Biljana Jovković, Aleksandra S. Vasić, Jasmina Bogićević: Determinants of Dividend Policy: A Case of Serbia’s Banking Sector 13 Determinants of Dividend Policy: A Case of Serbia’s Banking Sector Biljana Jovković University of Kragujevac, Faculty of Economics, Republic of Serbia bjovkovic@kg.ac.rs Aleksandra S. Vasić University of Kragujevac, Faculty of Economics, Republic of Serbia apesterac@kg.ac.rs Jasmina Bogićević University of Kragujevac, Faculty of Economics, Republic of Serbia jasminab@kg.ac.rs Abstract Dividend policy is one of the most controversial areas of corporate finance. The paper presents the results of the research in the banking sector of the Republic of Serbia. The specific characteristics of the financial sector make the research on dividend policy determinants additionally complex. This study aims to determine the factors of dividend policy in the Serbian bank- ing sector in the period 2009–2018. The model of random effects was cho- sen to test the relationship between dividend determinants and dividend payout. Empirical results show that previous years’ dividends have a signifi- cant positive effect on dividend policy. Individual investors can benefit from the research to a great extent, as well as bank managers, when creating dividend policies that would contribute to maximising profit and satisfying the needs of employees and shareholders in the long run. Keywords: corporate finance, banking sector, dividend policy, dividend payout, Serbia Introduction Profit as the primary business motive, driving power, and breaking point of var- ious interest groups and is at the centre of the corporate companies’ functioning. On the one hand, there are corporate assets entrusted to managers for managing, while on the other hand, there are shareholders that strive to realise the income on the invested capital. The dividend represents the only regulatory channel by which it is possible to transfer corporate assets to shareholders (Malinić, 1999, 226). The set of all dividend decisions is contained in dividend policy. The pol- icy of profit distribution is contained in the dividend policy. Dividend policy is one of the most controversial topics and researched areas of corporate finance. The profit generated by banks is faced with the challenges of choosing the optimal dividend policy that would reconcile the views on the distribution of disposable NAŠE GOSPODARSTVO OUR ECONOMY Vol. 67 No. 1 ORIGINAL SCIENTIFIC PAPER RECEIVED: FEBRUARY 2021 REVISED: FEBRUARY 2021 ACCEPTED: MARCH 2021 DOI: 10.2478/ngoe-2021-0002 UDK: 336.763.2(497.11) JEL: G21, G30 Citation: Jovković, B., Vasić S. A., & Bogićević, J. (2021). Determinants of Dividend Policy: A Case of Serbia’s Banking Sector. Naše gospodarstvo/Our Economy, 67(1), 13–22. DOI: 10.2478/ ngoe-2021-0002 pp. 1 3–22 2021 NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 67 No. 1 / March 2021 14 income to the part retained for reinvestment and the part that goes to shareholders through dividends. Dividend policy in the banking sector was not so often the subject of research in academic literature, as is the case with the companies that belong to some non-financial sectors (Baker et al., 2001; Agyei & Marfo-Yiadom, 2011; Dibia, 2018). The reason for this is the specific financial structure, the presence of stricter and specific regulatory demands, differentiation in the field of managing, etc. (Jabbouri, 2016; Dewasiri et al., 2019). The main goal of the paper is to investigate the determinants of dividend policy in the banking sector of the Republic of Serbia. This study considers the impact of six variables, namely, profitability, liquidity, leverage, the previous year`s dividend, bank size, and the growth rate of income on inter- est on the dividend payout ratios. The paper contributes to literature extension, as there are very few papers studying the dividend policies in the financial sector from developing countries. The findings are important for individual inves- tors and bank managers when creating dividend policies that would maximise profit and satisfy employees’ and share- holders’ needs in the long-term period. Apart from introduc- tory and concluding considerations, the paper is structured into three units. Within the first one, the review of theoretical and empirical literature will be presented. The development of research hypotheses, model description, and defining samples will make the content of the second unit, while the research results will be presented in the last one. Literature Review Many dividend theories have tried to explain how the divi- dend decisions are being taken, the trends in dividend move- ments, which factors affect them, and the connection of dividend payouts with the market value of companies. The research of Dewasiri & Weerakoon (2016) implies that the study of dividend policy cannot rely only on one theory, but that certain explanations of this subject can be provided only along with the development of a holistic observation model. The primary starting point of the famous Miller & Modigli- ani (1961) theory of irrelevancy of dividends is the presump- tion of perfect functioning of the capital market, perfect cer- tainty, and rational behaviour. In their paper, the creators of this theory express the attitude according to which dividend policy in the conditions