The Role of Collateralizable Asset, Profitability and Operating Cash Flow on Dividend Policy: A Study on ASEAN Non-Financial Companies

1. INTRODUCTION

In the current era of globalization, there is much economic competition in the world market, which usually knows no borders. Therefore, consumers face different choices when pursuing market offers (Kudeshia, C. & Kumar, A. 2017). Moreover, the relatively developed ASEAN country has become a very fiercely competitive centre (Wong & Chan, 2003). ASEAN itself is a livelihood for Southeast Asian countries. ASEAN is also expected to bring economic improvements to the world (Cite here) and establish its free trade area, known as ASEAN Free Trade Area (AFTA). This is being done through a gradual cut in tariffs, which this agreement itself has put into effect since January 1, 2003.

These ASEAN countries are a solution for investors to make investments, as several countries in ASEAN countries are included in the list of emerging countries. These countries are experiencing rapid economic growth (Salim et al., 2019). Most of these countries are ASEAN countries and have distinctive features: such as a young population and a developing middle-class (Hughes & Woldekidan, 1994). According to a study by the United Nations Conferences on Trade and Development (UNCTAD), ASEAN is classified as a host country, meaning it is directly exported for investment by foreign states (ASEAN Secretariat, 2021). Given this phenomenon, countries in the ASEAN region should compete to provide profitable capacity for investors so that this can be a special attraction for companies in the ASEAN region. The ability to deliver a performance that reflects the company's success is demonstrated by its ability to pay its dividends to shareholders (Sukmawardini & Ardiansari, 2018).

For this reason, it is very important to examine this dividend policy as it can reflect how much capital and profit is distributed by the company (Murtaza et al., 2018). In addition, dividend policy is also a problem that occurs in every country based on economic factors or investment decisions to be made (Jabbouri, 2016). The dividend policy is still a mystery because this dividend policy is very concerned about their policy. Because it creates attraction, for example, from investors, who will surely see that a company can pay its dividends before investing. Dividend policy appears to be a very important factor in meeting investor expectations.

However, there is a problem with companies in ASEAN. These companies still have an inconsistent trend in paying dividends to their shareholders. This means that companies in the ASEAN region, such as PT Astra International Tbk from Indonesia, still have fluctuations from 2014 to 2019 that tend to rise and fall. XL Axiata Group Berhad, a company from Malaysia, Wilmar International Limited from Singapore, and Thailand's Siam Cement PCL, which has similar things, still has its average dividend-paying abilities ups and downs every year.

Currently, the integration of capital markets in the ASEAN country region can lead to reduced information (Aney et al., 2016), because if it is wrong to integrate a policy, it can be a problem for a company. As a result, these companies themselves are still confused when deciding whether to save profits for business development or pay them out as dividends (Ross et al., 2015). Therefore, dividend policy is an issue that continues to grow locally and globally. Because this distribution policy is also closely related to the decision to be made whether the annual profit generated by a company after tax will be distributed to shareholders or converted into retained earnings, which will later be used to finance future investments.

This dividend policy is often opposed by two stakeholders: both shareholders and the interests of the company itself, who want the net profit after tax to be converted into retained earnings. In Indonesia, companies that have increased their dividend are those that are active in the non-financial sector (Jannah et al., 2019). This non-financial company is able to pay out consistent dividends to shareholders from year to year.

There are several things that can affect a company's ability to pay dividends. First, this dividend policy is influenced by collateral assets owned by a company, according to (Wijaya & Yamasitha, 2020), (Granda Carvajal, 2015), and (Henrekson & Stenkula, 2017). There are also those who note that this dividend policy is not affected by collateral assets owned by a company (Wahjudi, 2020). In addition, the ability to pay dividends is also closely related to a company's ability to make a profit (Dewasiri et al., 2019). In addition to this ability to pay dividends, we can see the ability to operate cash flow owned by a company. Tijjani & Sani (2016) propose that to increase the ability to pay dividends, the company must first increase its cash flow operating.

Based on the above explanation, we will try to focus on collateralizable assets, profitability, and cash flow operations. This has been done because we are very interested in the phenomenon because, as we know, this collateral asset is an asset that can be pledged and so we want to see if this collateral can be covered by a dividend policy. Since it must be high collateral, it must be able to signal to the company that the company has a high asset value, making it the job of management not to be afraid to pay dividends.

