Influence of Internal and External Governance Mechanisms on Corporate Governance Disclosure Among Islamic and Conventional Banks

The purpose of this paper is to investigate the extent of corporate governance disclosure in the annual reports of listed conventional and Islamic banks in Bangladesh. Out of fifty-six scheduled banks in Bangladesh, a sample of thirty-nine banks is selected, and data for the sample is extracted from the annual reports covering a period of 2011 to 2014. As such, the study focused on the extent of CGC after the stock market crisis in 2010 in Bangladesh. This results in the final observations of 116 which were used to perform balanced panel regression analysis. Fixed effect model is found significant for the balanced panel model which indicates that appointment of large audit firms negatively affects the extent of corporate governance compliance. Pooled OLS regression established that while profitability has a negative influence, the size of banks positively affects the extent of CGC. This study has focused on the commercial banks and thus results obtained from the study may not be representative for public and foreign banks operating in Bangladesh. Statistical evidence provided by the study provides guidelines for the policymakers toward necessary governance reforms required for banks to successfully operate in a post-crisis environment. Factors established by the study that influences corporate governance compliance using a balanced panel model are unique in the context of developing countries. Evidence of a difference in governance compliance between Islamic and conventional banks in Bangladesh establishes a new research arena and a necessary shift from the traditional performance comparisons.


Introduction
Corporate governance is a major concern in the South Asia-Pacific region, especially in the aftermath of the 1997 Asian financial crisis. The size and frequency of recent corporate governance debacles show that poor governance is not only a formidable hurdle to surmount but is also at the forefront of economic development issues. A dilemma has arisen from recent experience: it is possible for companies to appear to comply with the requisite corporate governance rules without complying with the principles and spirit of good governance.
The banking sector remains of enormous importance for Bangladesh who is striving hard to strengthen its developing yet fragile economy. To move from the agriculture-based economy to an industry-based one, Bangladesh needs its banking sector, which is the single largest element of the financial sector, to operate at its best with utmost efficiency. Anything Disclosure among Islamic and Conventional Banks short of that and even a slight instability in this area would wreak long-term havoc on Bangladesh's development. Moreover, sound corporate governance remains to be a key requirement for efficient and stable banking system. The uniqueness of banking companies and banking business require special corporate governance attention on a priority basis particularly for the developing countries where prudential regulation and supervision is inadequate to provide a safety net for the depositors and stakeholders of the banks.
In Bangladesh, however, there have been no serious corporate scandals which have been enough to send shock waves to undermine confidence in the financial system, nor has the country found that it has reached the limits of conventional corporate financing mainly through bank lending. The country report identifies that -the relatively low level of international investment in Bangladesh does not provide sufficient motivation for improving corporate governance, nor are there many traditional domestic motivations for improvement in corporate governance practices in Bangladesh‖.
Nevertheless, that this does not mean that Bangladesh should give low priority to corporate governance, as there are reasons other than capital market reforms to focus on corporate governance. The Bangladesh country report notes the significance of corporate governance for a competitive private sector in a global market as well as for efficiently utilising domestic investment to achieve greater economic development. Good corporate governance practices will help develop and stimulate better business management, strategic management, and risk management, which, in the long-term, will make Bangladeshi businesses more competitive. Also, the lessons from the experience of the neighbouring countries in South Asia are such that Bangladesh can deploy good corporate governance to prevent the problems which have afflicted other countries rather than to solve them after the event.
As is documented in this volume, in Bangladesh, failings in institutions, government agencies, legal enforcement, and market behaviour have resulted in weak corporate governance. In many cases, the current system in Bangladesh does not provide sufficient legal, institutional, or economic motivations for stakeholders to encourage and enforce good corporate governance practices. As a result, there are few rewards for companies that institute good corporate governance practices and no penalties for failing to do so. Targeted reforms in institutions or sectors can begin to provide the internal and external motivation for transparency and accountability that will lead to better corporate governance.

