EFFECT-OF-CORPORATE-GOVERNANCE-MECHANISMS-ON-TAX-AVOIDANCE-IN-DEPOSIT-MONEY-BANKS-IN-NIGERIA, Thesis of Accounting

Download EFFECT-OF-CORPORATE-GOVERNANCE-MECHANISMS-ON-TAX-AVOIDANCE-IN-DEPOSIT-MONEY-BANKS-IN-NIGERIA and more Thesis Accounting in PDF only on Docsity! EFFECT OF CORPORATE GOVERNANCE MECHANISMS ON TAX AVOIDANCE IN DEPOSIT MONEY BANKS IN NIGERIA BY AISHA NUHU MOHAMMED DEPARTMENT OF ACCOUNTING, AHMADU BELLO UNIVERSITY, ZARIA MARCH, 2017 ii EFFECT OF CORPORATE GOVERNANCE MECHANISMS ON TAX AVOIDANCE IN DEPOSIT MONEY BANKS IN NIGERIA BY Aisha Nuhu MOHAMMED Ph.D/ADMIN/14937/10-11 A THESIS SUBMITTED TO THE SCHOOL OF POSTGRADUATE STUDIES, AHMADU BELLO UNIVERSITY, ZARIA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE AWARD OF A DOCTORATE DEGREE IN ACCOUNTING AND FINANCE DEPARTMENT OF ACCOUNTING, FACULTY OF ADMINISTRATION, AHMADU BELLO UNIVERSITY, ZARIA, NIGERIA. MARCH, 2017 v ACKNOWLEDGEMENTS All praise is due to Allah, The Beneficent, The Merciful, The Supreme Creator and Fashioner of all things. May the Peace and Blessings of Allah be upon His messenger Muhammad (S.A.W.). I would like to acknowledge and extend my gratitude to all the members of my supervisory committee. The Chair, Professor S. A. Abdullahi and all the members Dr. A. B. Dogarawa and Dr. S. Mamman all gave constructive corrections and criticisms that together ensured that this work would come to fruition. For these I am ever grateful and my fervent prayer is that may Allah reward them abundantly. I would also like to acknowledge one–time Head, Department of Accounting, Ahmadu Bello University, Zaria; Dr. Ahmad Bello for his role in encouraging me to both partake in the PhD program and to look at the issue of corporate tax avoidance in Nigeria. Thank you for the stimulus Sir. The Co‟ordinator of the PhD program Prof. A. S. Mikaílu was also instrumental in encouraging the completion of this work. Special mention must also be made of Doctors M. H. Sabari, S. Abubakar, I. L. Chechet, M. A. Haruna, L. Mailafia, A. C. Dikki, N. S Aliyu, H. S. Kargiand M. S. Tijjanifor their encouragement which contributed to the successful completion of this work. Specifically the contributions of DoctorsI. L. Chechet, L. Mailafia and H. S. Kargiby way of reviews of the earlier drafts of this work must be appreciated. A work of this nature can also hardly be completed without the assistance of colleagues and peers for whom I am greatly indebted to. In particular course mates M. M. Bagudo and M. D. Tahir were really there for me in terms of both material and moral assistance. Mrs. Aisha Mahmoud H. was also a special breed in terms of assistance and moral support right from the onset till the culmination of the study. MessersN. Yunusa, L. Mohammed,A. A. Abdullahi, H. M. Mohammed, J. I. Yero, I. Yusufand A. Ahmad were also very instrumental in helping with theoretical and data analysis related issues. Other course mates such as Maryam I. Muhammad, YahayaZakari, Mahmoud Ibrahim and Abbas Sule also provided much needed advice on the conduct of the work. Haj. Halima S. Sambo of the department of Business Administration also deserves mention for her role in initiating me to real econometrics. Mention must also be made of Aisha O. Otori and AdamuMagaji for their help with data extraction. This acknowledgement would not be complete without my acknowledging the encouragement given to me by various family members. In particular mention must be made of my uncle Professor Dalhatu Mohammed as well as my elder brother and chief mentor Justice A. D. Yahayawho constantly enquired about my progress. Their constant questioning ensured that I actually did complete the work. For the concern and guidance they showed I am eternally grateful. The rest of my family members Mami, Inna, Guggos and Yayas also all encouraged the completion of this study through their shows of concern and prayers. For this I am grateful. Mention must also be made of my husband and children for their fortitude in respect of the time they forfeited to the actualization of this work. May God bless them all. vi DEDICATION This work is dedicated to the memory of Late Ambassador Nuhu Mohammed (Mallam), who was our beacon of knowledge and to Justice AbubakarDattiYahaya; a brother cum father like no other. vii ABSTRACT The issue of corporate tax avoidance hasreceived vast empirical examination in Western academe. This vast examination has however not been echoed in respect of research interest ontax avoidance in corporate entities in Nigeria.This study therefore sought to provide empirical evidence on whether internal corporate governance mechanisms such as board size, board independence,board ownership,high ownershipconcentrationas well as interactions between high ownership concentration with board size and independence are significantly associated with corporate tax avoidance in deposit money banks (DMBs) in Nigeria. A sample of fourteen out of the fifteen listed DMBs on the Nigeria stock exchange (NSE) as at December 2014 were examined. Data for the study were sourced solely from secondary sources in the form of annual financial statements of the studied DMBs for the period 2006 to 2014. In order to tackle the issue of endogeneity, arising from simultaneity in studies related to corporate governance outcomes, the Arellano-Bond Generalized Method of Moments (GMM) estimation technique was used. The study found that increase in last period‟s board ownershipsignificantly increased tax avoidance in the studied DMBs in the current period. Furthermore, increased board independence in the immediate preceding period was found to significantly decrease tax avoidance inthe DMBs in the current period. However, the relationship between board independence and tax avoidance in the DMBs is significantly moderated by high ownership concentration. Overall, the study concluded that internal corporate governance mechanisms combine to significantly affect tax avoidance in the DMBs. In light of the findings, the study therefore recommended thatin order to facilitate goal alignment between the interests of directors and that of DMB owners, in respect of tax avoidance, ownersinstitute more share-based remuneration for executive directorsand encourage non-executive directors to take up some minimum number of shares during their tenure. This will have the combined effect of incentivizing the board to render strategic x 2.4 Review of Empirical Studies……………………………………………………………..45 2.4.1 Corporate Governance Mechanisms and Corporate Tax Avoidance……...…………...46 2.4.2 Firm-specific Characteristics and Corporate Tax Avoidance…...……………………..51 2.5 Consequences of Corporate Tax Avoidance……………………………………………..57 2.6 Theoretical Framework…………………………………………………………………..60 CHAPTER THREE: RESEARCH METHODOLOGY 3.1 Introduction……………………………………………………………………………....64 3.2 Research Design………………………………………………………………………….64 3.3 Source of Data and Method of Collection ……………………………………………….65 3.4 Population and Sample …………………………………………………………………..65 3.5 Method of Data Analysis ………………………………………………………………...66 3.6 Variables, Measurements and Model Specification……………………………………...68 CHAPTER FOUR: DATA ANALYSIS AND INTERPRETATION 4.1 Introduction………………………………………………………………………………74 4.2 Descriptive Statistics……………………………………………………………………..74 4.3 Correlation Statistics...…………………………………………………………………...78 4.4 Robustness Tests…………………………………………………………………………80 4.5 Analysis of Results……………………………………………………………………….83 4.6 Tests of Hypotheses……………………………………………………………………...85 4.7 Discussion of Findings…………………………………………………………………...87 4.8 Implications of Findings…………………………………………………………………90 CHAPTER FIVE: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 5.1 Summary…………………………………………………………………………………92 5.2 Conclusions………………………………………………………………………………93 5.3 Limitations of the Study……………………………………………………………….....94 5.4 Recommendations………………………………………………………………………..96 5.5 Areas for Further Research……………………………………………………………….97 REFERENCESG.....00..0... 000000 cc cee cceee tee e cette cee eet eee ten ete cesta tte eeseneteeernneeeeeses 98 APPENDIX... 0000 ecient cette tenet e nett nets atte eens tteeetsneteeersneteeeses 114 xi xii List of Tables Table 4.1: Descriptive Statistics………………………………………………………….81 Table 4.2: Correlations……………………………………………………………………85 Table 4.3: Summary of OLS Results &Robustness Tests….……………………………88 Table 4.3: Summary of GMM Regression Results……………………………………….91 3 The question of what factors explain the ability of firms and corporations to avoid taxes is of particular interest to researchers. The reason for much of the focus on tax avoidance as opposed to tax evasion is because evasion as a criminal act has to be proven by court. Thus using the term avoidance is seen as less dyslogistic. In however considering corporate tax avoidance as a research issue, researchers such as Shackelford and Shevlin (2001) have earlier questioned the applicability of models of individual tax evasion and avoidance such as the Allingham and Sandmo (1972) framework in explaining and predicting corporate tax avoidance. They argued that the separation of ownership and control, a hallmark of corporate entities, means that existing individual tax non-compliance frameworks cannot adequately explain same for corporations. In furtherance of the argument, Slemrod (2004) stated that this same separation of ownership and control means that shareholders need the corporation to engage in some level of avoidance. Thus, ininvestigatingwhat factors explain tax avoidance by firms and corporations, earlier studies on corporate tax avoidance focused primarily on examining whether firm-specific characteristics such as size, leverage, growth, profitability,amongst others could explain the tax avoidance phenomenon for business entities (e.g., Gupta & Newberry, 1997). These earlier studies however failed to consider agency issues in their analyses. Pointing out why failing to do so is an anomaly, Desai and Dharmapala (2006, 2007) argued that since decisions about corporate tax avoidance are made by firm managers, then the analysis of such decisions is thus embedded in an agency framework. The agency framework is one that argues that managers are risk averse and self-serving in nature and that this risk averseness as well as self-serving nature means that managers will not typically act or make decisions in the best interest of owners. To ensure goal congruence between management and shareholders or owners, the framework suggests that managers should 4 be both incentivized and monitored. Corporate governance mechanisms represent the means by which monitoring managers can be achieved. Corporate governance mechanisms can however be internal or external. The internal mechanisms are those that have to do with the efficacy of the board of directors in appropriately advising on and overseeing the design and implementation of business strategies that will ensure that managers maximize shareholder wealth. In addition, such internal mechanisms include the role that shareholders themselves play in ensuring goal alignment. Prescriptions that have to do in particular with size, independence, remuneration and financial expertise of the board have therefore featured prominently in the various codes of corporate governance that have been issued both nationally and internationally as guides to what constitutes “best practice” in oversight. On the other hand, external governance mechanisms include all other stakeholder monitoring. Therefore mechanisms such as government regulation, debt covenants, takeovers, financial analysts and the like are all aspects of external governance. Internal governance mechanisms have, in particular, been touted to be foremost amongst the monitoring mechanisms that can ensure goal alignment between owners and managers. This is thought to be so because the board of directors is responsible for the strategic direction of the company. Charting a course for effective tax management is one of such strategic direction. Therefore, like any other board room stratagem, managing taxes requires a laid down philosophy which is usually determined by the board; documented, communicated and implemented as overall corporate strategy (Erle, 2007). Thus, board related governance mechanism such as size, independence, and ownership, amongst others can arguably play a key role in determining a company‟s tax planning. 