of cash flows from operations and issue new equities to finance for the dividend payment (ii) Pay dividend less than the cash flows from operations after making investments and the excess cash flows after paying dividends will be used to repurchase shares (Copeland and Weston, 1988) Under the dividend irrelevance theory, the pattern of cash flows provided by a company through the payment of dividends is irrelevant as shareholders are free to adjust the dividend patterns to suit their desired consumption patterns 8

20 through the capital market (Lumby and Jones, 1981). As dividend is irrelevant in enhancing shareholders wealth in a perfect capital market, the only source to enhance shareholders value is through the investment decision alone. (b) The Bird-in-Hand Fallacy Under the theory of Bird-in-Hand fallacy, risk-averse investors prefer dividends as compared to capital gains due to the uncertainties inherent with capital gains. However, there are a few arguments to this Bird-in-Hand Fallacy due to the following reasons:- (i) The choices between current dividends and the current share price appreciation. When dividend payment is announced by the company, share price of the company drops slightly lower than the dividend on the ex-dividend day. (ii) When firm increases its dividend without changing its investment policy, dividend payment has to be financed by issuing new shares and hence the increase in dividend payment is offset by losing an amount equivalent to the present value of price appreciation (Damodaran, 2001). As the risk of the firm is determined by the risk of the project cash flow, a decrease in dividend means greater investments and higher risk and therefore increase in the market rates of return (higher share price) (Copeland and Weston, 1988). 9

21 (c) Agency Theory When there are differences in ownership and control of the company, agency problem occurs. Managers as agents for shareholders must make their decisions based on the objective of maximizing shareholders wealth. In order to ensure their decisions are in line with the objective of maximization of shareholders wealth, shareholders will incur nontrivial monitoring costs. However, pursuing such monitoring action will create trade-off issue for both shareholders and managers. Shareholders will face the trade-off between monitoring costs and the forms of compensation involved to encourage the agents to act in the owners interest. At the same time, managers also face the trade-off between maximizing shareholders value and its own personal interests by pursuing non-pecuniary interest. Selfish managers may not pay out dividends but utilize the funds for personal compensation when the firms are making profits. Therefore, increase in dividend payout can reduce agency cost (Rozeff, 1982) as greater dividend payments serves as a mean in monitoring and bonding the performance of managers. Greater dividend payout may involve external financing i.e. through fund raising in the capital markets and hence put the firm under greater scrutiny by the supplier of capital besides the shareholders of the company. When a company is controlled by a majority of insiders, there is less need to pay dividends to reduce agency costs. At the contrary, agency cost will become higher when the shareholding structure of a company is dispersed and hence higher dividend payout (Rozeff, 1982). 10

22 (d) Clientele effect Contrary with the assumptions of perfect capital market and a tax-free world under the dividend irrelevance theory, the concept of clientele effect exists due to imperfection in the capital market (e.g. transaction costs, difference in interest rates and the presence of absolute capital rationing). The imperfection in capital market will impose certain costs to shareholders when adjustments in the dividend patterns are made to fit his preferred consumption pattern. In such an imperfect capital market, simple wealth maximization may not be a unique desire for the shareholders and therefore a continuous and stable supply of dividend is viewed positively as a source of cash inflows to match the desired consumption pattern of the shareholders (Lumby and Jones, 1981). The fulfillment of the shareholders wealth creates a consequential cost to the company as the company may leave with insufficient funds to finance for profitable investments. Alternatively, the company has to finance its investment needs from external resources which involve issuance cost. As such, companies start to maintain stable dividend policy with the hope that the stable dividend policy will not incur heavy cost penalties. (e) Information asymmetry The concept of information asymmetry is related to the differences in the amount of information held by the management and the external shareholders. The management, being the insiders of the company has more priori information as compared with external shareholders who have limited access to the information of the company. Due to such differences in a world of asymmetry information, any changes in the dividend will be interpreted as a 11