State of Corporate Governance Practices in Bangladesh
Corporate has become a top priority for the regulatory bodies with the objective of providing better and effective protection to all stakeholders and also to make the market confident as research reveals a positive correlation between corporate governance and share price. Corporate governance practices are gradually increasing among private and public organisations with the issuance of Corporate Governance Guidelines by the Securities and Exchange Commission of Bangladesh. The current CG Guidelines were issued on a ‗comply or explain' basis but has resulted in compliance among 66.7 percent of the companies and rest have compliance policy with national and international benchmarks (Khan 2006b). Khan (2006a) reports that the past few years have witnessed a silent inclination toward CG due to a variety of forces i.e. deregulation, disintermediation, institutionalisation, globalisation that are an acting toady and would become stronger in years to come. Karim, Sarkar, & Fowzia (2010) conducted a study on fifty listed companies in Bangladesh and found that all companies under review reported corporate governance in the annual reports. The study extracted data from annual reports of 2007 with a disclosure index for randomly selected listed companies but limited the findings to the extent of governance reporting and did not provide any factors resulting in such governance disclosure. Integration of the Code of Corporate Governance for Bangladesh in developing a disclosure standard has allowed this study to contribute to the existing literature with a possible factor of such compliance among two separate forms of banks in Bangladesh.
In an exploratory study, Wise & Ali (2009) looked into the relationship between corporate governance and corporate social responsibility in the banking sector. Content analysis was used to classify the disclosures into various categories such as environment, energy, human resources, product and safety, community involvement and others following (Hackston and Milne 1996). The study found evidence of economic and public responsibility responsiveness emerging in the corporate social responsibility disclosures contained in the annual reports of Bangladesh's banks. Wise & Ali (2009) reported evidence of the inclusion of women within the boards of directors of selected banks as an indicator of bank's attitude towards corporate social responsibility. Huq & Bhuiyan (2012) identified several problems relating to corporate governance compliance by conducting a questionnaire survey among twenty-six randomly selected banks. Among others, the dominance of family members in ownership structure, the mixed impact of accounting standards on CG, the weak regulatory system is identified as main problematic areas. Such limitations in CG compliance can be minimised by improving the capacity of the board of directors, ensuring a sound legal and regulatory framework and recommending best practices for corporate governance (Huq and Bhuiyan 2012).

Literature Review and Hypothesis Development
Corporate governance compliance has established itself as a trending research agenda among researchers in various parts of the world (Gandía 2008;Lokman, Mula, and Cotter 2011). The primary objective of this paper is to determine whether corporate governance mechanisms are associated with corporate governance reporting. Firms' Disclosure behaviours have already been linked with governance mechanisms by propositions of agency theory (M. C. C. . Corporate governance mechanisms are involved in monitoring and determining a firm's overall information disclosure policy (Kelton and Yang 2008).
The role of governance mechanisms in determining disclosure policy may be either complementary or substitutive (Ho & Wong, 2001). Several governance mechanisms have been studied in the past literature that has a significant association with the extent and quality of corporate financial disclosure. Among them, board size (Cheng & Courtenay, 2006), board independence (Mittal 2011;G. Chau and Gray 2010), audit committee (Johl, Johl, Subramaniam, & Cooper, 2013) and external auditing (Firth 1979) have made their mark in the governance literature. Tsamenyi, Enninful-Adu, & Onumah (2007) have identified significant between association between disclosure and governance and disclosure practices among 22 listed companies on the Ghanaian Stock Exchange.
Ntim, Opong, Danbolt, & Thomas (2012) focused on good governance practices among South African corporations and found a significant difference in the compliance score among 169 listed corporations. Also, the study reports that block ownership is negatively associated with CG compliance, while board size and audit firm size have a positive influence on CG compliance. Such findings support evidence provided by (Barako 2007;Mangena and Chamisa 2008;Haniffa and Cooke 2002;Eng and Mak 2003;Tsamenyi, Enninful-Adu, and Onumah 2007). Board independence has been left from the empirical model in the study of Ntim et al. This limitation leads us toward the study of Parsa, Chong, & Isimoya (2007). The paper focuses on AIM-listed companies over a period of three years starting from 2002 and concentrates on their extent of compliance with the corporate governance disclosure guidelines as per regulations. The study reports a positive association between the board independence and the extent of CG disclosure. Although these findings are in line with Cheng & Courtenay (2006) Disclosure among Islamic and Conventional Banks and Haniffa & Cooke (2002), the fact that this study on small and medium-sized companies (SME) do not allow us to generalise the findings for banks.
Propositions of agency theory (Fama 1980) and signalling theory (Spence 1973) holds that larger audit firms have a stronger incentive to maintain their independence and to impose more stringent and extensive disclosure standards (DeAngelo 1981). Audit firm size has been suggested to have a positive influence on corporate disclosure (Eng and Mak 2003) because larger audit firms with greater financial capabilities can limit opportunistic activities of managers. As larger auditors have greater incentives to demand higher quality disclosure, the appointment of larger auditors for external monitoring by a company would provide a signal of their acceptance of demand of higher quality disclosure (Healy and Palepu 2001). Ntim, Opong, Danbolt, & Thomas (2012) and Ismail & El-Shaib (2012) reports such positive association between audit firm size and voluntary CG disclosure among South African corporations. Chau & Gray (2010) found audit firm size as a non-significant predictor of voluntary disclosure in Hong Kong and Singapore. Akhtaruddin & Haron (2010) reported a non-significant negative association while Barako (2007) found a significant negative association between audit firm size and voluntary disclosure.
H 1 : There is a positive association between audit firm size and the level of corporate governance compliance by banks.
Ntim, Opong, Danbolt, & Thomas (2012) have studied the Voluntary corporate governance disclosures by pose-Apartheid South African corporations and found board size to have a positive association with voluntary CG disclosure. While investigating Internet-based CG disclosure, Gandía (2008) reported similar positive association. Although Haniffa & Hudaib (2006) argues that larger boards have greater managerial monitoring and thus will positively affect disclosure and performance, Jensen (1993) argues that larger boards are associated with poor communication and monitoring, which can have a negative effect on disclosure and performance. Such argument of Jensen (1993) is backed by Chaganti (1985) and reports that the board size for large corporations tends to be inversely related to the degree of effective monitoring and extent of disclosure information to stakeholders. Ho & Williams (2003) and Mangena & Chamisa (2008) report no link between board size and performance. We consider a positive association between board size and the level of CG compliance. Hence, the following hypothesis is formulated: H 2 : There is a positive association between board size and the level of corporate governance compliance by banks.
The proportion of independent directors on board has been reported by researchers (Haniffa and Cooke 2002) to have a significant positive influence on voluntary disclosure. Evidence also suggests that large proportion of independent directors on board could result in increased conflicts and reduced disclosure (Chaganti 1985;Eng and Mak 2003). While studying the extent of corporate governance compliance among US SMEs, Parsa, Chong, & Isimoya (2007) recommended the presence of more non-executive directors. The study of Cheng & Courtenay (2006) provides evidence that boards with a higher degree of independence are associated with higher levels of voluntary disclosures. Ho & Shun Wong (2001) have used a direct measure of voluntary disclosure based on analyst perception but could not identify a significant relationship between voluntary disclosure and board independence. Gul & Leung (2002) identified a negative relationship between voluntary disclosure and board independence, measured by a percentage of expert non-executive directors on the board. Hence, following hypothesis is formulated: H 3 : There is a positive association between Board independence and the level of corporate governance compliance by banks.