5 Board size simply refers to the number of persons that constitute the board of directors of a corporate entity. The number of persons that constitute the board of directors is thought to influence the advisory capacity of the board as well as its monitoring effectiveness. However, what constitutes the optimal board size to achieve this effectiveness has been a subject of debate. While some argue that a large board is more desirable because the larger the number on the board the more the level of diversity, skill and expertise; others argue that larger boards stifle discussion, therefore smaller boards are more effective because communication within a smaller group is easier (Jensen, 1993). Given that managing the tax expense (tax avoidance) is thought to be beneficial for corporate owners, this study therefore finds it necessary to examine whether board size as an internal governance metric affects tax avoidance by deposit money banks (DMBs) in Nigeria. Board independence, the proportion of members of the board who are non-executive directors, is another internal governance metric that influences board oversight. For this reason the codes of corporate governance give it categorical mention. Empirical studies to date, are however yet to consistently document either significant associations or signs when relating board independence to corporate outcomes. The intuition that board independence can influence tax avoidance by DMBs is however appealing. This is because in relation to other firm performance metrics such as profitability and firm value, it has been argued that the independent judgments that non-executive directors can bring to bear on corporate outcomes enhances oversight. Whether this is so in relation to tax avoidance by Nigerian DMBs is however an empirical question that is yet to be examined. Ownership structure dimensions such as having an individual with sizeable number of shares in a company (block shareholding), the level of managerial shareholding as well as the 8 Arising, in part, from concerns over the aforementioned lost amounts of revenue; researchers have also upped the ante on the analysis of tax avoidance. While however still acknowledging the legality of tax avoidance, several researchers such as Potas (1993), Christensen and Murphy (2004), Fuest and Riedl (2009), Sikka (2010), Prebble and Prebble (2013) and Fischer (2014), amongst others, bring to the fore the argument that the legality of a phenomenon does not necessarily translate to that phenomenon being fair or ethically and morally acceptable. This argument is particularly evident regarding the phenomenon of tax avoidance. This is because tax avoidance contributes to a situation whereby tax payers of the same income bracket will not pay taxes on the same marginal tax rate. This has the effect of making the tax system appear unfair to those who are unable to avoid taxes. Research on corporate tax avoidance in comparison with that of tax avoidance by individuals could however be said to be of relatively recent focus. The said recent focus of research on the phenomenon has also tended to be conducted principally in developed climes. The earlier empirical studies on corporate tax avoidance such as Gupta and Newberry (1997) had focused more on the interplay between firm-specific characteristics such as size, leverage, profitability, capital intensity, amongst others in determining corporate tax avoidance. Given that the results on the association between the studied firm-specific characteristics and tax avoidance turned out to be far from consistent (e.g., Richardson&Lanis, 2007 and Hsieh, 2012); researchers further broadened the scope of investigation, on the determinants of corporate tax avoidance, to include other factors such as corporate transparency (Wang, 2010), CEO/manager effects (Dyreng, Hanlon &Maydew, 2010; Chyz, 2010), ownership structure (Badertscher, Katz &Rego, 2009; Chen, Chen, Cheng &Shevlin, 2010), external auditor effects (Mcguire, Omer & Wang, 9 2010; Huseynov&Klamm, 2012), incentives (Philips, 2003; Armstrong, Blouin&Larcker, 2012) and a host of other characteristics. The basis for broadening of the scope of investigation has however been questionedby Hanlon &Heitzman (2010). They contend that the choice, by researchers,of what variables to study as determinants of corporate tax avoidance has seldom been backed by theory. Thus, in order to provide a theoretical base for the study of corporate tax avoidance, a relatively young but burgeoning literature which seeks to situate the determinants of corporate tax avoidance within an agency theoretical framework has emerged. According to Hanlon and Heitzman (2010), theoretical arguments for the study of corporate tax evasion and avoidance as agency issues have been pioneered by Crocker and Slemrod (2004), Slemrod (2004), as well as Chen and Chu (2005). The pioneers however failed to back their arguments with empirical data. Thus other researchers have sought empirical evidence in support of the framework. Specifically Desai and Dharmapala (2007), Minnick and Noga (2010), Khaoula and Ali (2012), Zemzem and Ftouhi (2013), Khaoula (2013) and Armstrong, Blouin, Jagolinzer and Larcker (2015) are amongst a spate of relatively recent researches that have directly examined the interplay between corporate governance mechanisms and tax avoidance. A common theme across the aforementioned studies that directly relate governance mechanisms with corporate tax avoidance is that while they refer to governance broadly, they habitually study only board related mechanisms to the detriment of other governance mechanisms. Governance is a myriad of mechanisms which are both external to the corporation as well as internal to it. For instance, shareholder and stakeholder monitoring, competition, the markets for both managerial and corporate control as well as debt covenants are all governance mechanisms. Therefore narrowing such a multi-faceted concept down to only board mechanisms 10 does not appropriately proxy the phenomenon. The said empirical studies have also not been consistent in documenting either significant associations or similar signs of association in their studies. One possible reason for the inconsistencies may be because none of the past studies has considered whether interactions with other variables are the actual force behind the effects of the studied mechanisms on tax avoidance. As Hermalin and Weisbach (2003) and Adams, Hermalin and Weisbach (2010) point out, the effect of board mechanisms on firm performancemay not be direct. Therefore failure to take this into consideration is likely to distort interpretations. In this study, it is therefore proposed that concentrated ownership moderates the relationship between board structure proxies such as size and independence on corporate tax avoidance. Considering this moderating effect is important because shareholders with concentrated ownership (blocks and institutions) usually leverage upon their concentrated holdings to either directly sit on the board of directors or indirectly have a representative as director (La Porta, Lopez-de-Silanes, Schleifer&Vishny, 1998). This representation on the board is likely to affect both board advisory capacity (size) as well as monitoring capacity (independence). Furthermore, Connelly, Hoskisson, Tihanyi and Certo (2010) have argued that concentrated ownership leads to concentrated decision rightsand this leads to superior monitoring. Localizing the focus to Nigeria, the researcher observes that the discourse as well as study of corporate tax avoidance is yet to gather full momentum. Most existing studies such as AbdulRazaq (1985), Alabede, Ariffin and Idris (2011) and Ibadin and Eiya (2013), to name a few, focused on examining individual tax compliance issues. This is despite assertions by researchers such as Adegbie and Fakile (2011a, 2011b), that in Nigeria, the contribution to overall tax revenue by companies has continually fallen far short of expectations. This lack of 13 i. Does board size have a significant effect on corporate tax avoidance among DMBs in Nigeria? ii. Does board independence have a significant effect on corporate tax avoidance among DMBs in Nigeria? iii. Does high ownership concentration have a significant effect on corporate tax avoidance among DMBs in Nigeria? iv. Does board ownership have a significant effect on corporate tax avoidance among DMBs in Nigeria? v. Does high ownership concentration significantly moderate the relationship between board size and tax avoidance among DMBs in Nigeria? vi. Does high ownership concentration significantly moderate the relationship between board independence and tax avoidance among DMBs in Nigeria? 1.4 Objectives of the Study The overall objective of this study is to determine to what extent internal corporate governance mechanisms determine tax avoidance in Nigerian deposit money banks. The specific objectives of the study are to: i. assess whether board size has a significant effect on corporate tax avoidance among DMBs in Nigeria. ii. ascertain whether board independence has a significant effect on corporate tax avoidance among DMBs in Nigeria. iii. estimate whether high ownership concentration has a significant effect on corporate tax avoidance among DMBs in Nigeria. 14 iv. evaluate whether board ownership has a significant effect on corporate tax avoidance among DMBs in Nigeria. v. examine whether high ownership concentration moderates the relationship between board size and tax avoidance among DMBs in Nigeria. vi. examine whether high ownership concentration moderates the relationship between board independence and tax avoidance among DMBs in Nigeria. 1.5 Statement of Hypotheses In line with the aforementioned objectives of the study, the following hypotheses; stated in null form,were formulated for testing: Ho1: board size has no significant effect on corporate tax avoidance among DMBs in Nigeria. Ho2: board independence has no significant effect on corporate tax avoidance among DMBs in Nigeria. Ho3: high ownership concentration has no significant effect on corporate tax avoidance among DMBs in Nigeria. Ho4: board ownership has no significant effect on corporate tax avoidance among DMBs in Nigeria. H05: the relationship between board size and tax avoidance among DMBs in Nigeria is not significantly moderated by high ownership concentration. H06: the relationship between board independence and tax avoidance among DMBs in Nigeria is not significantly moderated by high ownership concentration. 15 1.6 Scope of the Study This study empirically examined the effect of internal corporate governance mechanisms on corporate tax avoidance in Nigeria. The study covered only listed Deposit Money Banks (DMBs) in Nigeria. The focus on banks was in part, hinged on the role of banks as financial intermediaries in the economy. This role therefore means that banks tend to drive activities in the corporate sector. There are also assertions that banks structure transactions both for customers and themselves that are likely to facilitate tax avoidance (OECD, 2009). The study covered the period 2006 to 2014. The year 2006 was chosen as a base year so as not to allow the effects of the 2005 banking sector consolidation exercise to distort results while 2014 represents the latest year beforethe CBN classification of some DMBs as strategically important (CBN 2014a). The new classification, effective March 1, 2015 imposes stringent requirements on the said banks in respect of liquidity and capital adequacy requirements. The requirements are likely to affect some financial statement items and therefore introduce some distortion in analysis. The period overlaps with the 2001-2010 previously noted by the GFI (2012) report as the period in which Nigeria lost substantial revenue through avoidance activities. 