Board busyness and new insights into alternative bank dividends models

This study examines the possible opposing effects of the board function of busyness (i.e. the presence of busy independent non-executive directors serving on multiple boards) on bank dividend payout patterns between two alternative payouts models (i.e. conventional and Islamic). Using an international sample for listed banks during the periods of 2006–2018, we show that the busyness of boards of directors can explain differential dividend payouts behaviour between two banking systems. For conventional banking dividend model, a busy board has a significantly positive impact on the bank’s dividend payout level. However, during the financial crisis of 2007/2009, the positive impact of board busyness on dividends payouts is tempered for these banks. In contrast, Islamic banks operating under a more constrained dividend model, report significantly lower levels of payouts and lower likelihood when they have busy directors on board. We find insignificant evidence for the effect of the financial crisis in Islamic banks. These results highlight a potential challenge for the unique agency conflicts arising from the complex payout model of Islamic banks (in terms of profit distribution principles, motives, mechanics and techniques, and flexibility of payouts), which is subject to the demand for greater monitoring and additional rulings when compared to the conventional.


Introduction
The 2007-2009 global financial crisis appears to have brought a more controlled operational environment to banking and increased complexity in governance with additional calls for effective monitoring by the board of directors. This was followed by increased public calls and support from policymakers in designing effective board governance in banks to create a more ethical and sustainable value and to align the interests of managers with those of shareholders and other stakeholders . Although dividend payouts strategies have been investigated over 50 years, since Miller's seminal work (1958, 1961), it remains a 'puzzle' from an agency perspective. Dividend payout is an implicit governance tool in reducing agency costs between shareholders and managers (Sharma 2011;Onali et al. 2016;Mulyani et al. 2016). This is because the monitoring needs of capital providers are lower since the amount of free cash flow is reduced after distributing dividends, leading to a lower probability of managers wasting excess available cash (e.g. DeAngelo and DeAngelo 2000; Harford et al. 2008). In line with the risk aversion perspective, managers are likely to have a lower risk tolerance than firm shareholders since they might have substantial personal gains/incentives tied up with the firm's performance. As identified by Easterbrook (1984), shareholders might have the preference for higher dividends payouts, which reduce retained earnings, and forces managers to raise external funds.
Managers commonly use dividend payment to lessen agency costs (DeAngelo and DeAngelo 2000) but a primary concern is related to the fact that those managers have a control and can exercise discretion over the dividend strategies and payout levels. Because dividends reduce the proportion of discretionary funds available, those managers might have opportunities/incentives to manipulate the payout ratios to guarantee that excess profits will be retained within the company to meet self-objectives, ceteris paribus (Easterbrook 1984). Therefore, a board of directors (BOD) provides an essential internal governance mechanism to prevent such managerial discretion. This board has an ultimate oversight responsibility to scrutinise payouts policies, including the levels of payments before announcing dividends to the capital markets (White 1996). Such responsibility involves considerations of various factors related to a firm's growth opportunities, current leverage, and potential emergencies before approving a payout. Therefore, the BOD has an essential role in influencing and controlling agency costs associated with payouts process of dividends (Sharma 2011).
Appointing a busy BOD (i.e., independent directors holding multiple board seats across many firms) can jeopardise the board's effective control over payout policies. In line with the resource dependence theory, effective monitoring by BOD is vital for efficient resource allocation and risk mitigation (Meng et al. 2018;DeBoskey et al. 2019;Trinh et al. 2020a). However, an intense monitoring by boards is argued to be costly and complex for some institutions particularly in financial institutions (John et al. 2015), and a busy board might be challenged to fulfil their supervisory/advisory roles due to the limited time/efforts available from working across several banks' boards (Fich and Shivdasani 2006;Trinh et al. 2020b). Under the expectations of weak oversight by busy board, conflict of interests between shareholders and managers are likely to emerge. Board busyness is likely to lead to high agency costs, related to free cash flow and optimal choices of financial policies. Such associated conflicts have adverse impacts on dividend payout levels, investors' perceptions and stock market valuations (Sharma 2011;Chou and Feng 2019;Elnahass et al. 2020a). However, employing a busy BOD might still bring some reputational benefits to firms such as promoting extended business networking as well as connections, and quick access to market resources ). As such, banks employing busy outside directors might positively influence their dividends policy through flexible access to capital markets to raise funds at lower costs. Also, multiple directorships are likely to enhance internal board monitoring and reduce the agency problems of a firm's liquid assets. They may also result in using the cash more effectively and providing direct benefits to shareholders and dividend policy.
Under these two opposing views for board busyness, there is a relatively limited evidence on the impact of effective systems of governance and the influence of busy boards on the dividend policy within the banking sector. Dividend payouts strategy in the banking industry makes up a crucial pillar for their rigorous and prudent risk management (Kanas 2013); which became subject to stricter scrutiny by policy makers (Lepetit et al. 2018). In conventional banking, prior studies provide mixed evidence and they mainly focus on industrial firms (i.e. Sharma 2011; Akhigbe and Whyte 2012;Onali et al. 2016;Kutubi et al. 2018;Chou and Feng 2019). 1 Alongside this inconsistent finding, there is a clear gap in the literature to study alternative dividends payout models by different bank types (i.e. Islamic and conventional banks). While comparative assessments across the two bank types have been evolving over the past few years (e.g., Mollah and Zaman 2015;Mollah et al. 2017;Elnahass et al. 2020a;Trinh et al. 2020a, b), however, none of these studies have explicitly examined the association between board busyness and dividend payouts under the different banking business models like the Islamic and conventional models. Thus, investigating the influence of busy boards across Islamic and conventional banks will contribute to the ongoing debate identifying the impacts of different institutional characteristics and the importance of an effective governance system within the domain of dividend strategies in financial institutions.
Islamic banks operate on a non-interested based model which is governed by Shariah rulings with the aim to promote profit-loss sharing between the bank and depositors and to reduce uncertainty and speculations in trading/allocation of fund resources which are prohibited in Islam. The governance structures adopted by Islamic banking are more complicated compared to their conventional counterparts (Mollah and Zaman 2015). In both bank types, the BOD is responsible for the execution of strategic decisions, protection of the shareholders' interest and maximisation of the bank value. Based on some theoretical arguments by prior studies, there are some differences between the dividend policies of Islamic and conventional banks (e.g. Athari et al. 2016;Safiullah and Shamsuddin 2019). These differences are mainly related to dividend distribution principles, motives, mechanics and techniques, and flexibility of payouts. In general terms, the distributions of profits in Islamic banks must be compliant with Shari'ah principles (Duqi et al. 2020) and hence, their payouts involve a nexus of contracts between the bank, depositors and shareholders (Alhabshi 2002). The distributions of profits and the payouts policy within Islamic banks are more complex and less flexible than that of conventional banks. Islamic banks are usually challenged by liquidity management issues and accessing short-term borrowings from external sources (Čihák and Hesse 2010;Beck et al. 2013). Subsequently, they hold substantial excess free cash flow or other liquid assets at a low rate of return to meet expected/unexpected capital challenges and regulatory capital requirements (Elnahass et al. 2018). These constraints can have implications on the dividend payout strategies in Islamic banks, leading to low payouts ratios and less stable dividends distributions in the long-term (Athari et al. 2016). In contrast, conventional banks have quicker access to market sources and can use alternative financial instruments such as derivatives and options, which are prohibited in Islamic banking. These liquidity instruments are likely to promote greater flexibility to support dividend payouts strategies (Bitar et al. 2017).
Given the constraints imposed on the Islamic banking business model, the extended agency costs associated with this banking model and under its complex governance structure (see Abdelsalam et al. 2016), employing busy outside directors can have adverse implications on board of directors' monitoring abilities over dividend payouts. For example, board busyness might encourage managers to ex-post deviate from the payout policy and engage in poor or risky investment decisions. The negative impact of board busyness is likely to be more pronounced in Islamic banks than conventional banks (Elnahass et al. 2020a;Trinh et al. 2020a). Weak monitoring by the busy board can thus result in adverse selection and moral hazard problems on both sides of the balance sheet of Islamic banks (Visser 2009). Contrary, conventional banks operating on a less constrained business and dividend model and single-layer governance structure would have lower agency costs. Therefore, conventional banks can obtain the reputational benefits from busy boards through their improved internal monitoring ability. Accordingly, our premise in this study is that having a busy board in Islamic banks is likely to be associated with a lower payout ratio than that in conventional banks.
We use a comprehensive sample of 742 bank-year observations (70 listed banks) in 11 countries from the periods of 2006-2018. We focus on fully-ledged Islamic and conventional banks while dropping banks with Islamic windows. 2 Our final sample focuses on countries classified as emerging capital markets, which have long been ignored in empirical examinations by prior research. This set of countries operates on a dual banking system and report remarkable discrepancies in their dividend payouts policy, both in nature and characteristics of payments, relative to developed countries (Jabbouri 2016). Previous studies (e.g. Aldoseri and Worthington 2016) also document that these emerging capital markets have lower information asymmetry, more volatile, and smaller size than developed markets. Our focus on these set of countries also responds to calls by prior literature to investigate dividends policies in line with the several particularities reflected by emerging countries (e.g. Lagoarde-Segot 2013; Jabbouri 2016). Islamic banks in these countries are highly concentrated and well-established (Aldoseri and Worthington 2016). Hence, this study offers an ideal setting for comparative assessments among different banking systems.
Findings show that for the full sample (i.e. conventional and Islamic banks together), banks with busy boards exhibit, on average, higher payout ratio. This finding is consistent with the reputation hypothesis, suggesting that busy directors can use their expertise and connections to support effective dividends policy. Analyses conditioned on the bank type support our expectation and show that a conventional bank with a busy board offers significantly higher cash dividends and are more likely to pay dividends relative to Islamic banks. Having busy boards in Islamic banks leads to a detrimental effect on the dividend payout levels and likelihood. These results suggest that dividend strategies of Islamic banks are significantly influenced by board busyness, unlike their conventional counterparts. These results are consistent with expectations and attributable to the constrained dividend model used in Islamic banking. Our extended analyses show that conventional banks with busy boards have a higher likelihood to pay a dividend than Islamic banks. We also examined the effect of the financial crisis of 2007-2009 and find that during the crisis period, the impact of board busyness on dividends payouts is significantly lower within conventional but we observed insignificant effect within Islamic banks. Moreover, we find a non-linear linkage between busy boards and dividend level within conventional banks. That is, the reputation effects diminish proportionally as the outside board directorships increase. However, we merely find a simple linear relation between board busyness and Islamic bank dividends policy.
This study contributes to prior literature in several ways. To the best of our knowledge, this study is the first to examine the impact of board busyness and dividend policies in the banking industry and within a broader international context representing emerging capital markets. Hence, we offer updated evidence to prior studies in dividends policies for emerging countries (e.g. Baker et al. 2011;Jabbouri 2016). Our results also contribute to the inconclusive evidence within the U.S context, which only examined non-financial industries (e.g., Sharma 2011;Chou and Feng 2019). This study is also the first to investigate the possible differential impacts on payout policies across different bank types by utilising an important board attribute such as busyness and exploiting a unique dividend model of Islamic banking. Such comparative assessments between the alternative dividend models employed by Islamic versus conventional banks is necessary to extend both prior theoretical studies within the Islamic banking context (e.g., Al-Gurrah Daghi 2009;Essa 2010) and conventional banking studies (e.g. Sharma 2011;Jiraporn et al. 2011;Chou and Feng 2019). Moreover, we contribute to the growing stream of Islamic-conventional banking literature (e.g., Abdelsalam et al. 2016;Mollah et al. 2017;Alandejani et al. 2017;Alqahtani et al. 2017;Safiullah and Shamsuddin 2019;Duqi et al. 2020). Finally, our study adds to the ongoing debate about the effect of institutional characteristics and stricter governance mechanisms on several firm outcomes such as firm performance, risk-taking, capital structure, and cost of debt and cash holdings (e.g., Brown and Caylor 2006;Harford et al. 2008;Cheng et al. 2008;Meng et al. 2018).
The findings in this study provide important implications to bank regulators, investors and stock markets which engage with both bank types. Among the two bank types, we observe that dividend strategies are substantially affected by this attribute of the BOD (i.e., board busyness). For conventional banks, regulators and market participants can benefit from the empirical evidence presented in this study, which highlights that BOD busyness promotes several reputational benefits, which are likely to enhance dividend payout strategies for banks. This offers important implications for wealth creation and the proper investment decisions of investors. Nonetheless, such reputational benefits associated with recruiting busy boards might not be invoked in the presence of unique systems of governance and constrained banking model, as presented by Islamic banks. Accordingly, the findings are likely to inform the investment decisions of market participants who engage with the two bank types and for policy makers who govern countries with dual banking systems. Furthermore, as the international capital markets and regulatory standards are continuously revisited to promote high financial reporting quality systems and effective mechanisms of governance, our results might assist regulators in explaining the differential payouts patterns, which are conditional on the bank type. Moreover, we offer new venues into the examinations of dividend policy in emerging economies, which can support current regulatory reforms for multiple directorships and board diversity (Lagoarde-Segot 2013; Jabbouri 2016). Finally, Islamic banks can learn from their conventional counterparts on how to promote reputational benefits associated with payouts policies, including effective mitigation of extended agency issues when recruiting busy boards.
The next section presents the background and theoretical framework. Section 3 outlines the study hypotheses. Section 4 presents data and sample. Section 5 reports methods and measures. Section 6 discusses the findings and additional sensitivities. Finally, Sect. 7 closes the study.