of a perfect capital market is com- pletely irrelevant in terms of its impact on the market value of a company. The absence of dividend payouts proves that shareholders are entirely indifferent between dividends and capital gains. The critics of the Miller & Modigliani theory (1961) imply the groundlessness of assumptions on which the expressed conclusions are based and raise the question of the validity of expressed theses in real market conditions. In the following years, theoretical and empirical literature predominately focused on examining the dividend policy in companies in the conditions of market uncertainty. Free of the assumptions on perfect market functioning, simulta- neously recognising market irregularities, Gordon (1963) and Lintner (1962) presented their literature theories pop- ularly called Bird-in-the-hand Theory. Starting from the assumption that there is a natural aversion to risk, Gordon (1963) emphasises that the investors (shareholders) prefer relatively certain dividends concerning uncertain capital gains. In the real and imperfect world, investors do not have the same information at their disposal that is available to managers in terms of business risk and business in gen- eral. Further theoretical research was aimed at the possibil- ity that the investors are sent certain messages (signals) by the market by carrying out a specific dividend policy. The theory of dividend signals connects dividend policy with informational asymmetry (Shchurina & Prunenko, 2018, 992). It starts from the assumption that the dividend pay- out ratio signals information to investors about the future performance of firms (Al-Kayed, 2017, 119). The growth of dividends directs investors to the conclusion that the management of the company believes that its position is better than the current prices reflect, and that it is realistic to expect that the growth of share prices will follow the growth of dividends. Significant efforts in academic circles to solve the contro- versy of dividend policy have resulted in plenty of research in the past few decades. The first in line, Lintner (1956), tried to obtain the answers to the questions related to div- idend policy through the research of American companies that belong to the industrial sector. He concluded that the decisions on dividend payouts are based on the current profitability and last year’s dividends, which follows the relevance of dividend policy on the value market of com- panies. After expressing opposite attitudes by Miller & Modigliani (1961) in the theory of irrelevance, a high num- ber of researchers followed in this field. Since the theory of irrelevance was based on the hypotheses that are not char- acteristic of real business conditions, the critics (Gordon, 1963; Lintner, 1962; Litzenberger & Ramaswamy, 1979) started from modified hypotheses. On the basis of obser- vation of American companies that are listed on the New York stock exchange market, Baker & Powell (1999) came to the findings that dividend payouts are following theories – Bird-in-the-hand and dividend signals. The justification for the theory of dividend signals is offered by the research results of Arko et al. (2014). Their work is based on consis- tent evidence that the decisions on dividends are affected by the level of profitability, taxes, investment opportuni- ties, leverage, and risk. In the paper of Yarram & Dollery (2015), the hypothesis on signalisation of dividend payouts Biljana Jovković, Aleksandra S. Vasić, Jasmina Bogićević: Determinants of Dividend Policy: A Case of Serbia’s Banking Sector 15 was confirmed on the example of Australian companies. The results show that the companies that pay out dividends are bigger on average, more profitable, and more influen- tial, with a lower rate of growth and risk compared to the companies that do not prefer dividend payouts. In an effort to answer the question about what factors af- fect dividend policy, a comprehensive piece of research by Jabbouri (2016) that encompassed a ten-year observation of 533 companies listed in the stock markets of Middle Eastern countries was carried out. Banks and financial companies were excluded from the analysis due to their specific financial structure, accounting methods, and cor- porate management. A positive relation of dividend policy with company size, profitability, and liquidity was spotted, while the growth rate, leverage, and economic develop- ment expressed a negative connection with dividend poli- cy. According to what was said, the paper of Dewasiri et al. (2019) confirmed the relation of profitability and growth rate with the policy of dividend payouts. Apart from the mentioned determinants, based on the data of 191 compa- nies that were observed in the period from 2010 to 2016, the authors emphasise that dividend policy is also affected by the previous year’s dividends, investment opportunities, corporate management, ownership, and the branch of in- dustry which the company belongs to. Starting from dif- ferent economic development in which companies operate and comparing dividend policies of the companies of de- veloping countries with the companies in the USA, Aiva- zian et al. (2003) concluded that national factors affect both the structure of dividend policy and its sensitivity to the effect of different