1.7 Significance of the Study The study is significant for the following reasons: Firstly, taxes are a key source of revenue for governments and often feature prominently in government‟s income redistribution function. Government, through its relevant tax agencies, 18 phenomenon. The perspectives, however, cannot be said to be independent of one another but are rather intertwined. Given the said multifaceted perceptions as well as subject-oriented nature of the tax avoidance phenomenon, Lietz (2013) suggested that a conceptual framework for tax avoidance must of necessity be capable of not only aggregating existing views but should also be able to leave room for possible future views and/ dimensions to the phenomenon. However, whether or not such aggregation is possible depends on our ability to actually understand the divergence points of the existing views. A look at them individually is therefore necessary. It seems that the very term tax avoidance is a legal creation. This is because the term was coined to distinguish between two categories of tax dodging. The first category-evasion is defined as outright illegal and therefore punishable while the second category, avoidance is deemed legal but not desirable; therefore ordinarily not punishable. Thus, the focus of the law would be to establish that an act has been committed that contravenes the law. Given the said legal distinction, on the face of it, the difference between the two forms of tax dodging appears clear cut. In practice however as several researchers (e.g., Potas, 1993; Slemrod&Yitzhaki, 2002; Mclaren, 2008; and Murphy, 2008) point out, the dividing line between tax evasion and avoidance is not so clear. Tax avoidance is therefore severally referred to as a “grey area” between outright tax evasion and being tax compliant (Slemrod&Yitzhaki, 2002; Murphy, 2008). It is a process of getting round taxation law without actually breaking it (Murphy, 2008: 3). Mclaren (2008) further lamented that in Australia, the distinction between tax evasion and tax avoidance has continually become blurred due to laws that refuse to demarcate what constitute 19 either the former or the latter; but rather define activities as “tax exploitation”. Mclaren cites for instance Division 290 of the Taxation Administration Act 1953 (Cth) which ignores the distinction between tax avoidance and tax evasion and deals instead with „tax exploitation schemes‟ as well as the Anti-Money Laundering and Counter Terrorism Financing Act 2006 (Cth) (AML/CTF Act) as examples of laws that blur any clear cut demarcation of the two activities.Potas (1993) had earlier pointed out the increased blurring of the difference between the two terms; which he contends had led to the use of „non-compliance‟ and „compliance‟ respectively in lieu of the said terms. Tax Analysts (2010) however quoted Franklin D. Roosevelt, one time president of the US, as commenting on the subject of tax evasion and avoidance thus: Methods of escape or intended escape from tax liability are many. Some are instances of avoidance which appear to have the color of legality; others are on the borderline of legality; others are plainly contrary even to the letter of the law. All are alike in that they are definitely contrary to the spirit of the law. All are alike in that they represent a determined effort on the part of those who use them to dodge the payment of taxes which Congress based on ability to pay. All are alike in that failure to pay results in shifting the tax load to the shoulders of others less able to pay, and in mulcting the Treasury of the Government‟s just due. For Roosevelt, therefore, tax evasion and tax avoidance are both opposite sides of the same coin. They amount to the same thing; a reduction in tax revenue accruable to the government purse and a truncating of the fairness which tax regimes are meant to achieve. This argument falls neatly in place with the position of various aid and advocacy groups such as Action Aid, 2013; Oxfam (2014) and Tax Justice Focus (2009) as well as corporate social responsibility (CSR) proponents such as Christensen and Murphy (2004), Sikka (2010; 2012), and Fischer (2014), amongst others. 20 However, the ambiguity of the term tax avoidance does not end with the legalese of differentiating it from tax evasion. The term tax avoidance has often also been used interchangeably with other terms such as tax planning, tax mitigation, tax aggressiveness, tax sheltering, and tax minimization by researchers in their study of the tax avoidance phenomenon (Potas, 1993; Desai &Dharmapala, 2009b; Chen, Chen, Cheng &Shevlin, 2010; Badertscher, Katz &Rego, 2011; Armstrong, Blouin&Larcker, 2012; Huseynov&Klamm, 2012). Strictly speaking, each of these terms that have been adopted as synonyms of tax avoidance, connotes a different aspect of behavior. Arguably, the terms, perhaps, closer in meaning to tax avoidance are tax mitigation, tax minimization and tax planning. Even with these terms, as Kwaghkehe and Samuel (2011) observe, in some jurisdictions such as The United Kingdom (UK) and New Zealand, a distinction is made. The Scholes, Wolfeson, Erickson and Maydew(2009) framework on effective tax planning has often been referred to in defining tax planning in contrast to tax avoidance. According to the framework, effective tax planning is that strategy that identifies all parties to a transaction; all tax costs and benefits as well as all non-tax costs and benefits. Within this framework therefore the efficient tax planner would of necessity avoid any conduct that would tend to increase his tax and non-tax costs. He would however take advantage of allowances and reliefs permitted by law to their full extent as doing so is perfectly legal. The difficulty in delineating the two terms however arises with practices such as use of tax havens, which theoretically do not appear to be tax evasion but in all sincerity cannot be categorized as tax planning. This is because such activities lack any economic substance. Therefore, for want of another term, such activities are categorized as tax avoidance. 23 incorporate the several associated terminology or from a narrow lens by which specific behavior can be measured. Thus, just like with many other social science phenomena there is no consensus of definition in the matter of conceptualizing tax avoidance. It all depends on the context in which the phenomenon is being studied. This lack of an agreed consensus on what exactly the term tax avoidance portends, while not in itself exactly an impediment to research, nonetheless presents a great drawback in terms of the consistency of measuring, evaluating and comparing studies. This is because while on the face of it the titles of the research refer to the same term, avoidance; in reality the researchers may not be studying the same aspect of the phenomenon. The Hanlon and Heitzman (2010:137) definition of tax avoidance, “broadly as the reduction of explicit taxes” has been adopted by various other researchers. They contend that this broad definition, which had earlier been suggested by Dyreng, Hanlon and Maydew (2008), is in order to reflect the fact that tax avoidance or planning is a continuum.On one end, the continuum reflects transactions whose legality is not in question (e.g. real activities that are tax favored) while on the other end, transactions such as engaging tax shelters as well as actions that tend toward deception and fraud are described as “aggressive” tax planning and are therefore nearer to outright tax evasion. Their definition, therefore, embraces a wide spectrum of activities. This study therefore also adopted the Hanlon and Heitzman (2010) definition. This is because as a definition of the tax avoidance phenomenon, the definition avoids a moralistic tune to wit that avoidance is either good or bad. The definition also allows the use of a broad measure of tax avoidance for the study. 2.2.1 Methods of tax avoidance 24 Several methods are available both to individuals and corporations to both take advantage of legally allowed tax savings as well as to exploit loopholes in the tax laws in order to reduce amounts payable by them in the form of taxes. Such opportunities range from ordinarily taking advantage of statutorily deductible allowances, applications for statutorily approved reliefs, use of tax favored investments to various other ingenious schemes. Stiglitz (1985) in putting forth his general theory of tax avoidance itemizes three basic principles that apply in relation to avoidance of income taxes- the ability to postpone taxes, tax arbitrage across individuals facing different tax brackets or an individual facing different marginal tax rate at different times and tax arbitrage across income streams facing different tax treatments. Researchers have however specifically documented several methods, also referred to as schemes, by which corporations in particular can specifically take advantage of tax minimization and therefore plan their strategies accordingly. For instance, Sikka (2010) documented the use of transfer pricing, royalty programs, off shore tax havens and structured transactions as methods of tax avoidance by corporations. Gravelle (2013) on the other hand documented, in addition to transfer pricing, other methods such as allocation of debt and earnings stripping, contract manufacturing, check-the-box, hybrid entities and/ instruments as well as cross crediting and sourcing rules for foreign tax credits. An individual examination of the mechanisms is therefore necessary. 2.2.1.1 Income shifting Income shifting is a general term which refers to the ability of an individual or corporation to move income between different tax bases or bands. This follows the tax arbitrage principle of tax avoidance as outlined by Stiglitz (1985). Taking advantage of tax arbitrage as a 25 tax avoidance strategy allows incomes to be passed either across different individuals who are on incomes facing different tax treatments, the same individual facing a different marginal tax rate at different times or across income streams facing different tax treatments. For instance, income shifting by hiring family members to work in the businesshas often been employed by small businesses. Family owned limited partnerships have been known to use income splitting mechanisms to shift income. Trusts, annuities, gifts and interest free or below market loans have been known to be proven methods by which individuals as well as families shift income and therefore avoid taxes. Specifically, in the case of corporations, Gordon and Slemrod (2000) pointed out that income can be shifted through increasing the use of debt financing in order to increase interest deductions for firms in higher tax jurisdictions and to also increase interest income for those in lower tax jurisdictions. Other contrived means of corporate tax avoidance through income shifting include changing the form of employees or executive compensationfrom stock-based to cash-based as well as shifting business activity income between corporate and non-corporate bases. Kwaghkehe and Samuel (2011) have in particular identified private companies as perpetrators of income shifting in a manner that allows for shareholders to be executives and hence receive jumbo amounts, which are tax deductible, in the form of executive compensation. Kwaghkehe and Samuel also pointed at the capitalization of profits by means of bonus issue as an additional income shifting mechanism by corporations. Gordon and Slemrod (2000) however drew attention to the fact that the various income shifting, if not adequately tackled and monitored play havoc with the usual interpretation of many kinds of data, because it blurs the return to capital and the return to labour. Financing constraints are however thought to moderate the extent of corporations‟ income shifting. For instance,Dyreng and Markle (2013) documented 28 Fisher (2014:343) described tax havens as „financial conduits that, in exchange for a fee, use their one principal asset- their sovereignty- to serve a nonresident constituency of accountants and lawyers, bankers and financiers, who bring a demand for the privileges that tax havens can supply‟. Tax havens satisfy both deferral and arbitrage principles of tax avoidance put forth by Stiglitz (1985). With regards to tax haven use as a corporate tax avoidance mechanism, MNCs in particular have been fingered as the main customers. The multinational companies utilize the tax havens by locating either the head office or a major subsidiary of the group in the said jurisdiction and ensuring that the bulk of the profits of the group are declared there. In a study reported by Christian Aid (2013), it was found that MNCs with tax haven connections report 1.5 per cent less profits, pay 17.4 per cent less in taxes per unit of asset, pay 30.3 per cent less in taxes per unit of profit and have 11.4 per cent higher debt ratios than MNCs with no connection to tax havens. Tax havens have therefore come under severe criticism by various stakeholders because it has been argued that most tax havens‟ practices do not involve a transfer of real economic activity to those jurisdictions, but merely a shift on paper that allows for tax avoidance (Henn, 2013). 2.2.1.4 Royalty programs Using royalty programs is yet another tax avoidance activity that uses income shifting mechanism. In order for the royalty programs/ schemes to concretize in respect of tax avoidance purposes, they need to be channeled through tax havens. The tax haven(s) then facilitate the requisite tax savings by providing cover from taxes for such royalty income. Depending on the 29 amount of the purported royalty payment, the business can benefit from significant tax savings. An example is given by Henn (2013) of the use of royalty programs to facilitate tax avoidance by IKEA, the worlds‟ biggest furniture maker. Under the said royalty program, IKEA is touted to have channeled its royalties to a holding company under the name Inter IKEA Systems; which is based in the Netherlands and where the royalties can be amassed tax-free. Not much has however been documented in Nigeria on the use of royalty programs to facilitate corporate tax avoidance. 