The Determinants of Capital Structure of Listed Retailing Sector Companies on Indonesia Stock Exchange (IDX)

Introduction. Capital structure must be the concern of the company's financial managers in expanding and financing the company's operations. The financing decisions that will be used must have strong implications for the value of the company in the future. The data used in this study were secondary data obtained from the Indonesia Stock Exchange website and from the HOTS PT application. Mirae Aset Sekuritas as well as from the official website of each company in the researched samples. The researched population was listed retail ing sector companies on the Indonesia Stock Exchange (IDX) over the period of 2012-2016. The number of companies used was 18. The data analysis used panel data regression on the fixed effect assumption. Purpose. The purpose of this study was to conduct an analysis and obtain empirical evidence of the effect of the Profitability, Tangibility, Firm Size, Growth Opportunity, Liquidity, Business Risk and Non -Debt Tax Shield variables on the capital structure. Results. The results of the study showed that the capital structure (leverage) was influenced by variables of profitability, firm size, liquidity and non debt tax shields (NDTS). Variables of assets tangibility, growth opportunity and business risk did not affect the leverage. Conclusion. The company finance managers should prioritize the internal funding sources and need to be thoroughly in utilizing external funding. The results of the analysis showed that the Pecking Order (POT) theory was more influential than the Trade-Off (TOT) theory. Internal financing sources were preferred to avoid high loan interest rates in Indonesia, foreign exchange risk losses for loans in foreign currencies, government regulations that limit the amount of interest costs that can be charged as tax deduction fees and the Indonesian economy which is vulnerable to global influences.

Problem statement. The hypothesis publication of Modigliani and Miller (1958) was considered as a research milestone in capital structure and its effect on company value. After that, many researches have been conducted to determine the effect of determinants of capital structure on the selection of capital structures and their effect on company value. Some theories that have become references in further research include Jensen and Meckling (1976) known as "theory of agency cost", Breadley et al. (1984) was known as "static trade-off theory", and Myers and Majhluf (1984) suggested"pecking order theory" (Quan, 2002). Trade-off theory (TOT) and pecking order theory were the two most prominent theories.
The retailing sector is one of the fastest growing sectors in Indonesia. Factors driving the growth of the retail sector are Indonesia's demographic structure dominated by young people who will increase the number of productive workforce with disposible income and increasing needs, changes in lifestyle and consumption patterns of the Indonesian people, the number of middleincome people who continue growing, the incessant campaign of non-cash transactions and the development of entertainment combined with primary needs.
The developing companies need large funds. There are two funding sources, internal and external. Internal funding sources come from company profits and from external sources of debt and the issuance of new shares. Managers must choose the best funding source to avoid financial distress.
Researches conducted in several countries showed inconsistent effects of several variables on capital structure. Variables that still show inconsistent results include: profitability, assurance, firmsize, growth opportunity, liquidity, business risk and non debt tax shield.
Test Results.The research samples were selected by purposive sampling method because the retailing sector of companies did not always submit annual reports. The criteria for selecting samples were as follows (table 1). The list of companies met the criteria as follows (table 2). The number of companies met the criteria were 18 companies with an observation period (t)=5 years (2012-2016) and the number of companies (n) = 18 so that the total number of observational data becomes 90 units of data. From those units of data, 1 (one) unit was expelled, data from PT Sona Topas Tourism Industry Tbk in 2016, because the value of the business risk variable (X6) was too large, so the data observed becomes 89 unit of data.
This study used panel data regression with the following equation (1).
The variables used in this study were independent variables and dependent variables. Independent variable is a variable considerd to affect other variables. The independent variables in this study were profitability, assurance, firm size, growth opportunity, likquidity, business risk and non debt tax shield. Dependent variable was capital structure (leverage).
Descriptive statistics aimed to provide an overview or description of the data collected in the study. The results of the disk statistics are as follows (table 4). Panel fund regression have 3 (three) models, namely Common Effect (CE) or Pooled Least Square (PLS), Fixed Effect (FE) and Random Effect (RE). The common effect model (CE) is the simplest panel data regression approach, only integrates time series data and cross section data without regard to time dimensions and individual entities or it is assumed that the behavior of company data is the same in various time periods. The Ordinary Least Square (OLS) approach is commonly used in the CE model. The fixed effect model (FE) is a panel data regression model considering the differences among individuals can be accommodated from the difference in its intercepts. The fixed effect model uses dummy variables to capture intercept differences between individuals so that it is often called the least square dummy variable (LSDV). The difference in interception is caused by managerial, intensive and work culture.
The third panel data regression model is the random effect (RE) model. It carries out panel data estimation assuming that interfering variables may be interconnected between time and between individuals. In this model the difference in intercepts is accommodated by the error terms of each company.
To determine the best model whether it is common effect, fixed effect or random effect, a Chow Test was performed (choosing between the best CE and FE models), Hausman Test (choosing the best model between FE and RE) and Langragge Multiplier Test (choosing the best model between (CE and re). The Chow Test results are as follows (table 5).  The P-value (Probability)> 0.05, it can be concluded that the research data was normally distributed.
The symptoms of autocorrelation in Eviews 9 tested by looking at the AC-PAC plot. The results were as follows ( fig. 1

