This study was aimed at examining the differences between dividend policy determinants pre- and post-Malaysian Code on Corporate Governance (MCCG) 2012. Several factors, including profitability, lagged dividend, free cash flow, debt, firm size, investment opportunities and market risk were tested. The study investigated a total number of 631 non-financial firms in Malaysia that covered 7830 firm-year observations from 2005 to 2011 (pre-MCCG) and from 2013 to 2019 (post-MCCG). The study used pooled Ordinary Least Square (OLS) and random and fixed effect, with a robust standard error. The results demonstrated that from seven factors tested only four factors were found to be significant in determining dividend policy in pre-MCCG, and five factors in post MCCG. The pre-MCCG test revealed that before the revised MCCG 2012, the factors determining dividend policy were as follows: profitability, lagged dividend, debt, and firm size. However, there were slight changes in the range of determinants affecting dividend policy, Post-MCCG 2012. The post MCCG test revealed that profitability, lagged of dividend, and firm size consistently determined firm dividend policy; however, debt was no longer a significant determinant of dividend policy post MCCG. Additionally, investment opportunity and market risk were found to be significant determinants of dividend policy post-MCCG in 2012.
In general, at the end of each financial year, the firm with a profit will decide on whether to distribute its profit or not. If so, how much will this profit be distributed in the form of a dividend? For more than a decade, the dividend issues remained a mystery among researchers and practitioners. For instance, the firm is required to decide its optimum dividend policy, and this is because, the dividend will affect other the firm’s other policies, such as firm investment and capital structure, which will be directly linked to firm-specific characteristics (Smith & Watts, 1992; Gaver & Gaver, 1993; Braclay et al., 1995). The complexity of the dividend often makes it being visualised as a puzzle which is difficult to piece together (Black, 1976). The puzzle of the dividend remains as potential issues that require further research. There are issues such as the regulation, policy, and economic changes that significantly influence how a firm determines its dividend policy throughout a year. Thus, to identify what has changed over the past year, research to update the body of knowledge in this aspect is crucial. This will enable stakeholders to understand whether the determinants of dividend policy has changed over the years, and whether these have impacted the firm’s dividend policy. One of the earliest studies with regards to dividend determinants was conducted by Lintner (1956). This study revealed that a firm’s profitability and past year dividend had a major influence over the firm’s dividend policy. The study posited that the reasons for this influence were that the investors preferred a dividend with a stable payment rate. The finding was also supported by much empirical evidence from other later studies, which also showed that the dividend was significantly influenced by profits and past year dividend (Pruitt & Gitman, 1991). Since the classic study by Lintner (1956), more studies have been conducted to examine the factors that may influence the dividend policy. One of the earliest theories relating to dividend policy is the signalling theory. According to Miller and Modigliani (1961), a dividend announcement would convey some information about a firm’s performance, making the shareholders and investors likely to react to the announcement. Bhattacharya (1979) also supported this theory, when they showed that there was a profitability correlation with dividend policy. Furthermore, the dividend announcement reflected the stability of a firm’s financial performance in the future (Kale & Noe, 1990). The dividend reflected a firm’s financial performance stability because a firm generally would only announce a dividend payment or any dividend increase. Consider the scenario in which a firm has decided on a different dividend policy, for example, a dividend omission or decrease. In this instance, it has conveyed information to the shareholders and investors that the firm has become less profitable or could suffer a potential loss. As a result, the market would punish the firm that has made dividend omission, or has decreased their dividend by mitigating the demand for their stocks, or share sell by those firms, resulting in a low firm valuation. Besides signalling theory, agency cost theory has also been related to dividend policy. The theory is based on the principal and agent relationship, in which a manager act as an agent and the owner or shareholder act as a principal. The problem arises when the manager’s action is not aligned with the best interest of the shareholder. Agency cost theory suggests that the dividend can be used as a tool for the mitigated agency problem. Easterbrook (1984) and Jensen (1986) in their study indicated that the dividend could be used as a mechanism to divert the attention of insiders from using the excess cash to invest in an unprofitable project, or for personal use. Agency cost theory also suggested that the dividend can be used to mitigate the agency’s problem by reducing agency cost related to free cash flow, debt, firm growth, investment opportunities, firm size and risk (Jensen & Meckling, 1976; Rozeff, 1982; Jensen, 1986; Utami & Inanga, 2011).