2.2.1.5 Structured transactions, hybrid entities and instruments Structured transactions are an umbrella term used to describe a multitude of transactions that have been designed to be complex. Hybrid entities and instruments are one of such complex transactions used as tax avoidance mechanism by corporations. According to Needham (2013: 4), hybrid entities revolve around obtaining a deduction of the same cost, such as loan interest, from two different countries based on the company‟s affiliates‟ structures. Thus such entities are subject to different tax rates in one national jurisdiction and a different tax rate in the other. This could facilitate overall tax savings for such entities. Financial instruments such as derivatives provide a similar opportunity to avoid taxes. This can be made possible through the exploitation of different tax treatments afforded the derivatives in different jurisdictions; just like that described above for hybrid entities. As such derivatives can be used to facilitate a hedging pattern that allows a MNC to shift profits between subsidiaries and parent companies to reduce the overall tax burden for the company. Other structured transactions such as conduits and shell companies, collateralized debt obligations, as well as mortgage-backed securities have also been found to be variously used for corporate tax avoidance purposes (Needham, 2013).DMBs in Nigeria are particularly prone to using structured transactions. This is evident by the mention, without necessary explanation, of such transactions in several of the DMBs‟ annual reports. 30 2.2.1.6 Allocation of debt and earnings stripping Allocation of debt and earnings stripping mechanisms are also used to actualize tax avoidance. According to Gravelle (2013), companies within a group structure domiciled in lower tax jurisdictions can borrow more from related companies situated in higher tax jurisdictions or alternatively unrelated foreign borrowers not subject to tax on interest income in the country in which the company is domiciled may borrow sizeable amounts. Gravelle further pointed out that such a practice, referred alternatively to as earnings stripping has been touted to lure corporations into engaging in corporate inversion practices. There is also evidence that firms alter debt/equity ratios across countries in order to incur more interest expense in high tax countries and more interest income in low tax countries (Clausing, 2009). Where the debt/equity ratio is in favour of debt, the practice is referred to as thin capitalization. According to Taylor and Richardson (2013a) MNCs in particular have an incentive to fund their foreign direct investments in their related and subsidiary companies situated in jurisdictions with higher tax rates by debt. The logic behind having such a high debt/ equity ratio lies in the fact that the interest payments become deductible in the jurisdiction with higher tax rate while they become receivable in the jurisdiction with lower tax rates. Thus, while the interest payment will still be taxed at the headquarters, it however effectively reduces the amount chargeable to tax for the MNC as a single entity- hence tax avoidance is facilitated. There are therefore multitudes of ways by which companies in particular can perpetrate tax avoidance. As can be surmised from the preceding discussion, a key hallmark of most of the various activities and schemes is that they are not easily discernable. Hence tax avoidance continues to flourish. 33 even within affiliates, affiliates where the repatriation costs are higher tend to hold more cash than other affiliates where repatriation costs are lower. Further support for the financing constraints motive is found in Troncoso and Vergara (2010) who proposed a theoretical model of firms facing financial constraints. They based their model on the accelerator framework. In their model, financially constrained firms evaluate expected costs and weigh them against expected benefits of the avoidance/ evasion. Thus as financing constraints become tighter after a worsening in internal/ external financing conditions, firms will tend to avoid/evade more taxes. They however, concluded that even though tax compliance depends on both internal and external financing conditions, firms that are more dependent on external financing are more likely to engage in tax avoidance and/ evasion because of the effect of interest rate increases. Empirical support is documented by Chen and Lai (2012) found that financially constrained firms are highly tax aggressive. 2.2.3 Measuring tax avoidance Ordinarily, deciphering tax avoidance for companies would not constitute a difficult affair. One would merely compare data between a company‟s financial statements and its tax returns with any differences between the two associated with tax avoidance. Tax returns are however proprietary in nature and are not readily accessible to the public. Thus researchers have considered several measures, often based solely on financial statement data, as proxies in studies of corporate tax avoidance. Relying on the financial statement data to compute tax avoidance measures has however come under criticism. For instance, Hanlon and Heitzman (2010: 139) pointed out that “a lack of disclosure in financial statements about taxable income and/ or the actual cash taxes paid or to be paid on the current year‟s earnings” represents a great drawback in the use of proxies based on financial statement data as appropriate measures for corporate tax 34 avoidance. However, despite this criticism, in the absence of alternative and more reliable information, financial statement based proxies of corporate tax avoidance continue to be used. Hanlon and Heizman (2010) have earlier identified and reviewed twelve (12) measures from the literature as having been used to proxy for corporate tax avoidance. The identified measures were summarized as falling into one of four categories; effective tax rate measures, discretionary or „abnormal‟ measures, unrecognized tax benefits and tax shelter firms. The effective tax rate measures are broad measures usually capturing both permanent and temporary differences while the other three measures all capture specific forms of tax avoidance.The review by Hanlon and Heitzman (2010) concluded that all currently existing measures of corporate tax avoidance capture only non-conforming tax avoidance. They therefore caution that the choice of what measure to use as proxy should depend on the nature of the research questions of a study. Given the broad definition of taxes adopted by the study, of the aforementioned four categories of measures of tax avoidance, the effective tax rate (ETR) measures seem more apt. The effective tax rate measures include the GAAP ETR, cash ETR and long-run ETR. Typically, all the ETR measures capture the average tax rate per monetary (₦ ) value of pre-tax book income. Data needed to measure ETRs are therefore readily gleaned from financial statements. The GAAP ETR, usually measured as current tax expense divided by pre-tax book income, is one of the most popular measures of tax avoidance. This may be, in part, because it is easy to compute, the necessary data used in its computation are directly discernable from the annual published financial statements and the fact that GAAP matching rules ensure that once a firm generates income for a period, a corresponding tax expense must also be reported irrespective of when taxes are actually paid (Hanlon &Heitzman, 2010). Furthermore, AbdulWahab (2010) citing Hanlon (2003) posits that the GAAP ETR proxy avoids measurement errors in relation to 35 the effect of tax expense on foreign income and tax credits. However, despite these touted advantages, GAAP ETR has been criticized for not appropriately recognizing particular tax deferral strategies such as more accelerated depreciation charges and for absorbing some tax planning strategies and treating them as avoidance (Lee, Dobiyanski& Minton, 2015; Hanlon &Heitzman, 2010). The cash ETR and its long-run variant differ from the GAAP ETR in respect of their numerators. While GAAP ETR uses total taxes as represented by the tax expense in the income statement, cash ETR and long-run cash ETR focus only on the actual cash taxes that were paid as given by the statement of cash flows. While on the face of it, this may seem more reflective of the company‟s tax burden, the use of cash taxes paid as numerator against earnings as denominator means that where the cash taxes paid include taxes from a different period, there will be a mismatchbetween the numerator and denominator. Where there is a mismatch,thiswill present as one form of measurement error. Furthermore Rego and Wilson (2009) posit that cash ETR measures do not control for non-discretionary sources of book-tax differences such as intangible assets. According to Dyrenget al (2010), which of the ETR variants better captures tax avoidance by the corporation is likely to depend on whether the focus of managers is in reducing the tax expense or in reducing cash taxes. They argue that where managers are concerned with reducing the tax expense for financial accounting purposes, GAAP ETR suffices as a measure of tax avoidance. If however the focus of the managers was to reduce actual taxes paid, the cash ETR serves as a better proxy of tax avoidance.Notwithstanding the drawbacks earlier highlighted, this study used GAAP ETR as a proxy for the corporate tax avoidance phenomenon 38 corporations. The study focused only on internal governance mechanisms, therefore a review of the known mechanisms is given as follows: 2.3.1.1 Board Mechanisms The board of directors could be said to be the primary internal governance structure. This is because “the board sets the rules of the game for the CEO” (Jensen, 1993:862). As a prime mover of affairs therefore there is high interest in the effectiveness of board mechanisms in fulfilling its general oversight and advising role. In particular, issues that have to do with optimal board structure in terms of size, independence, CEO/ Chair duality and stock ownership by directors of a firm have been noted by Booth, Cornett and Tehranian (2008) as having enjoyed much debate in the literature. Other board mechanisms and characteristics such as diversity, board committees and board remuneration/ incentives have also been considered extensively (Bøhren&Strøm, 2005). Board size:simply refers to the number of people who make up the board. The focus on board size as an internal governance mechanism centers on concerns about cohesion, co-ordination and timely intervention of the board in respect of advice over relevant organizational matters. Section 4.1 of the Securities and Exchange Commission (SEC) code of corporate governance for all public listed entities in Nigeria (2009) states that “the board should be of a sufficient size relative to the scale and complexity of the company‟s operations…” Section 4.2 of the same code further suggests that minimum board membership should be five (5) but makes no such stipulations as to the maximum number. For banks in Nigeria, the Central Bank of Nigeria (CBN) code (2014) also stipulates a minimum board number of five (5). Unlike the SEC code, the CBN code sets the maximum number of board members for banks at twenty (20). The ideal board size has thus been 39 debated by researchers. No real consensus has however been achieved as to what constitutes optimal board size. Theoretically some researchers such as Jensen (1993) advocate for smaller board numbers. Their arguments are based on assertions that smaller board sizes are more effective in advising and monitoring because expressing opinions as well as communication within a smaller group is generally easier and less time consuming. On the other hand, it has been argued that larger boards suggest a larger pool of talent and a wealth of varied experiences. The larger pool of talent and varied experience is thought to make it easier for the board to tackle issues and makes it better poised to be able to advice management. All the above arguments are not however based on any empirical evidence. Yermack (1996) therefore empirically explored whether board size has an effect on the market valuation of companies. The findings of the study revealed that companies with large board sizes have lower market valuations and are on average less able to efficiently utilize assets. Consequently, companies with larger board sizes earn lower profits. In a similar study Beiner, Drobetz, Schmid and Zimmerman (2003) while confirming their initial proposition that board size is an independent corporate governance mechanism, found that board size is negatively related to firm performance as measured by Tobin‟s Q. The results are however not significant. Conversely, Sanda, Mikailu and Garba (2010) found a statistically positive association between board size and firms‟ financial performance. However, despite the documented positive association, Sandaet alstill recommended that board size should be kept at a moderate ten (10). Board independence:is also thought to play a key role in internal governance. Independence reflects the willingness of the boards to monitor management (Adams et al 2010). This feature of the board is largely unobservable but has been most commonly proxied by looking at the 40 appropriate mix of executive and non-executive directors sitting on any one board (Dalton, Daily, Johnson &Ellstrand, 1999). The emphasis on having a board with more non-executive directors (NEDs) stems from theoretical argument which has it that, a board dominated by executive directors (insiders) is likely to facilitate actions by management that are most favorable for management as opposed to shareholders. Therefore, theoretically, it is argued that a greater percentage of non-executives or outside directors on a board ensures better monitoring of the agent. Given this perspective, board independence requirements linked to the number of NEDs on a board feature prominently in the various codes of corporate governance. For instance, Section 4.3 of the Securities and Exchange Commission (SEC) code of corporate governance (2009) for all public listed entities in Nigeria states that “the Board should comprise a mix of executive and non-executive directors… … the majority of members should be non-executive directors, at least one of whom should be independent director”. Furthermore, Section 5.4 (a) states that NEDs should bring independent judgement as well as necessary scrutiny to the proposals and actions of management and executives especially on issues of strategy, performance evaluation and key appointments.Empirical studies such as Yermack (1996) and Beineret al (2003) while not categorically studying board independence, nonetheless examined the effect of board independence on firm valuation and performance. The results of these earlier studies suggest that the percentage of outside/ independent directors on the board has a significant effect on market valuation and performance of companies. Board remuneration:is also a focal internal governance mechanism. Interest in board remuneration stems from the fact that remuneration packages ought to be structured in such a manner as to provide both maintenance and incentive effects to board members. According to agency theory, incentives play a key role in facilitating the requisite goal alignment between 43 remuneration and audit (Brown et al, 2011). In respect of Nigerian banks, the CBN code (2006) specifies that Nigerian banks should have a minimum of the following committees: risk management, audit, as well as credit. The revised CBN code (2014) however stipulates the minimum board committees as risk management, audit and board governance & nominations committees. Several committee characteristics such as size, independence, expertise, remuneration and activity (meetings) are argued to influence the effectiveness of committees. While however there exist empirical studies that examined the influence of the characteristics of the various committees on company performance, arguably the committee that has received the most attention by researchers is the audit committee (e.g., Hayes, Mehran & Schaefer, 2004; Laux&Laux, 2006; Fauzi& Locke, 2012). According to Aldamen, Duncan, McNamara and Nagel (2012) the audit committee is a key monitoring mechanism, both in respect of shareholders‟ and for other stakeholders‟ interests. Xie, Davidson and DaDalt (2001) point out that with regards to issues such as the management of earnings, the role of the executive committee is an indirect one; it is the audit committee that has a more direct role. This is because the principal role of the audit committee is to monitor a firms‟ financial reporting and performance. The strength of its monitoring capacity is however contingent on its possession of certain characteristics; amongst which are its size, independence, activity (meetings) and financial expertise. Thus the UK Financial Reporting Council (FRC, 2012) stipulates that minimum audit committee membership should be three; two of which shall be independent. Audit committee meetings are recommended to be frequent (i.e. no fewer than three times during the year), and audit committee members should be given further remuneration to compensate them for their additional tasks. The Audit Committee News (2013) however posits 44 that the appropriate audit committee size will depend on “the needs and culture of the organization and the extent of responsibilities delegated to the committee by the board”. In respect of financial expertise, this has been emphasized upon by virtually all the various corporate governance codes as well as reports such as The Audit Committee News (2013) and the FRC (2012). What is mostly emphasized upon is that at least one member of the audit committee should have “recent” and “relevant” experience in finance, accounting or auditing. This is because having experts on the committee is thought to be critical to both the monitoring and advisory roles that the board of directors render (Robinson, Xue& Zhang, 2012). However, in a survey that intended to document what directors themselves consider most important among several corporate governance attributes, Adrian, Wright and Kilgore (2013) found that audit committee composition ranks as the second most important governance mechanism that matters to directors. Adrian et al. (2013) also documented that audit partner tenure and audit committee size were less emphasized on by directors as they were ranked bottom of the pack, coming in eighth and ninth respectively. Committee characteristics are not studied by this research because Anderson, Deli and Gillian (2003) have previously documented that elements of audit committee characteristics such as independence and expertise are explained by overall board characteristics. This study does not consider committee characteristics largely because sufficientdisclosure of committee-related mechanisms that have to do with knowledge and financial expertise are lacking in so many DMBs‟ annual reports. 2.3.1.2 Ownership structure 45 Ownership structure is also central to any corporate governance discourse (Lietz, 2013). Theoretical assertion has it that concentrated ownership is one of the most direct ways to align cash flow and control rights of outside investors (Shleifer &Vishny, 1997). According to Brown et al(2011: 114), “where ownership is concentrated and there is a difference between the cash- flow rights and voting rights of shares, owning a relatively small portion can be enough to control the firm.” La Portaet al (1998) also posited that having concentrated ownership allows shareholders to exercise power in firms that in most instances significantly exceeds the power vested in them by their cashflow rights and that one such way of exercising power is through participation in management. How well concentrated ownership may influence outcomes may however be contingent on ownership identity. This is because ownership identity has implication in terms of risk aversion of owners, monitoring incentive as well as the ability to defend shareholder rights (Shleifer &Vishny, 1997; Holderness, 2003). Concentrated ownership can be by individuals, management or other institutions. Thus separate terminologies have usually been employed to distinguish the identities of such concentrated ownership. Block shareholders is the term usually identified with individuals whose share of a company‟s equity amounts to five percent and above. Corporations or other businesses that have such a similar stake in a company are otherwise referred to as institutional shareholders. According to Holderness (2003) two motivations, which are not mutually exclusive, drive the existence of such block holders- shared benefits of control and private control benefits such as increased voting power and board participation. These two benefits are at the heart of arguments that block holders can and do appropriately monitor management. Increased pressure from institutional shareholders has also been noted as one of the facilitators of more diligent and independent boards (Adams et al, 2010). 48 Governance is primarily in existence to ensure satisfactory goal alignment between shareholders (principal) and management (agent). It is therefore a key mechanism within agency theory. The analysis of corporate tax avoidance within an agency framework is however a relatively recent phenomenon. Hanlon and Heitzman (2010) posited that the theoretical foundations for studying corporate tax avoidance as an agency issue were laid by Slemrod (2004), Chen and Chu (2005) as well as Crocker and Slemrod (2005). The literature preceding these papers had failed to see that due to the separation of ownership and control rights within companies, it could not to be assumed that both the principal and agent had the same motivation to avoid taxes. Of necessity therefore, if owners are desirous of some amount of tax savings, they have to find a way of communicating this to the agent. Thus the board of directors as the spokespersons of the shareholders assumes strategic importance in achieving this goal alignment in terms of tax savings. Within the tax-related agency framework, Slemrod (2004) and Crocker and Slemrod (2005) thus argued that the key to achieving requisite goal alignment lies in incentivizing the agent to do so through his compensation contract. Another perspective has been added to the agency view of corporate tax avoidance by Desai and Dharmapala (2006, 2007) who argued that even though theoretically better governance should result in greater tax avoidance, the fact that tax avoidance demands some degree of “complexity” and “obfuscation” means that mangers can hide behind tax avoidance to perpetuate opportunistic behavior. This argument is given empirical backing by Erickson, Hanlon and Maydew (2004) who found that their sample of US firms had overpaid taxes in order to conceal fraud. Armstrong, Blouin, Jagolinzer and Larcker (2014) also see corporate tax avoidance as an agency issue. However their view differs from that of Desai and Dharmapala, in the sense that they consider tax avoidance as an investment decision. Armstrong et al(2014) 49 therefore contended that as with any other investment decision, managers have personal incentives to engage in some degree of tax avoidance that may not be in the interest of shareholders, thereby giving rise to agency problems. Given the rising profile of tax avoidance as an agency issue, researchers have therefore begun to seek empirical evidence in order to test the said theoretical arguments. In particular, internal governance variables such as board characteristics as well as ownership structure proxies have therefore begun to be looked at. 2.4.1.1 Ownership structure and corporate tax avoidance Ownership structure has been of particular interest to researchers studying firm-level tax avoidance. In a treatise on corporate tax selfishness, Slemrod (2007) conjectured that privately held firms may sacrifice reporting higher financial accounting earnings in an attempt to reduce taxes. The conjecture was predicated on the argument that since such firms have fewer outside capital pressures, such a sacrifice would not be difficult to consider. Empirical evidence such as that provided by Badertscher, Katz and Rego (2009) however, showed that private companies which are majority owned by private equity firms engage in significantly less non- conforming tax planning when they are privately held than when they are publicly traded. Similarly, Chen, Chen, Cheng and Shevlin (2010) examined whether private family-owned firms; defined in the study as firms in which founding families retain controlling interest by having over 10% of the issued shares of the said firms, are less or more tax aggressive than their non family-owned counterparts. The study reported that family-owned firms are less tax aggressive than their non family-owned counterparts. The result is consistent with the notion that oversight of activities is 50 greater in such family firms and that such firm owners are usually risk averse. Their results are corroborated by Rego and Wilson (2009). McGuire, Wang and Wilson (2012) examined tax avoidance in