determinants. However, besides the differences, in principle, there is an agreement regarding the effect of profitability and leverage on dividend policy. Dividend policy is in economic theory most often studied from the aspect of companies that belong to some of the non-financial sectors. As the main reasons for excluding banks and other financial companies from empirical re- search, Dewasiri et al. (2019) stated high leverage was present at those companies, as well as specific regulato- ry demands related to their business. Dividend payouts of banks depend greatly on strictly prescribed regulations in this sector, which makes the research of the factors that affect dividend policy become a more complex challenge in relation to the same problems of companies from the industrial sector. Baker et al. (2001) noticed the differenti- ation in dividend policy attitudes between the managers of financial and non-financial sectors. The results have shown a statistically significant difference in 9 out of 22 separate factors in total, whereby the most important are earning stability, the level of current income, and expected income. Larger dividend payouts weaken financial power and the ability of banks to take on more risk (Basse et al., 2014, 6). This is the main reason why the increase of dividends in the banking sector is not always interpreted as a positive signal to investors. Specific characteristics of the financial sector have affected the decreased scope of empirical research on key determi- nants of dividend policy in this field. Analysing financial companies in Ghana, Agyei & Marfo-Yiadom (2011) con- cluded that profitability, leverage, previous year’s divi- dends, growth, and risk are the main factors that show a statistically significant impact on dividend policy. Accord- ing to the above, Zameer et al. (2013) revealed a positive connection of profitability and previous year’s dividends with dividend policy on the example of the banking sector in Pakistan, which is in the context of the theory of divi- dend signals. Apart from the aforementioned, the research singled out liquidity and ownership structure as equally influential factors. Dibia (2018) establishes that dividend policy in Nigeria’s banking sector is subjected to the im- pact of profitability, total assets at the disposal of a bank, and the amount of leverage, while the variable of previous dividend payouts has not shown satisfying statistical sig- nificance. In a ten-year observation (2005–2015) of banks listed on the Dakar stock market, Hosain (2016) singles out liquidity and leverage as the most important determinants of dividend policy. On the other hand, with determinants like risk, ownership structure, and the amount of total as- sets at the disposal of a bank, the direct impact has not been noticed. Unlike the previously mentioned research, Basse et al. (2014) discarded the findings of the theory of dividend signals by observing the European banking sec- tor in the period of the financial crisis. This paper has not supported the attitudes according to which dividend reduc- tions are a reliable sign of future problems for investors and financial analysts. On the contrary, it is considered that European banks should turn their policy to the model of dividend reduction to strengthen their financial position in the period of facing a financial crisis. Methodology of Research The development of research hypotheses In the literature review, the key explanations and relations of certain factors to dividend policy were emphasized. The purpose of the research is to establish the critical determi- nants of dividend policy on the example of banks that oper- ate in the territory of the Republic of Serbia. Banks are the cornerstone of a country’s financial system, especially in the emerging market economies where capital markets are NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 67 No. 1 / March 2021 16 underdeveloped (Zhang, Jiang, Qu, & Wang, 2013). The specificity of the financial sector of the Republic of Ser- bia is reflected in the dominant presence of banks, which makes the area of dividend policy research in this sector attractive. In this paper, dividend policy, as a dependent variable, is observed via dividend payout ratio, defined as the ratio between total dividends and net profit. Potential variables whose impact on dividend policy is examined are profitability, liquidity, leverage, previous year’s dividend, bank size, and growth rate of income on interest. The decision on dividend payout in companies starts with the realised profit; hence it is not surprising that profitabil- ity is considered one of the most critical variables of div- idend policy. In his paper, Dibia (2018) finds a positive connection between profitability and dividends. Zameer et al. (2013) emphasise that highly profitable companies distribute a larger share of their profit through dividends. Starting from previous research, the profitability of banks will be measured with the ROA variable (Return on assets). According to everything mentioned previously, the follow- ing hypothesis is formed: H1: Profitability has a statistically significant positive im- pact on dividend policy. Liquidity is considered one of the most critical factors of dividend policy. And, while a strand of literature (¬¬Jab- bouri, 2016; Hosain, 