Theoretical framework
The agency theory represents one of the dominant views on dividend policies (see Nohel and Tarhan 1998;Renneboog and Szilagyi 2015). Prior studies have long argued that managers (e.g. CEOs) have their strong incentives to engage in discretionary acts (Williamson 1964), take excessive compensation and perquisites (Sharma 2011), and utilise excess free cash flow in unprofitable projects. Easterbrook (1984) highlights that the monitoring and risk-aversion preferences might lead to agency problems between managers and shareholders and hence result in the arisen of cash dividends. Managers are likely to manipulate and shift the amounts of dividends across future periods (i.e. the earnings smoothing effect) if they have motives to increase the dividend ratios despite the low level of permanent earnings. A managerial discretion to establish the payout policy can exacerbate the agency problems between managers and shareholders; such conflict is notably more severe in banks because of their highly levered capital structure (John et al. 2010). Moreover, according to Filbeck and Mullineaux (1993) and Collins et al., (1994), dividends are usually employed as a signalling mechanism by banks. For example, banks can convey useful information to investors about the bank growth opportunities through dividend payouts (Abreu and Gulamhussen 2013).
The context of intense regulation and higher asymmetry in the banking sector lead to the unique relevance/role of the BOD, which has a legal responsibility in approving a bank's policies, procedures and business strategies. This board would have an ultimate oversight function for bank decisions (Elyasiani and Zhang 2015;Lu and Boateng 2018). The duties and obligations of the bank directors (i.e. inside and outside directors) serving on the board may arise in two primary contexts. First, they need to bring a discrete decision to the board for approval, which results in a rise of directors' legal responsibility on bank safety and soundness. Second, they must provide an effective bank oversight for the bank operations (Elyasiani and Zhang 2015). Also, many stakeholders (e.g. authorities) have placed additional expectations on bank BODs that delineate their responsibilities even further. Outside directors serving on the BODs should have either advisory or oversight role, or both over executives. They should also perform their tasks independently from inside directors in which they can provide vigilant scrutiny over inside board members on behalf of shareholders and, thus, may reduce agency problems (Fama and Jensen 1983;Harkin et al. 2019). To monitor managers more effectively, those outside directors might be required to invest their time, attention and efforts to analyse any information provided by managers, banks and consultants .
Prior studies provide mixed evidence for the effect of corporate governance on dividend policies. Within the industrial sector, the literature identifies the impact of BOD characteristics on firms' dividend payout. This includes board size (Van Pelt 2013), independent directors (Setia-Atmaja 2010; Boumosleh and Cline 2015), CEO duality (Sawicki 2009), age and experience (Custódio and Metzger 2014), CEO entrenchment (Hu and Kumar 2004;Elyasani and Zhang 2015), and board gender diversity (Saeed and Sameer 2017;Chen et al. 2017). Other studies find a significant and positive impact of corporate governance index (G-Index) on the likelihood of paying dividends and/or dividend yield in the US market (e.g., Jiraporn et al. 2011). Hu and Kumar (2004) show that CEO entrenchment is likely to increase dividend payout ratios. Setia-Atmaja (2010) also shows a positive influence of board independence and dividends paid in family-controlled firms. Renneboog and Trojanowski (2011) argue that the voting power of executive directors has a significant relation to the propensity to pay dividends or the combination of dividends and share repurchases. Deshmukh et al. (2013) find that an over-confident CEO pays lower levels of dividends than a rational CEO to accumulate higher financial slack for future investment needs. Caliskan and Doukas (2015) document that inside debt induces CEOs to pay dividends while convex CEO compensation is related to the lower payouts.
Board busyness and dividends policies have been previously investigated in non-financial firms and within the US context. For a sample of US non-financial firms, Sharma (2011) shows that the decision to pay out dividends is associated with the strength of board governance, which is measured through the level of board busyness. Chou and Feng (2019) find that when industrial US firms have more limited investment opportunities, board busyness is positively associated with higher dividend payouts. They explain that multiple directorships are likely to enhance internal board monitoring and reduce the agency problems of a firm's liquid assets. They also suggest that board busyness results in using the cash more effectively and providing direct benefits to shareholders. Other studies identify the influences of busy BODs while focuses on firm performance, market value, cost of debt and/or risk-taking (e.g. Ferris et al. 2003;Field et al. 2013;Chakravarty and Rutherford 2017).
Studies in the banking sector offer limited evidence of the association between governance and dividends payouts. In conventional banking, Theis and Dutta (2009) examine the relationship between inside ownership and dividend payout policies after controlling for the levels of bank capitalisation. Akhigbe and Whyte (2012) find a negative effect of managerial stock ownership and payouts across the financial firms. Onali et al. (2016) find a negative impact of director ownership and CEO power on the dividends of European-listed banks. Recently, Duqi et al. (2020) provide evidence on the important role of ownership structure (i.e. government, family, foreign) in explaining dividend strategies of Islamic and conventional banks, albeit in different patterns. For Islamic banking studies, prior literature on payouts of dividends is scarce. For example, Hassan (2003) use the signalling theory and show that dividends are only relevant financial information, which helps managers to signal returns on investments to the stock market. They emphasise the importance of investigating an Islamic bank's dividend model determinants. Al-Gurrah Daghi (2009) and Essa (2010) describe from a theoretical context the accounting process used in the profit distribution of Islamic banks and refer to the relevant financial/accounting standards.
Accordingly, evidence on the board busyness and dividends payouts within the banking industry in general terms and among different bank types in specific terms, is scant. None of the prior studies examined the influence of different bank types as a mediating factor for this possible association between board busyness and dividend payout decisions. We, hence, seek to fill in these gaps through a comparative assessment of Islamic and conventional banks.