The theory indirectly indicates that these factors can potentially influence the dividend policy (Yusof & Ismail, 2016). To investigate free cash effect based on agency cost theory, the study used “FCF” as a proxy for free cash flow. In general, the higher the profitability, the more cash available, and the more dividend paid to mitigate agency cost. In this regard, agency cost theory, pecking order theory, and signalling theory might help to explain the relationship between dividend and profitability (Fama & French, 2002; Yarram & Dollery, 2015). Kuzucu (2015) discovered that dividend and profitability had a negative relationship. In contrast, al-Malkawi (2007) Bokpin (2011), Patra et al. (2012) identified profitability as a key determinant which has a positive association with dividend policy. To investigate this effect on dividend policy, present study has used ROA as a proxy for profitability and FCF to proxy for free cash flow. To examine the impact of profitability on a firm’s dividend policy, and following Fang et al. (2014) and Jiang et al. (2017), the present study used “ROA” which is a return on asset as a proxy for profitability. Whereas to examines the cash availability on dividend determinants, the study used FCF following Yusof and Ismail (2016). In the context of the debt and dividend policy relationship, a firm with more debt requires excess cash to settle its debt obligations to prevent from default. It will then lead to a reduction, in terms of available cash to sustain the operating expenses. As a result, the management team may reduce the funds available to shareholders and reduce the term of dividend pay outs. In other words, a debt and a dividend have a negative relationship. Supporting this idea, several empirical studies have been conducted in the past years. For example, al-Malkawi (2007) discovered a negative association between debt financing and dividend policy when examining the determinants of dividend policy in Jordan. Similarly, the same findings were also found in Yusof and Ismail (2016); al-Shubiri (2011) and Ramli (2010). However, not all previous studies showed a negative association between debt level and dividend policy; for instance, Appannan and Sim (2011) found a positive association between debt level and dividend pay outs. In contrast, Singla and Samanta (2019) documented an insignificant relationship between leverage and dividend. To investigate the firm’s debt-level effect on dividend policy, the present study used “Debt” to refer to the total liabilities to total asset ratio as a proxy for debt level.
Generally, when the firm size increases, it will pay more dividend to mitigate the agency problem, which is likely to increase. The reasons for the increasing agency problem among larger firms are that a large firm has more widespread ownership and therefore, difficulty in controlling its internal and external financing activities, relative to small firms (Yusof & Ismail, 2016). The positive relationship between firm size and dividend policy has been well documented in numerous past empirical researches such as in Ramli (2010), Mehrani et al. (2011), Hashemi and Zadeh (2012), and Yusof and Ismail (2016). Besides the finding on the positive association between dividend policy and firm size, past studies also recorded a negative association between firm size and dividend policy, such as in the studies by Ahmed and Javid (2009) and Huda and Farah (2011). To investigate firm size effect on dividend policy, the present study has used “Size” to refer to a natural logarithm of a total asset to proxy for the firm size, following past studies such as those by Yusof and Ismail (2016); and Dewasiri et al. (2019). In the context of investment opportunity and its relationship with dividend policy, agency cost theory has been able to provide a rational explanation. According to this theory, a firm with no or less growth and investment opportunities would be prone to being exposed to agency costs related to free cash flow (Yusof & Ismail, 2016). According to Jensen (1986), to reduce firm agency cost, a firm with less investment and growth opportunities would pay more dividend. In contrast, a firm with greater investment and growth opportunities would be paying less or no dividend because they would have required more cash to finance its investment opportunity. The negative association between investment and growth opportunities, and dividend policy were recorded in many previous studies, such as those by Rozeff (1982), Jensen et al. (1992), al-Kuwari (2010), and Yusof and Ismail (2016). To investigate this effect, the present study used “INV” to refer to the ratio of retained earnings over total asset, following Yusof and Ismail (2016). The factor of risk was also often associated with dividend policy in many past studies. A greater need for external sources of financing demonstrates a massive fluctuation in firm cash flow, which also indicates a higher firm risk. Thus, to mitigate the concerns regarding external sources of financing, a firm would often pay less dividend
