6. Data and Methodology

#### 6.1. Data

This study analyzes 402 non-financial listed Indonesian firms between 2000 and 2014 (4737 total observations) with data extracted from the Datastream and annual reports of firms. As a normal practice in capital structure research, financial firms (banks, insurance companies, and investments trusts) are excluded from the sample. The 402 sample firms consist of 75% out of 537 listed firms on the IDX (as of December 2016), and this proportion could be regarded as the whole population of firms for generalization purposes. The sample firms cover firms from the various industries of listing that include agriculture, consumer products, industrial, infrastructure and utilities, mining, properties, trade and services, and miscellaneous industry.

Only firms with a minimum of three consecutive observations toward the end of the study period are included in the dataset [25], meaning that the firms should at least be listed on the IDX from the year 2012. Unbalanced panel data are utilized due to the different listing dates of firms within the study period of 2000–2014. Table 1 presents the structure of the panel data on sample firms for this study.

#### 6.2. Methodology

Debt financing (leverage) in this study, as applied in other capital structure studies, is defined as the ratio of total debt to total asset (at book value) [17, 30, 46]. To examine the determinants of leverage, this study employs the Generalized Method of Moment (GMM). This technique has the advantage of addressing bias due to the presence of lagged dependent variables, or endogeneity of other explanatory variables, associated with the fixed effects in short panels [47], and GMM can be used to control for these issues [48].

To test the hypotheses, the following regression model is employed:

$$\begin{split} \mathbf{L}\mathbf{e}v\_{it} &= \beta\_0 \mathbf{L}\mathbf{e}v\_{it(-1)} + \beta\_1 \mathbf{O}\mathbf{W}\mathbf{N}\_{it} + \beta\_2 \mathbf{O}\mathbf{W}\mathbf{N}\mathbf{D}\_{it} + \beta\_3 \mathbf{N}\mathbf{D}\mathbf{T}\mathbf{S}\_{it} + \beta\_4 \mathbf{S}\mathbf{I}\mathbf{Z}\mathbf{E}\_{it} + \beta\_5 \mathbf{R}\mathbf{I}\mathbf{S}\_{it} \\ &+ \beta\_6 \mathbf{T}\mathbf{A}\mathbf{N}\mathbf{G}\_{it} + \beta\_7 \mathbf{L}\mathbf{Q}\_{it} + \beta\_8 \mathbf{P}\mathbf{R}\mathbf{O}\_{it} + \beta\_9 \mathbf{N}\mathbf{T}\mathbf{A}\mathbf{N}\mathbf{G}\_{it} + \beta\_{10} \mathbf{G}\mathbf{R}\mathbf{O}\mathbf{W}\_{it} + \beta\_{11} \mathbf{A}\mathbf{G}\mathbf{E}\_{it} \\ &+ \beta\_{12} \mathbf{S}\mathbf{P}\mathbf{P}\_{it} + \varepsilon\_{it} \end{split} \tag{1}$$

where the dependent variable, Levit, represents the leverage level of firm i at time t. Levit(�1) represents the lag leverage and firm-level determinants comprising of ownership concentration (OWN), ownership identity (OWNID), non-debt tax shield (NDTS), firm size (SIZE), business risk (RISK), asset tangibility (TANG), liquidity (LIQ), profitability (PROF), intangibility (INTANG), growth (GROW), firm age (AGE), and share price performance (SPP), and εit is the error term.

This study takes the first difference of Eq. (1) to eliminate the firm's fixed effects, thereby avoiding any correlation between unobserved firm-specific effects and the explanatory variables.

$$
\Delta L \varepsilon v\_{it} = \beta\_0 \Delta L \varepsilon v\_{it(-1)} + \sum\_{n=1}^{N} \beta\_k \Delta X\_{kit} + \Delta \epsilon\_{it} \tag{2}
$$

Eq. (2) denotes the model estimated based on the GMM (first difference). One of the advantages of GMM is that it can handle important modeling concerns, namely the fixed effects and endogeneity of regressors, while avoiding dynamic panel bias. It is important to note that the flexible GMM framework accommodates unbalanced panels, a characteristic of micropanel dataset in this study, as well as endogenous variables [25]. Hence, this study uses GMM for the purpose of estimation.