Alternative dividend models
There are two major features which distinguish the Islamic from conventional business model; the dominance of risk sharing between the bank and shareholders alongside the existence of additional monitoring required through a separate board; the Shari'ah supervisory board (SSB). 3 Under the conventional banking finance paradigm, a bank is likely to shift credit risk to the depositors under an interest-based contractual arrangement (Safiullah and Shamsuddin 2019). Contrarily, in line with the Shari'ah guidelines, Islamic banks are expected to perform their intermediation functions through profit-and-loss sharing (PLS) contractual agreements between the banks, depositors and investment account holders (IAHs) (Alandejani et al. 2017). According to the PLS paradigm, entrepreneurs share their profits and losses with Islamic banks according to a pre-determined ratio. Islamic banks pool together all profits and losses from different investments and then share the profits with depositors of funds taking into account the relative contributions of capital and equity and the investment deposits (Olson and Zoubi 2008). A proportion of the remaining earned profits is used to pay dividends to equity holders, for which dividends on common equity is discretionarily allocated and distributed by the bank managers (Khan and Mirakhor 1989).
Within the Islamic banking context, dividend policies vary across different countries because of their variations in the government regulations and tax policies. For example, in the six Gulf States, which operate on a dual banking system, there is no dividend tax required and Islamic banks must follow the Shari'ah principles related to Zakah payments (i.e. charity donations) (Athari et al. 2016). Islamic banks operating in Bahrain and Qatar adopt specific Islamic accounting standards developed by the Auditing Organization for Islamic Financial Institutions (AAOIFI). 4 In other countries (e.g. Turkey, Malaysia and Indonesia), Islamic banks do not strictly follow the Islamic accounting standards, the financial reporting systems should comply with IFRS besides the central banks' local standards for Islamic banking (Chong and Liu 2009).
There are several structural differences between the Islamic and conventional dividends models which can represented in terms of the distribution principles, the extent of flexibility of payouts and the mechanics and techniques (e.g. Ayub 2007;Beck et al. 2013;Athari et al. 2016). We summarise these key differences in Table 1. These differences are expected to influence the governance monitoring effectiveness of both bank types and the overall levels of dividend payouts.
First, a payout policy in an Islamic bank is likely to be less flexible than that of a conventional bank. While the dividend distribution decisions of the former are significantly affected by their challenges in managing liquidity and accessing short-term borrowings from outside sources (Beck et al. 2013;Elnahass et al. 2014), the latter has better liquidity opportunities promoted by their ease and quick access to external market sources and the availability of alternative instruments to raise funds such as hedging and derivatives (Bitar et al. 2017;Deng et al. 2017). Islamic banks, therefore, are likely to hold higher capital buffers to mitigate their liquidity challenges and preserve their regulatory capital ratios (Elnahass et al. 2018). This is consistent with regulations of Islamic banks with Basel III aiming to strengthen global capital and liquidity rules for banks to make them more

Motives of payouts Preferences of both investors (shareholders) and depositors Preference of investors (shareholders)
Conflicts between depositors and shareholders towards dividend payouts ratio for the latter High Low Depositors' return is linked to the return on assets Yes

No
Banks' pooling of depositors' funds to provide depositors with professional investment management

Yes No
Process Activeness High. Profit distribution is a more active process involving a nexus of contracts between the bank, depositors and shareholders. Hence, the profit distribution of Islamic banks is agreement among such three parties including depositors Low. Depositors will receive interest payment from the banks. Interests paid for depositors are treated as expenses when calculating net profits and dividends for shareholders. Hence, the profit distribution of conventional banks is only an agreement between shareholders and the bank Complexity of payouts mechanics and techniques High. Dividend decision subjects to the interaction between PSIA and dividend distributions. It depends much on the effectiveness of profit distribution among parties under the PLS arrangements Low. Dividend decisions are not subject to the interaction between PSIA and dividend distributions; however, they are associated with current profitability, future growth opportunities, and optimal capital budget and equity amount needed to finance the optimal budget through retained earnings