(Rozeff, 1982). The negative relationship between risk and dividend policy were recorded in several past studies such as in Ramli (2010) and al-Shubiri (2011), which discovered a negative relationship between dividend policy and firm risk. In contrast, Al-Shabibi and Ramesh (2011) found a positive relationship between a firm’s risk and the firm’s decision to pay a dividend. In contrast, not all research discovered a negative association between risk and dividend payment. For example, Franc-Dabrowska et al. (2020) found an insignificant relationship between risk and dividend policies. To examine the influence of risk on a firm’s dividend policy, the present study used “Risk” to refer to one year of market beta, following Yusof and Ismail (2016). Based on the postulates of signalling theory, agency cost theory, the Lintner model and empirical evidence discovered in previous studies, the hypotheses for the Malaysian context of dividend policy were developed as follows: H1 : There is a positive relationship between profitability and firm dividend policy pre- and post-MCCG 2012. H2 : There is a positive relationship between lagged of dividend and firm dividend policy pre-and post-MCCG 2012. H3 : There is a positive relationship between free cash flow and firm dividend policy pre- and post-MCCG 2012. H4 : There is a negative relationship between debt level and firm dividend policy pre- and post-MCCG 2012. H5 : There is a positive relationship between firm size and firm dividend policy pre- and post-MCCG 2012. H6 : There is a negative relationship between investment opportunities and firm dividend policy pre-and post- MCCG 2012. H7 : There is a negative relationship between market risk and firm dividend policy pre- and post-MCCG 2012. METHODOLOGY Sample Selection and Data Collection The present study used a non-financial firm sample which was made available in DataStream within Malaysia. The reason the present study has elected to exclude a firm within banks and in non-financial sectors was because of the issue of the high leverage and industry regulations (Dewasiri et al., 2016). This study also included a one year lagged of dividend per share to mitigate the concerns of serial correlation. As a result, the study lost some firm-year observations. There were two periods covered in this study; the first is the period between 2005 to 2011 (pre-MCCG 2012), and the second is the period between 2013 to 2019 (post-MCCG 2012). The main justification for selecting these two time frames for this study was the aim of examining the pre-andpost MCCG 2012 impact on dividend policy determinants. between 2013 to 2019 (post-MCCG 2012). The main justification for selecting thes between 2013 to 2019 (post-MCCG 2012). The main justification for selecting thes between 2013 2013 to to 2019 2019 (post-MCCG (post-MCCG 2012). 2012). The The main main justification justification for for selecting selecting these thes between for this study was the aim of examining the pre-and-post MCCG 2012 impact oo between 2013 to 2019 (post-MCCG 2012). The main justification for selecting thes for this study was the aim of examining the pre-and-post MCCG 2012 impact for this study was the aim of examining the pre-and-post MCCG 2012 impact o for this study was the aim of examining the pre-and-post MCCG 2012 impact on Data Analysis determinants. for this study was the aim of examining the pre-and-post MCCG 2012 impact o determinants. determinants. determinants. determinants. The data were analysed using Stata (version 13). More specifically, Data Analysis Data Analysis Analysis Data the Dataregression Analysis method was conducted using pooled OLS, random as Data Analysis well as fixed-effect analysis. Before analysing the data, the study also the The data were analysed using Stata (version 13). More specifically, regres The data were analysed using Stata (version 13). More specifically, the regres The data data were were analysed analysed using using Stata Stata (version (version 13). 13). More More specifically, specifically, the the regress regres The carried out several diagnostic tests to ensure the data were free from conducted using pooled OLS, random as well as fixed-effect analysis. Before analy The data were analysed using Stata (version 13). More specifically, the regres conducted using pooled OLS, random as well as fixed-effect analysis. Before analy conducted using using pooled pooled OLS, OLS, random random as as well well as as fixed-effect fixed-effect analysis. analysis. Before Before analy analy conducted any multi-collinearity and heteroscedasticity issue. The diagnostic study also carried out several diagnostic tests to ensure the data were free from any conducted using pooled OLS, random as well as fixed-effect analysis. Before analy study also carried out several diagnostic tests to ensure the data were free from any study also carried out several diagnostic tests to ensure the data were free from any study alsoidentify carried multi-collinearity out several diagnostic testsincluded to ensure the the data were free from any test to issues correlation and heteroscedasticity issue. The diagnostic test to identify multi-collinearity is study also carried out several diagnostic tests to ensure the data were free from any and heteroscedasticity issue. The diagnostic test to identify multi-collinearity is and heteroscedasticity heteroscedasticity issue. The The diagnostic diagnostic test to identify identify multi-collinearity iss is and issue. test to multi-collinearity matrix analysis and Variance Inflation Factors analysis. Furthermore, correlation matrix analysis and Variance Inflation Factors analysis. Furthermore, the s and heteroscedasticity issue. The diagnostic test to identify multi-collinearity is correlation matrix analysis and Variance Inflation Factors analysis. Furthermore, the correlation matrix analysis andVariance