Theoretical Perspective
Agency theory introduced the conflicting relationships between two parties with the difference in interest (Fama and Jenses 1983;Fama 1980). The theory is concerned with resolving the problems between two parties, i.e. agent and principal through increased monitoring as the theory views agents as opportunistic. Agency theory suggests that where there is a separation of ownership and control of a firm, the potential for agency costs arises because of conflicts of interest between contracting parties (Chau & Gray, 2002). Due to greater agency problems (M. C.  and higher political costs (Tsamenyi, Enninful-Adu, and Onumah 2007), firms are expected to disclose more information voluntarily. In terms of corporate governance disclosure, as predicted by agency theory (Leftwich, Watss, and Zimmerman 1981;Raffournier 1995), an active policy of disclosure on the part of the company can contribute to reducing the supervision cost for small shareholders and in this was will favour their interest in the development of the business. Traditionally, most studies of the principal-agent theory have focused on investor as the principal. This study focuses on the most relevant agency relationships for banking industry by concentrating on corporate governance disclosures by traditions means, i.e. annual reports.
Managers are motivated to disclose more detailed information to support the continuance of their positions and remunerations and to signal institutional confidence (Gandía 2008). Signalling theory (Spence 1973) proposes that companies become more interested in signalling their good position to stakeholders. The use of disclosure as a signal of firm value have been widely researched (Hughes 1986). Signalling theory suggests that voluntary disclosures are on means for companies or managers to distinguish themselves from others on such dimensions as quality and performance (Xiao, Yang, and Chow 2004). Inchausti (1997) have employed signalling theory and found that management discloses more information in the situation of high profits or good news than bad news. Signalling theory has been utilised in the investigation of voluntary internet financial (Xiao, Yang, and Chow 2004), corporate governance (Gandía 2008) and environmental disclosure (Joshi and Gao 2009). This study thus plans to implement propositions of signalling theory to understand the corporate governance compliance among listed banks in Bangladesh.