Difficulties in payouts
High. It is difficult to determine the actual (Shari'ah) profits for any financial year because some investment projects may not be finished before the end of the accounting year. In addition, Islamic banks cannot use all the available fund to undertake investment activities which challenges their profit/dividend distribution Low. Interest amounts are treated as expenses which are paid to depositors. Such expenses do not depend on the completion of investments and conventional banks can pool and employ all available capital. Net profits (after all expenses) will be distributed to shareholders according to the shareholding percentages Flexibility of payouts policy Low. Dividend decisions appear to be significantly affected by Islamic banks' challenges in managing liquidity and accessing Shari'ah short-term borrowings from outside sources High. Higher liquidity position as they enable quicker access to external market sources and the use of hedging and financial instruments The stricter new capital requirements are also likely to impose a discipline on the better utilisation and maintenance of capital in those Islamic banks. In addition, new liquidity requirements under Basel III might create some difficulties and serious challenges for Islamic banks. That is, they need to meet the minimum liquidity coverage ratio (LCR) but it is difficult to find the high-quality liquid Shari'ah -compliant instruments because of the lack of supplying in High Quality Liquid Assets, non-existence of secondary capital markets, and Interbank and money markets (Ozkan and Iqbal 2015). The existence of limited sources of finance, such as issuing Islamic bonds, to enhance the liquidity and capital position leads to substantial restrictions imposed on the bank business model and dividends strategies (Elnahass et al. 2014). As a result, conventional banks are better positioned to offer more frequent payouts of dividends at higher rates when compared to Islamic banking (Athari et al. 2016). Second, Islamic banks encounter additional challenges related to their actual (Shari'ah) profit determination compared to conventional banks. Under the constrained dividends model, any fraction of earnings which are generated from investments that do not comply with the Islamic principles cannot be distributed to shareholders or used to acquire assets (Safiullah and Shamsuddin 2019). 5 Given that an Islamic bank's contracts should, in principle, be backed by underlying assets or investment activity, in many occasions it is too complex to determine the estimated profits when some projects have not yet been realised before the end of the fiscal period. This can have implications on the bank's dividends payouts. Also, unlike their conventional counterparts, Islamic banks cannot employ all the capital available to undertake investment opportunities, either because the regulations do not allow them, or because the capital available for investment is higher than the Islamic banks' investment portfolio (Ahmed 1996). However, such related complexities and issues are not raised in a conventional banking business model as Islamic rulings will not constrain its distributable profits. Depositors in this bank type obtain their returns in the form of regular/composite interest payments, which are treated as expenses when conventional banks compute their net profits and dividends for shareholders. As such, an important difference between Islamic and conventional banks in this respect is the shift in treating returns payable to depositors as a distribution of shared profits and not an expense (Alhabshi 2002;Saeed and Izzeldin 2016). In contrast to Islamic banks, the interest expenses paid for depositors in conventional banks should be independent of the completion of investment projects. These banks, hence, may have lower difficulties in calculating profits distributable for shareholders. Accordingly, the Islamic banking financial structure of a dividend-based model differs from conventional banks (Schaik 2001;Safiullah and Shamsuddin 2019), which may lead to different payout levels between two bank types.
Third, with the restrictions imposed on the Islamic banking dividends model, which must comply with the Shari'ah principles, profit distributions by Islamic banks reflect an active process involving a nexus of contracts between the bank, depositors and equity holders (Schaik 2001;Alhabshi 2002;Safiullah and Shamsuddin 2019;Abdelsalam et al. 2020). Thereby, the profit and dividend payout decisions of Islamic banks are associated with an agreement among these three parties. In other words, the basis and manner of profit distributions could change subject to the contract agreement among parties. This adds to the main structural differences in the distribution motives of Islamic banks relative to their conventional counterparts. The payouts decisions by Islamic banks' managers are ultimately driven by the preferences of both investors and depositors. 6 In contrast, a sound distribution policy in conventional banks depends solely and mainly on the preference of investors (shareholders) to enhance the bank market value (Al-Hunnayan and Hashem 2011). Therefore, compared to conventional banks, additional monitoring costs imposed on Islamic banks might be needed to avoid investors/depositors' disappointment. 7 Finally, the mechanics and techniques of Islamic bank dividend distributions are likely to be more complicated than those of conventional banks (Athari et al. 2016). A survey by Al-Hunnayan and Hashem (2011) defines a commonly used dividend model in an Islamic bank and summarises its key structures based on four steps; (i) revenues and expenses allocation; (ii) reserves and provisions deductions; (iii) distributions for profit and loss saving and investment accounts (PSIA); and (iv) distribute dividends (see "Appendix A"). At each step of this payout process, there are potential variations in the practices of Islamic banks. Moreover, under the PLS paradigm, the dividend decisions by Islamic banks managers are subject to the interactions between PSIA and dividend distributions. In contrast, conventional banks are known as intermediates between depositors and borrowers, and their revenue is defined as the difference in the interest gains between the two parties. Thus, their net profit is calculated by the deduction of expenses from revenues (Saeed and Izzeldin 2016). Payout decisions in conventional banks, nevertheless, are related to current bank profitability, future growth opportunities and optimal capital budget as well as equity amount needed to finance the optimal budget via retained earnings (e.g., Deshmukh et al. 2013;Onali et al. 2016). more opportunities to amend the reserves and provisions, profit distribution rates on PSIA and the dividends, which could lead to severe agency conflicts. Such a wide latitude of discretion further adds to the complex structure of the dividend model employed by this banking sector.
The BOD's characteristics and attributes are likely to affect the dividend strategies for both bank types. However, the monitoring needs in Islamic banks by BOD are likely to be higher than conventional banks due to their complex governance structure and strict trading process (Safieddine 2009). Under the constrained Islamic banking model, BOD has additional responsibilities related to the establishment of the appropriate Shari'ah governance framework besides the development of relevant policies to ensure that all activities are conducted in compliance with the Shari'ah law.
Therefore, the extent and effectiveness of systems of governance within the two banking sectors are expected to have implications on their dividend payouts. Unlike conventional banks, there is a scarcity of outside directors who have expertise and knowledge in Shari'ah legitimacy to support the complex payout structure within Islamic banks. Hence, busy BOD in Islamic banks are expected to be less capable of providing the necessary level of oversight and monitoring functions. This is because a busy board might have less involvement (i.e. time, attention and efforts) to thoroughly review the long-term strategies and investment opportunities, which must be compliant with the Islamic principles, and make indicative decisions for dividend distributions. Hence, the pitfall related to less effective monitoring can lead to lower dividend levels as managers can pursue their interests at the expense of shareholders. According to the busyness hypothesis (e.g., Sharma 2011;Cashman et al. 2012), busy outside directors are less likely to effectively monitor managers' risk-taking and expropriation behaviours for banks as they overstretch themselves across too many companies and spend less time on each board. Moreover, busy boards may not have sufficient reputational benefits to contribute to their institutions ). As such, an increase in their workload is closely associated with a decline in dividend payouts (Sharma 2011). Thereby, an inverse association between busy BOD and dividend payout in Islamic banks is predicted.
Unlike Islamic banks, conventional banks operating on a single layer of governance and a more flexible/stable dividend model tend to encounter relatively lower agency costs. For this specific banking business model, busy boards have several opportunities to promote additional reputational benefits for their banks (e.g., provides advising services on payouts policy, brings flexible and alternative funding sources) to the conventional banks' dividend models by enhancing board internal monitoring and mitigating the agency problems of a firms' liquid assets, managerial opportunism and uncertainties (e.g., Chou and Feng. 2019). However, these benefits are less likely to be obtained by Islamic banks because their business model is marked with greater complexity. Therefore, board busyness in conventional banks is predicted to be more beneficial from a reputational and expertise perspectives when compared to their Islamic counterparts (Elnahass et al. 2020a). Such reputational benefits are likely to influence the dividends business model of conventional banking, leading to possibly higher payouts levels.
Based on the identified differences for the dividends models among the two bank types, we conjecture that board busyness has adverse impact(s) on dividends payouts within Islamic banks unlike conventional banks. Having a busy board is anticipated to cause pronounced negative consequences on the payouts levels in Islamic banks relative to conventional banks. This forms our hypothesis stated in an alternative form, as below: H 1 Islamic banks with a busy board of directors pay lower levels of dividends than conventional banks.