23 costly signaling tool used by insiders of a company to convey information on the firm s future prospects (Bhattacharya, 1979; John and Williams, 1985; Miller and Rock, 1985). The problem of information asymmetry is mitigated when the ownership of a company becomes more concentrated, as such the importance of dividend signalling is reduced (Vieira and Raposo, 2007). Dividend policy of a firm can be measured using 2 methods: (1) dividend yield and dividend payout ratio (Damodaran, 2001; Investopedia online). Changes in these 2 measures provide some information in relation to risks and future growth in earnings of the company. Previous research studies showed that shares with high dividend yields will result in excess returns, after adjusting for the market performance and risk (Damodaran, 2001). Dividend payout ratio is used in estimating future dividends and expected growth in earnings. When dividend payout ratio increases, the amount of free cash flow decreases and fewer investments can be made from the available cash flow, therefore the firm is expected to have lower growth in earnings. In other words, high retention ratio (1-dividend payout ratio) will result in higher growth in earnings (Damodaran, 2001). 2.2 THE DIVIDEND SIGNALLING THEORY Dividend signalling theory suggests that dividend announcements convey information on the firm s future prospects (Investopedia online) by stimulating changes in share prices which further generate returns to the shareholders. In other words, we refer this as the information content of dividend as suggested by Miller and Modigliani (1961). According to Miller and Modigliani, a company s value is determined by its 12

24 expected future earnings and not on current earnings. If dividends are dependent on the permanent component of the earnings, dividends would serve as a surrogate for expected future earnings. The classic study on dividend signalling suggests that current dividend is dependant on future as well as current and past earnings (Lintner, 1956). Although changes in dividends do contain some information to the investors, dividend signalling is not universally applied to all firms (Chin, 2005). An early study by John Lintner (1956) on dividend signalling showed that changes in earnings will affect dividend payout and managers rarely change their dividend payout in order to achieve the target payout ratio. Subsequent study by Fama and Babiak (1968) also confirmed the findings by Lintner (1956) in which changes in dividend lagged changes in earnings. While early scholars suggested that firms use changes in dividends to convey information on the firms financial prospects to the investors, some argued that firms rarely change their dividends regardless of the earnings of the firm. The reasons of such sticky dividend can be explained by 2 factors (Damodaran, 2001):- (a) (b) Concern of firms in maintaining higher dividends in the future; and Negative views on dividend decrease, which is associated with decrease in share price Based on the assertion of firms reluctant to change dividends, an increase in dividend signals a favorable expectation on the firm s future prospects and vice versa. 13

25 There are two important hypotheses related to the dividend signalling theory, namely the free cash flow hypothesis and the maturity hypothesis. The free cash flow hypothesis suggests that dividend signals information on investment policies of overinvestment firms (Litzenberger and Ramaswamy, 1979). An increase in dividend payment signals lack of investment opportunities for the firm and vice versa. The maturity hypothesis suggests that an increase in dividend conveys information on decreased investment opportunities, decreased return on assets and future earnings growth rate as well as decrease in systematic risks. (Grullon, Michaely, Roni and Swaminathan, 2002). Using dividend as a mechanism to convey information on the firm s profitability involves signalling cost, especially in countries that impose taxation on both dividend income and capital gains such as United States. When the cost of signalling becomes higher due to higher tax rate imposed on dividend as compared to capital gains, an increase in dividends will involve higher cost and therefore a higher return is required to compensate for the cost of taxation involved (Brennan, 1970; Litzenberger and Ramaswamy, 1979). In other words, firm value is more sensitive to a more costly signal under the signalling model (Bernhardt et al, 2005). Contrary with Bernhardt s view, subsequent studies showed that market responses more favorably to dividend increase when the tax rate on dividend is reduced, as experienced after the implementation of the Tax Reform Act The decision whether to use dividend to signal firm s prospects to the investors may be determined by the quality and characteristics of the firms. Firms with reputation may rely on other lower cost communication channel rather than using dividend 14

26 signalling to convey information to the shareholders. If changes in dividends signal information to the investors in an efficient capital market, such changes will be reflected in the share prices of the firms immediately after the dividend announcements. According to Fama (1970), a market is efficient when it fully reflects all available information and is characterized by availability of investment data, large pool of investors and fund managers, active trading, welldisseminated business and financial information, appropriate degree of market regulation and reasonably sophisticated communication system. (Lian, 2000) Previous literatures that support earnings and past dividends determine current dividends The signalling effect of dividend has become the major debates amongst the finance researchers and scholars. The famous dividend-signalling model in 1950s by Lintner (1956) showed that earnings of a company can be subdivided into permanent earnings and temporary earnings. He observed that only changes in permanent earnings affect changes in dividends, while temporary earnings will not have any influence on a company s dividend policy. Due to the nature of dividend which functions as a lagging indicator of changes in a company s permanent earnings, dividend payout ratio rises when a company begins a period of bad times and falls when a company reaches a period of good times. Further study by Aharony and Swary (1980) supported Lintner s model. According to their findings, quarterly dividend announcements have information content beyond the earnings announcement, which further supported the semi-strong form of efficient market hypothesis. 15