Data and Research Methodology
The sample for the study is drawn from twenty-nine banks listed on the Dhaka Stock Exchange as at the end of 2014 and Table 1 contains a summary of the sample selection procedure. There are 56 scheduled banks operating in Bangladesh. Among them, specialised, state-owned and foreign commercial banks are not considered while selecting the sample. Among 39 private commercial and Islamic banks, ten could not be selected due to lack of availability of required annual reports at bank websites. Thus, the final sample of the study consists of twenty-nine banks (21 private commercials and 8 Islamic) with four years of data for each bank providing a total number of 116 observations. The CG compliances were extracted from the sampled companies' annual reports through a self-developed disclosure index with Bangladesh Bank's guidelines for corporate governance. We set two criteria for bank selection in our final sample: the availability of a company's full four-year annual reports from 2011 to 2014 inclusive and the availability of a bank's corresponding accounting data for the same period. Following the studies of Barako (2007) and Eng & Mak (2003), such criteria were deemed helpful in meeting the requirements of a balanced panel data analysis. Panel data provides both cross-sectional and time-series observations, more degrees of freedom and less multicollinearity among variables (Gujarati 2003). Thus, examination of four-year balanced panel data mat help in determining whether the observed cross-sectional link between voluntary CG compliance and its drivers also hold over time. Disclosure among Islamic and Conventional Banks

Research Methodology: Definition of Variables and Model Specification
We have constructed a disclosure index to measure the extent of corporate governance compliance named (CGCI) with the guidelines provided by Bangladesh Bank. The CGCI contains 87 CG provisions based on the seven broad sections covering: duties to depositors and customers, disclosures, the board of directors, credit assessment and asset monitoring, debt recovery, risk management and corporate governance compliance. The CGCI is constructed by awarding a value of -1‖ in any of the 87 CG provisions is disclosure in the annual report and 0 otherwise. Thus, bank's CG compliance score varies between 0 to 100 percent, with higher index level indicating better compliance and disclosure.
Some control variables are included in the study to reduce potential omitted variable bias. Independent variables included in this study are common across prior studies on CG compliance. Table II provides summary definitions of all variables employed, including the dependent (CGCI), independent and control variables. Assuming that all relationships are linear, our main ordinary least squares (OLS) regression equation to be estimated is specified below: Where, CGCI is the CG compliance index; α 0 the constant term; Big5Audit is the reputation of the external audit firm; BS is the board size; IND is board independence, CONTROLS include gearing (GEAR), firm size (SIZE), profitability (ROA); ε the error term. Empirical results relating to the descriptive statistics and panel regression results are discussed in the following section of the paper. Table 1 provides a summary of descriptive statistics for all variables included in our analysis. Board size ranges from a minimum of 6 to maximum 22; leverage ranges from -3.37 to 24.17 and size ranges from 23.36 to 27.20. The figure of audit quality and board independence as well as control variables includes substantial variations, hence minimising possibilities of sample selection bias. Corporate governance compliance score has a standard deviation of 0.28 while having a mean score of 0.66. This indicates on average banks comply 66 percent of the 87 CG provisions examined. Correlations table provided in Table IV indicates that there is some degree of correlations among independent and dependent variables, thus indicating toward a valid regression model.  Table 2 reports the reports the panel data regression analysis of the determinants of CG compliance. The study has utilised pooled, fixed effect and random effect designs. Column 3 and 4 provides the pooled OLS regression results. It can be observed that audit quality has a significant negative impact on corporate governance compliance while board size and board independence are found to be non-significant predictors of CG compliance. Among the control variables, ROA has a negative impact while size has a positive impact on CG compliance. Leverage is found to have a non-significant impact on CG compliance. Note: ***, ** and * significance at 1, 5 and 10% level, respectively

Descriptive statistics
Column 5 and 6 provides the fixed effect regression results. It can be observed that only audit quality has a significant negative impact on corporate governance compliance while all other variables are found to be non-significant predictors of CG compliance. The random effect regression provides many similar results compared to the pooled OLS regression results. It is observed that the adjusted R 2 is highest for fixed effect model (.466) while lowest for the random effect model (0.088). Next, Hausman test is performed to determine whether fixed or random effect model is significant for the study. Table VI indicates with significant value that fixed effect model provides better estimates for the developed balanced panel regression model.