Data and sources
Dividend and other consolidated financial data (in thousand U.S. dollars) are collected from DataStream, Bankscope and Bloomberg. Governance-level data is obtained from annual reports, which reflect board members' profile for both listed Islamic and conventional banks, including the number of directorships of outside directors, the number of directors and independent directors on boards, among others. 8 Macroeconomics and country governance indicators used in our tests are obtained from the World Bank database. We started the data collection process with the list of Islamic banks trading on global stock markets. Consistent with prior Islamic-conventional banking literature (e.g., Beck et al. 2013;Mollah et al. 2017;Elnahass et al. 2020a, b;Trinh et al. 2020a, b), four criteria are applied to filter our sampled banks which are: (a) banks located in countries which have both bank types, and at least two listed banks; (b) banks with annual reports published in their official website and of the financial year of 31 December; (c) banks are classified as commercial full-ledged; and (d) banks have at least three consecutive years of data availability. The availability of corporate governance, dividend and financial data for banks located in countries with dual banking systems have reduced the study's sample size to 70 listed banks operating in 11 countries. 9 These countries are classified as emerging economies where dividend payouts policies are different in either nature or characteristics from those adopted in developed markets. These differences are mainly associated with challenges in raising equity fund, the effect of controlling shareholders who prefer re-investing into future projects than distributing dividends, the extent of investor protection, the quality of market infrastructure and legal uncertainties (see Baker et al. 2011;Jabbouri 2016).
The final sample represents 27 Islamic (295 firm-year observations) and 43 conventional banks (447 firm-year observations) during the period from 2006 to 2018. We are challenged by the availability of high-quality of data for Islamic banks, which are very limited before 2006. Hence, we identified year 2006 as our cut off beginning year for data collection. We also emphasise that Islamic banks have been subject to several regulatory changes before 2006 including the mandatorily adoption of Basel II capital adequacy requirements as of the end of 2006 (see Elnahass et al. 2018). Therefore, our study's sample period offers simultaneous and comparative examinations across both Islamic and conventional banks while recognising the effect of during the financial crisis of 2007-2009. Table 2, reports the sample distribution by country and bank. The percentage of bank representations is 39.8% for Islamic banks and 60.2% for conventional banks. Both Bahrain and Bangladesh have the highest concentration of Islamic banks, while Indonesia and Bangladesh represent the highest concentration of conventional banks.

Bank dividend policy and board busyness
We follow prior literature to measure our dependent variable; cash dividends over total net income (DIV/NI) (e.g. Jiraporn et al. 2011;Chen et al. 2017). This proxy represents the proportion of cash dividends paid to the shareholders over the earnings reported in a given period. We treat DIV/NI as a censored variable since it cannot be below zero (Jiraporn et al. 2011). Following previous literature (Onali et al. 2016;Jiang et al. 2017;Chen et al. 2017), we further use alternative measures of dividend strategies, comprising of the propensity to pay dividend (LIKE_PAY), dividends over total assets (DIV/Assets), dividends over sales (DIV/Sales) and dividends per share (DIV/Share), in our sensitivity tests. Our main variable of interest is busy BODs. A busy outside director is defined as an individual who serves in at least two outside firms (Ferris et al. 2003;Fich and Shivdasani 2006;Elyasiani and Zhang 2015). We focus on outside directors since those directors' primary duty is to monitor the management board while inside directors serve in the BOD for several other purposes (Cashman et al. 2012). Based on this, we measure a busy BOD as the ratio of outside directorships per outside director (i.e. ABOD), representing the average number of other outside board seats held by each outside director. It is computed as the total number of additional (outside) boards occupied by outside directors divided by the number of outside directors on the board (Ferris et al. 2003;Chou and Feng 2019).

Model specifications
To the extent that dividends are expected to mitigate agency costs of managerial expropriation and overinvestment (Easterbrook 1984;Jensen 1986) and under the predictions of relatively higher agency conflicts arisen from the payout process in Islamic banks compared to their conventional counterparts, we conjecture that conventional banks with busy BOD are more likely to pay higher dividends to shareholders than Islamic banks. In our estimations, we test the possible relation between busy BOD (ABOD) and dividend payout ratio (i.e. DIV/NI). However, since managers pay dividends to shareholders in ways that align interests between shareholders, managers and directors, board busyness and payouts decisions are likely to be determined endogenously. For example, busy outside directors can choose to work for banks with high dividend payout (e.g. Sharma 2011). Also, banks could simultaneously select busy outside directors and dividend policies to address agency problems of free cash flow. Therefore, we performed the Three-Stage Least-Square (3SLS) estimations and Instrumental Variables (IVs) to minimise such possible presence of endogeneity (see Elyasiani and Zhang 2015;Onali et al. 2016;Trinh et al. 2020a). 10 This estimation requires the identification of suitable IVs. Following previous studies (e.g. Elyasiani and Zhang 2015;Trinh et al. 2020a;Trinh et al., fcos2020b), we use the number of public firms headquartered in the country as our first IV. It is argued that outside directors of the banks having headquarters located in nations with more public firms can easily find additional jobs. Therefore, we predict that the higher the number of public firms headquartered in the same country, the more the number of busy outside directors. We also include the country-level income-generating category as a second IV. 11 This is a dummy variable taking the value of 1 if "home" bank is headquartered in a middle and high-income generating country and otherwise, taking the value of 0 if such bank is in a low-income generating country. We contend that outside directors of banks located in high-income generating nations will have more opportunities to find high-skill jobs in other institutions, and hence, they are more likely to have multiple directorships Trinh et al. 2020a). Both IVs might be correlated with possible endogenous variables (board busyness) and should predict bank dividend policy only indirectly through their impacts on endogenous variables (Black et al. 2006). Within our study's setting and sampled banks, those IVs can indirectly influence bank dividend payout patterns since the country-level variables are less likely to affect individual firms' payout strategies endogenously. Under 3SLS estimations, we treat both ABOD and DIV/NI as endogenous variables and establish the simultaneous equations as follows 12 : where DIV∕NI i,t represents the cash dividends over net income. We estimate the dividend payouts using busy BOD ABOD i,t . ϕP is a vector of control variables in the dividend regression model that accounts for the effect of firm-level and country-level characteristics on the dividend payout; Year effects capture the year-fixed effects. ε it is the error term.
Consistent with prior studies (e.g. Jiraporn et al. 2011;Sharma 2011;Mulyani et al. 2016;Chen et al. 2017), we control for other board characteristics, firm-specific and country-specific characteristics to mitigate potential omitted variables bias and capture other factors affecting the bank dividend payouts. Managerial entrenchment is likely to affect dividend policy (Hu and Kumar 2004), and dividend payout is considered as a collective decision of the board (Saeed and Sameer 2017). Therefore, we control for a set of boardrelated variables to capture the quality of bank governance structure such as board size (LogBSIZE) and board independence (%INDEP). The former is measured by the number of directors on the board while the percentage of outside non-executive directors measures the latter (Hu and Kumar 2004;Chen et al. 2017). We additionally control for other firm-specific characteristics, which may affect corporate dividend payouts. This includes firm size measured by the total assets in the logarithm form (LnTA) which may positively related to a payout (Mulyani et al. 2016;Saeed and Sameer 2017). We also control for bank financial leverage (LEV) measured as the ratio of total liabilities (long-term and short-term) to total equity. This measure affects dividend payouts due to its role in reducing agency problems and due to debt covenants on dividends imposed by debtholders (Sharma 2011). Higher LEV reflects lower values for capitalisation; thus, it signals weaker corporate financial health and is expected to be linked to lower dividend payouts (Abreu and Gulamhussen 2013; Saeed and Sameer 2017). We capture growth opportunities (CAPEX/ASSETS) through the ratio of capital expenditures to total assets We control for the availability of cash/cash reserves through the ratio of cash and marketable securities divided to net assets (total assets minus cash and marketable securities), CASH/ASSETS (Jiraporn et al. 2011). According to DeAngelo et al. (2006), retain earnings are essential determinants of dividend payouts. Therefore, we control for the ratio of retained earnings to total equity (RETAIN/EQUITY). We also control for profitability performance measured by the ratio of net income to total assets (ROA) which is expected to positively affect dividend payouts (Sharma 2011;Mulyani et al. 2016;Saeed and Sameer 2017). We also control for the possible impact of banking sector concentration (i.e. activity diversification) on dividend policy by using Herfindahl-Hirschman Index (HHI) (Mollah et al. 2017).
Moreover, we use the natural logarithm of the gross domestic product per capita (GDP-CAPITA) to capture the economic development of the region/country ). Finally, we control for the difference in the national quality of governance across countries by including determinants of regulatory quality (REGULATORY) (Bitar et al. 2017). This indicator measures the quality of governance performance that reflects perceptions of the ability of the government to formulate and conduct good policies and regulations to promote the private sector. It is estimated by ranging from − 2.5 (weak) to + 2.5 (strong) (World Bank 2016). "Appendix B" provides definitions of all variables used in our models. Table 3 reports the descriptive statistics for the full sample and both, Islamic banks (IBs) and conventional banks (CBs) sub-samples. We find that CBs (0.334) have a higher mean ratio of payouts relative to IBs (0.313), which is confirmed by the paired mean comparisons t test which is significant. The results provide some primary indications supporting our expectations that IBs are likely to report lower dividend payouts relative to CBs. For board busyness, CBs show a higher board busyness (ABOD) than IBs; with higher means of 2.389 (1.995) for CBs (IBs) respectively. The two bank types show significant differences for the mean t test. 13 For other governance control variables, CBs have a significantly lower mean (2.109) for the BOD size (LogBSIZE) than IBs (2.293). The board independence (%INDEP) mean is higher in CBs (36.6%) as compared to IBs s (32.3%). Furthermore, consistent with prior literature (e.g. Beck et al. 2013, Abedifar et al. 2013, CBs are larger and more profitable than IBs. They also tend to hold less cash (CASH/ASSETS) and retain higher income relative to their total equity (RETAIN/EQUITY) than IBs. These results are supported by the paired mean t-test across the two bank types. Table 4 presents the Pearson Pair-Wise correlation matrix among all tested variables for the IBs (lower diagonal) and CBs subsample (higher diagonal). CBs report significant positive correlations between ABOD and DIV/NI while there is a negative correlation between ABOD and DIV/NI in IBs. These results suggest that an increase in BOD busyness is associated with higher levels of dividends in CBs, which is an opposite case for IBs. The correlations between DIV/NI and others are generally in line with prior literature. All correlations among explanatory variables are considerably lower than the 0.80 thresholds, which   Table 4 Pairwise correlation matrix for islamic banks and conventional banks