27 Lintner (1956) discovered for the first time that firms maintain a target dividend payout ratio and adjust their dividend policies to such target. His studies showed that current earnings of the firm and dividend in the previous years determined the firms dividends. Lintner also pointed that managers believed that investors prefer firms with stable dividend policies. Subsequent surveys by Baker, Farrelly, and Edelman (1985) on the listed companies in the New York Stock Exchange supported Lintner s views with the conclusion that the major determinants of dividend payments are current earnings and past dividends. Empirical evidences had shown that dividends do signal to the market on the financial position of a company. A research study by Garrett and Priestly (2000) showed significant evidence of dividend smoothing and dividends convey information on unexpected positive changes in current permanent earnings. No evidence was found to support the notion that the dividend signals future permanent earnings. In the same year, Guay and Harford (2000) found that relationship exists between dividend distribution with the past and contemporary cash flow shock. Their studies were further tested by Al- Sharaks (2005) and his findings supported the previous research done by Guay and Harford (2000). 16

28 2.2.2 Previous literatures that support changes in dividends do not provide signals to the market A well-known finance theory on irrelevance of dividends without any influence on the share prices of the company was introduced by Miller and Modigliani (1961) who supported the market efficiency theory. While some researchers have different views that the capital market is inefficient and therefore changes in dividends provide signal to the market, Miller and Modigliani theory suggests that changes in dividend might have information content if there exists insider information (managers are better informed than investors). Further studies by Watts (1973) and Gonedes (1978) showed that there is no relationship between current dividends and future earnings. However research done by Watts (1973, 1976a, 1976b) showed that the hypothesis on the information content of dividend was trivial. Although Marsh and Merton (1987) briefly considered the dividend signalling hypothesis, they argued that dividend signalling unlikely to occur as the firm s specific information will be washed-out. In other words, the market efficiency theory exists to support the study by Marsh and Merton (1987). They argued that the dividend decisions of individual firms are not independent of the decision of other firms in the same industry. Managers respond to the dividend announcements of their peers, regardless of the company s financial position and future investment requirements. Further study by Healy and Palepu (1988) and Benartzi et al (1997) proved the earnings reversal phenomena. The results by Benartzi et al (1997) showed no significant relationship between changes in dividends and changes in earnings in the subsequent years. However, changes 17

29 in dividends indicate changes in earnings in current year. Using Marsh and Merton model, Kao and Wu (1994) discovered marginal evidence of dividend signalling. Following Kao and Wu s studies, Fudenberg and Tirole (1995), and Vieira and Raposo (2007) proved that dividends are sticky as managers tend to maintain their dividend per share even when the company faces temporary net losses. Managers cut down their dividend payment only when they are sure that the earnings will not revive. There are researchers who viewed a dividend cut as a good news to the investors as it shows managers decision to solve the firms financial problems (Abeyratna and Power, 2002). The linearity and non-linearity of the mean reversion of earnings are found to be the elements that distinguish between the effects of dividend signalling. Earlier study on dividend signalling by Nissim and Ziv (2001) supported for the dividend signalling theory, however the reverse was found after considering the non-linearity of earnings (Grullon et al, 2005). Initial test on the relationship between changes in dividend and changes in future earnings by assuming the linearity in the changes in earnings and controlling on the uniform mean reversion and momentum in earnings showed that changes in dividends convey some information about future earnings. The study by Grullon et al (2005) showed that changes in dividends are strongly related to concurrent earnings. His result was consistent with the empirical evidence that changes in dividend policy occur only when changes in earnings are substantial (Brav, Graham, Harvey and Michaely, 2003). In other words, 18

30 changes in dividends are considered as surrogate for non-linearity of earnings under a uniform mean reversion model. The result by Grullon et al (2005) was proven earlier by DeAngelo, DeAngelo and Skinner (1992) and Benartzi et al (1997). The dividend signalling theory holds true for dividend increase based on the study by Nissim and Ziv (2001) who found the lack of correlation between dividend decrease and future earnings as the information content of dividend decrease is already captured by current year earnings. Indeed, they found that current year earnings and dividend decrease are highly correlated, which is consistent with the result of DeAngelo et al (1992) and Benartzi, et al (1997). Study by Benartzi et al (1997) supported the dividend signalling theory, showing the evidence on changes in dividends provide information on the current and past level of earnings. According to Nissim and Ziv (2001), the negative relationship between these two variables is due to accounting conservatism. Besides, market characteristics also determine the extent of dividend signalling. The usage of dividend as a signalling tool is less prominent in less developed markets as compared with developed markets such as United Kingdom in which current dividends are determined by lagged dividends (Vieira and Raposo, 2007). The fact that a developed stock market with larger pooled of investors is characterised by less concentrated ownership and the management may use changes in dividends to convey information on firm s prospects to the external shareholders. In other words, the pattern of firm ownership determines the effect of dividend signalling. Firms with concentrated ownership may not need dividend as signalling tool (Goergen, 19