A no less important point is profitability because this seems to be a very important factor in increasing the dividend. Because from the standpoint of the rule, if the company has positive retained earnings, it can pay dividends. This positive earnings record itself stems from the achievable profitability of the company, which is why we think it needs to be looked at and reconsidered. The last thing we want to learn is the impact of operating cash flow on dividend policy, as it can provide insight into the quality of a company's earnings, as operating cash flow reflects all of a company's business activities.

This study also uses several regression analyses using the econometric engineering approach of Dynamics Panel Data (DPD), called System GMM (Blundell & Bond, 1998). According to researchers, this approach will be much more flexible because it is based on an approach used in financial economics that requires few assumptions, often called moment conditions. The Sys-GMM method also allows a more detailed estimation of data that does not have confidence parameters.

2. LITERATURE REVIEW

Several theories have been developed to explain the reasons why companies pay dividends. One such explanation is signal theory, which proposes that companies pay dividends to signal information favorable to capital markets (Nguyen & Bui, 2019). According to (Ross et al., 2015) introduced signal theory as a theory that describes a signal as an action performed by senior management. This signal is in the form of information about what management has done to achieve the principal's wishes, which may be reflected in the company's financial statements as a form of accountability from the agent to the principal.

When a company experiences a decline in earnings, this indirectly influences the determination of the dividend policy to be awarded. Failure to adopt an appropriate dividend policy may reflect the company's management skills less favorably, sending a bad signal to investors that performance will decline and the company will not generate profits for investors.

According to (Murtaza et al., 2018), a dividend policy is a policy that organizations use to decide how much profit to pay out of shareholder profits. This dividend policy also takes various forms, namely the presence of a managed policy. This dividend policy is also characterized by the tendency to distribute profits from the profits made (Baker et al., 2019).

A company's dividend policy is all about how the company uses after-tax profits, as this is highly correlated with dividends paid to shareholders. The company must decide to use the profits generated as a source of financing or pay them out to shareholders in the form of dividends. If paid out, the company must be careful in determining the percentage that will be generated. There is another alternative that can be used by companies, which is that companies can use these profits to invest in new projects that are expected to add value to the company. This is supported by the statement of (Mamduh & Abdul, 2016), which states that dividend policy refers to the profit the company generates at the end of the year, which must be paid out to shareholders in the form of dividends or withheld as retained earnings for reuse as financing for future investments.

This dividend policy stems from the scope of matters to be decided by both the board of directors and shareholders. This problem often arises when companies determine their dividend policy. because this influences the assessment of investors and the performance that a company will deliver. This dividend policy is reflected in the Dividend Payout Ratio (DPR) (Gitman et al., 2015), the percentage of profit that is paid out in cash. This can be used as a reference to see a company's calculation when providing guidelines for its dividends.

According to some researchers, this dividend policy is closely related to the amount of collateral assets owned by a company, as reported by (Henrekson & Stenkula, 2017) and (Wahjudi, 2020). However, these two things have opposing influences when hedgeable assets positively or negatively influence a company's dividend policy.

Collateralizable assets is the amount of an asset owned by a company that can later be used by a company as collateral for creditors when (Darmayanti & Mustanda, 2016) borrow. Collateral assets are also fixed assets owned by a company and can be used as collateral for loans to creditors (Ross et al., 2015).

This is very useful for investors because creditors need this guarantee as collateral when making a loan. Since the funds to be allotted by the creditors are usually of high value, they need this asset as a handle for the creditors to reduce the risk of default. Therefore, companies with collateral assets are likely to be good companies because they have a low liability risk for their debts. With this high value, creditors don't have to worry about this company failing to meet their obligations later on.

Wahjudi (2020) examined the impact of collateral assets on the company's dividend policy. Here he found that hedgeable assets have a negative but insignificant impact on dividends. But (Granda Carvajal, 2015), and (Henrekson & Stenkula, 2017) found several things that they found to have a positive and significant impact of Hedgeable assets on a company's dividend policy. For this reason, since there are still differences, this should be reconsidered. As a guide or evidence of the impact of collateralised assets on the dividend policy itself.