firms with dual class ownership; a situation whereby some firms have two classes of stock- one class with voting rights and the other with no voting rights. Such firms are characterized by a marked separation of voting rights from cash flow rights. The study found that dual class owned firms tend to avoid more tax. Their research findings, taken together suggest that the costs of risky tax avoidance increase as the separation of voting rights and cash flow rights grows. In another study, Cheng, Huang, Li and Stanfield (2012) examined businesses that have been targeted by hedge fund activists. Their findings taken together suggest that hedge fund activism lessens tax avoidance for the targeted firms. However, tax avoidance increases when the activists have a successful track record of tax knowledge and/ interest. Cheng et al(2012) contended that their findings are consistent with hedge fund activists having concentrated investment portfolios whereby the performance of an individual firm may significantly affect their portfolio return. Therefore, with hedge fund acquisition comes increased monitoring. Their study appears to be a first as no previous literature exists regarding the effect of this form of ownership on avoidance. With regards the issue of CEO ownership, Desai and Dharmapala (2006) found that the incentive to engage in tax avoidance activities is greater when a CEO has a large ownership stake in the firm. This is in addition to the incentive he has to minimize the burden of his personal taxes. This is consistent with the notion that having an ownership stake in the firm, helps align the goal of the executive with that of the owners, thereby mitigating agency costs and if the CEO 53 From the preceding review, it appears therefore that board size and independence have so far been the most frequently studied board-related corporate governance variables in relation to tax avoidance. However the documented empirical associations between the said variables and tax avoidance are far from consistent. 2.4.2 Firm-specific characteristics and corporate tax avoidance Firm-specific characteristics such as size, profitability, leverage, growth and capital intensity, among others, have been found to feature prominently in determining firms‟ financing and investing decisions. Based on this, it has also been argued that firm-specific characteristics may have the potential to play a pivotal role in determining a firm‟s tax avoidance stratagem. Researchers have therefore traditionally examined the issue of corporate tax avoidance from the lens of how well firm-level characteristics explain the phenomenon. 2.4.2.1 Firm size and corporate tax avoidance Of the firm-specific characteristics thought to impact tax avoidance, a perusal of the literature suggests that firm size is the characteristic mostly examined or controlled for in the study of the corporate tax avoidance phenomenon. This is because two competing hypothesis in the literature suggest that firm size may present as a factor to contend with in deciphering firm- level avoidance. The political cost hypothesis, as postulated by Watts and Zimmerman (1978) and Zimmerman (1983), argues that larger firms, due to their greater visibility tend to bear more regulation and by juxtaposition therefore pay more taxes. The association, therefore, given the said high visibility, between firm size and corporate tax avoidance, according to proponents of the political cost hypothesis is that larger firms pay more taxes and by doing so thus avoid/ evade less than smaller firms. Conversely, the alternative hypothesis- the political power hypothesis 54 argues that larger firms have more financial muscle to manipulate tax payments to their advantage. Thus, under the political power hypothesis, larger firms are seen to avoid more. Several researchers (e.g. Gupta & Newberry, 1997; Philips, 2003; Richardson &Lanis, 2007; Taylor & Richardson, 2013; Edwards, Schwab &Shevlin, 2013) measure firm size as the natural logarithm of the book value of total assets of the firm whereas studies such as Armstrong, Blouin and Larcker (2012) and Gaertner (2013) measure firm size using change in market value of assets and the natural logarithm of market capitalization respectively. Alternatively, Crabbe (2010) argued in favour of measuring firm size as number of employees of the firm. The argument by Crabbe (2010) in favour of the number of employees measure was in order to avoid colinearity between firm size and other variables scaled by total assets. Different measures of firm size therefore abound. Arising in part from the different measures of firm size, the findings of various empirical studies on the association between firm size and corporate tax avoidance have been found to be mixed. Gupta and Newberry (1997) who performed a pre and post 1986 US tax reform period analysis of the determinants of corporate ETRs in the US, found a positive association between ETR and firm size in the pre-reform period and an inverse association in the post-reform period. Similar studies such as Richardson and Lanis (2007), Md Noor, Fadzillah and Matsuki (2010) as well as Hsieh (2012), who replicated Gupta and Newberry to specifically study firm level determinants of ETR in different countries, have all documented a significant positive association between ETR and firm size, implying that larger firms are less inclined to tax avoidance. Other studies that focused on other determinants of tax avoidance, of necessity have to control for the various firm-specific characteristics. Therefore studies such as Rego (2003), Loretz and Moore (2009), Minnick and Noga (2010), Dyreng, Hanlon and Maydew (2010), 55 Crabbe (2010), Harrington and Smith (2011), as well as Cheng, Huang, Li and Stanfield (2012) all control for firm size. All the aforementioned studies found a positive association between firm size and avoidance, thereby lending credence to the political cost view of the firm size effect. In contrast, studies such as Philips (2003), and Wang (2012) found a statistically significant inverse relation between firm size and avoidance. Yet other studies such as Armstrong, Blouin and Larcker (2012), Taylor and Richardson (2013a), and Gaertner (2013) all found no significant association between firm size and tax avoidance. The results on the association between firm size and corporate tax avoidance are therefore mixed. 2.4.2.2 Profitability and corporate tax avoidance Profitability is another firm characteristic that theory posits is likely to determine firm- level avoidance. From an intuitive point, this view is premised on the fact that profitability implies the firm having more funds at its disposal for both investments and paying dividends; thereby mitigating the need for engaging in tax avoidance. Given the said intuitive hypothesis, therefore, profitability should, theoretically, have the effect of decreasing firms‟ tax avoidance. Profitability has been often measured in the literature on corporate tax avoidance as return on assets (ROA). For instance, several studies such as Gupta and Newberry (1997), Philips (2003), Minnick and Noga (2010), Armstrong, Blouin and Larcker (2012), Taylor and Richardson (2013b), Edwards, Schwab and Shevlin (2013); amongst others, all proxied profitability by ROA. Dyrenget al(2010) however measured profitability as earnings before interest and tax while Gaertner (2013) measured profitability as return on equity (ROE). The empirical evidence provided by studies such as Gupta and Newberry (1997), Rego (2003), Richardson and Lanis (2006), Frank, Lynch and Rego (2010), Wang (2012), Khaoula 58 ETR, discretionary permanent tax differences and shelters) found that greater debt financing of private equity- backed firm years substantially reduces the marginal tax rate for the said firms as well as their need for more aggressive tax strategies. Studies that specifically focused on firm characteristics and tax planning such as Gupta and Newberry (1997), Richardson and Lanis (2007) and Md Noor, Fadzillah and Mtsuki (2010), have all found a negative and significant association between financial leverage and tax avoidance (GAAP ETR). Other empirical studies that includefinancial leverage in their models as a control variable such as Gaetner (2013) and Edwards et al(2013) have also documented a negative and significant association between financial leverage and tax avoidance. Conversely, a study by Philips (2003) that examined the effects of incentives on tax planning in US companies, while controlling for various firm-specific characteristics including financial leverage, found a positive association between financial leverage and tax avoidance. The positive result documented in Phillips (2003) is also echoed in Ding et al(2013). The empirical results on the association between corporate tax avoidance and leverage are therefore also mixed. 2.5 Consequences of Corporate Tax Avoidance A focal reason why corporate tax avoidance is such a topical issue lies in its consequences. According to Dyrenget al(2010), this aspect (consequences) of corporate tax avoidance has, however, received less attention from researchers. One of the foremost issues amongst the consequences, is the fact that corporate tax avoidance leads to a substantial loss in revenues accruable to the purse of government. These lost revenues have implications in the provision of public goods, general government administration as well as fiscal policy. Cobham (2005) estimated that US$285 billion per year is lost by developing countries because of tax 59 evasion in the domestic informal (shadow) economy. A recent media report states that a non- government organization (NGO), Christian Aid estimated that tax evasion alone costs developing countries $160 billion annually (CFM News mobile, 2013). These losses significantly affect GDP for the developing world. Fuest and Riedel (2009) documented the effect of these losses and highlighted that tax to GDP for developing countries as at 2005 was between 12% - 15%. A substantially low figure when compared with the thirty-five percent average for the developing world. In the US, a 2001 IRS Tax Compliance Measurement Programme (TCMP) report quoted in Slemrod (2004) states that corporations avoided taxes to the tune of $29.9 billion, representing an under reporting rate for corporations of about 17.4%. The under-reporting rate for individuals, in that same year, stood at 13.8%. Harnessing all these lost revenues, especially for developing countries, is seen as a most promising medium and long-term source of new funds for development (Cobham, 2005). Another consequence of corporate tax avoidance lies in the fact that it is likely to impact firm value. Various studies suggest that investors are able to factor in the avoidance activities and this affects firm value. This is because the lower ETRs of companies that engage in tax avoidance activities actually serves as a signal to the discerning investor as to the occurrence of any such activity. The nature of the effect is however, not clear. Desai and Dharmapala (2009b) examined whether investors fully capture the value of corporate tax avoidance activity in their analysis. Using a sample of 862 US firms, over the period 1993-2001. They regressed firm value, as measured by Tobin‟s q, against book-tax gap- their measure of corporate tax avoidance, a vector of control variables and corporate governance variables which were modelled as mediators. They found that firm value is affected by corporate tax avoidance and that corporate governance does indeed act as a mediator of the relationship between the two. Robinson and 60 Schmidt (2011) also examined whether firms‟ disclosure practices under Financial Interpretations Number 48 (FIN 48), which requires disclosure of uncertain tax benefits (possible tax avoidance)- a move at making firms more transparent to outsiders, affects investor‟s response to such disclosures. Having examined variations in disclosure quality for their sample firms, they found that investors appear to reward firms for low disclosure quality, suggestive of investors being primarily concerned with proprietary costs of disclosure rather than increased transparency. Distributive justice is a key function of government which it seeks to achieve by the imposition of taxes. The logic of it is simple, and is one of the arguments for government intervention in the economy. The government aims to reduce inequity in income within the populace by taxing those with more income in order to supply those with less income with the basic necessities, which without the government intervention; they may not afford to have. This cannot be achieved in the presence of individual or corporate tax avoidance. This failure to achieve distributive justice has welfare implications from the economical viewpoint. Another consequence of corporate tax avoidance is that it is liable to engender reputational loss for managers as well as the firm. Reputational loss for managers may dent their careers because it gives the impression that they are untrustworthy. Whereas reputational loss for the firm may, in addition to depressing firm value, compromise relations with customers because of some perceptions they develop regarding the avoidance. This consequence has been verified by the work of Loretz and Moore (2009). The possibility of reputational loss to customers, by companies with well-known brands is also explored in a very recent working paper by Austin and Wilson (2013). 