31 Renneboog and Silva, 2005) as concentrated ownership reduces information asymmetries (Vieira and Clara, 2007). The size of the dividend announcing firms also has some level of influence on the extent of dividend signalling. According to Bajaj and Vijh (1990), dividend announcements by large firms will not trigger large market reaction Previous literatures that support changes in dividends provide signals to the market Some previous studies on dividend behavior found evidence that managers use changes in dividend as a signalling device to convey information about unexpected shock in earnings (Bhattacharya, 1979; John and Williams, 1985; Miller and Rock, 1985; Aharony and Dotan, 1994; Chen and Wu, 1999; Nissim and Ziv, 2001; Arnott and Asness, 2001 and 2003; Harada and Nguyen, 2005; Baker, Mukherjee and Paskelian, 2006; Staceseu, 2006; Vivian, 2006). The reason behind changes in dividends as a clear and unambiguous tool to convey the future prospects of the companies lies on the fact that financial reports of the firms only reflect past financial performance of the firms and are manipulated by the management especially when the firms faced financial and operational difficulties (Kaplan and Roll, 1972). Research conducted by Yoon and Starks (1995) supported dividend signaling, which is in turn supported by the evidence of payout asymmetries. In the same year, Bernheim and Wantz (1995) found evidence in support of signalling rather than agency explanations on the reasons dividends are paid. 20

32 Further, two research studies by Fama and French (1998) explained current dividend payout signals future expected earnings. Other researchers who used financial models to test the dividend signalling and dividend behavior in the stock market such as Brickley (1983), Healy and Palepu (1988) and Aharony and Dotan (1994) found that an increase in dividend leads to the increase in future earnings. Research done by Aharony and Dotan (1994) showed that firms that increase (decrease) their dividends experience greater (smaller) unexpected changes in earnings in the subsequent years as compared with firms that do not change their dividends. However, the magnitude of the relationship becomes smaller when the earnings change events move further away from the event quarter. Differences in the magnitude of the changes in the unexpected earnings yield are discovered under different categories of dividend change. Some researchers tested the relationship between dividends and future earnings with and without controlling for the effect of past and current earnings. Nissim and Ziv (2001) proved a positive relationship between current dividend and changes in earnings in the subsequent 2 years following the dividend change year by controlling a particular (linear) form of mean reversion in earnings. However, their results showed that dividend decrease was not related to future profits. Brickley (1983) found a positive relationship between dividend increase (decrease) and earnings increase (decrease) without controlling for the effect of past and current earnings announcements. 21

33 From the above review on the historical research findings, we can conclude that there are different views on dividend as signal on the future prospects of the company. The reasons on the differences in findings are partly due to geographical differences (different stock exchanges with different market characteristics / sophistication / market liquidity), differences in companies characteristics (different corporate culture or dividend policies) and industry effect. Many research studies on dividend signalling have been done in developed countries, but lack of research studies are conducted in developing countries such as Malaysia. As such, there is a need to further discover on the dividend signalling of the listed companies in Bursa Malaysia. 22

34 CHAPTER 3: RESEARCH METHODOLOGY 3.1 DEVELOPMENT OF THE HYPOTHESES The information content of dividend with the underlying assumption that earnings follow a random walk (Benartzi et al, 1997; Nissim and Ziv, 2001) is tested using the following hypotheses:- (A) Relationship between changes in dividends in current year with changes in earnings in the concurrent and subsequent years: HA o : Companies that change their dividends in year 0 will not experience any changes in unexpected earnings in the following years, i.e. year 1 to year 5. Instead, there is a positive and significant relationship between changes in dividends and unexpected earnings in the concurrent year. HA 1 : Companies that change their dividends in year 0 will experience changes in unexpected earnings in the following years, i.e. year 1 to year 5 with positive and significant relationship (Dividend signalling exists). The above hypothesis is further extended to test the relationship between the type of dividend change (dividend increase and dividend decrease) with change in unexpected earnings in the concurrent and subsequent years as follows:- (B) Relationship between increase in dividend in current year with increase in earnings in the concurrent and subsequent years: HB o : There is no relationship between increase in dividends in the 23