Variance Inflation Factorsanalysis. analysis. Furthermore, thesss correlation matrix analysis and Inflation Factors Furthermore, the the study also used thecalculation robust standard errors calculation analysis to free from robust standard errors analysis to ensure that the data were correlation matrix analysis and Variance Inflation Factors analysis. the s robust standard errors calculation analysis to ensure that the data were free from robust standard standard errors calculation analysis to ensure ensure that the data dataFurthermore, were free free from from robust errors calculation to that the were ensure that the data were free fromanalysis heteroscedasticity issues. issues. robust standard errors calculation analysis to ensure that the data were free from issues. issues. issues. issues. The dependent variable used in this study was the dividend per share. The dependent variable used in this study was the dividend per share. Whereas the inde The dependent variable used in this study was the dividend per share. Whereas the ind Thedependent dependent variableused usedvariables inthis thisstudy study was the dividend pershare. share.Whereas Whereasthe theinde ind Whereas the independent used inthe this study per included the The variable in was dividend used in this study included the following: profitability, lagged of dividend, free cas The dependent variable used in this study was the dividend per share. Whereas the inde used in this study included the following: profitability, lagged of dividend, free cas following: lagged of dividend, free cash flow, debt used in in this thisprofitability, study included included the following: following: profitability, lagged oflevel, dividend, free free cash cas used study the profitability, lagged of dividend, firm size, investment opportunities and market risk. The study also added industries an used in this study included the following: profitability, lagged of dividend, free cas firm size, investment opportunities and market risk. The study also added industries an firm size, investment opportunities and market risk. The study also size,investment investmentopportunities opportunitiesand andmarket marketrisk. risk.The Thestudy studyalso alsoadded addedindustries industriesan an firm firmsize, which used dummies to control for industry and year effect. The model used in this s firm size, investment opportunities and market risk. The study also added industries an which used dummies to control for industry and year effect. The model used in this added industries and to year fixedfor effect which used dummies to control which used used dummies dummies to control control for industry and year effect. The The model used usedin inthis thisstss which industry and year effect. model which used dummies to control industry yearstudy effect. The model used in this s for industry and year effect. Thefor model usedand in this is as follows: 𝐷𝐷𝐷𝐷𝐷𝐷 = 𝛽𝛽 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝛽𝛽 𝐹𝐹𝐹𝐹𝐹𝐹 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝐷𝐷𝐷𝐷𝐷𝐷 = 𝛽𝛽 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝛽𝛽 𝐹𝐹𝐹𝐹𝐹𝐹 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = 𝛽𝛽000 + + 𝛽𝛽𝛽𝛽11111𝑅𝑅𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛽𝛽222𝐷𝐷𝐷𝐷𝐷𝐷 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽333𝐹𝐹𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝛽𝛽444𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝐷𝐷𝐷𝐷𝐷𝐷 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 = 𝛽𝛽00 𝑖𝑖,𝑡𝑡 + 𝛽𝛽22 𝑖𝑖,𝑡𝑡−1 + 𝛽𝛽33 𝑖𝑖,𝑡𝑡 + 𝛽𝛽44 𝑖𝑖,𝑡𝑡 + 𝛽𝛽 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝛽𝛽 𝐼𝐼𝐼𝐼𝐼𝐼 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛿𝛿 + 𝜇𝜇 + 𝜀𝜀 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖,𝑡𝑡 = + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷 + 𝛽𝛽 𝐹𝐹𝐹𝐹𝐹𝐹 + 𝛽𝛽 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 + 𝛽𝛽 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝛽𝛽 𝐼𝐼𝐼𝐼𝐼𝐼 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛿𝛿 + 𝜇𝜇 + 𝜀𝜀 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 +𝛽𝛽𝛽𝛽55555𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽 𝐼𝐼𝐼𝐼𝐼𝐼 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛿𝛿 + 𝜇𝜇 + 𝜀𝜀 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 + + 𝛽𝛽 𝐼𝐼𝐼𝐼𝐼𝐼 + 𝛽𝛽 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛿𝛿 + 𝜇𝜇 + 𝜀𝜀 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 + 𝛽𝛽5 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽66𝐼𝐼𝐼𝐼𝐼𝐼𝑖𝑖,𝑡𝑡 + 𝛽𝛽77𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑖𝑖,𝑡𝑡 + 𝜇𝜇𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 Where, Where, Where, Where, Where, Where, DPS = Dividend per share i,t DPS Dividend per share i,t DPS == per share DPSi,ti,t ==Dividend Dividend per share i,t Dividend per share i,t ROA = Return on asset DPS Dividend per share i,t ROA = Return on asset i,t i,t ROA = Return on asset ROAi,ti,t Returnon onasset asset i,t i,t ==Return DPS = Lagged of dividend ROA Return on asset i,t−1 DPS = Lagged of dividend i,t i,t−1 = Lagged of dividend DPSi,t−1 = Lagged of dividend i,t−1 DPS = Lagged of dividend i,t−1 i,t−1 FCF = Free cash flow DPS Lagged of dividend i,t FCF = Free cash flow i,t−1 = Free cash flow i,t FCFi,ti,t Free cash cash flow flow FCF == Free i,t i,t Debt = Total liabilities FCF Free cash flow over = Total liabilities overtotal totalasset asset i,t Debt = Total liabilities total asset i,t i,t Debti,ti,t = Total liabilities over over total asset Debt i,t i,t = Total liabilities over total asset Size = Natural logarithm of total asset = Natural logarithm of total asset Debt Total liabilities over total asset Size Natural logarithm of total asset i,t i,t Sizei,ti,t = Natural Natural logarithm logarithmof of total total asset asset i,t Size == i,t i,t INV = Investment opportunity Size Natural logarithm of total asset INV Investment opportunity i,t i,t INVi,ti,t = Investment Investment opportunity opportunity i,t INV == i,t i,t Risk = Market risk INV Investment opportunity i,t Risk = Market risk i,t i,t == Market Riski,ti,t Market risk risk Risk i,t i,t δ = Industries fixed effect Risk Market risk i,t δ = Industries fixed effect i,t i,t Industries fixed fixed effect effect δδi,ti,t == Industries i,t i,t