63 the owner. Bonding costs may or may not be incurred by the agent in facilitating the goal convergence between him and the principal. In practice, the incentives are usually provided through the use of cash or share-based bonus schemes or a combination of both; while the need for monitoring is accomplished by having a board of directors as a part of corporate governance mechanisms. In reviewing the theory of corporate tax avoidance, Hanlon and Heitzman (2010) posit that the theory of corporate tax avoidance within an agency framework is still in a developing stage; thereby suggesting that as an agency issue corporate tax avoidance is yet to gain prominence. The separation of ownership from control within companies is however at the heart of arguments of corporate tax avoidance as an agency issue (Slemrod, 2004; Chen & Chu, 2005, Desai &Dharmapala, 2007). This is because such separation means that “… in a large, publicly held corporation, decisions about taxes (and inter alia, accounting) are not made by shareholders directly but rather by their agents, whether that is the chief executive officer or the vice president for taxation” (Slemrod, 2004). Desai and Dharmapala (2006) further argued that agency issues crop up in relation to corporate tax avoidance because of the complexity and obfuscation (concealment) necessary to proliferate the said avoidance. Armstrong et al(2014) further the agency perspective argument by positing that corporate tax avoidance can be viewed from the perspective of being an investment like any other investment made by a company through its managers. As an investment, managers can therefore over or under invest in the tax avoidance issue thereby leading to agency conflicts. In addition to the above varied nuances of how corporate tax avoidance constitutes an agency issue, the average effective tax rate (ETR) is a measure by which shareholders can explicitly or implicitly tie the compensation of managers to in order to align the incentives of 64 decision makers with that of owners (Slemrod, 2004). Hanlon (2003), further pointed out that the information in the tax expense, while strictly not an avoidance issue, nonetheless serves as an alternative benchmark/ measure of performance in respect of financial accounting earnings. These preceding arguments, in addition to the study‟searlier adoption of a broad conceptualization of corporate tax avoidance as being any activity that results in the reduction of a firm‟s explicit taxes (Hanlon &Heitzman, 2010), mean that, from the shareholders‟ (owners) perspective, tax avoidance is a desirable act as it serves to increase after tax returns. Thus, corporate tax avoidance can be studied within an agency framework. Eisendhart (1989: 70) also lends strength to this study‟s adoption of the agency perspective by arguing that “due to the contributions of agency theory to organizational theory, the testability of the theory as well as the fact that the theory has empirical support; it seems reasonable to urge for the adoption of an agency theory perspective when investigating the many problems that have a principal-agent structure”. 65 CHAPTER THREE RESEARCH METHODOLOGY 3.1 Introduction This chapter presented and explained the research methodology adopted by the study. Specifically, the chapter discussed the research design that was adopted for the study; nature and type of data that was sourced as well as the method by which the said data was collected. In addition, the population of the study, sampling criteria, model specification, data analysis techniques, variable definition and measurement as well as justification(s) for the adoption of each of the techniques so chosenwere discussed. 3.2 Research Design Research design influences what materials need to be collected (source of data), how such materials are actually collected (method of data collection) and how the materials are 68 augmented by correlation analysis and the use of descriptive statistics. The correlation analysis was deemed necessary as a first step to assessing the existence of anassociation between the variables of the study. The descriptive statistics on the other hand served as a first step to assessing the nature of the sampling distribution from which the variables were drawn. Assessing the nature of the sampling distribution for the individual variables is particularly important because the nature of the sampling distribution (i.e. whether Gaussian, Poisson, Chi-square, or other) is a key assumption for the class of parametric tests of which regression belongs to. The specific descriptive statistics employed were analysis of minimum and maximum values, means, medians, skewness, kurtoses, ranges and standard deviations. The regression technique used by the study was a dynamic panel data estimator; the Generalized Method of Moments (GMM) of Arellano and Bond (1991). The GMM estimator wasused because of its ability to tackle the issue of endogeneity (due to simultaneity) said to be present in studies related to corporate governance outcomes (Hermalin&Weisbach, 2003; Roberts &Whited, 2012) and the fact that similar studies such as Minnick andNoga (2010) employ it in their estimation. Several methods exist to tackle endogeneity concerns amongst which are fixed effects modelling, simultaneous equations, differencing, use of instruments as well as the GMM (Roberts & Whited, 2012). However Minnick andNoga (2010) argue that an instrumental variables approach may not be feasible in corporate governance-related ETR studies because finding suitable exogenous instruments may be very difficult. In addition, they also posit that fixed effects models may be biased because such models ignore past performance and past governance choices (Minnick &Noga, 2010: 712). The GMM estimator is therefore superior to the other two methods of tackling endogeneity because it allows for the introduction of lags of 69 the dependent variable in addition to lags of potentially endogenous variables into the estimated function and also controls for both year and company effects. As a prelude to the GMM estimation, ordinary least squares (OLS) multiple regression was performed to facilitate regression diagnostics such as testing for heteroskedasticity and multicollinearity. A further OLS regression was also conducted as a necessaryprelude to assessing whether the moderators included in the final model contribute to model explanatory power. According to Hair, Black, Babin and Anderson (2010) whenever a moderator is to be included in a model, the model must first be estimated without the moderator and its coefficient of determination (R 2 ) noted. Secondly, the model with the moderator should be estimated, its R 2 also noted. Finally an assessment is made of the change in R 2 . If the change in R 2 is statistically significant, then a significant moderator effect is present whether the individual variables are significant or not. 3.6 Variables, Measurements and Model Specification The dependent variable (outcome) of the study is the variable of interest. In this study, the outcome variable is corporate tax avoidance. Corporate effective tax rates (ETRs) were used to proxy for individual banks‟ tax avoidance. The independent variables of this study are six (6) internal governance mechanisms; four “predictors” and two “moderators”. According to Baron and Kenny (1986: 1174) “moderators and predictors are at the same level in regard to their role as causal variables antecedent or exogenous to certain criterion effects.” The predictors therefore studied were: board size, board independence, board shareholding,and ownership concentration while the moderators were the effects of two interaction variables; the interaction between ownership concentration and board 70 size as well as the effect of interaction between ownership concentration and board independence on tax avoidance among DMBs. Four firm-specific characteristics were selected as control variables and included in the models of the study. These were firm size, profitability, capital intensity and financial leverage. Firm size was included as control because of several assertions in the literature that larger firms are on the one hand likely to bear more taxesdue to their visibility (Political cost hypothesis of Watts & Zimmerman, 1978, Zimmerman, 1983 and Watts & Zimmerman, 1990). On the other hand, it is argued that larger firms are likely to bear less taxes due to their high level of sophistication which in turn gives them greater financial muscle to manipulate tax payments in their favor (Zimmerman, 1983; Gupta & Newberry, 1997; Mills, Erickson &Maydew, 1998; Richardson &Lanis, 2007). Profitability was also controlled for because ETRs are computed by dividing tax liability by profit. To a large extent therefore ETR is contingent on the amounts of profits made by an entity. Not controlling for such an association is therefore likely to bias parameter estimates for the independent variables. Several other related studies such as Minnick and Noga (2010), Hsieh (2012) and Armstrong et al (2015) also control for it. Capital intensity was controlled for because of the effect of depreciation charges on profit and therefore by association ETR. Since ETR is a book value measure of tax preferences and capital allowances are granted to entities with qualifying capital expenditure, it is necessary to control for the effects of capital intensity on estimated ETRs. Studies such as Gupta and Newberry (1997) and Richardson and Lanis (2007) have earlier found significant associations between a firm‟s level of capital intensity and ETR. 73 measurement is consistent with the requirements of the various corporate governance codes as well as other researches such as de Andres and Vallerdo (2008) and McKnight and Weir (2009). Bsizeit-1: board size of bank i at time t-1. Board size is measured as the total number of directors serving on the board. This measure is the norm with other researches including Ibrahim (2009) and Muhammad (2009) amongst others. Bshareit-1:board of directors‟ shareholding of bank i at time t-1. This is measured as the proportion of total share ownership of directors of the board to total number of shares outstanding for the banks in the same period. This measure is consistent withMuhammad (2009) and Byrd (2010). Oconcit-1: high ownership concentration. This is a dummy variable coded 1 for bank-year observations with ownership concentration >= the mean ownership concentration ratio value of 0.2767 for the banks in the studied panel and 0 for bank-year observations with ownership concentration less than the mean value. The actual ownership concentration ratio for each bank-year is arrived at by taking the aggregate percentage of shares owned by outside individuals and institutions, in each of the studied years, if five percent and above for the examined banks. Oconc*Bsizeit-1: interaction between the dummy for ownership concentration and board size of bank i at time t-1. Oconc*Bindpit-1: interaction between the dummy for ownership concentration and board independence of bank i at time t-1. 