From the explanation and opinions above, it can be concluded that the influence of this collateralizable asset on dividend policy lies in the high sense of trust felt by the company if it has a high level of collateralizable asset, with which it will not hesitate to distribute dividends to shareholders. Because with the high ownership of collateralizable assets, can convince the company not to hold profits but to distribute profits to shareholders.

H1: Collateralizable assets have a positive significant effect on the dividend policy in a company.

This dividend policy is also closely related to a company's profitability ratio. Profitability is a metric used to measure a company's ability to generate profits. The profitability ratio is a measure used to assess the company's ability to generate profits (Kasmir, 2017). Profitability can also be used as a measure of comparison between the various financial components of the statement regarding the company's ability to generate profits.

This ability to generate profits is closely related to a company's equity or dividend policy. The higher the company's ability to generate profits, the higher the dividends that will be paid by a company, and vice versa, if the company's ability to generate profits is not good, the dividends generated by a company will not be good. That brings profitability with a dividend policy. When measuring profitability, you can use the return on investment, which is one of the indicators of the profitability ratio, which evaluates the company's ability to generate profits (Kasmir, 2017).

This profitability measure is also a crucial factor in determining the dividend policy of companies (Dewasiri et al., 2019). In their research, profitability is the most important determinant in the decision-making about dividend policy. With a high profitability, the high and large dividend that is given has an impact. Similarly, in their study, (Tekin & Polat, 2021) found that this profitability has a negative effect on a company's dividend policy, where profitability can also reduce the dividends a company has to pay. However, in this study, we argue that this profitability measure will have a positive impact on a company's dividend policy, which is supported by research by (Wahjudi, 2020).

This profitability relationship is very close because we know that the payment of dividends is based on the company's ability to generate profits. With these profits, the company can generate funds which are later distributed to the shareholders. The company becomes indirect. This will benefit the company by generating high confidence from shareholders so that the company will receive positive things for the growth of the company itself. This profitability can therefore have a positive influence on a company's dividend policy.

H2: Profitability has a positive significant effect on the dividend policy in a company.

Looking at the previous explanation, this profitability can be used as a signal to shareholders when determining a company's dividend policy. There is something very interesting about determining a company's dividend policy. So if we look at the quality of a company's profits, we can see the company's operating cash flow. This is because this operating cash flow reflects the company's ability to regulate the inflows and outflows of its business activities, indirectly reflecting the company's activities in its business operations.

(Suhardianto & Kalanjati, 2014) mentioned that cash flow operating is a statement that shows the cash inflows and outflows of operating activities over a period of time. The relationship between cash flow operating and dividend policy itself has received mixed responses from several researchers, such as (Supardi, 2018), who found that this operating cash flow negatively affects a company's dividend policy. Because the presence of high operating cash flow can reduce the specified dividend policy.

But other things were discovered by (Tijjani & Sani, 2016), who argue that operating cash flow positively and significantly impacts the dividend policy of oil and gas companies in Nigeria. As a result, it can be argued that increasing the cash flow from such operations can increase the payment of dividends. In addition, cash flow from operating activities is a very critical variable in dividend policy, as companies that can pay dividends are better cash flow companies. This is also supported by research by (Ifada & Kusumadewi, 2014) and (Hidayat, 2019), who argue that this operating cash flow positively affects a company's dividend policy.

From the explanation, we can understand that the cash flow operation is something that describes all the operating activities of a company, and this cash flow operation, it can influence the dividend policy that the company will follow because if the company can generate positive cash flow operations, it will generate a signal to cash flow controllers that the company can easily determine its dividend policy. Because the company is not dependent on external financing.

H3: Cash flow Operating (CFO) has a positive significant effect on the dividend policy in a company.

3. METHODOLOGY

In this study, researchers used data obtained from the OSIRIS database, which processed data from each stock exchange in the country we studied. In this study, the researchers used three independent variables: collateralizable assets, profitability, and operating cash flow. In contrast, the dependent variable was a dividend policy. as well as overcoming the unfamiliarity of researchers using several control variables, namely the current ratio, debt to equity ratio, and growth in net assets. In this study, the researchers used three independent variables: collateralizable assets, profitability, and operating cash flow. In contrast, the dependent variable was a dividend policy. as well, to overcome the unfamiliarity, researchers used several control variables, namely the current ratio, debt-to-equity ratio, and growth in net assets.