2016) relates high liquidity to a bet- ter financial position that provides the possibility of more significant dividend payouts, others (Banerjee et al., 2005; Dewasiri et al., 2019) find a negative relation between these two variables. Most often, this type of connection is present in the banking sector. Zameer et al. (2013) con- sider that banks have more significant needs for liquidity compared to the non-financial sector. And, when they mark high liquidity, banks strive to maintain the current level to respond to all challenges readily. Following the research of Marfo-Yiadom & Agyei (2011), liquidity is expressed as the relation between cash and net assets of a bank. Ac- cording to the aforementioned, we start from the following hypothesis: H2: Liquidity has a statistically significant negative impact on dividend policy. The level of indebtedness is one more variable that is relat- ed to dividend policy. Increased exposure of a company to risks leads to higher levels of leverage. High indebtedness decreases the possibility of sending dividend signals to in- vestors (Jabbouri, 2016, pp. 292). Al-Kayed (2017) empha- sises that banks with a high level of leverage are subject- ed to stricter regulations, which is negatively reflected on dividend payout. As in Hosain’s (2016) paper, we express leverage by putting into relation total liability with the to- tal assets of banks. The hypothesis we started from is the following: H3: Leverage has a statistically significant negative im- pact on dividend policy. A high growth rate of income on interest represents the signal to investors that a company is in the phase of open investment opportunities, whereby each new investment decreases the amount that remains for the dividend pay- out. Dewasiri et al. (2019) emphasised that the growth ob- served through the availability of investments represents an essential determinant of dividend payout. Al-Kayed (2017) finds negative relation between these two variables. The growth rate will be followed through the growth of income on interest, as in the paper of Marfo-Yiadom & Agyei (2011). According to the arguments presented, the following hypothesis is formed: H4: Growth rate has a statistically significant negative im- pact on dividend policy. Lintner (1956) singled out previous dividends as one of the determinants of dividend policy. The results of Dickens et al. (2002) emphasise dividends from previous years as the critical factor in determining dividend policy in the bank- ing sector. The amount of last-year dividend payouts de- fined in a national currency as an absolute number is used for expressing the variable of previous dividends. Based on the research of Al-Kayed (2017) and Dewasiri et al. (2019), the following hypothesis is formed: H5: Previous year’ s dividends have a statistically signifi- cant positive impact on dividend policy. Investors estimate larger companies as less risky, with a better financial market position, easier approach to assets, and larger dividend payouts. Bank size was singled out as an important factor in determining dividend policy in the research of ¬¬Jabbouri (2016). The papers of Dickens et al. (2002) and Dibia (2018) also confirm the positive con- nection of these two variables. For the research of bank size, following the research of Hosain (2016), the natural logarithm of total assets will be used. Based on the afore- mentioned, we started from the following hypothesis: H6: Bank size has a statistically significant positive impact on dividend policy. Biljana Jovković, Aleksandra S. Vasić, Jasmina Bogićević: Determinants of Dividend Policy: A Case of Serbia’s Banking Sector 17 Empirical model and data Empirical research is based on the analysis of the panel data series, as the combination of cross-section data and time se- ries (Dragutinovic Mitrovic, 2002, 12). The research focus- es on the banking sector of the Republic of Serbia, with 27 banks operating in 2018 (National Bank of Serbia, 2018). Out of the total number, the basis of this research consists of ten banks chosen according to the criteria of total assets at the end of 2018 (Graph 1). Given that the market share of those ten banks was 78.1% at the time, the authors con- sidered it a sufficient choice for the sample. The variables examined in the paper consist of secondary data collected from annual reports of the chosen banks, published deci- sions on the use and distribution of profit, and the reports published by the National Bank of Serbia. The time span of the research is ten years, from 2009 to 2018. Unlike the model of fixed effects, random effect mod- el assumes that there is no individual specific effect of companies. However, all factors that affect the dependant variable are unified and contained in random error term. Simply put, the factors that can manifest their impact on a dependent variable, and are not contained in some of the independent variables, are approximately presented by the value of random error. The model of random effects can be presented in the following way: Y it = β 0 + β 1 X 1 it + β 2 X 2 it + β 3 X 3 it + β 4 X 4 it + β 5 X 5 it + β 6 X 6 it + β i + β it whereby Y represents the dependent variable (dividend payout ratio), X 1 , X 2 , X 3 , X 4 , X 5 , and X 6 are independent variables (profitability, liquidity, leverage, growth