Institutional Ownership , Board of Directors and Yield Spreads in Long-and Medium-term Corporate Bonds and Sukuk

Previous studies documented that type of investor significantly affects the performance of bonds and sukuk. These studies showed that the yield to maturity (YTM) of bonds and sukuk are significantly associated with institutional investors. This association is because institutional investors actively monitor the performance of bonds and sukuk. Apart from the type of investor, the roles played by the board of directors (BOD) in decision making significantly influence the performance of bonds and sukuk, especially the YTM. This study aims to investigate the relationship between institutional ownerships and the BOD and yield spreads of longand medium-term corporate bonds and sukuk. Data are obtained from firm issuers’ annual reports, Bond Info Hub of Malaysia Central Bank, Department of Malaysia Statistics and Bloomberg from 2000 to 2014. The study employed unbalanced panel data approach for multivariate robust regression, OLS, fixed-effect, and random-effect models. Results revealed that the presence of top-six institutional investors and characteristics of the BOD exert a significant negative effect on the yield spreads. The findings are also consistent with the agency cost of debt theory, which suggests that long-term bonds carry a lower cost of defaults than medium-term bonds.


Introduction
Bonds have been increasingly but slowly supplementing bank lending as a source of finance for the private sector in emerging market economies.In the United States, bond financing has long ago overtaken borrowing from domestic banks (Hawkins, 2002).Hawkins (2002) mentioned that this competition between public issued bonds and banks significantly affect banking systems.Therefore, the development of a bond market was viewed as a priority because it served as an alternative source for raising capital (Sharma, 2001;Zakaria, Hussin, Noordin, & Mohamed Sawal, 2010).In recent decades, the development of Islamic capital market (ICM), especially sukuk has given another alternative of source of financing.Hesse, Jobst, and Sole (2008) stated that the development of ICM assets and structured finance in several Islamic jurisdictions countries, such as Bahrain, Malaysia, and Dubai, has encouraged participation from shariahcompliant debt issuance in a bid to promote themselves as centers for Islamic finance.The leading financial centers (e.g., Hong Kong, London, New York, and Singapore) are making significant progress in establishing legal and prudential foundations to accommodate Islamic finance alongside the conventional financial system, including sukuk market.
In Malaysia, sukuk share similar features to conventional bonds.Sukuk has maturity period, coupon rate, and trades on normal yield price relationship.The difference between these two instruments is the underlying asset for conventional bonds is money (debt) and that of sukuk is an asset (Afshar, 2013).Sukuk is structured such that the issuance is not an exchange between paper and money and does not impose interest (yields), unlike conventional bonds.The issuance of sukuk is based on an exchange of approved assets for financial consideration that allows investors to earn profits from the transactions.Approval of the assets and the contract of exchange are based on shariah principles.On the contrary, conventional bonds are standard bonds bearing a coupon, that requires payment of interest either yearly, twice a year, or quarterly.Conventional bonds also have a maturity date at which investors will redeem their bonds at face value or par value.The price of conventional bonds changes primarily with the interest rate, which can include trading activities that are deemed prohibited and are regarded unsuitable for Muslim investors.Unlike bonds, the yields on sukuk are based on profit from transaction and not from interest or any prohibited activities.Past studies showed that the yield on sukuk and bonds are associated with several factors, such as ownership structures, corporate governance, macroeconomic factors, and others.This study aims to examine whether the institutional ownership and characteristics of the board of directors (BOD) can influence the performance of sukuk and bonds.The study also intends to investigate whether tenure plays a crucial role in determining yield to maturity (YTM) and risk of default.
The remainder of this paper is organized as follows.The next section briefly discusses corporate governance, institutional investor, BOD, and agency cost of debt theory.The section after that presents the the research methods and findings.