The data collection method in this study is carried out through documentation studies, namely by reviewing various literature in the form of literature, journals, previous research, and published reports to get an overview of the problems to be studied and through secondary data in this study. In the form of reports published by stock exchanges in the ASEAN Region. In this study itself, the object of the study was a non-financial company listed on the stock exchange in ASEAN countries for the 2014-2019 period. The data used is selected with the following criteria:

1. Non-financial companies listed on stock exchanges in ASEAN 2014 - 2019.

2. Non-financial companies in ASEAN countries registered as 30 emerging market countries 2014-2019.

3. Non-financial companies that are able to generate profits successively for the 2014-2019 period.

Based on the results of the criteria above, there are 367 companies that have successfully passed the criteria within a period of six years; therefore, our dataset comprises 2,202 firm annual observations.

NO Keterangan Description Formula Satuan
1 Dividend Policy Dividend policy is a policy used by an organization in deciding how much profit will be paid from the profits earned to shareholders (Murtaza et al., 2018). This dividend policy is reflected by the Dividend Payout Ratio (DPR) (Gitman et al., 2015) DPR= (Dividen Per Share (DPS))/(Earning Per Share (EPS)) (Gitman et al., 2015) Rasio
2 Collateralizable Assets Collateralizable Assets is the amount of assets that can be guaranteed by the company to creditors (Wahjudi, 2020). So, the higher the assets that can be used will reduce the conflict of interest between shareholders and creditors. Collas= (Total Asset Tetap)/(Total Assets) (Wahjudi, 2020) Rasio
3 Return On Assets Return on assets is a ratio that is often used in looking at a company's ability to make profits, which is also used in looking at the good or bad financial performance of a company. Return on assets is also known as the return on investment, which is seen from the results on the amount of assets used in a company (Kasmir, 2017). ROA = (Net Profit)/(Total Assets) (Kasmir, 2017) Rasio
4 Cash Flow Operating According to N Suhardianto, Devi S (2014) Cash Flow is a report that reports cash inflows and major cash outflows from a company over a given period CFO =(Brigham & Houston, 2010)CFO = (Operating Cash Flow)/(Total Assets) (Brigham & Houston, 2010)
5 Growth In Net Assets Growth in net assets is the growth that occurs in the number of assets owned by a company. This asset growth is also a ratio that can indicate asset growth, where assets are located assets used for the company's operational activities (Riyanto, 2013) NAST_G=(Total Assets (t) - Tottal Assets (t-1))/(Tottal Assets (t-1)) (Wahjudi, 2020) Rasio
6 Current Ratio According to (Brigham & Daves, 2018), the current ratio is a ratio that will indicate the company's ability to cover current liabilities compared to the company's assets, which is expected to be converted into cash in the near future. The current ratio is calculated by dividing asset flows by current liabilities. CR = (Current Assets)/(Current Liabilities) (Brigham & Houston, 2018) Rasio
7 Debt To Equity Ratio Debt to equity ratio is the ratio used in looking at the ratio of the amount of debt to equities held by a company. (Wahjudi, 2020) argues that this debt to equity ratio is the ratio of total debt to equities, which will show a company's ability to pay off all its debts. DER = (Total Liability)/(Total Equity) (Gitman et al., 2015) Rasio
Table 1.Description of Variables

The analysis method used in this study is to conduct a quantitative analysis expressed by numbers that in their calculations use statistical methods assisted by static data processing programs. The method used is descriptive analysis and panel data modeling, which is a combination of data between cross-section and time-series data. From this data, there are two data models, namely static data models and dynamic data models.

Here is an originality of the dynamic panel data model that can solve the problem of endogeneity related to the use of lag of dependent variables, where in static data models the use of lag is more likely to use the dependent variable lag so as to cause estimation. So the researcher in this case wants to use dynamic panel data, which becomes biased due to a lack of consistency and reliability, so the researcher in this case wants to use dynamic panel data. The first step in estimating unknown parameters in this dynamic panel data regression model can be done by utilizing the orthogonality conditions contained in the lag value and error value, so that the dynamic panel data method becomes the Arellano and Bond model. Dynamic panel data is very popular in handling short panel structures (N<T) (Roodman, 2009).