Descriptive statistics
The where DIV∕NI i,t represents the cash dividends over net income. The diagnostic tests show that LM Statistics P-value is lower than 1% and Sargan test P-value is higher than 10% across all models, suggesting that the chosen IVs for busy boards are valid and all models are not over-identified. P-values are shown in parentheses, *P < 0.10, **P < 0.05, ***P < 0.01. See "Appendix B" for all variable definitions. suggest that multicollinearity is not dominant and it is mitigated in our model (Gujarati 2003). 14 6 Empirical results

The level of cash dividend payouts within Islamic and conventional banks
The regression results examining the effect of BOD busyness on dividend payouts are presented in Table 5 for Panel A (full sample), Panel B (IBs sub-sample) and Panel C (CBs sub-sample). We control for year fixed-effects in all models. 15 In Panel A, we find that the coefficient on the board busyness indicator (ABOD) is positively associated with the dividend payout ratio (DIV/NI) for the full sample. The coefficient on busy BOD is also economically significant; an increase by 1% in BOD busyness, on average, is leads to an increase in the dividend payout ratio of banks by 0.062%. This result suggests that banks with busy BOD are likely to support a higher cash dividend payout policy. Such finding is also in line with the resource dependence theory indicating that outside directors working in multiple companies can promote stronger governance mechanism and bring valuable resources (i.e. expertise, skills, experience, and access to external resources) to their firms. The reputational benefits associated with busy BOD appear to reduce the conflicts between managers and shareholders related to the usage of free cash flows (see Sharma 2011) and hence, mitigate the probability that managers misuse available cash. This, in turn, leads to a high dividend ratio. With respect to control variables, results are in line with expectations and prior studies (Jiraporn et al. 2011;Chen et al. 2017;Chou and Feng 2019) indicating that dividends are employed to disgorge free cash flow to investors in the absence of other devices. In fact, LogTA and ROA have significantly positive impacts on the payouts ratio. Larger and more profitable firms exhibit however, in contrast, board independence (%INDEP) and retain earnings (RETAIN/EQUITY) have negative and significant effects on the dividend ratio. Although the negative coefficient on %INDEP does not support our predictions, they are consistent with Hu and Kumar (2004) which find that board independence is only positively linked to payout if it exceeds 40%. Finally, we obtain significant results for country indicators, with a positive coefficient of GDP growth and a negative coefficient of regulatory quality.
When examining the effect of the bank type on the expected association between board busyness and dividend payout, in Table 5 panels B and C, we find that IBs with busy BOD exhibit lower cash dividend payout ratios, with a significantly negative coefficient on ABOD. In contrast, CBs having busy BOD tend to pay out significantly high levels of cash dividends to their shareholders, supported by positive coefficients on the test variable ABOD. These results indicate that the reputational benefits of busy BOD for banks dividend payout strategies tend to be more pronounced in CBs rather than IBs. In term of economic significance, for CBs, the coefficient of busy BODs is economically significant, as a 1% increase in board busyness reflects 0.8% increase in the bank dividend payout ratio.
Meanwhile, a 1% increase in board busyness reduces dividend payout ratio by only 0.085% within IBs. 16 For the control variables, we find that the board independence (%INDEP) it is negatively associated with DIV/NI. For both bank types, we find that the effects of LogTA and ROA on DIV/NI are significantly positive. Notably, we find opposing results for LEV between IBs and CBs. We find that higher leverage is negatively associated with dividend payouts within IBs. This is consistent with our predictions developed for the IB business model, which is marked by free-interest payments and profit-loss sharing, as well as prior literature; suggesting that both debt and high dividend payout are ways to reduce free cash flow problem (Benito and Young 2003;Chen et al. 2017). Indeed, LEV affects dividend payouts due to its role in reducing agency problems and due to debt covenants on dividends imposed by debtholders (Sharma 2011). Higher LEV reflects lower values for capitalisation; thus, it signals weaker corporate financial health and is expected to be linked to lower dividend payouts (Abreu and Gulamhussen 2013;Saeed and Sameer 2017). Although such an association is found to be positive for CBs and full sample, it is still in line with prior studies in conventional literature (e.g., Jiraporn et al. 2011;Chou and Feng 2019) showing a positive relation between leverage and payouts. This can be explained that financial leverage can increase shareholders' return on investment and usually have tax advantages. Therefore, when LEV increases, this may result in an increase in return and in turn positively affects dividend policy. Moreover, according to Jiraporn et al. (2011) and Cooper and Lambertides (2018), dividend-paying firms are more leveraged than non-dividend paying firms. This is consistent with management increasing the dividend to use up excess debt capacity that may be particularly true for CBs which can easier access to debt market than IBs. The coefficients for RETAIN/EQUITY in both bank types are significant and negative, which is in line with those reported in Panel A. 17 Taken together, the findings provide strong evidence that board busyness positively affect dividend cash payouts on average. Having a busy board tend to have differential impacts on the dividend payouts across the two bank types. Busy board increases the levels of cash dividends for CBs relative to IBs. Such positive impact of board busyness within CBs suggests that busy boards seem to offer preferential access to funds and other networking benefits to support higher payouts. The finding is consistent with the reputation view of busy boards . The negative effect of busy boards on dividend payouts within IBs can be justified by their constrained dividend model which is less flexible and more restrictive. Managers in IBs have more opportunities for discretion and control over the payout process. Hence, having busy board lead to a detrimental impact on cash dividends payouts. Such result support earlier arguments regarding the negative consequences for board busyness (e.g. Sharma 2011; Elyasiani and Zhang 2015), indicating ineffective monitoring ability by busy boards to review a complex dividend model like IBs. In addition, the results showing the distinct impacts of BOD busyness on the Islamic 17 Following Elnahass et al. (2018), we further introduce an indicator variable: The Auditing Organization for Islamic Financial Institutions (AAOIFI) to control for financial reporting regulatory differences across IBs in our sample, which apply either AAOIFI or IFRS standards. This variable takes a value of one if an IB is located in Bahrain, Jordan, or Qatar and applies AAOIFI, and zero for an IB in another country and applies IFRS. Results remain relatively unchanged. Tables will be provided upon request. 16 In unreported sensitivities, we captured cross-country variations in governance perceptions for our sample. We followed Čihák and Hesse (2010) to develop a country governance index (COUNTRY_GOV) as an additional control variable. This variable is estimated as the average of six key country-governance measures developed by the World bank which are: corruption, government effectiveness, political stability, and regulatory quality, the rule of law, and voice and accountability. We relatively obtained similar results.

