University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA UNIVERSITY OF GHANA BUSINESS SCHOOL TAX STRUCTURE AND ECONOMIC GROWTH IN GHANA BY ABENA DUROWAH HOPE (ID: 10280301) A LONG ESSAY SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF DEGREE OF MASTERS IN BUSINESS ADMINISTRATION IN FINANCE DEPARTMENT OF FINANCE JULY, 2019 University of Ghana http://ugspace.ug.edu.gh DECLARATION I, hereby, declare that this work is the result of my own research carried under supervision. This long essay has not been presented by anyone for any academic award in this institution or any other institution. All references used in the work have been fully acknowledged. I am solely responsible for any shortcomings in this work. …………………………...... ………………………………… Abena Durowah Hope Date (ID: 10280301) …………………………...... ………………………………… Dr. Patrick O. Asuming Date (Supervisor) i University of Ghana http://ugspace.ug.edu.gh DEDICATION I dedicate this work to the Almighty God and my entire family for their love and support. ii University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT First and foremost, praises to God Almighty for seeing me through the successful completion of this study. My profound gratitude goes to my supervisor, Dr. Patrick O. Asuming for his guidance. I am extremely grateful to my parents for the love and prayers and all their sacrifices towards my education for a better future. Special thanks to my siblings Ms. Michaelina Serwaah Owusu and Ms. Eunice Pokuaah Hope for their sacrifices, encouragement, and support throughout my study. And to everyone who contributed to the successful completion of my research work, I am very grateful. iii University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION ..................................................................................................................................... i DEDICATION ........................................................................................................................................ ii ACKNOWLEDGEMENT ..................................................................................................................... iii TABLE OF CONTENTS ....................................................................................................................... iv LIST OF ABBREVIATIONS ................................................................................................................ vi LIST OF FIGURES .............................................................................................................................. vii LIST OF TABLES ............................................................................................................................... viii ABSTRACT ........................................................................................................................................... ix CHAPTER ONE ..................................................................................................................................... 1 INTRODUCTION .................................................................................................................................. 1 1.1 Background of the Study......................................................................................................... 1 1.2 Statement of the Problem ........................................................................................................ 2 1.3 Objectives of the Study ........................................................................................................... 3 1.4 Research Questions ................................................................................................................. 4 1.5 Significance of the Study ............................................................................................................ 4 1.6 Chapter Outline ....................................................................................................................... 5 CHAPTER TWO .................................................................................................................................... 6 LITERATURE REVIEW ....................................................................................................................... 6 2.1 Introduction ............................................................................................................................. 6 2.2 Conceptual Review ................................................................................................................. 6 2.2.1 Meaning of Taxation and Related Concepts ....................................................................... 6 2.2.2 Meaning of Economic Growth and Related Concepts ........................................................ 7 2.2.3 Taxation in Ghana ............................................................................................................... 8 2.3 Theoretical Review ............................................................................................................... 10 2.3.1 Exogenous Growth Theories ............................................................................................. 10 2.3.2 Endogenous Growth Theories ........................................................................................... 11 2.4 Empirical Review .................................................................................................................. 12 2.5 Chapter Summary and Gaps in the Literature ....................................................................... 15 CHAPTER THREE .............................................................................................................................. 17 RESEARCH METHODOLOGY ...................................................................................................... 17 3.1 Introduction ........................................................................................................................... 17 3.2 Research Design .................................................................................................................... 