In addition, dynamic panel data is often used as a specification for linear regression where there is a persistence property that will be caused by bound variables, the presence of endogeneity in explanatory variables, the presence of fixed effects on cross-sectional and the presence of autocorrelation and heterochedasticity in intra-cross-section units. For this reason, in other words, the dynamic data model is quite generalized. Due to the large nature of economic data, this data is autoregressive.

3.1. Emperical Model

From the explanation above, the idea of using the GMM System is to estimate the system of equations either at the differentiation that occurs first or the level where the instrument used at that level is the lag first difference of the series. We estimate that variables from COLLAS, ROA, and CFO may be endogeneous because they are defined and measured from variables of interest to dependent variables. According to (Blundell & Bond, 1998) say that it is important to take advantage of the initial conditions in producing efficient recovery of dynamic panel data models when the T value is small. For this reason, the models that will be analyzed in this study are as follows:

(1)

Information :

DPR : Dividend Payout Ratio as dependent variable

a : Constanta

COLLAS : Collateralizable Assets as independent variable

ROA : Return on Assets as independent variable

CFO : Cash Flow Operating as independent variable

NAST_G : Growth in Net Assets as independent variable

CR : Current Ratio as control variable

DER : Debt to Equity Ratio as control variable

𝛽 : Beta

u : Error reggresi data panel

v : Cross section residual

e : Error

The analysis method using the GMM system requires or requires multiple model specifications to provide valid and consistent results where there is no serial correlation with errors. As a first test, the feasibility of the model is examined, which can be seen from the values of AR (1) and AR (2). The most important thing to see is the AR value (2), where this value must reject the null hypothesis. The next test is the Hansen test, or Sargan test, which is used to determine the validity of the instrumental variables used as a whole. The feasibility of the model of this test itself is that if the chi-square probability value has a significant range from 0.1 to 0.9, it obtains H0 or can be interpreted as validating the instrumental variable used.

4. RESULTS / ANALYSIS

In the first test, we first tested using the panel data method and then tested using the OLS, Fixed Effect, and Random Effect methods. For testing, we first show the results of the best descriptive statistics, then we show the results of the tests using the GMM system.

4.1. Descriptive Statistics

Table 2 shows descriptive statistics used in the study, where the statistics include the number of observations, means, standard deviations, minimum, maximum, and percentiles (1%, 5%, 95%, and 99%) in the study. In this table, we focus on explaining the variables to be studied as well as the variables of interest (VIR) in this study itself.

For this reason, it can be seen from Table 4.1 that from 2,202 observations we can know that the 2014-2019 DPR has a mean value of 39.34% with a standard deviation value of 42.84%, the minimum value of the data is 0%, and the existing maximum value is 238.8%. Meanwhile, COLLAS itself has an observation value of 2,202 with a 2014–2019 mean value of 52.53%, with a standard deviation value of 20.51%, a minimum data value of 21.54%, and an existing maximum value of 91.68%. Secondly, for the second VIR variable, namely ROA, with an observation of 2.202, with a 2014-2019 mean value of 8.56%, with a standard deviation value of 7.66%, a minimum data value of 1.52%, and an existing maximum value of 44.64%. And in the third VIR, the CFO with a total observation of 2,202 and a 2014-2019 mean of 14.72%, with a standard deviation value of 11.64%, the minimum value of the data is 3.12% and the existing maximum value is 58.28%.

DPR COLLAS ROA OCF NAST_G CR DER
Obs 2,202 2,202 2,202 2,202 2,202 2,202 2,202
Mean 39.345 52.540 8.563 14.721 9.703 2.221 113.662
Standard Deviation 42.841 20.515 7.661 11.645 14.789 1.648 106.549
Min 0 21.54 1.52 3.12 -4.71 0.85 25.04
Max 238.8 91.68 44.64 58.28 89.76 10.29 710.32
1% 0 21.54 1.52 3.12 -4.71 0.85 25.04
5% 0 21.54 1.52 3.12 -4.71 0.85 25.04
95% 109.58 86.73 21.59 37.64 33.78 5.26 303.94
99% 238.8 91.68 44.64 58.28 89.79 10.29 710.32
Table 2.Descriptive Statistics

This table reports descriptive statistics, statistics reported are number of observations, mean, standard deviation, min, max and percentiles (1%, 5%, 95%, and 99%). All variables are in percentage terms.