rate of income on interest, previous year’s dividends, and bank size). The error term is ε i + ε it and the regression param- eter is β. The model is restricts the co-efficient of the ex- planatory variables to be common across the units (i) and the time period (t). Unlike the basic model, the component of random error ε i , which is specific for individual obser- vation units, is added to random error ε it , referring to the combination of time series and comparative data. Table 2 gives a summed review of previously presented variables used in this research. Figure 1. Total Asset Value (in 1,000s RSD) Source of data: National Bank of Serbia, 2018. The application of panel regression requests the choice be- tween the model of fixed effects and the model of random effects. For that purpose, the Hausman test is used in the paper. The Zero hypothesis of this test implies the accep- tance of a random effects model, in contrast to the alter- native hypothesis that speaks in favour of the fixed effects model (Hosain, 2016, 8). If it is supposed that there is no correlation between a random error and independent vari- ables, the random effects model is more adequate, while otherwise, it is considered that in the assessment of param- eters, it is better to use a fixed effect model (Dibia, 2018, 6). In Table 1, the results of the Hausman test were shown. The amount of 7.84 Chi-Square with the probability of 0.2500 implies that the model of random effects is more adequate in the analysis of collected data. Table 1. Hausman T est Test summary Chi. Sq. Stat p - value 7.840421 0.2500 Table 2. Variables, description and expectation Variables Symbol Description Expectation Depended Variable Dividend payout ratio DPR Cash dividend / Net profit Independed Variable Profitability PROF Return on Asset (ROA) + Liquidity LIQ Cash and cash equivalent / Net asset - Leverage LEV Total liability / Total asset - Growth rate GR_R (Current interest income – Last year interest income) / Last year interest income + Previous year`s dividends PREV_DIV Last year dividend + Bank size BSZ Natural logarithm of total asset + NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 67 No. 1 / March 2021 18 Testing the set of research hypotheses was carried out with the help of corresponding statistical methods and with the support of the statistical package EViews. Results and Discussion The collected data will be processed and analysed in four stages. Firstly, the research started from the presentation of dividend policy that chosen banks with headquarters in the Republic of Serbia had pursued in the last ten years. After that, a statistical description of the observed variables in the sample will follow. Then, by applying correlation and regression analysis, the relation and impact of dependent variables on independent ones will be examined within the third and fourth stages. The overview of dividend policy through the amounts of paid out dividends represents often used and at the same time the simplest form of observing. Table 3 presents the amounts of paid out dividends in the last ten years on the example of the ten biggest banks in the Republic of Serbia. There are three banks singled out in the table above, which in the observed time interval did not pay out dividends – Erste bank, Eurobank, and OTP Group. The losses from previous years, retained earnings in the company, growth, and development financing are just some of the reasons for the absence of dividend distribution to their shareholders. The continuity in dividend payouts for the observed period was noticed only with Kombank. АIK and Raiffeisen bank are following. Table 3 shows that good results of the first six largest banks for the last two years (2017 and 2018) pro- vided the opportunity of consecutive payout of dividends to Table 3. Dividend Payout (in 1,000s RSD) Source of data: National Bank of Serbia, 2018. Intesa Unicredit Kombank Societe Raiffeisen AIK Erste Eurobank Postanska OTP 2009 0 0 44,820 0 0 455,208 0 0 1,967 0 2010 0 0 44,822 0 0 756,762 0 0 1,967 0 2011 0 0 37,575 0 0 149,917 0 0 1,967 0 2012 0 1,135,000 40,264 0 4,936,731 200,000 0 0 7,598 0 2013 0 0 37,351 0 4,657,614 210,000 0 0 0 0 2014 0 0 604,620 0 5,616,555 0 0 0 0 0 2015 0 2,500,000 1,962,751 0 5,227,634 3,120,182 0 0 224,211 0 2016 0 2,600,000 23,531 0 4,341,952 3,377,934 0 0 0 0 2017 18,110,986 1,250,000 16,808 5,748,046 5,411,291 3,743,569 0 0 0 0 2018 20,034,339 1,255,000 2,535,916 4,864,010 6,626,785 10,049,237 0 0 110,100 0 Table 4. Descriptive statistics Dividend payout ratio Profitability Liquidity Leverage Growth rate Previous year`s dividends Bank size Mean 0.1779 0.0117 0.6909 0.5179 0.0509 0.1225 18.9634 Median 0 0.0158 0.5876 0.7933 0.0084 0 19.0702 Maximum 1.7808 0.0577 4.2227 0.9119 1.8136 1.5286 20.1683 Minimum -0.3116 -0.1462 0.0693 -8.1391 -0.2875 -0.3116 17.2772 Standard deviation 0.3901 0.0241 0.5320 1.3126 0.2336 0.3105 0.7094 Skewness 2.3197 -3.6911 3.3311 -5.3774 4.7886 2.5813 -0.6790 Kurtosis 7.8302 22.1889 21.2285 32.533 34.8241 9.0822 2.9304 Biljana Jovković, Aleksandra S. Vasić, Jasmina Bogićević: Determinants of Dividend Policy: A Case of Serbia’s Banking Sector 19 the shareholders. The most significant benefit in the given period was realised by the shareholders of the bank Intesa, whose payouts exceeded the amount