“Debt policy of military-connected firms in Indonesia”

Indonesia has a thin capitalization policy since 2015. It restricts the maximum interest expense that can be deductible from corporate tax payable. This paper discusses the association between boards with military background and the debt policy of firms, taking into account the thin capitalization policy. This study used a sample of 2,330 firm-year observations from companies listed on Indonesia Stock Exchange during 2010–2019. A moderated analysis regression was employed to analyze the association of each variable. The result reveals a significant positive correlation with a t-value of 2.14 at a confidence level of 95% between military-connected firms and debt policy. The same correlation also occurred between board of commissioners with the military background and debt policy with a t-value of 2.18 at a 95% confidence level. Meanwhile, the correlation between these variables became significantly negative after the imple- mentation of thin capitalization policy. CEM and Heckman’s two-stage method were used to validate the findings. This study is for a listed company to consider the appoint- ment of military background in a board of commissioner position after a period of thin capitalization policy. This study aims to examine boards with military backgrounds and the debt policy of firms listed on Indonesia Stock Exchange during 2012–2019. This paper specified the uniqueness of the contrasting characteristic of the military-connected board and its consequences for firm debt policy, taking into account the before and after thin capitalization policy. This paper shows that thin capitalization policy has not effectively minimized debt intensity of Indonesian listed firms. The results are expected to assist the government in assessing the effectiveness of thin capitalization policy. However, this paper shows the combined roles of commissioners with military background and thin capitalization policy can effectively encourage lower debt policy. The results can be taken into account by stakeholders in making decisions related to the appointment of the military background of the board of commissioners after the implementation of the thin capitalization policy. However, the findings of this paper should be used carefully, given the limitations of the study. This paper measured the past experience of boards in military-related positions, so this can lead to the tip of the iceberg. This may be an opportunity for further research to more fully use measurements of military background boards.


INTRODUCTION
For more than 50 years, Indonesia has been led by military personnel such as Soeharto and Susilo Bambang Yudhoyono (Harymawan, 2018). The presence of retired military personnel in business and government also continues to this day. In business, board with the military background has two contrast characteristics. Benmelech and Frydman (2015) describe the character of ex-military personnel who is conservative, organized, and tactical in each decision. On the other hand, Elder et al. (1991) found other possibilities of former military personnel to use power excessively, to take risks, to make aggressive decisions, and to be overconfident because of their strategic position in a firm.
Previous research has documented mutual benefits from board with the military background such as low interest rate (Harymawan, 2018), minimizing corporate tax avoidance (Mills & Law, 2015;Law & Mills, 2017), increasing good corporate governance (Lin et al., 2012; Koch-Bayram & Wernicke, 2018), easiness of financial resources (Agrawal & Knoeber, 2001; Goldman et al., 2013). However, previous research has not already connected between former military personnel's characteristic and mutual benefits. The need for these connections is motivated by two contrast characteristics from ex-military personnel that supposed to lead to different mutual benefits. This might be explained by Upper Echelon Theory (Hambrick & Mason, 1984).
Furthermore, Indonesia has been implemented thin capitalization policy since 2015, as declared in the regulation of Ministry of Finance 169/PMK.010/2015. It restricts the maximum of interest expense that can be deducted from corporate tax payable. The results of this paper discover that military-connected firms use higher loans compared to non-military-connected firms. Otherwise, after the issuance of PMK No. 169/PMK.010/2015 showing that military-connected firms are more likely to use lower loans compared to non-military-connected firms. In particular, the study shows that the significant relation between the military background boards and debt policy of the firms are only applied to the position of board of commissioners. This paper is contributed in an academic and practical way. First, it expands the literature on board with the military background and firm debt policy in a two-tier system. Second, this study is expected to be a source of evaluation from Indonesian thin capitalization policy. The firms' debt does not show significant changes after the implementation of the Regulation of the Minister of Finance No. 169/PMK.010/2015 through the descriptive statistics results. This means that this policy has not effectively minimized debt intensity of Indonesian listed firms. However, this paper shows that the roles of commissioners with the military background can effectively thin capitalization policy to encourage lower debt. The results of this paper can inform interested parties in making decisions related to the appointment of the military background of the board of commissioners after the application of thin capitalization policy.  (Harymawan, 2018). It is categorized as military if the board has educational background and/or experience in the army, navy, air force, and police (Fisman, 2001). Based on a two-tier system, this paper divided the measurement of military-connected firms in three parts as follows: To gain the maximum possible observations, pooled panel crossed-section regression data are used. This study used moderated regression analysis. The general form of moderated regression analysis following J. Cohen and P. Cohen (1983) can be specified more compactly as:

LITERATURE REVIEW AND HYPOTHESES
where Y represents a dependent variable in the model. X 1 represents an independent variable, while X 2 is a moderated variable. X 1 and X 2 represent an interaction between independent and moderated variables. β 0 is a constant value of the model. The empirical model takes the following form: , 01 ,

RESULTS AND DISCUSSION
The results of descriptive statistics indicate there is no significant difference in the firms' debt policy both before and after thin capitalization policy. The maximum debt amount of 1,498 in the sample shows there are some firms that use debt more than allowed. This means that thin capitalization is still ineffective.    Table 5 shows a significant positive association between military-connected firms and debt policy. This paper also found that military-connected commissioners are positively and significantly associated with corporate debt policy. On the other hand, there is no significant association between military-connected directors and debt policy. These results indicate the first hypothesis is not rejected and confirm the assumption that military-connected firms have lower cost of debt (Harymawan, 2018)  According to test results in Table 6, another interesting fact has been found that thin capitalization policy is not significantly related to debt policy. This finding indicates the application of thin capitalization policy is not effective enough to stimulate lower debt usage. Interestingly, the first column of Table  6 shows a significant negative association between military-connected firms and debt policy after the implementation of thin capitalization policy. It can be interpreted that thin capitalization policy is effective in stimulating lower debt utilization when firms have boards with military backgrounds.

Robustness test
This paper used endogeneity tests to ensure endogeneity problems were minimized in the results. It is important to carry out endogeneity tests in order to gain confidence that this paper has met the necessary condition to obtain valid results (Roberts & Whited, 2013 Tables 5 and 6. Thus, it can be confirmed the main findings did not change when using the matched approach.

Heckman's two-stage regression
Following Harymawan (2020), this paper used Heckman's two-stage regression to overcome the problem of the potential sample-selection bias. This paper estimated the possibility of potential sample-selection bias problems, since the firms have discretion to hire boards with military experience or not. Following Heckman's (1979) twostage procedure, this paper estimated the probit model to predict factors related to military connections but not having a direct association with    Table 8. The main variables of interest were positive and significant coefficients on the instrumental variables. Models 2 and 3 show the results of the second stage of regression. These results confirm the main findings in Tables 5 and 6 after considering sample selection bias. The MILLS ratio showed insignificant results in all the models used so there were no endogeneity problems in the results of the main analysis.

Group test for robustness
To confirm the validity of the results, this paper conducted a group test based on thin capitalization policy before implementation (2012-2014), initial implementation (2015-2016), and after implementation (2017-2019). The results are listed in Appendix A. According to Table A1, MCON coefficient is positive and significant in the year before the implementation of thin capitalization pol-icy. Meanwhile, in the initial implementation and after implementation of regulations, the results are not significant. Subsequently, this paper found similar results on the association between a board of commissioners with a military background and debt policy. These results confirm the main finding in Tables 5 and 6.

Group test for additional analysis
Furthermore, the sample of firms is divided on firms' size. First, this paper calculated the median firm size for each group of years, then classified the sample into large or small firms. Appendices B and C show that firms with total assets below the median are rated 1, and 0 otherwise. As a result, small firms have lower debt policies than large ones. This is because small firms are more vulnerable to changes in economic conditions, having fewer assets, higher default risk, and tend to be less profitable (Utami, 2021), thereby reducing creditor confidence in providing loans (Wahyuni, 2019). Therefore, small firms have a higher cost of debt, thereby encouraging lower use of debt. However, the existence of boards with military background in small firms prior to the thin capitalization policy helped them to obtain lower interest rates (Harymawan, 2018) and easier access to finance (Agrawal & Knoeber, 2001;Goldman et al., 2013). Therefore, the presence of boards with military background in small firms before thin capitalization policy led to higher use of debt.

CONCLUSION
This study aims to examine boards with military backgrounds and the debt policy of firms listed on Indonesia Stock Exchange during 2012-2019. This paper specified the uniqueness of the contrasting characteristic of the military-connected board and its consequences for firm debt policy, taking into account the before and after thin capitalization policy. This paper shows that thin capitalization policy has not effectively minimized debt intensity of Indonesian listed firms. The results are expected to assist the government in assessing the effectiveness of thin capitalization policy. However, this paper shows the combined roles of commissioners with military background and thin capitalization policy can effectively encourage lower debt policy. The results can be taken into account by stakeholders in making decisions related to the appointment of the military background of the board of commissioners after the implementation of the thin capitalization policy.
However, the findings of this paper should be used carefully, given the limitations of the study. This paper measured the past experience of boards in military-related positions, so this can lead to the tip of the iceberg. This may be an opportunity for further research to more fully use measurements of military background boards.