Corporate Governance, Institutional Investor, BOD, and Agency Theory
Several researchers have studied the impact of corporate governance mechanisms on the reduction of default risks of conventional bonds by mitigating agency costs and monitoring managerial performance between the firm and the lenders (e.g., Berle & Means, 1932;Bhojraj & Sengupta, 2003).Given the role of institutional investors in monitoring the performance of companies (Mallin, 2008;Elyasiani & Jia, 2010), these investors are considered as advanced investors and are better in utilizing current information to predict future earnings than other types of shareholders (Jiambalvo, Rajgopal, & Venkatachalam, 2002) as well as in bidding to increase liquidity and enhance transparency (Jobst, 2006).
The relationship between agent and principal in the issue of separation and control leads to agency conflict (Fama & Jensen, 1983;Jensen & Meckling, 1976).Jensen and Meckling (1976) emphasized the diversion of the managers' interest from that of the owners, which leads to the agency conflict of interest often referred to as agency problem (Fama & Jensen, 1983).Agency problem arises because any financing contract is not costless, written, and enforced (Fama & Jensen, 1983).Agency costs include the cost of structuring, monitoring, and bonding a set of contracts among agents with conflicting interests (Jensen & Meckling, 1976).Control of agency problems in the decision process is important when the decision made by the BOD and managers who initiate and implement important decisions are not the major residual claimants and therefore do not bear a major share of the wealth effects of their decisions (Fama & Jensen, 1983).
Without effective controlling and monitoring mechanism, such a management decision of the BOD is likely to take actions that deviate from the interests of residual claimants.An effective system for decision control implies that the control focus on the ratification and monitoring of decisions and should be separate from the management during initiation and implementation of decisions to a certain extent.Individual decision agents can be involved in the management of specific decisions and the control of others, but separation means that an individual agent does not exercise exclusive management and control rights over the same decisions (Jensen & Meckling, 1976;Fama & Jensen, 1983).
As a governance mechanism, institutional investors and the BOD are beneficial to bondholders (Jensen & Meckling, 1976;Fama & Jensen, 1983;Bhojraj & Sengupta, 2003;López-iturriagaa & García-meca, 2015).The presence of institutional investors who have purchased corporate bonds and sukuk rather than that of individual investors helps reduce the cost of debt and increases the benefit of monitoring with the duration of the investment.In addition, their share ownerships in the issuer firm are strong incentives to monitor the firm's management, thereby reducing the cost of debt and consequently mitigating default risk (Barry, Lepetit, & Tarazi, 2011) and enhancing the firm performance in the long run toward debt policy planning decisions (Qiu, 2006).Instead of a negative relationship with yield spreads, the presence of this institutional ownership exerts a significant positive effect on yield spreads in multinational firms (Boubakri & Ghouma, 2010).
Past studies documented that an issuer firm with effective BOD characteristics, as highlighted in the Malaysia Code of Corporate Governance (MCCG) code of best practices, can reduce default risk.BOD characteristics, including CEO duality, BOD composition, and board size, are important because they influence corporate bond markets (Tanaka, 2014).A CEO who has considerable decision-making power is associated with high yield spreads. Powerful CEOs tend to maintain an opaque information environment and become a critical determinant to the cost of bond financing (Liu & Jiraporn, 2010;Shailer & Wang, 2015).CEO duality positively affects firm performance, and separating the roles of CEO and chairman is associated with low yield spreads (John & Senbet, 1998;Liu & Jiraporn, 2010).
In addition, excellent board experience may lead to good quality advice to management and effective terms for debt.The quality of the board includes BOD composition, and this characteristic exerts a material impact on low corporate debt yield spreads.A large number of independent directors can reduce a firm's cost of capital because many directors can provide multitasking skills and qualifications that generate a broad perspective in decision making (Bhojraj & Sengupta, 2003;Ertugrul & Hegde, 2008;Fields, Fraser, & Subrahmanyam, 2012) that can enhance efficient monitoring (John & Senbet, 1998) and advising functions as well as create much value (Andres & Vallelado, 2008).Consequently, board size has a significant negative relationship with debt financing (Huang & Wang, 2015;John & Senbet, 1998) and bond yield spreads (Anderson, Mansi, & Reeb, 2004;Ertugrul & Hegde, 2008;Fields, Fraser, & Subrahmanyam, 2012).However, despite the importance of BOD characteristics, no prior research has explored their influence on sukuk yield spreads in long and medium terms.