DPS Lagged of dividend i,t−1 = DPS = Lagged of dividend i,t−1 = DPS Lagged of dividend DPSi,t−1 = Free Lagged offlow dividend i,t−1 = FCF cash i,t FCF = Free cash flow i,t FCF = Free cash flow FCFi,t Free cash flow over total asset i,t Debt == Total liabilities Thei,t International Journal of Banking andtotal Finance, Vol. 16, Number 2 (July) 2021, pp: 1–22 Debt = Total liabilities over asset i,t Debt = Total liabilities over total asset Debti,t = Natural Total liabilities over total asset i,t = Size logarithm of total asset i,t Size Natural logarithm of total asset i,t Size == Natural of Sizei,t = Investment Natural logarithm logarithm of total total asset asset i,t INV = opportunity i,t INV = Investment opportunity i,t INV = Investment opportunity = opportunity i,t INV = Investment opportunity i,t Risk == Market risk i,t Risk Market risk i,t Risk == Market i,t Risk Market risk risk δδi,t i,t == Industries fixed effect = Industries fixed effect
= fixed effect i,t δδi,t = Industries fixed effect = Industries fixed effect i,t μμi,t === Year fixed effect Year fixed effect Year fixed effect i,t μμi,t == Year fixed effect Year fixed effect i,t εεi,t === Error terms Error Error terms i,t εεi,t = Error terms = Error terms i,t Table Table Table Table
| Definition |
| and |
| Proxy |
| of |
| Variables |
| Definition |
| and |
| Proxy |
| of |
| Variables |
| Definition |
| Definition and |
| and Proxy |
| Proxy of |
| of Variables |
| Variables |
| Variables |
| Constructs |
| Constructs |
| Dividend |
| Constructs |
| Constructs |
| Dividend |
| Dividend |
| Profitability |
| Dividend |
| Dividend |
| Profitability |
| Lagged of dividend |
| Profitability |
| Profitability |
| Profitability |
| Lagged |
| dividend |
| Free cashof |
| flow |
| Lagged |
| of |
| dividend |
| Lagged |
| of |
| dividend |
| Lagged |
| of |
| dividend |
| Free |
| cash |
| flow |
| Debt cash |
| Free |
| cash |
| flow |
| Free |
| flow |
| Free |
| cash |
| flow |
| Debt |
| Firm size |
| Debt |
| Debt |
| Debt Size |
| Firm |
| Investment |
| opportunity |
| Firm |
| Size |
| Firm |
| Size |
| Firm |
| Size |
| Market beta |
| Represented by Definitions |
| Represent |
| by |
| Proxy |
| variables |
| Represent |
| by |
| Proxy |
| variables |
| DPS Represent |
| DividendProxy |
| per share |
| by |
| variables |
| Represent |
| by |
| Proxy |
| variables |
| DPS |
| Dividend |
| Per |
| Share |
| DPS |
| Dividend |
| Per |
| Share |
| ROA DPS |
| Return on |
| asset |
| Dividend |
| Per |
| Share |
| DPS |
| Dividend |
| Per |
| Share |
| ROA |
| Return |
| on |
| Asset |
| LDPSROA |
| DividendReturn |
| per share |
| ROA |
| Return |
| on |
| Asset |
| on |
| Asset |
| ROA |
| Return |
| on |
| Asset |
| Per |
| Share |
| FCF LDPS |
| Free cashDividend |
| flow |
| LDPS |
| Dividend |
| Per |
| Share |
| LDPS |
| Dividend |
| Per |
| Share |
| LDPS |
| Dividend |
| Per |
| Share |
| FCF |
| Free |
| Cash |
| Flow |
| Debt FCF |
| Total liabilities/total |
| assets |
| FCF |
| Free |
| Cash |
| Flow |
| Free |
| Cash |
| Flow |
| FCF |
| Free |
| Cash |
| Flow |
| Total |
| liabilities/ |
| Total |
| assets |
| Size Debt |
| Natural logarithm |
| of total assets |
| Debt |
| Total |
| liabilities/ |
| Total |
| assets |
| Debt |
| Total |
| liabilities/ |
| Total |
| assets |
| Debt |
| Total |
| liabilities/ |
| Total |
| assets |
| logarithm |
| total |
| assets |
| INV Size |
| RetainedNatural |
| earnings/total |
| assetsof |
| Size |
| Natural |
| logarithm |
| of |
| total |
| assets |
| Size |
| Natural |
| logarithm |
| of |
| total |
| assets |
| Size |
| Natural |
| logarithm |
| of |
| total |
| assets |
| Risk |
| 1 Year of market beta |
| FINDINGS AND DISCUSSIONS |
| Descriptive Statistics |
The descriptive statistics presented in Table 2 showed the mean, standard deviation, and minimum and maximum value for each variable tested in this study. As demonstrated in Table 2, the average dividend per share and return on assets were 0.05 and 3.57, respectively. The average lagged of dividend, free cash flow, debt, size, investment opportunities, and market risk were 0.05, 0.02, 0.37, 12.70, 0.08, and 1.05 respectively. Before examining the main analysis, the study conducted several diagnostic tests. These tests were conducted to ensure the robustness of the result. They included tests on normality, heteroscedasticity, an autocorrelation test and lastly a multi-collinearity test. To mitigate the concern about any outlier, the study winsorizing the data at 1 and 99 percentiles. Next, the study examined the potential heteroscedasticity of the data using the White heteroscedasticity test. To reduce this concern, as indicated by the test, the study used the robust standard errors calculation.