of RSD 38 billion in total. At the same time, this represents the most significant amount that any of the observed banks paid out in the ten- year period that was analysed in the paper. The amounts of paid out dividends in previous years are used to calculate the rate of dividend payout, a dependent variable that serves as a representative of dividend policy in this research. Table 4 shows the descriptive statistics for each of the variables whose impact on dividend policy is being examined along with the dependent one. By the analysis of arithmetic mean value, standard devia- tion, asymmetry, and flattening are interpreted as sample characteristics. The average amount of earnings paid out in the form of dividends is 17.79% (Mean = 0.1779). The average rate of income to assets is 1.17%, while the aver- age growth rate is about 5%. The biggest standard devia- tion from arithmetic mean value is marked with variable Leverage (Std. Dev. = 1.3126). The most significant stan- dard deviation from the arithmetic mean value is marked with variable Leverage (Std. Dev. = 1.3126). The obtained values of distribution asymmetry show both positive and negative asymmetry in relation to the mean value. All flat- tening results are positive, which implies that the distribu- tion is more peaked than a normal one. The analysis of the strength and direction of observed vari- ables is performed with the help of correlation analysis. For determining statistical importance, the significance level ά=0.05 was used. The value of the correlation co- efficient up to 0.3 implies weak correlation; from 0.3 to 0.5 the mean value, while the values above 0.5 imply a strong correlation connection between variables (Pallant, 2017, 134). The direction of correlation implies positive or negative signs of the correlation coefficient. A positive val- ue implies that with the growth of one variable the grows other as well, and vice versa. On the other hand, negative values represent an inverse movement of variables, with the growth of one comes the reduction of the other, and vice versa. In Table 5, the correlation matrix of all exam- ined variables is given. The correlation analysis results showed a statistically important correlation between certain pairs of variables with the probability of 95% and 99%. Specifically, in this research, the greatest importance will be to examine the correlation between dividend policy and independent variables. The strongest, statistically significant relation is noticed between the growth of dividend payout ratio and previous year`s dividends ( = 0.663; p = 0.000). In the case of a somewhat weaker relation and the probability of 99%, the positive relation between dividend payout ratio and bank size was observed ( = 0.291; p = 0.003). Apart from the aforementioned, the rate of dividend payout is correlated with the profitability of banks as well ( = 0.222; p = 0.027). By observing pairs of independent variables, the mean correlation relation was observed between prof- itability and bank size ( = 0.431; p = 0.000). Also, anal- ysis results imply weak, positive relation of bank size and previous year`s dividends ( = 0.233; p = 0.019), as well as between bank size and leverage ( = 0.209; p = 0.037). Table 5. Correlation matrix Dividend payout ratio Leverage Liquidity Previous year`s dividends Profitability Growth rate Bank size Dividend payout ratio 1 Leverage 0.0895 1 Liquidity 0.0397 -0.0103 1 Previous year`s divi- dends 0.6635** 0.0793 0.1049 1 Profitability 0.2218* -0.0147 -0.0371 0.1603 1 Growth rate -0.1225 0.1547 0.0750 -0.1343 -0.0061 1 Bank size 0.2906** 0.2089* 0.0763 0.2332* 0.4311** -0.0675 1 The results significant at 5% significance level are followed by * and at 1% significance level by **. NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 67 No. 1 / March 2021 20 A regression analysis was applied to determine the associa- tion between the explanatory and dependent variables. The testing of the research hypothesis has been done with the help of the random effects model. The obtained results are presented in the Table 6: The value of determination coefficient R 2 is 0.466, which means that 46% of the variability of dividend policy is ex- plained by the regression model, while the rest of the vari- ability is under the impact of other factors. The Adjusted R square is 0.431. The F statistic value is 13.525 (p-value = 0.000), which at the significance level of 5%, implies that the hypothesis of the existence of significant linear relation between dependent and independent variables is accepted. The values of the Durbin-Watson test of 2.153 imply that there is no autocorrelation. By observing the values (β coefficient, t value, and p sta- tistical significance) given in Table 6 on independent vari- able Profitability, a positive (β = 1.176), but not a statisti- cally significant impact on the dependent variable can be established (p = 0.396). On that basis, the hypothesis H1 is rejected. The obtained results are consistent with the re- search of Puspitaningtyas (2019). As expected, Liquidity shows negative, but like in the previous explanation, it is about the value that has no statistically significant impact on dividend payout. The hypothesis H2 is also easily re- jected. Leverage and growth rate of income on interest do not show a statistically