Methodology
The data of this study are gathered from various sources, including Bank Negara Malaysia, RAM, SC, Bloomberg and Department of Statistics, Malaysia.Specifically, data are retrieved from Bond Info Hub's website from BNM for issuance characteristics for conventional bonds and sukuk, including yield, number of tranches, issuer name, price of debt, issue date, maturity date, issue amount in millions (MYR), and debt instrument categories.Table 1 summarizes the variables used and their measurement as along with the data sources.
All the data are sorted, screened, and matched.Firms with missing data are omitted, leaving the total usable observation data for the long-term debt instruments of 140 issuer firms with 256 tranches from 2000 to 2014.In medium-term debt instruments, the total usable observation data are 160 issuer firms with 395 tranches from 2003 to 2014.

Multivariate Panel Robust Regressions (MPRR)
The study used MPRR to test the relationships between the dependent and independent variables.
The OLS model has treated standard error of estimations represented by ε as identically and independently distributed disturbances that are uncorrelated with the correlations of standard error for independent variables, χ . In this case, the data can be pooled, and OLS can be used to estimate the model by denoting the estimator of the slope OLS β .The intercept and slope coefficients are constant across N and T represented by tranche issuances of each issuer, postulating that both the intercept and slope are the same across observations.However, these assumptions may be restrictive and lead to heterogeneity bias that must be handled for the robustness check analysis.Otherwise, the model does not require any additional technique for such estimations.The regression model equation for pooled OLS can be represented as follows: The fixed-effect model is used when the constant value for each tranche of issuances, is correlated with the independent variables of the issuers for the year, and within variation in the data only but is the most flexible in that it allows for the endogeneity of regressors.Where The random-effect model assumes that the tranche of issuances has intercepts while restricting the slope to be homogenous for yield spreads.Their spread is probably in random effect as liquidity movement requires regressions to accommodate such a heterogeneity.The random-effect model decomposes into two composite error terms as

Analysis of Normality Tests and Descriptive Results
The empirical data used in this study have been tested for normal or symmetrical distribution.De Vaus ( 2002), Tabachnick and Fidell (2007), and Howell (2007) stated that for secondary data to be non-symmetrical is normal.Following Tabachnick and Fidell (2007) and Howell (2007), options for handling non-normal data are several.The data can be transformed to enable further statistical analyses, including natural log transformation, to solve normality data problems especially for secondary data.Data transformation is based on the types and degree of violation as well as the randomness of the missing data.In this study, three variables, namely, size of issuances, total assets, and current prices of GDP have been transformed to natural log transformation due to large range of data.Other variables, such as CEO duality and BOD composition, are dichotomous variables (two types of categories valued as either 0 or 1), and, thus, natural log transformation is inappropriate for these two variables.
Table 2 shows that most of the variables are normally distributed in long and medium terms.However, the Kolmogorov-Smirnov test is biased by sample size, and the test may be statistically significant from a normal distribution in any large sample.Therefore, this study reconfirms the results of normality test by performing another two tests (i.e., skewness and kurtosis).Based on the skewness and kurtosis test results, the shape of a probability distribution for both skewness and kurtosis is positively and negatively skewed.Ten variables show p-value > 0.01, indicating that the data are non-normal according to the Z-score, which is proxied by the standard score value.The variables involved are yield spreads, top-six IO, other IO, BODR2, BODC, volatility, tenure, leverage, firm value, and sustain.The kurtosis test results show that the probability of data distributions are fat or short-tailed (platykurtic) and slim or longtailed (leptokurtic) as the standard score values which represent the Z-scores are not within the range of -3.29 and +3.29 (normal distribution range).The affected variables for LTCBS are yield spreads, top-six IO, BODR2, BODC, volatility, tenure, profit, firm value, firm size, sustain, and lnGDP.Another five variables have a normal distribution or mesokurtic, i.e., other IO, BODS, BODM, lnSize, and leverage.The Z-score values fall within the range of -3.29 and +3.29, suggesting that the distribution of the corporate bonds and sukuk tranche spreads are positively and negatively skewed and that they are flatter than expected for a normal distribution.In addition, although the data are transformed to logarithm, the normality test by kurtosis continues to capture non-normality for variables such as firm size (log of total assets) and curren1t prices of GDP (log of GDP).Regardless whether the 10 variables in long-term issuances are non-normal in terms of probability distribution for data according to skewness test results for p-value > 0.01, the number of variables increases to 12 in mediumterm issuances.The MTCBS for kurtosis test shows that the probability distribution of tranche spreads for the 11 variables as reported earlier also show that the standard score values are not within the range of -3.29 and +3.29.