4.2. Regresi Data Panel

After performing descriptive statistics, we first process panel data. The results can be seen in Table 4.2, which is the result of regression using Ordinary Least Square (OLS), Fixed Effect (FE), Random Effect (RE) and their conditions. The regression results indicate that this study is aimed at FE users, as the Chow test shows that for this model the p-value is 0.000, which is a value less than alpha 0.05. It can be concluded from this, that when comparing the OLS model and FE, the best model is FE.

In addition, we want to compare FE with RE, this test is often called Hausmann test, where this test is used to find out which model is better between FE and RE, the results of Hausmann test itself show a significant value of 0.000 , which is a smaller value than the alpha value of 0.05, which can be interpreted as the best model from this study being one that uses FE.

From the results of the regression using the FE model, we found that only ROA and CFO influence DPR owned by non-financial companies, but the influence exerted is a negative influence. For this reason, we suspect that this study is an endogeneity issue, so we conducted a study using the GMM system method.

VARIABLES OLS FE RE
Coeff SE Coeff SE Coeff SE
COLLAS 0.170*** -0.057 -0.004 -0.143 0.189** -0.080
ROA 0.635*** -0.133 -1.142*** -0.271 0.021 -0.182
CFO -0.109 -0.101 -0.609*** -0.182 -0.362*** -0.129
NAST_G -0.390*** -0.061 -0.057 -0.056 -0.168*** -0.055
CR 3.142*** -0.647 0.676 -1.054 2.495*** -0.804
DER 0.032*** -0.009 -0.021 -0.020 0.0149 -0.013
Constant 19.74*** -4.123 59.85*** -9.587 28.97*** -5.695
Observations 2,202 2,202 2,202
R-squared 0.04 0.036
Number of ID 367 367
Chow Test 5.05*** 0.0000
LM Test 721.7*** 0.0000
Hausman Test 142.4*** 0.0000
Table 3.Regresi data panelStandard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

4.3. System GMM

With the previously explained explanation of where in this study there is a possibility of bias and this study proves that this study has dynamic data, it is proven by containing the coefficient value of the lag variable which has a statistically significant value of 0.010 in Table 4 For this reason, this study uses the Generalized Method of Moments (GMM) to solve problems caused by dynamic data panel models. The results of the regression using the GMM system, the results of this study itself, are shown in Table 4.

Table 4 explains some of the best model specifications used in this study. In model 1, the performed conditions assume that all variables are endogenous; in the 2nd model condition, all independent variables are assumed to be endogenous variables, but we add the lag of DPR as endogenous variables. However, of these two models, they still failed the validity test; model 1 itself failed the validity test itself because it has a Hansen value of 0.094, less than the required 0.100. And the AR value (2) of model 1 is still below a significant 0.100, indicating an autocorrelation problem. This is the same problem encountered in Model 2; although the AR value (2) is above the significance value of 0.100, but the Hanse test value is still below the estimated significance value.

Since the problems of models 1 and 2 still passed the validity test, we went through several ways to find the best estimate. Finally, we find the estimate for model 3, assuming that the endogenous variables themselves are COLLAS, ROA and the CR control variable. In contrast, the exogenous variable is a CFO variable with one of the DER control variables.

To ensure the appropriateness of using method 3, we have again subjected this method to a plausibility check. Validity checks can also use two things, namely the Hansen test, which is useful as a test to ensure that there is no endogeneity, which used to be a big problem. From Hansen's own results, shown in Table 4.3, he himself had a p-value of 0.155 in this study, the value of the Hansen test in this study being greater than the 0.100 requirements. In addition, for autocorrelation tests with Arellano binding AR(2) tests under the condition that the p-value > 0.100, this study also found that AR(2) has a higher value than the required p-value. This proved or supported that all model specifications are dynamic and make the proposed dynamic model stable.