significant impact on the depen- dent variable; thus, it comes to the rejection of hypotheses H3 and H4. Zameer et al. (2013) came to similar results in their research when it comes to these two independent variables. Unlike the previously mentioned ones, the vari- able Previous year’s dividends confirms the correctness of the hypothesis H5, according to which last-year dividend payouts have a statistically significant (p = 0.000), positive (β = 0.784) impact on dividend policy. The results speak in favour of using the previous year’s dividends as the type of signal of what potential investors can expect in the future. The value of p = 0.000 implies a highly statistically signif- icant relation, which is confirmed in numerous works of research (Hosain, 2016; Al-Kayed, 2017; Dewasiri et al., 2019). When it comes to the last examined determinant, the value of β = 0.061 and p = 0.210 do not show a statisti- cally significant impact of bank size on dividend policy, by which the hypothesis H6 is rejected. Table 6. Regression results Depended Variable Independed Variable β t - value Sig. Dividend payout ratio Profitability 1.1762 0.8519 0.3965 Liquidity -0.0216 -0.3828 0.7027 Leverage 0.0039 0.1694 0.8658 Growth rate -0.0439 -0.3391 0.7353 Previous year`s dividends 0.7844 7.8784 0.0000** Bank size 0.0612 1.2628 0.2098 R square (R2) = 0.4660; F = 13.5246 (p (F statistic) = 0.000); Durbin-Watson stat. = 2.153 Significance: * p ≤ 0.05; ** p ≤ 0.01 Biljana Jovković, Aleksandra S. Vasić, Jasmina Bogićević: Determinants of Dividend Policy: A Case of Serbia’s Banking Sector 21 Conclusion As the material basis in which the interest of various groups of corporate companies is being diffracted, the profit has been the subject of studies in economic theo- ries for decades. Researchers’ special attention is attract- ed by dividend policy, especially the one implemented by the companies that belong to the finance sector. This re- search focuses on determining dividend policy factors of the banking sector in the Republic of Serbia by applying the regression model of random effects. By a choice of six determinants (profitability, liquidity, leverage, growth rate of income on interest, previous year’s dividends, and bank size) whose impact on dividend policy was examined in the paper, we concluded that only one of them could ex- plain the movements of dividend payouts. Previous year’s dividends have shown as the most significant variable in predicting future dividend payouts. The positive impact of the last-year dividends offers support in defining dividend policy that should rely on previous payout patterns. A sta- tistically significant impact of profitability, leverage, divi- dend payout rate, and bank size on dividend policy has not been found in the paper. Deficiencies/limitations of the research. Research limita- tion refers to the size of the sample. In the observed period, a significant number of observation units were noticed that did not contain dividend payout; hence the recommendation is to expand spatial and time observation of variables to decrease mistakes and increase the correctness of findings. Practical implications and the directions of future re- search. Individual investors can benefit from the research to a great extent when choosing the banks whose shares they invest in. If an investor prefers companies that pay out dividends, the first indicator that they observe on that occasion is dividend payouts in previous years. Research results can also help bank management make efficient and reliable decisions on dividend payouts that, in the long- term period, could contribute to maximising profit and sat- isfying the needs of employees and shareholders. When it comes to future research directions, they should be direct- ed to the inclusion of additional variables into the model, such as ownership, risk, economic conditions, life cycle, investment opportunities, etc. Apart from this, the recom- mendation for future research refers to the observation of determinants of dividend policy in other finance compa- nies, such as insurance companies. References Agyei, S. K., & Marfo-Yiadom, E. (2011). Dividend Policy and Bank Performance in Ghana. International Journal of Economics and Finance, 3(4), 202-207. doi: 10.5539/ijef.v3n4p202 Aivazian, V., Booth, L., & Cleary, S. (2003). Do Emerging Market Firms Follow Different Dividend Policies From U.S. Firms? Journal of Financial Research, 26(3), 371-387. doi: 10.1111/1475-6803.00064  Al-Kayed, L. T. (2017). Dividend payout policy of Islamic vs conventional banks: case of Saudi Arabia. International Journal of Islamic and Middle Eastern Finance and Management, 10(1), 117-128. doi: 10.1108/imefm-09-2015-0102  Arko, C. A., Abor, J., K.D. Adjasi, C., & Amidu, M. (2014). What influence dividend decisions of firms in Sub-Saharan African? Journal of Accounting in Emerging Economies, 4(1), 57-78. doi: 10.1108/jaee-12-2011-0053  Baker, H., & Powell, G. H. (1999). Dividend policy issues in regulated and unregulated firms: a managerial perspective. Managerial Finance, 25(6), 1-20. doi: 10.1108/03074359910765975  Baker, H. K., Veit, E. T., & Powell, G. E. (2001). Factors Influencing Dividend Policy Decisions of Nasdaq Firms. The Financial Review, 36(3), 19-38. doi: 10.1111/j.1540-6288.2001.tb00018.x Banerjee, S., Gatchev, V. A., & Spindt, P. A. (2005). Stock Market Liquidity and Firm Dividend Policy. SSRN Electronic Journal. doi: 10.2139/ssrn.391663  Basse, T., Reddemann, S., Riegler, J.