3
versus conventional banking business models further support arguments by Elnahass et al. (2020a) and Trinh et al. (2020a) showing that busy BODs are likely to exacerbate agency conflicts in IBs, leading to lower bank stability and poorer market valuations relative to conventional banks. There overall findings are in line our expectations and support the study's main hypothesis. 18

Additional analyses and robustness checks
This section presents several sensitivity and robustness tests for our main results. We aim to examine whether or not our main findings hold when using: (i) alternative measures for dividend policy; and (ii) alternative model specifications/estimation procedures. We further test for the possible non-linear relationships between busy boards and bank dividend payout among the two dividend models.

The likelihood of dividends payouts
In line with our main predictions, we further examine whether CBs having a busy BOD are more likely to pay cash dividends when compared to IBs. Because the dependent variable is a dummy variable, we use a Logit function where the probability of LIKE_PAY variable is estimated using the functional form π(x) = e. g.(x) /(1 + e. g.(x) ). This function represents the propensity to pay cash dividend (e.g. Sharma 2011;Chen et al. 2017). Our likelihood model is specified as follows: where LIKE_PAY i,t takes the value of 1 if the bank pays cash dividends in year t and otherwise 0. Robust standard errors are employed to account for potential correlation in errors. 19 Analyses for both bank types in Table 6 (Panels B and C) indicate that IBs with busy boards are less likely to pay dividends, with a significant and negative coefficient on LIKE_ PAY. In contrast, CBs having busy boards are positively associated with a high likelihood of a payout. Busy BODs are likely to recommend the payment of a cash dividend in CBs. These results imply that the reputational benefits of a busy BOD for payout decisions might be more pronounced in CBs relative to IBs, which further explains our main findings in Table 5 and offer additional support. Overall, findings confirm the existence of differential impacts of BOD busyness on the propensity to pay dividends across two banking models.

Alternative measures for dividend payout ratios
In this section, we re-estimate our baseline models in Eqs. 1 and 2 but we use other alternative measures for dividend payout ratios. These include: (i) dividends over total assets (DIV/Assets); (ii) dividends over sales (DIV/Sales); and (iii) dividends per share (DIV/ Share). 20 Those measures are widely employed in literature such as Jiraporn et al. (2011), Chen et al. (2017, and Saeed and Sameer (2017). In Table 7, we find consistently across all regressions that coefficients on ABOD within IBs are significantly negative while those in CBs are significantly positive. These results are in line with our main findings and confirm that the main findings are not sensitive to alternative indicators for dividends payouts.

Alternative measures for board busyness
To check if the measures of board busyness affect our main results, we extended our analyses to use two other alternative proxies for board busyness: (i) the percentage of busy outside directors serving on the board (%BBOD) estimated by the number of busy outside directors serving on two or more additional (outside) firms divided by the total number of outside directors on the board; and (ii) the dummy board busyness variable (DUMMY_BOD) taking the value of one if at least 50% of outside directors on board are busy and zero otherwise. All of these measures are widely utilised in previous board busyness studies (e.g., Ferris et al. 2003;Fich and Shivdasani 2006;Chou and Feng 2019;Trinh et al. 2020b). We report the results in Table 8, which show consistent findings to the main analyses which are presented in Table 5. We, therefore, conclude that ABOD is a robust measure for board busyness.

Test for a non-linear relationship
We examine whether there is a possible non-linear association between busy BOD and dividend payouts in banks, under the two opposing views of having a busy board (i.e. reputation versus busyness view). We follow Trinh et al. (2020a) and Trinh et al. (2020b), which suggest a simple linear link may not fully describe the relation between busy BOD and firm outcomes. For instance, at lower degrees of busyness (i.e. the number of outside directorships), the reputational impact is likely to outweigh the cost of busyness impact because the reputational effect may increase greater than the proportional rise in BOD busyness. Nonetheless, when board busyness increases, this effect tends to grow less than proportionately with a rise in BOD busyness, resulting in the dominance of the busyness view.
We, therefore, add the square of busy BOD, i.e. (ABOD 2 ), into our baseline models to check if such non-linear association exists for both bank types. The simultaneous equations models are specified in Eqs. 4 and 5 as below:   Table 9 provides the 3SLS results for the non-linear check, with Panel B (for IBs) and Panel C(CBs). We find that for CBs, there is an opposite direction on the coefficient of ABOD 2 relative to its original form (ABOD), which suggests a non-linear relationship between busy BOD and bank dividend level. More specifically, we find that within CBs, the reputation effects tend to diminish proportionally as the outside board directorships increase. That is, the reputation effect of busy BOD in CBs only outweigh the cost of busyness effects at the lower degree of outside directorships. However, for IBs, we find an insignificant and positive sign on the coefficient of the square term of busyness; ABOD 2 . This result implies a simple linear relation between busy BOD and bank dividends policy. Such additional analyses provide evidence supporting the detrimental role of having a busy director on board when the bank's business model is unique and marked by a constrained dividend model. The results also support our main findings for the adverse effects of busy boards on the payout ratio of IBs.
where ϕP is a vector of control variables in the bank dividend payout level model including bank-level indicators and country-level indicators. The diagnostic tests show that LM Statistics P-value is less than 1% and Sargan test P-value is greater than 10% across all models, indicating that the chosen IVs for board of directors' busyness are valid and the models are not over-identified. P-values in parentheses, *P < 0.10, **P < 0.05, ***P < 0.01. See "Appendix B" for other variable definitions.

The effect of the financial crisis
In this analysis, we examine if the relationship between board busyness and dividend policy changes will still holds under the macro-economic shock of the financial crisis of 2007-2009. This global crisis may prevent banks from distributing dividends or paying lower levels to retain cash to support distressful periods of low liquidity and bank survival. We introduce to our baseline model in Eqs. 1 and 2, an interaction term (ABOD*Crisis) between the time dummy variable of financial crisis (Crisis) and our main board busyness variable (ABOD). The simultaneous equations models are specified in Eqs. 6 and 7 as below: Table 10 The effect of financial crisis on the relationship between boards busyness and bank dividend payout level -Within Islamic and Conventional Banks The table reports 3SLS results for the effect of financial crisis on the associations between busy boards of directors and bank dividend payout level for the IB subsample (Panel A) and CB subsample (Panel B). We build simultaneous equations models: DIV∕NI it = 0 + 1 ABOD it + + 2 CRISIS + 3 ABOD*CRISIS + P + Yeareffects + it (4) ABOD it = 0 + 1 DIV∕NI it + + 2 CRISIS + 3 ABOD*CRISIS + P + Yeareffects + it (5) where ϕP is a vector of control variables in the bank dividend payout level model including bank-level indicators and country-level indicators. The diagnostic tests show that LM Statistics P-value is less than 1% and Sargan test P-value is greater than 10% across all models, indicating that the chosen IVs for board of directors' busyness are valid and the models are not over-identified. P-values in parentheses, *P < 0.10, **P < 0.05, ***P < 0.01. See "Appendix B" for other variable definitions.  Abedifar et al. 2013). Table 10 reports the results for IBs (Panel A) and CBs (Panel B). We find consistent results for the coefficient on ABOD for IBs and CBs. We find no significant evidence for board busyness on IBs during non-crisis years. However, during the non-crisis periods, board busyness is positively associated with cash dividends payouts in CBs. Within the crisis period, the observed positive association within CBs is mitigated; the effect of a busy board on dividends payouts is lower. Within IBs, there is an insignificant association between the interaction term and the dividend payout ratio (DIV/NI), showing insignificant association. These findings imply that CBs reduce their dividends payouts (i.e. holding more cash) during periods of financial distress to meet expected losses and possibly meet regulatory capital requirements, which is in line with life cycle theory showing that firms may behave differently according to their existing stage (Fama and French 2001). Both the Crisis variable and the interaction term show insignificant results within IBs. This further confirms previous arguments in literature (e.g., Abedifar et al. 2013;Beck et al. 2013) which suggest that the Islamic banking business model is more stable during episodes of financial distress. Unlike conventional banking system, which promotes pro-cyclicality in lending, the Islamic banking business model is counter cyclical in line with its interest-free nature (Elnahass et al. 2018). This business model represents risk sharing rather than risk shifting, which is more dominant in conventional banking.