Table 1. Panel data structure.

firm will not have time to retain funds and may be forced to borrow. Consequently, age is likely to be negatively related to debt financing [45]. Older firms have longer track records and therefore a higher reputational value. Age of firm is measured from the year of listing on the stock exchange [44]. The hypothesis is that H11: age has a negative influence on debt financing.

Equity issuance will be preferred if a firm accumulates a strong share price performance with the present market values comparatively higher than the past market values. On the other hand, firm will repurchase equity if the situation is otherwise. This notion is based on the market timing theory, indicating a negative relationship between share price performance and debt financing. Haron [25] find significant negative relationship between share price performance and debt financing on Indonesian firms. Share price performance is represented by yearly change in year-end share price [24, 25]. The hypothesis for this variable is that H12: share

This study analyzes 402 non-financial listed Indonesian firms between 2000 and 2014 (4737 total observations) with data extracted from the Datastream and annual reports of firms. As a normal practice in capital structure research, financial firms (banks, insurance companies, and investments trusts) are excluded from the sample. The 402 sample firms consist of 75% out of 537 listed firms on the IDX (as of December 2016), and this proportion could be regarded as the whole population of firms for generalization purposes. The sample firms cover firms from the various industries of listing that include agriculture, consumer products, industrial, infrastruc-

Only firms with a minimum of three consecutive observations toward the end of the study period are included in the dataset [25], meaning that the firms should at least be listed on the IDX from the year 2012. Unbalanced panel data are utilized due to the different listing dates of firms within the study period of 2000–2014. Table 1 presents the structure of the panel data on

Debt financing (leverage) in this study, as applied in other capital structure studies, is defined as the ratio of total debt to total asset (at book value) [17, 30, 46]. To examine the determinants of leverage, this study employs the Generalized Method of Moment (GMM). This technique has the advantage of addressing bias due to the presence of lagged dependent variables, or endogeneity of other explanatory variables, associated with the fixed effects in short panels

ture and utilities, mining, properties, trade and services, and miscellaneous industry.

price performance has a negative influence on debt financing.

[47], and GMM can be used to control for these issues [48].

To test the hypotheses, the following regression model is employed:

5.13. Share price performance

14 Financial Management from an Emerging Market Perspective

6. Data and Methodology

sample firms for this study.

6.2. Methodology

6.1. Data

To ensure the efficiency of the GMM estimator, this study performs three diagnostic tests which are the Wald test to assess the joint significance of the determinants of leverage (null: all coefficients on the determinants of leverage are jointly equal zero); the AR(2) or secondorder autocorrelation test (null: no second-order autocorrelation in the residuals); and the J-test, a test for the validity of the instrumental variables representing Levit(�1) (null: instrumental variables are valid). Estimates derived from the GMM are only consistent if there is no secondorder autocorrelation in the residuals and instrumental variables representing Yit(�1) are valid. To check for multicollinearity, this study performs the variance inflation factor (VIF) on each independent variable in the regression model. As a rule of thumb, the VIF for each independent variable should be less than 10 to avoid multicollinearity problem.

#### 7. Analysis and findings

#### 7.1. Descriptive statistics

Table 2 summarizes the descriptive statistics of all variables in this study. Indonesian firms employ mean leverage of 0.3691 in their capital structure. Ownership concentration shows on average 47.64% ownership exceeds 5% shareholding with the maximum and minimum of 100% and 0, respectively. This statistic shows that the ownership structure of public Indonesian firms is highly concentrated. Utama et al. [11] posit that it is quite prevalent for public firms in Indonesia to have only a few shareholders with substantially large holdings (i.e., at least 5%). Profitability shows a mean of 0.0654 ranging from �2.9565 to 2.8310. Business risk on firms as represented by yearly change in firms' EBIT is found to be substantial, shown by


Notes: Number of all firms = 402; Number of observations = 4737.

Table 2. Descriptive statistics.

the standard deviation of 3.0502, bigger than the mean of �0.0594. Tangible asset has a much higher mean of 0.3922 compared to intangible asset, 0.0164 as recorded. On firm age, firms in Indonesia have been listed on the stock exchange for 15.4104 years on average, with the longest and shortest listing of 38 years and 3 years, respectively.