Source: processed data
The results of the ACF and PACF plots showed that no margin appeared except the lag-3, so this condition can be concluded that the residual model was not autocorrelated because most ACF-PACF plots were within the error margin line.
Heteroscedasticity test. Tests were carried out with the White Test as follows (table 8). From the output results in the t-test, all P-values (Sig.) were obtained greater than 5% (not significant). On the other hand in the F-test, it was also obtained that the Pvalue (Sig.) was greater than 5% (not significant). Thus, the results of this test concluded that the Residual model was not Heteroscedaticity.

Multicollinearity
Test. For Multicollinearity assumptions, VIF values were tested from the estimation results of the regression model predictor X1 to X7 on the response variable of Y (table 9). Source: processed Secondary Data Table 9 above shows that all independent variables had VIF values <10 so that there were no symptoms of multicolineirity.
From the results of the classical assumption of the regression model test, it was concluded that the model had passed the classic assumption problem. Furthermore, the panel data regression model was estimated on the fixed effect assumption. The results of testing with the EViews 9 program are as follows (table 10).

Simultaneous Testing on Effect Hypothesis.
To test the simultaneous effect of independent variables on the dependent variable, F. Test was conducted. Significant effects simultaneously when Pvalue ≤ α with α determined at 5%.
Based on the output results obtained F-statistic value of 133.1872 with Prob (F-statistic) of 0.000000 <5%) so it can be concluded that variables X1, X2, X3, X4, X5, X6, and X7 simultaneously had a significant effect on Y. Simultaneously these variables were able to explain 97.70% diversity of leverage variables and the remaining 2.30% explained by error variables or other variables not included in the model.

Partial / Individual Testing on the effect Hypothesis.
Testing the effect of partial / individual independent variables was done through testing the partial coefficient of the Test model where the conclusion of significant effect of a variable was obtained if the P-value ≤ α with α determined at 5%. H1: Profitability variable had a significant effect on capital structure.
The effect of X1 on Y with a regression coefficient (coefficient) of -5.221720 (negative influence) and P-value (Prob.) of 0.0000 so it can be concluded that there was a significant effect of X1 on Y which was -5.100254. In other words, if profitability increases by 1 unit, then leverage will decrease by 5,100254 units, and vice versa if profitability decreases by 1 unit, then leverage will increase by 5,100254 units. Therefore, H1was accepted and it can be concluded that the profitability variable had a significant effect on the capital structure. The results of this analysis were in accordance with the pecking order theory.
H2: Assettangibility variable had a significant effect on capital structure.
The test result of the effect of variable X2 partially on Y produced a regression coefficient of -0.004107 and Pvalue (Prob.) of 0.9925 greater than 0.05 (0.9925> 0.05), so it can be concluded that there was no significant partial effect of variable X2 on variable Y. The results of the test contradicted with the pecking order theory (POT) and trade-off theory (TOT), both predicted that positive effects caused by large assets available as collateral.