Robustness check: Controlling for firm fixed effects
Following the study of Bhagat and Bolton (2019), we control for firm fixed effects. We find that our main findings in Table 11 are relatively unchanged compared to those reported for the main analyses in Table 5 where we do not apply firm fixed effects but we use only year fixed effects. This test shows that our results are consistent and robust under different model specifications.

Robustness check: two-step system generalized models of moments (GMM)
To address potential endogeneity problems in estimated models, we investigate the robustness of our main findings by employing GMM technique (Arellano and Bover 1995;Blundell and Bond 1998). This approach captures for unobserved influences by transforming the variables into first difference to mitigate the effects of either unobserved heterogeneity or omitted variable bias. By using the GMM, we can treat all corporate governance and bank-level indicators as endogenous, and their IVs are lagged values (Hermalin and Weisbach 2003;Mollah and Zaman 2015;Mollah et al. 2017;Trinh et al. 2020a, b, c). Other country-level variables, including macroeconomics and country governance, are treated as strictly exogenous. GMM has an advantage of solving the endogeneity problem upon on "internal instruments" rather than "external instruments" or natural experiments and accounting for the dynamic nature of panel data by including bank dividend payouts as one of the regressors (see Meng et al. 2018). (6) ABOD it = 0 + 1 DIV∕NI i,t + 2 Crisis i,t + 3 ABOD i,t * Crisis i,t + P + Yeareffects + it In Table 12, findings are consistent with our main results. Busy BOD in CBs tends to be positively associated with the dividend payout ratio (DIV/NI) while IBs show significant negative impacts of board busyness on their dividends payouts. These results indicate that main findings remain unchanged even after using controlling for unobserved heterogeneity, simultaneity and dynamic endogeneity.
For a separate, unreported, sensitivity we check if our findings are sensitive to the selection of IVs under the 3SLS estimations. We use an alternative IV for our baseline models in Eqs. 1 and 2, which is the yearly average of the busy boards of other firms in the same country for our sample. This approach of instrumenting was adopted by Trinh et al. (2020a) and Safiullah and Shamsuddin (2019). A change in dividend payout level of one bank is less likely to affect the busy boards of other banks. Therefore, this IV is expected to be associated with board busyness but unlikely to correlate with error terms in the main equation. Our reported findings remain unchanged to the main findings.

Propensity score matching (PSM) method
We argue a possibility that busy BODs maybe not randomly distributed across banks within our sample. Also, some variables which are related to the appointments of those outside directors could be associated with banks' risk levels, which may affect the dividend payouts levels. To solve the problem, we utilise a propensity-score matching technique 21 to identify a control group of banks (whose BODs are not consisted of at least fifty per cent busy outside board members, but exhibit no observable differences in characteristics relative to banks with busy BODs) and the treatment group (banks whose BODs are consisted of at least fifty per cent busy outside board members). Therefore, the main purpose of matching banks from control and treatment groups is to ensure that each pair of matched banks is virtually indistinguishable from one another except for the function of BOD busyness.
We followed Chakravarty and Rutherford (2017) and Casu et al. (2013) to construct matches and computing the propensity score equal to the probability that a bank with a given function has a BOD that is consisted of at least fifty per cent busy outside board members. Indeed, we estimate propensity score as a function of LogBSIZE, %INDEP, LEV, return on assets (ROA), LogTA and GDPCAPITA. This propensity score test aims to balance all the covariates between the two groups ). The control variables included in this model should not be affected by the treatment; therefore, we lagged them by one year, which is consistent with Casu et al. (2013). Subsequently, we proceed to match with replacement each bank with a busy BOD (treatment group) with banks having non-busy BOD (control group) using the nearest neighbour technique.
Results in Table 13 (Panel B) show that CBs with a busy board have higher payouts ratio even after holding observable bank characteristics virtually constant between control and treatment groups. Differences between these two groups are statistically significant. In contrast, we find an opposite finding in the subsample of IBs (Panel A). Our results imply that the self-selection bias does not significantly afflict our analyses. Taken together, our main results appear to be robust across different model specifications. 22

Conclusion
This study offers a novel perspective regarding the effectiveness of an essential internal governance mechanism which is the BOD in association with firm dividend payout strategies. We offer new insights to the wide set of literature, covering both financial and nonfinancial firms, through exploring the impacts of having a busy board member on cash dividend payouts. The study comparatively assesses the use of alternative dividend models by different bank types (i.e. Islamic versus conventional banks). Our results show that for the full sample, a greater representation of busy members on the board has a positive and significant influence on the dividend payout ratio. The results provide support for resource dependence theory, which argues the reputational benefits are likely to emerge for banks with busy outside directors. When extending the analyses to identify the effect of different bank types, we find that the dividend level and likelihood of payouts are higher in conventional banks with busy BODs when compared to Islamic banks. We explain these findings structural differences between the dividends model employed across the two bank types, where for example, the Islamic dividend model is marked by more complexity and less flexibility for payout dividends under stricter Islamic principles associated with the noninterest return payments, the mechanics of profit distributions and the required effective Shari'ah governance.
Our extended analyses examined the effect of the financial crisis of 2007-2009 and we find that during the crisis period, the impact of board busyness on dividends payouts is significantly lower within conventional banks but we observed insignificant effect within Islamic banks. Furthermore, we find a non-linear relationship between board busyness and dividend level within conventional banks. Specifically, the reputation effects tend to diminish proportionally as the outside board directorships increase. However, we find only a  Overall, the findings imply that unlike conventional banks, the dividend policy of Islamic banks is likely to be more sensitive to the busyness level of outside directors serving on the board. The results indicating the positive impact of busy boards on dividend payouts within conventional banks suggest that busy outside directors tend to offer several preferential benefits to their banks which support high levels and high likelihood of dividend payouts. These results reinforce those of Chou and Feng (2019) showing that increasing the number of outside directorships of independent directors can enhance the board internal monitoring function. The results reported in this study highlight the need for policymakers who govern dual banking systems to consider board multiple directorships quotas particularly for Islamic banks. Investors and depositors dealing with different bank types and those located in emerging economies could benefit from our research to understand the underlying nature/structure of the dividend models employed within alternative banking sectors. For global banking systems, it might be the time for regulators and other stakeholders to address how institutional factors and additional governance requirements (e.g. Shari'ah governance) might have implications on optimal finance choices and distributions of profits, as presented in the case of Islamic banks. Future research, therefore, may extend our study to consider other board characteristics, such as financial expertise, education and tenure, and their impacts on long-term financial policies. Acknowledgement The authors thank Professor David Higgins, Professor M. Shahid Ebrahim, Professor Omneya Abdelsalam, Dr Stergios Leventis, and Dr Marwan Izzeldin for their valuable feedback and constructive comments. This research did not receive any specific grant from funding agencies in the public, commercial, or non-for-profit sectors.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not Table 13 Propensity-score marching estimators-board of directors busyness and Bank dividend payouts The table reports the propensity score matching estimates of the average treatment effects (ATE) of board busyness on bank dividend policy of busy board of directors for Islamic banks (Panel A) and Conventional banks (Panel B) subsamples. *P < 0.10, **P < 0.05, ***P < 0.01. See "Appendix B" for all variable definitions