| Variable | Obs | Mean | Std. Dev. | Min | Max |
|---|---|---|---|---|---|
| DPS | 7830 | 0.0415 | 0.0917 | 0 | 0.59 |
| ROA | 7830 | 3.5717 | 9.5172 | -35.41 | 31.96 |
| LDPS | 7830 | 0.0406 | 0.0897 | 0 | 0.59 |
| FCF | 7830 | 0.0114 | 0.1519 | -0.617 | 0.569 |
| Debt | 7830 | 0.3736 | 0.2007 | 0.0235 | 0.9308 |
| Size | 7830 | 12.7062 | 1.6258 | 5.7930 | 19.0014 |
| INV | 7830 | 0.0889 | 0.4020 | -1.8470 | 0.7086 |
| Risk | 7830 | 1.0470 | 0.6926 | -0.678 | 3.268 |
FCF
Debt
Size
INV
Risk
The study also used the lag of dependent variable to solve the potential autocorrelation in the data, as indicated by the Breusch Pagan Lagrange Multiplier (LM) test. Finally, the study also conducted a multi-collinearity test. Following Hair et al. (2010), any value higher than 4.00 in VIF Score and 0.60 in the Pearson correlation matrix was considered as having high multi-collinearity. As shown in Table 3 and 4, no variables, excluding lagged dependent variables, had a value higher than 4.00 and 0.60. This indicated that there was no risk of multi-collinearity.
| Policy | ||
|---|---|---|
| Variable | VIF | 1/VIF |
| INV | 1.71 | 0.5850 |
| Size | 1.66 | 0.6031 |
| ROA | 1.42 | 0.7059 |
| Debt | 1.37 | 0.7302 |
| LDPS | 1.3 | 0.7711 |
| Risk | 1.09 | 0.9161 |
| FCF | 1.06 | 0.9431 |
| Mean VIF | 1.37 | |
Risk
FCF
Mean VIF
ROA
FCF
Debt
Size
INV
Risk
ROA
Note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively.
DPS
DPS
Variables
Pearson Correlation Matrix Results
| Variables | DPS | ROA | LDPS | FCF | Debt | Size | INV | Risk |
|---|---|---|---|---|---|---|---|---|
| DPS | 1 | |||||||
| ROA | 0.3453*** | 1 | ||||||
| LDPS | 0.9321*** | 0.3150*** | 1 | |||||
| FCF | 0.0958*** | 0.2181*** | 0.0738*** | 1 | ||||
| Debt | -0.0289** | -0.1117*** | -0.0246** | -0.1073*** | 1 | |||
| Size | 0.3748*** | 0.2464*** | 0.3694*** | 0.0704*** | 0.2903*** | 1 | ||
| INV | 0.2717*** | 0.4842*** | 0.2678*** | 0.1466*** | -0.2820*** | 0.3637*** | 1 | |
| Risk | -0.1689*** | -0.1225*** | -0.1691*** Note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively. | -0.0248** | 0.0925*** | 0.1078*** | -0.1184*** | 1 |
The main analysis was as presented in Table 5. The Hausman test indicated that the fixed effect was the most appropriate choice, which referred to model III pre-MCCG and model VI post-MCCG 2012. From the seven variables tested in this study, four variables were found to be significant pre-MCCG 2012 and five variables in post-MCCG 2012. Referring to Model III, ROA, LDPS, Debt, and Size were found to be significant. From among the four significant variables pre-MCCG 2012, ROA, LDPS and Size, were significant at a 1 percent level, while debt was significant at 10 percent. From among the four significant variables, ROA, LDPS, and Size showed a positive relationship, whereas debt was negatively related. The post-MCCG 2012 analysis results presented in Table 6 showed that there were slight changes in the range of dividend policy determinants; it was found that from the original seven factors tested post-MCCG 2012, only five variables were found to be significant including ROA, LDPS, Size, INV and Risk. The five factors found to be significant were the following: ROA and LDPS were significant at 1 percent level, Size and INV were significant at 5 percent level and risk was significant at 10 percent. Among the five factors, three variables, namely ROA, LDPS and Size, were positively related to dividend whereas, two factors, namely Debt and Risk, were found to be negatively correlated. Hence pre-MCCG 2012, H1 (profitability), H2 (lagged of dividend), H4 (Debt), H5 (Size) were supported, and the remaining variables were not supported. On comparison, post-MCCG 2012, H1(profitability), H2 (lagged of dividend), H5 (Size), H6 (INV) and H7 (Risk) were supported, while the remaining variables were not supported. The positive and significant relationship between profitability and dividend pre- and post-MCCG 2012 indicated that as a firm’s profitability increases, it would increase its dividend payment, which indirectly supported the postulates of signalling theory. As the firm’s performance became better, it could reward its shareholders in the form of a dividend payment. The results in this study were consistent with many other previous research studies, such as those by al-Malkawi (2007), Ahmed and Javid (2009), al-Kuwari (2010), Ramli (2010), Mehrani et al. (2011), al-Shabibi and Ramesh (2011), Yusof and Ismail (2016), and Dewasiri et al. (2016). However, the findings regarding dividend and profitability were not corroborated in the results found in Anil and Kapoor (2008) and Appannan and Sim (2011) where the researchers discovered a negative association between dividend and profitability. The documented positive relationship between dividend and lagged of dividend pre-and post-MCCG 2012 were consistent with those in Lintner (1956) and Yusof and Ismail (2016). Lintner (1956) had suggested that firm profitability and the past year dividend (lagged dividend) were major determinants of firm dividend policy. According to Yusof and Ismail (2016), the influence of past year dividend over the current year dividend was the result of the investor preferring a much stable dividend rate. As a result, a firm tended to make the dividend decision based on the past year dividend policy. The negative association between debt level and dividend pre-MCCG 2012 was consistent with past studies such as those by Yusof and Ismail (2016), al-Shubiri (2011), Ramli (2010), and al-Malkawi (2007). This result also indicated that the debt level provided incentives to the firm to limit its dividend payment to fulfil its debt obligation. The finding also indirectly suggested that the debt level was influenced by the bank covenant, which might restrict the amount of dividend that could be paid by the firms. However, the result was inconsistent with several past studies such as in Appannan and Sim (2011), and Singla and Samanta (2019), which had demonstrated a negative and insignificant relationship. The significant positive relationship between firm size and dividend policy pre- and post-MCCG 2012, indicated that the agency cost theory was supported. According to this theory, the larger the firm size, the greater the percentage of widespread ownership. This most likely had created a difficult to control internal and external financing sources (Yusof & Ismail, 2016). Thus, to control agency costs associated with the greater widespread ownership found in a large firm, the large firm tended to increase dividend payment. The findings from this study with regards to firm size and dividend policy had been consistent with those in previous studies such as those in al-Malkawi (2007), al-Kuwari (2010), Ramli (2010), Yusof and Ismail (2016) and Dewasiri et al. (2019). The negative association between investment opportunities and dividend policy post-MCCG 2012 was consistent with agency cost theory. According to this theory, a firm with little to no ability to demonstrate a high potential growth rate would be highly exposed to agency cost, especially regarding free cash flow (Yusof & Ismail, 2016). Therefore, to mitigate those concerns, a firm with a greater investment opportunity represented by INV in the present study would most likely pay less dividend. The reasons were due to the firm requiring the cash to finance its investment opportunity. The negative relationship between dividend and risk has found support in past empirical evidence presented in Al-Shubiri (2011) and Ramli (2010). The results indicated that a firm with greater market risk had more cash flow issues or greater fluctuation with regards to the firm cash flow. As a result, the firm might reduce its dividend payment to ensure better cash flow to meet its obligation and finance its investment opportunity. However, the results were found to be not consistent with other past empirical studies such as those in FrancDabrowska et al. (2020) which had found an insignificant relationship between dividend and risk.
| Robust Standard Errors) | |||||||
|---|---|---|---|---|---|---|---|
| Model | Model I: | Model II: | Model III: | ||||
| Pooled Ordinary | Random effect | Fixed effect | |||||
| Least Square | (Robust Standard | (Robust Standard | |||||
| (Robust Standard | Errors) | Errors) | |||||
| Errors) | |||||||
| Regressors | Regression | t-statistics Regression | z-statistics | Regression | t-statistics | ||
| coefficient | coefficient | coefficient | |||||
| Constant | -0.0364 | -3.39** | -0.0344 | -4.99*** | -0.1031 | -3.62** | |
| ROAi,t | 0.0004 | 4.71*** | 0.0005 | 6.14*** | 0.0005 | 5.36*** | |
| LDPSi,t | 0.8896 | 28.44*** | 0.9138 | 32.41*** | 0.3994 | 5.00*** | |
| FCFi,t | 0.0091 | 1.43 | 0.0088 | 1.60 | -0.0087 | -1.10 | |
| Debti,t | -0.0134 | -3.22** | -0.0128 | -3.49** | -0.0208 | -2.57* | |
| Sizei,t | 0.0045 | 4.97*** | 0.0037 | 5.22*** | 0.0107 | 4.57*** | |
| INVI,t | -0.0030 | -2.41* | -0.0031 | -2.38* | -0.0050 | -1.91 | |
| Riski,t | -0.0031 | -3.02** | -0.0033 | -3.83*** | 0.0006 | 0.44 | |
| Industries | Yes | No | No | ||||
| Year | Yes | No | No | ||||
| R-Squared | 0.8321 | 0.8267 | 0.7543 | ||||
Note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively.