17 3.3 Data ....................................................................................................................................... 17 3.4 Model Specification .............................................................................................................. 18 3.5 Stationarity Condition of Autoregressive Process ................................................................ 21 iv University of Ghana http://ugspace.ug.edu.gh 3.6 Dealing with Seasonality ...................................................................................................... 22 3.7 Measurement of Variables .................................................................................................... 23 3.8 Method of Data Analysis ...................................................................................................... 25 3.9 Chapter Summary ................................................................................................................. 26 CHAPTER FOUR ................................................................................................................................. 27 RESULTS AND DISCUSSIONS ......................................................................................................... 27 4.1 Introduction ........................................................................................................................... 27 4.2 Descriptive Statistics ............................................................................................................. 27 4.3 Degree of Association Among Variables .............................................................................. 31 4.4 Tax Structure and Growth ..................................................................................................... 32 4.5 Individual Taxes and Growth ................................................................................................ 35 4.6 Model Diagnostics ................................................................................................................ 37 4.7 Chapter Summary ................................................................................................................. 42 CHAPTER 5 ......................................................................................................................................... 45 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS ........................................................ 45 5.1 Introduction ........................................................................................................................... 45 5.2 Summary of Findings ............................................................................................................ 45 5.3 Contribution of the Study ...................................................................................................... 46 5.4 Conclusions ........................................................................................................................... 47 5.5 Policy Recommendations ...................................................................................................... 47 5.6 Limitations and Suggestions for Further Research ............................................................... 48 5.7 Chapter Summary ................................................................................................................. 49 REFERENCES ..................................................................................................................................... 50 v University of Ghana http://ugspace.ug.edu.gh LIST OF ABBREVIATIONS ARDL Auto-Regressive Distribution Lag ARIMA Auto-Regressive Integrated Moving Average ARIMAX Auto-Regressive Integrated Moving Average with Extra Explanatory Variables CST Communication Service Tax FDI Foreign Direct Investment GDP Gross Domestic Product GETFund Ghana Education Trust Fund GRA Ghana Revenue Authority GSS Ghana Statistical Service NHIL National Health Insurance Levy OECD Organisation for Economic Cooperation and Development OLS Ordinary Least Squares PAYE Pay-as-you-earn PMG Pooled Mean Group VAR Vector Auto-Regression VAT Value Added Tax vi University of Ghana http://ugspace.ug.edu.gh LIST OF FIGURES Figure 2.1: Tax to GDP Ratio for Ghana, Africa, and OECD Countries for 2000-2016 . . 9 Figure 2.2: Tax Structure of Ghana as at 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 3.1: Seasonal subseries plot of GDP for 2009-2017 . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 4.1 Graph of direct taxes and indirect taxes from 2009-2017 . . . . . . . . . . . . . 29 Figure 4.2: Capital Expenditure and Recurrent Expenditure by Government of Ghana for 2009 – 2017 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Figure 4.3 Correlation Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 4.4 Multiple Correlation Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Figure 4.5: Plot of Forecast Errors for Equation 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 4.6: Plot of Forecast Errors for Equation 3.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 4.7: Normal Density Plot Overlaid on Histogram of Forecast Errors for Equation 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 4.8: Normal Density Plot Overlaid on Histogram of Forecast Errors for Equation 3.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 4.9: Correlogram of Forecast Errors at Lags 1-10 for Equation 3.3 . . . . . . . . . . . 41 Figure 4.10: Correlogram of Forecast Errors at Lags 1-10 for Equation 3.4 . . . . . . . . . . 41 vii University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 4.1 Descriptive statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Table 4.2 Coefficient Estimates for Tax structure versus Growth (Equation 3.3) . . . . 33 Table 4.3 Coefficient Estimates for Individual Taxes versus Growth (Equation 3.4). . 36 viii University of Ghana http://ugspace.ug.edu.gh ABSTRACT Taxation is not only a major economic policy tool but also a means to encourage positive social behaviour and discourage negative social behaviour. Unfortunately, in Ghana, taxation is almost reduced to a tool for winning political power. Majority of the time, political discourse and economic policy discourse on taxation is at best anecdotal. The objective of this study is to investigate and provide empirical evidence on the relation between tax structure and growth in Ghana in order to inform future tax policy discourse. The study used ARIMAX time series model to estimate the relation between tax structure and growth over the research period 2009-2017. The relation between some individual taxes and growth was also estimated over the same period. The results suggest that tax structure (direct and indirect taxes) significantly impact economic growth in a positive direction. However, the positive effect of indirect taxes on growth is mainly driven by National Health Insurance Levy. Other indirect taxes considered like Value Added Tax, Communication Service Tax, and Excise Tax negatively impact growth. Based on the research findings, we recommended that going forward, government shift from the current tax structure where indirect taxes dominate direct taxes to a tax structure where direct taxes dominate indirect taxes. ix University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 Background of the Study Tax policy debate continues to dominate economic policy discussions in the press, academia, and civil society advocacy (Mcbride, 2012). This observation is attributable to the fact that taxes are not only the largest source of revenue for governments, states or municipalities, but also a tool to effect fiscal policy and cause positive change in behaviour. For instance, carbon tax which is a tax imposed on fossil fuel, reduces the harmful effects of carbon dioxide on the environment. However, the impact of taxes on economic growth is not as definite as illustrated in the carbon tax example. The imposition of carbon tax may negatively affect economic growth via increased fuel prices (Zhou, Shi, Li, & Yuan, 2011). The uncertainty about how taxes influence economic growth has prompted several empirical evaluations analysing the relation between taxation and economic growth. Whereas some studies conclude that taxes negatively impact economic growth (Ahmad, Sial, & Ahmad, 2016; Alesina & Ardagna, 2010; Atems, 2015; R. Barro & Redlick, 2011; Blanchard & Perotti, 2002; Dladla & Khobai, 2018; Ferede & Dahlby, 2012; International Monetary Fund, 2010; Koester & Kormendi, 1989; Lee & Gordon, 2005; Mertens & Ravn, 2012; Padovano & Galli, 2001; C. Romer & Romer, 2010), other studies suggest taxes lead to economic growth (Arnold, 2008; Azam & Shinwari, 2015; Egbunike, Emudainohwo, & Gunardi, 2018a; Kalaš, Mirović, & Andrašić, 2017; Mercedes & Mehrez, 2004; Miller & Russek, 1997; Stoilova, 2017; Yahaya, n.d.). In addition, there are some studies which observe no significant relation between taxes and economic growth (Easterly & Rebelo, 1993; Gale, Krupkin, & Rueben, 2015; Katz, Mahler, & Franz, 1983; Mendoza, Milesi-Ferretti, & Asea, 1997; Mendoza, 1 University of Ghana http://ugspace.ug.edu.gh Razin, & Tesar, 1994). Thus, there is lack of consensus among researchers regarding the relation between taxation and growth. In spite of the mixed empirical results, taxation remains a major tool in fiscal policy administration. Governments increase taxes to pursue contractionary fiscal policy or decrease taxes to pursue expansionary fiscal policy. In developing economies where financial markets are largely informal and underdeveloped, the unpredictability of tax-led fiscal policy is high. Consequently, tax policy initiatives are circumstantial rather than concrete. 1.2 Statement of the Problem Empirical studies lack conclusive evidence regarding the relation between taxation and economic growth. According to Huang and Frentz (2014), the lack of concensus is partly explained by the fact that “many different studies examine different measures and types of taxes, across different political units, economic conditions, time periods, fiscal and monetary policies”. For instance, Stoilova (2017) studied the relation between tax structure and growth across 28 european economies using cross-country pooled data from 1996-2014. The results of the study were mixed. Stoilova (2017) found that social security contributions, personal income taxes, and import taxes leads to positive economic growth while consumption taxes like Value Added Tax (VAT) negatively impact economic growth. The results also suggested that there is neutral relation between property taxes and economic growth. Acosta-Ormaechea, Sola and Yoo (2018) used Pooled Mean Group (PMG) estimation to investigate whether changes in tax position over the period 1970-2007 increase economic growth or not across 69 courntries. Contrary to Stoilova (2017), Acosta-Ormaechea et al. (2018) concluded that economic growth is positively affected by consumption and property taxes and negatively affected by personal income taxes and social security contributions. 2 University of Ghana http://ugspace.ug.edu.gh The application of these results to inform policy in individual countries is weakened by differences in taxes, tax systems, social redistribution preferences, political administration, and economic conditions (Huang & Frentz, 2014; Shinohara, 2014; Stoilova, 2017). Hence, there is the need to study not only the relation between taxation and economic growth , but also the relation between tax structure and economic growth using country specific data. There are few empirical studies in this regard (Pakistan – Ahmad et al., 2016 and Azam & Shinwari, 2015; Hungary – Benczúr, Kátay, & Kiss, 2018; United States – Kalaš et al., 2017; South Africa – Dladla & Khobai, 2018; Nigeria – Egbunike et al., 2018 and Kizito, 2014). The trend in the empirical literature is clear – recent empirical studies of tax structure and economic growth are country specific. In the case of Ghana, the closest empirical evaluation is carried out by Egbunike et al. (2018). Egbunike et al. (2018) demonstrated that revenue generated from taxes increases Gross Domestic Product (GDP) in Ghana. However, they failed to decouple the relationship between taxation and GDP in Ghana. Specifically, Egbunike et al. (2018) did not show the relationship between individual taxes and GDP. In tax policy analysis, it is valuable to know how direct taxation and indirect taxation drives economic growth. This study adds to Egbunike et al. (2018) by investigating the effects of direct taxation and indirect taxation on economic growth in Ghana. 