#### 7.2. Determinants of leverage

To ensure the efficiency of the GMM estimator, this study performs three diagnostic tests which are the Wald test to assess the joint significance of the determinants of leverage (null: all coefficients on the determinants of leverage are jointly equal zero); the AR(2) or secondorder autocorrelation test (null: no second-order autocorrelation in the residuals); and the J-test, a test for the validity of the instrumental variables representing Levit(�1) (null: instrumental variables are valid). Estimates derived from the GMM are only consistent if there is no secondorder autocorrelation in the residuals and instrumental variables representing Yit(�1) are valid. To check for multicollinearity, this study performs the variance inflation factor (VIF) on each independent variable in the regression model. As a rule of thumb, the VIF for each indepen-

Table 2 summarizes the descriptive statistics of all variables in this study. Indonesian firms employ mean leverage of 0.3691 in their capital structure. Ownership concentration shows on average 47.64% ownership exceeds 5% shareholding with the maximum and minimum of 100% and 0, respectively. This statistic shows that the ownership structure of public Indonesian firms is highly concentrated. Utama et al. [11] posit that it is quite prevalent for public firms in Indonesia to have only a few shareholders with substantially large holdings (i.e., at least 5%). Profitability shows a mean of 0.0654 ranging from �2.9565 to 2.8310. Business risk on firms as represented by yearly change in firms' EBIT is found to be substantial, shown by

Variable Mean Maximum Minimum Median Standard deviation

TD/TA 0.3691 0.9020 0.0998 0.3355 0.1872 Ownership 0.4764 1.0000 0.0000 0.5700 0.3383 Ownership identity 0.5643 1.0000 0.0000 1.0000 0.4959 NDTS 0.0310 0.6045 0.0000 0.0244 0.0384 Firm size 11.5277 16.8969 4.1109 11.5955 1.7817 Risk �0.0594 28.5000 �29.7739 �0.0275 3.0502 Tangibility 0.3922 0.9852 0.0000 0.3677 0.2504 Liquidity 2.1793 29.8679 0.1027 1.4378 2.6678 Profitability 0.0654 2.8310 �2.9565 0.0672 0.1791 Intangible 0.0164 0.9650 0.0000 0.0000 0.0621 Growth 8.3666 97.8479 0.6000 2.9101 14.2480 Age 15.4104 38.0000 3.0000 15.0000 7.6098 Share price performance 0.0058 2.7810 �4.8121 0.0010 0.2038

Notes: Number of all firms = 402; Number of observations = 4737.

Table 2. Descriptive statistics.

dent variable should be less than 10 to avoid multicollinearity problem.

7. Analysis and findings

16 Financial Management from an Emerging Market Perspective

7.1. Descriptive statistics

Based on Table 3, nine determinants which are ownership, ownership identity, NDTS, size, risk, tangibility, liquidity, profitability, and age of firm are found to significantly influence the debt financing of Indonesian firms throughout the period understudy.

This study depicts a positive relationship between ownership and debt financing. Higher level of concentrated ownership has a positive influence on debt financing (p = 0.01), and H1 is thus supported. This finding supports the findings by [17, 34, 39]. The positive relationship depicted in this study reflects the power and authority of the large controlling shareholders in a highly concentrated ownership environment employing debt as controlling mechanism on


Notes: \*\*\*, \*\*, \*denote probability values significant at 1, 5, and 10% levels, respectively. The t-statistics in parenthesis are the t-values adjusted for White's heteroscedasticity consistent standard errors. The Wald test statistic refers to the null hypothesis that all coefficients on the determinants of debt financing are jointly equal to zero; the m-statistic for AR(2) refers to the null of no second-order correlation in the residuals; the J-test statistic for the null that the over identifying restrictions are valid. The VIF test of less than 10 confirms that there is no multicollinearity problem.

Table 3. Determinants of debt financing.

the managers. The positive relationship may also be explained by the reluctance of large shareholders to engage with equity financing as to avoid ownership dilution and thus can maintain the control of the firms.

In terms of ownership identity, the result shows that ownership identity has significant influence on debt financing (p = 0.10), thus H2 is supported. Family-owned firms is found to consume lesser debt financing compared to the non-family-owned firms. This is well explained by the fact that family-owned firms are known with the reputation of being risk averse [19]; thus, debt engagement is very much avoided. The lesser consumption of debt by the familyowned firms depicted could also be the result of the alignment of interest between shareholders and managers, which makes issuing debts as manager's disciplinary tool less crucial for family-owned firms. It is expected that family-owned firms do not suffer from agency cost considering that the owner and the management of firms are the same people, and hence no issue of diverging interests is between the two parties [7].