-J., & von der Schulenburg, J.-M. G. (2014). Bank dividend policy and the global financial crisis: Empirical evidence from Europe. European Journal of Political Economy, 34, 25-31. doi: 10.1016/j.ejpoleco.2013.09.001 Dewasiri, N. J., & Weerakoon Banda.Y.K. (2016). Why Do Companies Pay Dividends? A Comment. Journal of Corporate Ownership and Control, 13(2), 443-453. doi: 10.2139/ssrn.2699666 Dewasiri, N. J., Yatiwelle Koralalage, W. B., Abdul Azeez, A., Jayarathne, P. G. S. A., Kuruppuarachchi, D., & Weerasinghe, V. A. (2019). De- terminants of dividend policy: evidence from an emerging and developing market. Managerial Finance, 45(3), 413-429. doi: 10.1108/mf-09-2017-0331  Dibia, O. N. (2018). Determinants of Dividend Policy in Nigerian Banks. Archives of Current Research International, 15(2), 1-13. doi: 10.9734/ACRI/2018/44463 Dickens, R., Casey, K. & Newman, J. (2002). Bank Dividend Policy: Explanatory Factors. Quarterly Journal of Business and Economics, 41(1), 3-12. doi: 10.2307/40473341 Dragutinović Mitrović, R. (2002). Analiza panel serija. Beograd: Zadužbina Andrejević. NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 67 No. 1 / March 2021 22 Gordon, M. J. (1963). Optimal investment and financing policy. The Journal of Finance, 18(2), 264-272. doi: 10.1111/j.1540-6261.1963. tb00722.x  Hosain, Z. (2016). Determinants of the Dividend Payout Policy: A Study on Listed Private Commercial Banks of Dhaka Stock Ex- change Limited in Bangladesh. IOSR Journal of Economics and Finance, 7(5), 1-10. doi: 10.9790/5933-0705040110 Jabbouri, I. (2016). Determinants of corporate dividend policy in emerging markets: Evidence from MENA stock markets. Research in International Business and Finance, 37, 283-298. doi: 10.1016/j.ribaf.2016.01.018 Lintner, J. (1956) Distribution of Incomes of Corporations among Dividends, Retained Earnings, and Taxes. The American Economic Review, 2, 97-113. Lintner, J. (1962). Dividends, Earnings, Leverage, Stock Prices and the Supply of Capital to Corporations. The Review of Economics and Statistics, 44(3), 243-269. doi: 10.2307/1926397  Litzenberger, R. H., & Ramaswamy, K. (1979). The effect of personal taxes and dividends on capital asset prices. Journal of Financial Economics, 7(2), 163-195. doi: 10.1016/0304-405x(79)90012-6  Malinić, D. (1999). Politika dobiti korporativnih preduzeća. Beograd: Ekonomski fakultet Univerziteta u Beogradu. Marfo-Yiadom, E. & Agyei, S. (2011). Determinants of dividend policy of banks in Ghana. International Research Journal of Finance and Economics, 61, 99-108. Miller, M.H. & Modigliani, F. (1961). Dividend Policy, Growth, and the Valuation of Shares. The Journal of Business, 34, 411-433. doi: 10.1086/294442 National Bank of Serbia, Price and Financial System Stability. (2018). Annual report on activities and results in 2018. Puspitaningtyas, Z. (2019). Assesment of financial performance and the effect on dividend policy of the banking companies listed on the Indonesia stock exchange. Banks and Bank systems, 14(1), 24-39. doi: 10.21511/bbs.14(2).2019.03 Pallant, J. (2017). SPSS priručnik za preživljavanje. Beograd: Mikro knjiga. Shchurina, S.V. & Prunenko, M.A. (2018). The Dividend Policy for Banking Sector: Examples Of Russian And Chinese Banks. In: Se- lection and peer-review under responsibility of the Organizing Committee of the conference. Paper presented at the Proceedings of the International Scientific Conference Global Challenges and Prospects of the Modern Economic Development, 990-1000. doi: 10.15405/epsbs.2019.03.99 Yarram, S. R., & Dollery, B. (2015). Corporate governance and financial policies. Managerial Finance, 41(3), 267-285. doi: 10.1108/ mf-03-2014-0086  Zameer, H., Rasool, S., Iqbal, S. & Arshad, U. (2013). Determinants of dividend policy: A case of banking sector in Pakistan. Middle East Journal of Scientific Research, 18, 410-424. doi: 10.5829/idosi.mejsr.2013.18.3.12200 Zhang, J., Jiang, C., Qu, B., & Wang, P. (2013). Market concentration, risk-taking, and bank performance: Evidence from emerging economies. International Review of Financial Analysis, 30, 149-157. doi: 10.1016/j.irfa.2013.07.016 Dejavniki dividendne politike: primer bančnega sektorja v Srbiji Izvleček Dividendna politika je eno najbolj kontroverznih področij podjetniških financ. Prispevek predstavlja rezultate raziskave bančnega sektorja Republike Srbije. Zaradi specifičnih značilnosti finančnega sektorja je raziskovanje dejavnikov divi - dendne politike še bolj kompleksno. Cilj te študije je določiti dejavnike dividendne politike v srbskem bančnem sektorju za obdobje 2009-2018. Da bi preverili razmerje med dejavniki in izplačilom dividend, smo izbrali model slučajnih učinkov. Empirični rezultati so pokazali, da imajo dividende preteklih let občuten pozitiven učinek na dividendno politiko. Pričujo - ča raziskava bi lahko zelo koristila posameznim investitorjem in direktorjem bank pri oblikovanju dividendnih politik, ki bi pripomogle k ustvarjanju čim večjega dobička in zadovoljevanju potreb zaposlenih ter delničarjev na dolgi rok. Ključne besede: podjetniške finance, bančni sektor, dividendna politika, determinante dividendne politike, Srbija