Moderation Analysis
The moderation influence of the nature of banks on the extent of corporate governance compliance was tested using the expand command in Views. A dummy variable was created with a score of 1 for Islamic banks and 0 for conventional banks. Equation 2 provides the modified regression equation with the mediator. Regression results provided in Table VI indicates a significant moderating effect of the nature of the Bank.
CGCI it = α 0 + β 1 Big5Audit + β 2 BS + β 3 DUAL + β 4 IND + β 4 Moderator + ∑ CONTROLS + ε it -------(2) Note: ***, ** and * significance at 1, 5 and 10% level, respectively The constant tells us that if all variables were equal to zero, we would expect the extent of corporate governance compliance to decrease by 14.7 percent for Islamic Banks. The coefficient of the interaction term is negative and significant, indicating that corporate governance compliance is 12.7 percent lower for Islamic banks compared to conventional banks. Evidence of such low level of governance compliance can be seen from corporate governance compliance score provided in Table VII (see the appendix). It is also observed from the OLS regression results that, audit quality and profitability play a negative role in the extent of CGC for Islamic banks while size positively affects CGC. The modified model was able to predict 118 percent of the variations in the extent of CGC.

Summary and Conclusion
This paper has aimed at determining the factors that influence corporate governance factors among listed banks in Bangladesh. Also, the moderating impact of the nature of bank is provided to analyse whether there is a difference in the extent of governance compliance between Islamic and conventional banks. Seven hypothesis were formulated to test the impact of governance and firm-specific constructs on dependent variable (corporate governance compliance).
H 1 states that audit firm size has a positive association with corporate governance compliance by banks. Both OLS and fixed effect regression results have indicated toward a negative association between these two variables in the context of Bangladesh. So, hypothesis H 1 is rejected. However, these results provide justification to the results found by Barako (2007). In hypothesis H 2 the second governance variable, board size, is introduced and hypothesised to have a positive association between the level of corporate governance compliance. Table V indicates toward a non-significant association which is similar to the findings of Ho & Williams (2003) and Mangena & Chamisa (2008). As a result, H 2 is rejected. H 3 introduced a positive association between board independence and governance compliance. The non-significant association is found through balanced panel regression and thus H 3 is rejected.
Three firm-specific variables, i.e. leverage, profitability and size were also introduced in the regression model. H 5 states that leverage has a negative association with the level of governance compliance by banks. Fixed effect model in Table V provides a non-significant negative association and thus H 5 is rejected. Profitability is hypothesised to have a positive impact on CGC in H 6 . Although Ismail & El-Shaib (2012) reported a positive association, this study validates the negative association between profitability and corporate governance compliance following Xiao et al. (2004). Finally, H 7 states that size of the bank has a positive influence on the extent of CGC. Table V provide justification toward this claim by providing a significant positive result. Thus, H 7 is accepted.
The study has ended with a moderation analysis by a dichotomous variable. The new variable Islamic had values 1 for all Islamic banks and 0 for all conventional banks. Due to the smaller amount of observations for Islamic banks, separate regression was not conducted. Instead, the moderation analysis was conducted in EVIEWS using the @expand command. Table VI provides the OLS regression results for the modified equation with a moderating variable. The model was found to have a significant negative impact of CGC which indicates that the nature of the bank moderates the extent of corporate governance compliance. The level of CGC is decreased for Islamic banks as compared to conventional Banks which is evident from the scores obtained from the CGC index (see Table VII in the appendix).
This study has made several contributions to the governance literature. First, the propositions of agency theory that higher governance mechanism will lead to increased compliance is not found valid for banks operating in Bangladesh. The negative association of external audit size with governance compliance indicates that banks appointing big audit firms for annual audit function are less motivated to comply with the central bank governance compliance guidelines. It is a relief to see that big banks are complying more with the governance standards but while isolating the profitability issue, we can see that banks with higher profits are not the one with the highest compliance score. The agency conflict is evident in this scenario as the profit maximisation goal of a manager is affecting their decision toward governance compliance which is beneficial for the stakeholders. So, central banks should be focusing on enforcing strict laws to ensure a greater level of governance compliance among not only bigger but also profitable banks in Bangladesh.
Finally, the difference in governance compliance between Islamic and conventional banks, more specifically, the lower level of compliance among Islamic banks has established a separate research agenda which could not be discussed in this paper. The primary reason is that the index used to determine the governance compliance have focused on the conventional issues while a separate index grounded with Maqasid-Al-Shariah framework might provide better results for Islamic banks.