Analysis of Pairwise Correlations and Variance Inflation Factor (VIF) Coefficients
As reported in the correlation matrix in Table 3, mixed positive and negative correlation results are noted in both panels A and B between yield spreads and independent variables via conventional bonds or sukuk issuances for both observations either long or medium term.These findings are not only for reports on the correlation results between dependent and independent variables but also among variables as revealed by the coefficient of estimations.
The results show that none of the correlation coefficients regressors among variables in both panels A and B are larger than the value of 0.80.
Coefficient regressors with a value larger than 0.80 signifies a multicollinearity problem, which allows for omitting the variables (Damodar, 2004).The highest coefficient correlations regressors from both panels show less than the threshold (<0.80), so the multicollinearity is not a serious problem in multiple regression analysis (Nguyen et al., 2015).
No   Without a multicollinearity problem, the accuracy of model estimations for all models is unquestionable.Overall, the analysis indicates that multicollinearity is not detrimental to the results of multiple regression estimation analysis because both pairwise correlation and VIF coefficients do not report multicollinearity in this clustered.

Analysis of Robust Regression Results
As reported in

Conclusion
The presence of top-six and other institutional ownerships shows positive significant relationship with yield spreads in long-term issuances of corporate bonds and sukuk, but it has no relationship with yield spreads in medium-term issuances.With respect to BOD characteristics, such as CEO duality, board size, and directors' religion, appear to be significant determinants in influencing yield spreads except for the composition of independent directors in the firms.Evidence of CEO duality suggests that the separation of the role between the chairman of directors and CEO is an important determinant for long-term issuances.The implications of the study can be seen in five perspectives.For the theoretical aspects, sukuk investment matches with the conventional theory assumptions for YTM on conventional bonds investment despite differences in the financing structures.In addition, the study provides a significant contribution regarding shareholder, managerial, economic, and policy implications.Compliance with the corporate governance code of best practices by the BOD should be morally anchored to the concepts of trust, mutual consultation, justice, and integrity to ensure consistency with Islamic values.The BOD can be seen as experts in investment and working toward getting the blessing from God.These Islamic ethical values that relate to accountability and governance through religion make good agency relationships and strive to avoid default risk.The issue and issuer characteristics as well as current prices of GDP bear positive and negative implications toward issuers who issued long-term and medium-term debt, respectively.This finding shows a good and bad signal of default risk to an issuer.In particular, an issuer benefits from issuing both medium-term financing instruments, either conventional bonds or sukuk, if the current price of the GDP is high because the default risk will be low.Otherwise, the issuer may issue long-term financing instruments (both or either conventional bonds or sukuk) when the current price of the GDP is low.Issuing long-term debt when GDP is high results in high default risks.The issuer must properly analyze the big/ small size of issuance and long/medium term of issuances to alleviate default risk.
fe β = the coefficient estimates in fixed effect of the explanatory variables, the intercept for fixed effect, and it u = the error term for fixed effect.

Table 1 :
Data Description, Proxy/ berg 13 Firm Value Tobin's Q Total of market value of equity with book value preferred stock and book value of LTD/TA.+ 14 Firm Size lnTA Log of total short-term assets and long-term assets.my This journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE)

Table 2 :
Normality and Descriptive Results

Table 4 :
Robust Regressions Results for LTCBS and MTCBS Notes: The t-statistics of OLS, FE, and RE estimators are reported in parentheses.Asterisks indicate significance at 1% (***), 5% (**), and10% (*).myThis journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE)