“The impact of the supervisory board on bond ratings of non-financial companies”

Issuing bonds is one of the alternative ways for non-financial companies to get money from the public besides borrowing money from banks. Compared with getting money banks, obtaining money from the bond market is slightly economical because the companies are not essential to borne the intermediation cost anymore. As a consequence, the companies in the bond market will get the assessment from the appointed agency. Furthermore, the rating of bonds will determine their reputation. Mentioning the literature review, the bond ratings are affected by the features of the supervisory board: size, independence, and audit committee. Therefore, this research intends to attain two goals. Firstly, it aims to prove and analyze the impact of the supervisory board size and independence, as well as the audit committee size on the company’s possibility to get a high bond rating with profitability as the control variable. Secondly, it intends to know the accuracy rate of grouping the company bond ratings through the classification matrix. The population originates from the non-financial companies. The total samples are determined by the Slovin formula with a boundary of the fault of 10%. Based on this formula, the total samples are 36 companies. Furthermore, they are randomly grabbed from the population. The ordered probit regression model and the classification matrix are utilized to analyze the data. Based on the data analysis, this research finds out that the supervisory board size and independence, the audit committee size, and profitability positively affect the bond ratings. It means that the number of the commissioner board and the members of the audit committee have to be added until achieving the maximum level to monitor the performance of the directors so that the company can reach a high bond rating. To sum up, board governance is effective in improving the company’s bond rating.


INTRODUCTION
The assets are the resource of the firm to generate cash flow (Brealey, Myers, & Allen, 2006).They must be fixed assets because they can create profits (Gitman & Zutter, 2012) to the shareholders as the firm owners (Ehrhardt & Brigham, 2012).The determination of these types of fixed assets that companies invest and the ways to get profits becomes the responsibility of the financial managers (Brealey et al., 2006).To obtain it, they have to seek for an inexpensive source of funds at the convinced risk (Hanafi, 2017).
According to the pecking order theory, as disclosed by Brealey et al. (2006), using the debt is the second sequence if the retained earnings of the firms are not adequate to finance the available projects.The source of debt can be from a bank and a capital market.If the firms choose the loans from the bank, they have to endure the intermediary cost (the difference in the lending rate and the deposit rate), which is Investment Management and Financial Innovations, Volume 17, Issue 1, 2020 http://dx.doi.org/10.21511/imfi.17(1).2020.02more expensive than the issuing bonds in the capital market.On the other hand, if they issue bonds, they do not pay this intermediary cost (Husnan, 2009).The firms deciding to issue bonds are mandatory to be monitored and evaluated by the government-appointed rating agencies (Hartono, 2017), such as KASNIC Credit Rating and PEFINDO in Indonesia (Marfuah & Endaryati, 2016).
From previous literature, firm governance is vital to raise the bond rating.The mechanism of this governance covers the role of the supervisory board reflected by its size (Setyaningrum, 2005 Mariana, 2016), and the presence of individuals becoming the audit committee (Setyaningrum, 2005;Syakhroza, 2005;Rianingsih, 2009;Marfuah & Endaryati, 2016).
The contradictory impacts of committee audit size on the bond rating also still exist.This condition stands for the research illustrating a positive impact (Setyaningrum, 2005;Rianingsih, 2009;Marfuah & Endaryati, 2016) and the absent impact (Sareen & Vij, 2015;Mariana, 2016).
Because the previous research evidence does not yield the consensus yet, the research with the probit regression model is essential to do.The use of this model because of the ability to estimate the regression coefficients (Widarjono, 2013) and to get the accuracy rate of grouping the bond ratings through the classification matrix (Gray, Mirkovic, & Ragunathan, 2006).Hence, this study intends to examine and analyze the impact of supervisory board size and independence, audit committee size on company possibility to get a high bond rating, and know the accuracy rate of categorizing the bond rating through the classification matrix.

LITERATURE REVIEW
The company with good governance can assure the return of the invested money to their providers (Shleifer & Vishny, 1997): dividends to shareholders (Mitton, 2004), interest and principal to banks (Gitman & Zutter, 2012), coupon and the principal to bondholders (Tandelilin, 2010).To implement good governance, the company issuing the bonds needs to control top management (Abor, 2007) by appointing the supervisory board (Tjager, Alijoyo, Djemat, & Soembodo, 2003).The following section discusses the role of the supervisory board and its components with the bond rating.
Indonesia adopts the dual board system.As a consequence, the commissioner board and director board exist in the companies (Tjager, Alijoyo, Djemat, & Soembodo, 2003).In Indonesia, the board of commissioners and directors is appointed based on the decision of the general meeting of shareholders (Syakhroza, 2005).Furthermore, the board of commissioners supervises the board of directors (Sukamulja, 2004).The effectiveness of the supervisory board depends on the number.The resource dependence theory argues that a large number of the supervisory board consisting of many experts is needed to give advice to the board of directors to improve the company's per-formance (Pfeffer, 1972).According to the study of Setyaningrum (2005), Aman and Nguyen (2013), Elhaj, Muhamed, and Ramli (2017), this condition will make the company to get a high bond rating status.Based on this information, the first research hypothesis is as follows: H1: The supervisory board`s size has a positive impact on the company's possibility to get a high bond rating.
The presence of individuals becoming the independent supervisory board uses to create the virtuous monitoring of the board of directors (Bhojraj & Sengupta, 2003).Through it, directors are prevented from investing the money from the issued bonds on hazardous projects that the controlling shareholders like (Easterbrook, 1984) or distribute the cash from the issued bonds as dividends to the firm shareholders (Kalay, 1982).If this prevention is successful, the risk of outstanding bonds will decline so that the firm bond rating status will improve as Bhojraj and Sengupta (2003), Ashbaugh-Skaifea, Collins, and LaFond (2006), and Grassa (2016) declare.According to this information, the second research hypothesis is as follows: H2: The independence of the supervisory board has a positive impact on the company's possibility to get a high bond rating.
An audit committee is made by the supervisory board to ensure that board directors already implement the transparency and disclosure principle consistently (Tjager et al., 2003).Two principles intended are related to the accuracy of financial information delivered to its users.If this committee can function properly, the deduction of the opportunistic behavior of the board of directors happens (Ashbaugh-Skaifea et al., 2006).After that, the quality of financial statement reporting increases; hence, the issuers of bonds will have a low default risk and the improvement of their rating status (Setyaningrum, 2005;Ashbaugh-Skaifea et al., 2006).The effectiveness of the audit committee to do this function depends on the number of individuals in this position (Setyaningrum, 2005).The research of Setyaningrum (2005), Rianingsih (2009), and Marfuah and Endaryati (2016) affirms that the bigger the number of individuals in the audit committee, the better the bond rating status.
According to this information, the third research hypothesis is as follows: H3: Audit committee`s size has a positive impact on the company's possibility to get a high bond rating.
Following Setyaningrum (2005) H4: Profitability has a positive impact on the company's possibility to get a high bond rating.

RESEARCH METHOD
The population of this research is the non-financial companies issuing the bonds until June 2018 with a total of 56.Since one company does not have a rating, it is removed from the population so that the relevant population ( ) In this model, the association between the unobserved * BR and the observed BR is displayed in the second equation: Parameter µ shows the limit point to determine the area of the bond ratings.If BR is equal to 0 and 1, the companies having grade C and D is lower than 1 µ and companies having grade , BBB − , BBB and BBB + is between 1 µ and 2 , µ and so on.
Furthermore, the data are analyzed by the classification matrix to get the accuracy rate of categorizing all firm bond ratings (Gray et al., 2006).
Mentioning Ghozali (2016), the accuracy rate is calculated by dividing the number of the correct prediction of the firm bond ratings with the number of the actual firm bond ratings.a) for SBS, the number range of people becoming the board of commissioners is between 2 and 10, with a mean of 5.4167 and a standard deviation of 1.90301; b) for SBI, the range value of the independent commissioner's board ratio is between 0.14 and 0.60, with a mean of 0.3720 and a standard deviation of 0.1100; c) for ACS, the number range of people appointed as an audit committee is between 2 and 6, with a mean of 3.3611 and the standard deviation of 0.83333; d) for profitability, the range value of ROA is between -24.48% and 16.48%, with a mean of 3.0269% and a standard deviation of 0.6432.The maximum likelihood technique makes the standard errors of the ordered probit regression model to follow the normal distribution.Therefore, the normality of standardized errors has to prove by using the statistical test of the Jarque-Bera (JB) (Widarjono, 2013).Figure 1 exhibits the normality test result.In this figure, the probability value of the Chi-square of the JB statistic is 0.627967.Since this value exceeds the 5% significance value, the null hypothesis declaring that errors follow the normal distribution is not rejected and this accomplishes the requirement of this model.

RESULTS AND DISCUSSION
After the normality condition is reached, estimating the coefficients of the ordered probit regression model is the following stage.Moreover, the estimation result is in Table 4.To test the null hypothesis, we compare each probability value of the Z-statistic of the regression coefficient of SBS, SBI, and ACS, and ROA with a 5% significance value.If the probability value is lower than a 5% significance value, the null hypothesis is rejected.Instead, the alternative hypothesis is acknowledged.
In the statistical test, each research hypothesis in section 2, furthermore, will become an alternative hypothesis.From Table 4, the probability value of Z-statistic standing for SBS, SBI, ACS, and ROA is 0.0005, 0.0139, 0.0020, and 0.0001, respectively and each regression coefficient shows a positive  sign.This condition means research hypotheses one, two, three, and four are not rejected.
The test outcome of the first hypothesis pronounces that the supervisory board size has a positive impact on the company's possibility to get a high bond rating.This outcome confirms the resource dependence theory stating the big number of this board is needed to improve the performance of the company issuing the bonds so that its rating can go up.This positive impact verifies the research of Setyaningrum (2005), Aman and Nguyen (2013), Elhaj et al. (2017).
The test outcome of the second hypothesis pronounces that the supervisory board's independence has a positive impact on the company's possibility to get a high bond rating.The independent members of this board are important to protect the interest of bondholders by prohibiting the board directors to utilize money from the issued bond to be invested in the risky projects and paid as dividends.This positive impact verifies the research of Bhojraj and Sengupta (2003), Ashbaugh-Skaifea et al. ( The test outcome of the third hypothesis pronounces that the audit committee size has a positive impact on the company's possibility to get a high bond rating.By having many individuals in this position, the power of this committee becomes stronger to prevent the opportunistic actions of board directors.As a consequence, they improve the quality of financial reporting leading to an increase in bond rating status.This positive impact verifies the research of Setyaningrum (2005), Rianingsih (2009), Marfuah and Endaryati (2016).
The test outcome of the fourth hypothesis pronounces that profitability has a positive impact on the com-pany's possibility to get a high bond rating.It means that only companies with profits that can pay coupons for bondholders are needed to get a high grade of bond rating.Conversely, the companies with loss are assumed in financial distress and bankruptcy and get a speculative grade.This positive impact verifies the research of Setyangrium ( 2005 Ideally, to make this quantitative research have a high external validity, the sample selection is needed to create the generalization (Hartono, 2012).To realize it, the samples have to be randomly grabbed from the population (Sugiyono, 2012).This research already meets this required condition because of the use of the simple random sampling method.Therefore, the utilization of 36 companies as the sample is totally sufficient.
To answer the second purpose of this study, the classification matrix of the ordered probit is used.This matrix, according to Gray et al. (2006), has the function to measure the accuracy rate of grouping the ratings based on the explanatory variables used.Based on the matrix presented in Table 5, the accuracy rate is 58.333%.This rate is low and reasonable because this research focuses on theory testing and does not aim to make the prediction model that requires a high accuracy rate.
To get a high bond rating, the controlling shareholders are expected to increase the number of the supervisory board.The next question that may arise is the ideal maximum number of CB.To answer it, the research suggests that firms should have the number of CB in between 5 (rounded) as the average value and 10 as the maximum value (see Table 3).Regarding the bond rating being positively affected by the number of supervisory board and audit committee as well as the independent supervisory board, the investors who are interested in the bonds with low default risk are expected to choose the issuers having many independent supervisory and audit committee board members because they are effective to monitor the board of directors.

CONCLUSION
This research aims to prove and analyze the impact of supervisory board size and independence, as well as audit committee size on company possibility to get a high bond rating with profitability as the control variable and know the accuracy rate of grouping the firm bond ratings based on the explanatory variables used.Using 36 companies as samples, this research infers that supervisory board size and independence, audit committee size, and profitability have a positive impact on the company's possibility to get a high bond rating.It means that the monitoring of the supervisory board can improve the bond rating of companies when the board of directors can generate profits.In addition, the accuracy rate of grouping the bond rating is 58.333%.
This research owns some limitations.Firstly, the accuracy rate of grouping the bond rating is as high as 58.333%.It is due to the use of three main explanatory features of board governance.To make this rate increase, the next researchers are suggested adding the corporate governance index, institutional and managerial ownership, board duality, audit quality as the other main explanatory variables.Also, they are recommended using more other control variables, such as liquidity, leverage, size, interest coverage ratio.
Secondly, it only utilizes bond ratings resulted from PEFINDO, the domestic agency in Indonesia.As a consequence, it is impossible to compare the firm bond ratings in Indonesia with those of other countries in Southeast Asia.To compare the ratings among the countries, the next researchers are expected to search the same international agency assessing bond ratings.After that, they can use the countries as the moderating variable so that the theory related to bond rating determinant can be developed.

Table 1 .
Description to measure research variables N consists of 55.To get the representative total samples ( ),

Table 2
demonstrates the number of companies based on the group of bond ratings.Overall, 94.4% of the companies becoming the samples dominantly own the investment grade (BBB -until AAA).Only 33.3% of companies with grade A + , A, and A -have the largest domination.Conversely, the smallest one comes from the companies with grade D and C (5.6%).

Table 2 .
Number of the companies based on the group of bond ratings Source: Modified output of EViews 6.

Table 3 .
Descriptive statistics of SBS, SBI, ACS, and ROA

Table 4 .
Estimation result of the ordered probit regression model: the impact of supervisory board size and independence, audit committee size, and profitability on bond rating Source: Output of EViews 6.

Table A1 .
Investment Management and Financial Innovations, Volume 17, Issue 1, 2020 Names of the companies as the samples Source: Reprocessed from IBMD 2018-2019.