______________________________________________________________________________________ 31
Bank lending rates are quite high in Indonesia make retailing sector companies listed on the IDX avoid expanding and investing with financing from bank loans. High interest rates cause interest costs to be very large which can increase the company's financial risk. Besides high interest rates, the Indonesian economy is also vulnerable to world economic turmoil such as the volatility of the Rupiah exchange rate against foreign currencies. If a company decides to take loans from foreign banks in foreign currencies, it may be possible to obtain lower loan interest rates but face the risk of a decrease in the exchange rate which will cause a loss in foreign exchange differences.
To get tax benefits from charging interest fees in Indonesia today, it is no longer effective. Because the current tax regulations in Indonesia only allow to charge interest only for interest costs up to loan value of 4 (four) times the company's net capital. If the amount of the loan exceeds this provision, then the interest expense incurred cannot be charged as a deduction for corporate income tax.
From above description, the researcher concluded that the tangibility variable did not affect leverage because of high domestic loan interest rates, the risk of foreign exchange loss for foreign loans, and the existence of rules on limiting interest costs as deduction from income tax. H3: Firmsize variables had a significant effect on capital structure.
Variable X3on Y with a regression coefficient (coefficient) of -0.374651 and P-value (Prob.) of 0.0000 <0.05, so the test results showed that there was a significant effect of X3 on Y of -0,374651. The results of testing the significant effect of firmsize variables on retailing sector companies supported the pecking order theory where significant negative effects were caused by large companies experiencing asymmetric information problems and large companies did not need to send sensitive signals to investors besides the resulting profits were quite large (Al-Ajmi et al., 2009).
H4: Growth opportunity variable had a significant effect on capital structure.
Partially, the effect of variable X4 on Y based on the test results obtained regression coefficient (coefficient) of 0.002969 and a P-value (Prob.) of 0.9204> 0.05. The value of P-value (Prob.) showed that there was no partial significant from variable X4 to variable Y. The results were not significantly contradictory to theory, both the POT theory (Karadeniz et al. 2009;Chen et al. 2014) and the TOT theory (Gomez et al. 2014;Vo 2016).
The growth of the retailing sector company is not used for the convenience of obtaining credit because the concerns of company managers will decrease people's purchasing power. Consumers of retailing companies generally are people who consume goods and services. The purchasing power of the Indonesian people is very much influenced by the prices of several commodities such as the prices of agricultural and plantation products and will decline only by the issue of rising fuel prices. Community consumption usually increases in certain seasons such as religious holidays. Demand is unstable throughout the year so it is risky for companies to pay debt at a rateinterest in Indonesia is quite high every month. The manager's concern with the decrease in consumer purchasing power causes growth opportunities does not increase debt (no positive effect on leverage). H5: Liquidity Variabel had a significant effect on capital structure.
Testing the effect of variable X5 on Y obtained regression coefficient (Coefficient) of -0.069762 and Pvalue (Prob.) of 0.0000 <0.05. These results show that there is a significant negative effect of X5 on Y at -0.069762. This result is in accordance with the pecking order theory which states that high liquidity indicates positive working capital and the existing sources of funds are safe to finance long-term investments so there is no need for funding sources external (Myers and Majhluf, 1984;Haron, 2016 According to Abdillah et al. (2017) that the bussiness risk variable did not affect on go public companies in Indonesia because managers in making funding decisions pay less attention to business risk issues because they pay more attention to macroeconomic factors. Indonesia's macroeconomic conditions continue to be revised during the study period. Because managers had been able to measure the level of risk to be faced, these variables were not a concern in funding decisions.
H7: The NDTS variable had a significant effect on the capital structure.
The effect of Partial X7 Variable on Y resulted in a regression coefficient of -0.761305 and P-value (Prob.) of 0.0161<0.05, so the results of the test can be concluded that there was a significant partial effect of the X7 variable on Y. The significant effect due to the companies were afraid of facing the risk of default because the addition of debt, so that they used depreciation and amortization to obtain greater tax benefits. This result was in accordance with the theory of pecking order and trade off theory.
Conclusion.The determinants of the capital structure of this study were determined by profitability variables, firm size, liquidity and non debt tax shields (NDTS), it had a significant negative effect on capital structure.

______________________________________________________________________________________ 32
Tangibility variables, growth opportunity and business risk did not significantly influence the capital structure. In general, funding decisions on retailing sector companies listed on the Indonesia Stock Exchange (IDX) were more influenced by the pecking order theory. The analysis showed that the structure of the retailing sector company model in Indonesia was more affected by the pecking order theory. Therefore, managers before making funding decisions are expected to prioritize internal funding before using external funding to avoid financial distress. Interest on bank loans in Indonesia is quite high, so a manager must be careful in expanding and funding sources.
The data used in this study were limited to quantitative data sourced from the company's financial statements. It is expected for future researcher will include other variables out of the financial statements such as macroeconomic variables.