The results of the study itself showed that hedged assets, return on assets and operating cash flow, taken together, have a significant impact on dividend policy, as the p-value is < 0.010, which is a significant value greater than 1 %. Meanwhile, in the first VIR, we initially had a hypothesis (H1) that these collateral assets would have a positive and significant impact on the dividend policy. However, upon investigation, we have determined that these assets that can serve as collateral have no statistical significance, regardless of whether they have a significance of 1%, 5% or 10%. For this reason, this study has identified a rejection of (H1) for which this hedged asset will have no impact or impact on the dividend policy. For this reason, we found in our study that the previous hypothesis is rejected, indirectly supporting the research of (Wahjudi, 2020).

For the second VIR, namely ROA, we found the same with some previous research opinions and previously developed hypotheses. And what we had initially (H2) was that there was a significant positive effect of profitability using ROA proxies on dividend policy. In our research, we found that the ROA has a significant level that is lower than the significance level of 1%. This encourages us to accept hypothesis (H2). The reported coefficient percentage was positive at 5.227. Because the influence of ROA means that you can increase a company's dividend policy. And the opposite will also happen: if ROA falls, a company's dividend policy will also fall; this is also in line with the previous hypothesis that ROA can have a significant positive effect on dividend policy. This supports a statement by (Dewasiri et al., 2019) stating that ROA also has a positive and significant impact on dividend policy.

Furthermore, the third VIR, namely the CFO that we discussed earlier (H3), states that this CFO will have a positive and significant influence on the dividend policy. However, upon examination, it appears that this CFO refutes our previous hypothesis, as according to the results of our research, this CFO has a significance below 10%, with a coefficient value of -1.639. In this case, this CFO appears to have a significant but negative influence on the dividend policy. This, therefore, supports (Supardi, 2018) statement that this CFO will have a significant negative effect on the dividend policy. Because having the senior CFO owned by a company reduces the dividend policy that a company will pursue. The results of this discussion can be seen in Table 4 below:

VARIABLES Model 1 Model 2 Model 3
Coeff SE Coeff SE Coeff SE
DPR (-1) 0.392*** 0.168 0.158 ** 0.078 0.531*** 0.154
DPR (-2) -0.385 0.253 -0.046 0.071 -0.229 0.211
COLLAS 0.270 0.673 1.014 0.658 1.028 0.666
ROA 4.597 *** 1.412 3.544 *** 1.344 5.227*** 1.567
CFO -0.177 0.868 -0.634 0.802 -1.639* 0.860
NAST_G -0.559 *** 0.193 -0.280 ** 0.121 0.086 0.306
CR 2.690 3.066 5.841 * 3.106 -1.467 7.048
DER 0.071*** 0.028 0.061 *** 0.020 0.008 0.033
Constant -19.597 46.772 -56.274 43.747 -44.940 47.615
Observations 1,468 1,468 1,468
Number of ID 367 367 367
F Stat - p Value 48.35*** 0.000 55.50*** 0.000 59.71*** 0.000
Hansen - p Value 23.46 0.094 0.000 55.69 28.65 0.155
AR(1) stats - p Value -4.130 0.000** -4.990 0.000** -4.17 0.000**
AR(2) stats - p Value 1.930 0.099 0.490 0.625 1.54 0.124
Table 4.Regresi Data Panel DinamisStandard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

4.4. Robustness check

To bolster this research, we attempted to conduct a robustness test necessary to demonstrate the impact of each VIR we use. Here we do it by replacing and nominating some existing VIRs. In the testing process, we performed three experiments; in the first experiment, we tried to use COLLAS and ROA only as VIR; in the second experiment, we used COLLAS and CFO as VIR; and in the last experiment, we tried to use VIR ROA and CFO Die Results. We received, we thought the results of our analysis were strong enough, as shown in the table below:

VARIABLES Model 1 Model 2 Model 3
Coeff SE Coeff SE Coeff SE
DPR (-1) 0.519*** 0.193 0.475*** 0.224 0.781*** 0.249
DPR (-2) -0.357*** 0.198 0.049 0.181 -0.147 0.273
COLLAS 0.239 0.508 -0.078 0.466
ROA 3.795*** 1.175 5.766*** 1.565
CFO 1.644** 0.760 -1.024** 0.424
NAST_G 0.120 0.271 0.400 0.327 0.032 0.346
CR_W 2.919 2.331 0.607 2.691 -3.248 6.064
DER_W 0.051* 0.030 0.007 0.023 0.013 0.031
Constant -25.048 35.804 -8.167 30.255 -13.864 23.311
Observations 1,468 1,468 1,468
Number of ID 367 367 367
F Stat - p Value 64.33*** 0.000 80.77*** 0.000 91.51*** 0.000
Hansen - p Value 23.46 0.102 21.23 0.170 22.92 0.116
AR(1) stats - p Value -4.38 0.000** -3.32 0.000** -3.29 0.009**
AR(2) stats - p Value 1.93 0.154 0.49 0.625 1.11 0.268
Table 5.Robustness Sequential Inclusion Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

On several alternatives we used, we found that COLLAS still had no significant impact on dividend policy. The first, which we tried to replace with VIR COLLAS and ROA, found from the results of our analysis that the significant value of COLLAS was even greater than the alpha value, be it 1%, 5% and 10%. However, unlike ROA, which has been shown to still have a significant positive impact, which has a significant value of less than 1%, this reinforces that this ROA has a strong influence on the value of the company's dividend policy.

Then we tried to create an alternative by making COLLAS and CFO VIR; again, we found that COLLAS had no significant influence on the dividend policy. However, the value of the CFO is consistent with the previous analysis that this CFO has a significant influence on a company's dividend policy.

We have tried to reinforce the last step by using ROA and CFO as VIR in an alternative way, and the result can be seen in Table 4.4 that both ROA and CFO influence the dividend policy. Both agree with the first analysis we conducted that ROA will individually have a positive effect on the dividend policy and that CFO will individually have a negative effect on the dividend policy to be pursued.

Before concluding the results found, we also did not forget to pay attention to Hansen's test and autocorrelation. From this, we see that no significant changes have occurred, the results of which are almost the same as the first analysis. But only changes in the coefficients are common due to the number of variables used.

5. DISCUSSION

The regression results indicate that this study is aimed at FE users, as the Chow test shows that for this model, the p-value is 0.000, which is a value less than alpha 0.05. It can be concluded from this that the best model is FE when comparing the OLS model and FE. This study proves that this study has dynamic data, it is proven by containing the coefficient value of the lag variable, which has a statistically significant value of 0.010 in Table 4 For this reason, this study uses the Generalized Method of Moments (GMM) to solve problems caused by dynamic data panel models. The results showed that hedging assets, returning assets, and operating cash flows had a significant influence on dividend policies; however, after being investigated, we determined that assets that could be used as a guarantee had no statistical significance. For this reason, this research has identified the rejection (H 1) where this hedging asset will not impact dividend policy.

For this reason, we found in our research that the previous hypothesis was rejected, indirectly supporting the research. In our study, we found that ROA had a lower significance level than a significance level of 1%. This encourages us to receive hypotheses (H 2). This supports a statement from(Dewasiri et al., 2019) which states that ROA also has a positive and significant effect on dividend policies. Furthermore, this study found that this CFO had a significant but negative influence on dividend policy. Therefore, this supports the statement that this CFO will significantly negatively influence dividend policies. Because it has a senior CFO owned by the company, it reduces the dividend policy that the company will take(Supardi, 2018).

6. CONCLUSION

This study examines the impact of collateralizable assets, profitability and cash flow operating on dividend policy. The researchers use multiple control variables to ensure that this study is unbiased. This study itself shows that collateral assets do not affect a company's dividend policy. The opposite can now be observed, that profitability and operational cash flow have a positive and negative impact on the dividend policy of non-financial companies. Profitability influences the positive impact, while the operating cash flow determines the negative impact. With the research we conducted, we hope these companies can maintain the value of their profitability so that they can adopt a dividend policy that benefits the investors or the company itself. For cash flow operating itself, we think the negative impact of this dividend policy is due to using high cash flow for other things, so this may reduce the dividend policy that these companies should be able to manage their operating cash through the provision of shareholder benefits by using this high cash flow for dividend payments. Hopefully, this can serve as a reference or guide for companies outside the financial sector itself to do further analysis for their companies so that companies operating in the non-financial sector can do better. And if new researchers want to re-examine, consider extending the year and adding a more recent year so that the research is more current and can replace or find other variables that may influence dividend policy determination.