| Robust Standard Errors | ||||||
|---|---|---|---|---|---|---|
| Model | Model IV: | Model V: | Model VI: | |||
| Pooled Ordinary | Random effect (Robust | Fixed effect (Robust | ||||
| Least Square (Robust | Standard Errors) | Standard Errors) | ||||
| Standard Errors) | ||||||
| Regressors | Regression | t-statistics | Regression | z-statistics | Regression | t-statistics |
| coefficient | coefficient | coefficient | ||||
| Constant | -0.0267 | -3.43** | -0.0176 | -2.64** | -0.0399 | -2.01* |
| ROAi,t | 0.0005 | 6.24*** | 0.0005 | 6.34*** | 0.0004 | 4.28*** |
| LDPSi,t | 0.9075 | 46.28*** | 0.9217 | 51.09*** | 0.4343 | 8.20*** |
| FCFi,t | 0.0091 | 1.43 | 0.0094 | 1.52 | -0.0076 | -1.11 |
| Debti,t | -0.0031 | -1.04 | -0.0034 | -1.35 | -0.0098 | -1.85 |
| Sizei,t | 0.0020 | 3.45** | 0.0018 | 2.86** | 0.0054 | 3.32** |
| INVI,t | -0.0050 | -3.78*** | -0.0053 | -4.22*** | -0.0077 | -3.48** |
| Riski,t | -0.0022 | -3.56*** | -0.0016 | -2.59* | -0.0032 | -2.36* |
| Industries | Yes | No | No | |||
| Year | Yes | No | No | |||
| R-Squared | 0.9078 | 0.9060 | 0.8876 | |||
Note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively.
The results presented in Table 5 and 6 might be influenced by endogeneity with regard to omitted variables bias. According to Jiang, Ma and Shi (2017), the firm fixed effect could reduce the concern of endogeneity, especially regarding omitted variables bias. Thus, the pesent study re-examined the model using the firm fixed effect to mitigate omitted variables bias following Jiang et al. (2017) and Bakri et al. (2020), the results are as presented in Table 7. The results revealed that dividend policy determinants remained persistent even after controlling for endogeneity concerns regarding omitted variables bias both in pre-and-post MCCG 2012, as represented via model VII and model VIII. Additionally, FCF was also found to be significant post-MCCG.
| Model | Model VII: | Model VIII: | ||
|---|---|---|---|---|
| Firm Fixed Effects | Firm Fixed Effects | |||
| (Pre-MCCG 2012) | (Post-MCCG 2012) | |||
| Regressors | Regression | t-statistics | Regression | t-statistics |
| coefficients | coefficients | |||
| Constant -0.1031 | -3.39 -0.0399 -2.23 | |||
| ROAi,t | 0.0005 | 5.09 0.0004 5.58 | ||
| LDPSi,t | 0.3995 22.18 | 0.4343 30.00 | ||
| FCFi,t | -0.0087 | -1.82 -0.0076 -2.15 | ||
| Debti,t | -0.0208 | -2.92 -0.0098 -1.79 | ||
| Sizei,t | 0.0107 | 4.27 0.0054 3.78 | ||
| INVI,t | -0.0050 | -1.42 -0.0077 -2.81 | ||
| Riski,t | 0.0006 | 0.43 -0.0032 -3.34 | ||
| Industries | Yes | Yes | ||
| Year | Yes | Yes | ||
| R-squared | 0.7543 | 0.8876 | ||
Note. *, ** and * denote statistical significance at 1%, 5% and 10% level of significance respectively.
This study investigated the factors that might influence firm dividend policy in Malaysia. Examining a total of 631 non-financial firms listed in Malaysia, the study discovered that the range of determinants of dividend policy demonstrated slight changes pre-and-post MCCG 2012. Pre-MCCG results revealed that profitability, lagged of dividend, firm size, and debt significantly determined firm dividend policy in Malaysia. Post-MCCG 2012, the range of factors had changed; profitability, lagged of dividend, size, investment opportunity and market risk were found to be significant determinants of firm dividend policy. Specifically, the result was robust, even after controlling for endogeneity concerns, especially with regard to omitted variables bias. The present study has shown originality in its approach and this is important as it has provided empirical evidence on the range of dividend policy determinants that could change as a result of comparing two different timelines. Compared to previous studies in the field, this study has made a significant contribution by comparing pre-and-post MCCG 2012, which was neglected in the past. The study has also contributed to the growing body of knowledge based on empirical evidence obtained from studies of dividend policy. The present study has also added a corporate governance perspective within the Malaysian context. However, the study has some limitations within the context of the selected timeframe on the latest MCCG changes. For example, the study only examined pre-and-post MCCG 2012. Future research may want to include important recent changes on the MCCG, which happened in 2017. Such an updated analysis would be very beneficial for management to settle on a reasonable or ideal dividend strategy that would support shareholder capital without neglecting corporate governance regulation, which might jeopardise firm evaluation and reputation. ACKNOWLEDGMENT This research received no specific grant from any funding agency.