This study records a negative relationship between NDTS and debt financing (p = 0.01); thus, H3 is supported. Following the trade-off theory, since engaging to debts means bringing in risks and insolvency to the firms, firms may opt to NDTS. Frank and Goyal [36] argue that NDTS should be negatively correlated with debt financing as NDTS is the alternative to tax shields provided by debt financing. Significant negative relationship between NDTS and debt financing is reported in [42] on Indonesian firms. Looking from the lens of family-owned firms, debt is very much avoided being risk averse and the active monitoring by the family in the firm makes debt less needed as disciplinary tool on the managers and hence explains the negative relationship.

A negative relationship is reported between size and debt financing (p = 0.01), in contrast to H4 in which a positive relationship is expected. Haron [25] also depicts significant negative relationship between size and debt financing. Perhaps according to [25], the negative relationship is due to the effects of Indonesian financial market deregulation activities where the control over initial offering prices and the daily movement of stock prices were lifted and thus encouraged large firms to issue equity over debt. Nonetheless, looking at the nature of family-owned firms, the fear of power dilution and the concern over business risk and insolvency have hindered the firms to employ higher level of debt in the capital structure. The larger the firms, the more retained earning they have accumulated; thus, debt is the least choice of financing, following the pecking order theory.

Risk is also found to negatively related to debt financing (p = 0.10), and H5 is then supported. Trade-off theory explains that the higher the debt employment the riskier it gets for the firms in case of default payments; thus, debt financing should be avoided. Haron [25], Ameer [42], and De Jong et al. [43] find business risk having a significant negative relationship with debt financing among firms in Indonesia. One of the distinctive characteristics of family-owned firms is being risk averse for fear of losing the firm in case of bankruptcy risk and insolvency. It is therefore expected of these family firms to avoid debt employment in their capital structure as to avoid the risks that come with it.

This study depicts a positive relationship between tangibility and debt financing (p = 0.01). This finding supports H6. Tangible assets help firms obtain more debt from lenders as tangible assets act as collateral, making debt less risky. Moosa and Li [24], Bunkanwanicha et al. [30], and De Jong et al. [43] all share similar significant positive relationship between tangibility and debt financing in their studies on Indonesian firms. Degryse et al. [49] argue that the positive effect of tangibility on total debt comes entirely from long-term debt as these tangible assets are used to secure long-term debt. This argument confirms the characteristics of family-owned firms in Indonesia where for the case of family-owned firms, long-term orientation is very crucial in keeping the firm to the family for the next generation as suggested by [16].

Liquidity is reported to relate positively with debt financing (p = 0.05), in contrast to H7 in which a negative relationship is expected. When firms in Indonesia have high liquidity level, they seem to increase their debt consumption perhaps to reap the tax shield advantage supporting the trade-off theory. Looking at the case of family-owned firms in Indonesia, this finding is in line with the argument by [50] where liquidity is found to positively relate to longterm debt and again being family-owned firms, long-term debt ensures continuity of controlling power for the family which is the main concern for such firms, thus rationalizes the positive relationship depicted in this study.

Profitability is found to relate negatively with debt financing (p = 0.01). H8 is thus supported. Highly profitable firms in Indonesia choose to use their retained earnings to finance their investments, thus reflecting the influence of pecking order theory in their debt financing decisions. Supporting [24, 30] the negative relationship reported may be the results of the financial reformations taken place in Indonesia which have opened up and encouraged firms to turn to their retained earnings instead of merely bank loans to finance their investments. For the case of family-owned firms, using retained earnings can become the main agenda as to avoid dilution of power and unnecessary intrusion from outsiders via equity financing [13].

Age of firm is positively related to debt financing (p = 0.01), H11 is thus supported. Conforming to what has been argued previously in past studies, the older the firm, apparently, will have more impressive track record and thus become less risky comparative to new firms. Being considered as less risky encourages lender to lend more and firms can reap the benefits of debt following the trade-off theory. From the view of family-owned firms, with the impressive track record over the years, these aged family-owned firms enjoy greater availability of credit [16] and a lower cost of debt financing [23].

Nonetheless, results of this study show that three determinants (intangibility, growth, and share price performance) appeared to be insignificant in the debt financing decisions of Indonesian firms despite being reported as important factors in capital structure studies.