74 Fsizeit: firm size of bank i at time t; measured as the natural logarithm of the book value of total assets at the beginning of the period. This is consistent with Gupta and Newberry (1997), Richardson and Lanis (2007), Lanis and Richardson (2011), Edwards et al (2013) as well as several other researchers. ROAit-1: profitability of bank iat time t-1; measured as earnings before interest and tax divided by total assets. The measure is consistent with Gupta and Newberry (1997), Richardson and Lanis (2007), Edwards et al (2013) as well as several other researchers. Capintit: capital intensity of bank iat time t; measured as the beginning of period balance of property, plant and equipment divided by book value of total assets. This is also consistent with several other researchers, amongst which are Gupta and Newberry (1997), Richardson and Lanis (2007), and Edwards et al (2013). Flevit:financial leverage of bank iat time t; measured as the amount of total liabilities divided by total assets (both @ book value). This is consistent with Hsieh (2012), Keen and deMooij (2012), Edwards et al (2013), deMooij, Keen and Orihana (2013) and Armstrong et al(2015). β1, β2 … … β10: coefficients of the estimated variables. uit:error or disturbance term 75 CHAPTER FOUR DATA ANALYSIS AND INTERPRETATION 4.1 Introduction This chapter covers the analysis of the data extracted for the study on the effect of corporate governance mechanisms on tax avoidance by deposit money banks in Nigeria. Specifically, the chapter presented and analysed descriptive statistics and correlations for the variables under study. Furthermore, the chapter presented and analysed the regression estimation results. In addition robustness tests for the models being studied were also presented and discussed. Given the results, the hypotheses of the study were then tested. A discussion of the findings of the study was given as the concluding part of the chapter. 4.2 Descriptive Statistics Table 4.1 reports the descriptive statistics for both the dependent variable, ETR, independent variables and control variables that were studied. 78 differences. The skewness values of 1.4919 and kurtosis of 4.8074 suggest some little departure from symmetry for the variable. Ownership concentration for the period ranged from 0.00% to 99.00% suggestive of wide variations between the studied entities in concentrated ownership. Mean concentration was 27.60% well above the 5% necessary to have some degree of influence over decisions. The high ownership concentration dummy has a mean value of 0.3790 with standard deviation of 0.4871. The mean value of the dummy indicates the probability that 1 (i.e. reference to high ownership concentration values) will be observed. The value of 37.90% gives the proportion of DMB-year observations in the sample having high ownership concentration. Skewness and kurtosis values for the ownership concentration dummy are0.4987 and 1.2487 respectively. However, no inferenceson normalityare made regarding thevalues because dummy variables are typically binomial distributions. The interaction terms both yielded 118 bank-year observations each. The mean value for the interaction between board size and high ownership concentration is 5.2881 with standard deviation of 6.8239 while mean interaction between board independence and high ownership concentration was 0.2372 with standard deviation of 0.2993. Both moderators appear to be platykurtic having kurtosis values of 1.6757 and 1.5396 respectively for the board size and board independence moderators. The distribution of both moderators however have low positive skew values of 0.6501 and 0.5587. Taken together, the descriptive statistics for both moderators suggest some degree of non –normality with their individual distributions. The control variables of the study were profitability (ROA), capital intensity (CINT), financial leverage (Flev) and firm size (Ln Total Assets). ROA for the period ranged from a minimum of -17.1% to a maximum of 11%. Mean ROA was 4.73% while the median was 5.2%. 79 Standard deviation of ROA was 0.3456, skewness -2.79 and kurtosis of 16.41. Overall, there is some degree of wide variation in profitability among the studied DMBs. Capital intensity for the period ranged from 0.00% to 9.10% with a mean of 3.23% and median of 3.05%. On average therefore Nigerian DMBs are not highly capital intensive. The skewness and kurtosis values of 0.7103 and 3.5962 for CINT suggests little significant departure from symmetry for the said variable. Financial leverage for the DMBs over the study period ranged from 54.8% to 100% indicative of high variation in the sector and high debt capacity that can be exploited for tax planning purposes. The mean value was 84.3% with standard deviation of 0.0729. The mean is close to the median value of 85.30% and as such suggests that mean leverage is fairly representative for the industry. The high average financial leverage values are not surprising given the peculiarities of banks as institutions of financial intermediation. Skewness value of - 0.86990 and kurtosis of 5.2686 signify some departure from symmetry within the Flev data set but the figures are not so alarming as to suggest the presence of outliers. Firm size as measured by the natural logarithm of total assets has a mean of 20.3226 with a standard deviation of 0.9158 while the median value was also very close to the mean at 20.3805. The skewness value of -0.2533 indicates a slight negative departure from symmetry while the kurtosis value of 2.3302 indicates that the sampling distribution for total assets is relatively mesokurtic. 4.3 Correlations of Variables Table 4.2 reports the Pearson correlation coefficients (R),in a matrix form, among variables of the study. The correlation coefficients provide a preliminary assessment of both the direction and strength of relationship between the dependent variable and the explanatory variables as well as between the explanatory variables themselves. 80 Table 4.2: Correlation Matrix ETR Bsize Bindp Bshare Oconc ROA Capint Flev Oconc*Bind Oconc*Bsz Fsize ETR 1.0000 Bsize -0.0655 1.0000 Bindp 0.0093 0.1351 1.0000 Bodshare -0.0001 -0.0023 -0.0975 1.0000 Oconc -0.1907 -0.1451 0.0485 -0.0851 1.0000 ROA -0.3699 0.0523 -0.0011 0.0931 0.1352 1.0000 Capint 0.2416 0.0096 0.1067 -0.2459 0.0142 0.0177 1.0000 Flev 0.1469 -0.0470 -0.0919 -0.0850 0.1233 -0.2808 0.0561 1.0000 Oconc*Bind -0.2094 -0.1296 0.1486 -0.0852 0.9783 0.1721 0.0324 0.1125 1.0000 Oconc*Bsz -0.2099 0.0562 0.0591 -0.0671 0.9565 0.1392 -0.0132 0.0920 0.9412 1.0000 Fsize -0.0571 0.2055 -0.1540 -0.3680 0.0761 0.0166 -0.0993 0.2311 0.0627 0.1227 1.000 Source: STATA Output as per appendix From table 4.2 it can be seen that the correlation between ETR and board size is negative (i.e. R of -0.0655). Going by the benchmark for effect sizes as posited by Cohen (1988, 1992) and Cohen, Manion and Anderson (2000) the coefficient appears relatively small. The negative coefficient implies that board size and tax avoidance are moving in opposite directions. ETR is however positively correlated with board independence (i.e. R of 0.0093). The sign of the coefficient suggests that that board independence and ETR are moving in the same direction, that is to say as board independence increases ETR also increases. In terms of strength of association, the association between board independence and ETR appears weak. Board equity ownership is negatively correlated with ETR (i.e. R of -0.0001). The sign of the association suggests that as board equity ownership increases, ETR decreases. However the magnitude of the association appears to be very small. High block ownership concentration is also negatively correlated with ETR with a correlation coefficient of -0.1907. This is slightly higher than for board size, independence and ownership but also appears to be small. The negative sign suggests that as block ownership concentration increases ETR decreases. Both interaction terms are inversely related to ETR with the interaction term between high ownership concentration and board size having a correlation coefficient of -0.2099 while the interaction term between high ownership 83 From the results in Table 4.3, it can be seen that Model 1 has an R 2 of 15.26% while Model 2 has an R 2 of 15.70%. The higher R 2 of Model 2 indicates that the inclusion of the two interaction terms has increased model explanatory power by 0.44%. The increase in adjusted R 2 from 8.86% to 8.95% also corroborates the increased explanatory power of the moderators. Based therefore on the results, both interaction terms were then included in the final model estimation. A key assumption necessary for OLS regression results to be unbiased and consistent is that the variance of the error term should be constant (homoskedasticity). The Breusch-Pagan / Cook-Weisberg test for heteroskedasticity was therefore conducted for both models. Under the null hypothesis that assumes constant variance of the error term, the test reported a Chi-square statistic of 164.52 and 160.70 for Models 1 and 2 respectively which both have p-values of 0.000. The p-values which are statistically significant at 1% therefore suggest the rejection of the null hypothesis of constant variance in favour of the alternate hypothesis of non-constant variance (i.e. heteroskedasticity). Further robustness tests were conducted on both models in the form of tests for multi- collinearity. According to Hairet al (2010), multi-collinearity is a situation whereby two or more of the co-variates in a model are highly correlated. While the presence of multi-collinearity does not necessarily prevent OLS estimates from being best linear unbiased estimators (BLUE), high amounts may cause the OLS estimators to have large variances and covariances, making precise estimation to be difficult. (Gujarati & Porter, 2009: 327). To test for the absence of harmful multi-colinearity the variance inflation factor (VIF) test and corresponding tolerance values (1/VIF) were computed. The rule of thumb is that VIF should be less than 10 and values of 84 1/VIF approaching 1 are evidence that a co-variate (Xij) is not collinear with the other co-variates (Gujarati & Porter, 2009). Table 4.3 gives the mean VIF for both Model 1 and Model 2. Mean VIF for Model 1 is 1.23, a value which is also lower than the accepted benchmark while that for Model 2 is 12.43, quite high due to the presence of the moderators. Individual variable VIF and Tolerance values are given in the appendix. From the results in the appendix it can be seen that for all the individual co-variates VIF values are consistently lower than the benchmark of 10 except those for the two interaction terms which are both above 10.In addition, with the exception of the two moderated terms, all other Tolerance values are greater than 0.5 and are thus closer to 1 than to 0, indicating no obvious collinearity issues. High VIF and low Tolerance values for interaction terms that appear to indicate multicollinearity do not however present much cause for concern and as such can be disregarded (Voss, 2004; Allison, 2012). 4.5Presentation and Analysis ofGMM Results Table 4.4 reports the results of the generalized method of moments (GMM) estimation of the effect of internal corporate governance mechanisms on ETRs of the studied DMBs. The results were gotten using the Arrellano-Bond dynamic panel GMM estimator in STATA 13. The use of the GMM estimator is in response to endogeneity concerns raised by researchers such as Hermalin and Weisbach (2003) with regards studies on corporate governance outcomes. GMM is also used by a related study (Minnick &Noga, 2010) to address the endogeneity issue. Following Minnick and Noga (2010), two lags of the dependent variable were estimated together with 1 lag each for the six co-variatesthought to be endogenous. The lagged co-variates were: board size, board independence, board shareholding and ownership concentration as well as the two moderating variables; ownership concentration*board size and ownership 85 concentration*board independence. In addition, one of the control variables ROA was also lagged. The introduction by the estimator of lags of the said independent variables meant that current values of the variables were automatically dropped from analysis. This was done by the estimator so as to avoid multi-collinearity. It is therefore the estimated lagged values of the endogenous independent variables that have been reported. Another issue that arises is that having lags of the dependent variable within the model introduces some degree of autocorrelation into the model. To address the autocorrelation concerns and simultaneously address the issue of the presence of heteroskedasticity, the robust option in STATA 13 was employed. The robust option uses Huber-White sandwich estimators to correct the standard errors of the estimated coefficients such that both heteroskedasticity and autocorrelation are kept to the minimum. Table 4.4: Summary of GMM Results Coefficient Robust Std. Err. Z P>|z| Intercept 3.2133 1.1476 2.8000 0.0050 *** ETR_L1 -0.2926 0.0793 -3.6900 0.0000 *** ETR_L2 -0.2606 0.0792 -3.2900 0.0010 *** Bsize_L1 -0.0083 0.0095 -0.8800 0.3810 Bindp_L1 0.4541 0.2476 1.7500 0.0790 * Bshare_L1 -0.4255 0.1979 -2.3000 0.0220 ** Oconc_L1 0.0754 0.1346 -1.0100 0.3130 ROA_L1 -3.6085 0.6086 -5.9300 0.0000 *** Oconc*Bindp -0.9329 0.3487 -2.6800 0.0070 *** Oconc*Bsize 0.0269 0.0102 2.6300 0.0080 *** Capint -0.0031 1.1992 0.0000 0.9980 Fsize -0.1616 0.0719 -2.2500 0.0250 ** Flev 0.5315 0.5250 1.0100 0.3110 AR(1) test -1.6827 0.0924 * AR(2) test 1.5813 0.1138 Wald 714.3