1.3 Objectives of the Study We propose to pursue the following objectives in the study; i. Ascertain the relation between real GDP and direct taxes ii. Ascertain the relation between real GDP and indirect taxes. iii. Ascertain the relation between real GDP and the individual tax types. 3 University of Ghana http://ugspace.ug.edu.gh 1.4 Research Questions The study provides empirical evidence on tax structure and economic growth in Ghana which answers the following questions: i. What is the relation between real GDP and direct taxes? ii. What is the relation between real GDP and indirect taxes? iii. What are the relationships between real GDP and individual tax types? 1.5 Significance of the Study Differences in tax systems, economic conditions, and social preferences across countries has limited the policy implications of existing empirical results on taxation and economic growth (Huang & Frentz, 2014). There are few studies which performed country specific evaluation (Pakistan – Ahmad et al., 2016 and Azam & Shinwari, 2015; Hungary – Benczúr, Kátay, & Kiss, 2018; United States – Kalaš et al., 2017; South Africa – Dladla & Khobai, 2018; Nigeria – Egbunike et al., 2018 and Kizito, 2014). Egbunike et al. (2018) carried out the closest empirical evaluation which investigated relation between tax revenue and economic growth in Ghana; however, they failed to decouple the relation between tax revenue and economic growth. This study addresses this drawback in the Ghanaian Literature. In this direction, the study adds to empirical literature on taxation and economic growth. Endogenous growth theories proved that governments can affect economic growth through fiscal policy – government spending and taxation. In Ghana, tax policy discussions are politically contentious and often based on circumstantial evidence. This study represents a shift from the status quo. It produces concrete empirical evidence to inform tax policy debates and guide future tax reform initiatives by government. 4 University of Ghana http://ugspace.ug.edu.gh 1.6 Chapter Outline The study is divided into five chapters. The first chapter introduces the research. It presents the background of the study, the research problem, and the research objectives as well as research questions. Chapter one ended with a discussion of the relevance of the study. Chapter two reviews extant literature on taxation and economic growth with the purpose of identifying gaps that the research attempts to fill. The literature review included a conceptual review, a theoretical review, and an empirical review. Chapter three presents and provides justification for the methodology employed to achieve the research objectives while chapter four discusses the research findings. Chapter five concludes the research. It summarizes the research findings, discusses the drawbacks of the research, and offers recommendations for policy and future studies. 5 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO LITERATURE REVIEW 2.1 Introduction Chapter two reviews extant literature on tax structure and economic growth to identify research gaps for the study. The literature review contains four main sections. The first section provides a conceptual review which defines concepts in taxation and economic policy. The second section presents a theoretical review where main theories on taxation and economic growth are discussed. The next section critically examines extant literature on taxes and economic growth, and the final section summaries the gaps identified in the literature. 2.2 Conceptual Review 2.2.1 Meaning of Taxation and Related Concepts Tax is a derivative of the Latin word taxo which means a compulsory financial charge levied by a government, a state or a municipality on an individual or any legal entity to fund public expenditure. According to Goode (1984) a tax is a mandatory payment made by a business entity or an individual to government. The legal definition of tax is different from the economic definition of tax. Legally, a tax is any levy instigated by a statute. Some statutes imposing taxes in Ghana include the Income Tax Act, 2015 Act 896, the VAT Act 2017 Act 948, and the Excise Duty Act, 2014 Act 878. Economist view tax as a mandatory payment made to the public sector by the private sector. Tax structure describes the general classification of taxes in a tax system. Mostly, taxes are classified as direct taxes or indirect taxes. In Ghana, direct tax is an aggregation of corporate tax, income tax, gift tax, and capital gains tax while indirect income tax is an aggregation of 6 University of Ghana http://ugspace.ug.edu.gh VAT, Ghana Education Trust Fund (GETFund) Levy, and National Health Insurance Levy (NHIL). Other classifications of taxes are property taxes, consumption taxes, and income taxes (Acosta-Ormaechea et al., 2018; Arnold, 2008). According to this classification, income taxes comprise personal income tax, social security contribution, and corporate income tax; consumption taxes include VAT and other levies like excise duty, and custom duty; property taxes include gift tax and capital gains tax. Other researchers like Bleaney and Kneller (2001); Gemmell, Kneller, and Sant (2011) and Kneller, Bleaney, and Gemmell (1999) classified taxes as distortionary taxation or non-distortionary taxation. Distortionary taxation refers to taxes that impede economic growth. For example, income taxes and property taxes are considered to be distortionary taxes. Non-distortionary taxation refers to taxes that do not impede economic growth. For example carbon tax and plastic tax are considered to be non- distortionary taxes. Governments impose taxes to administer governance, to fund social projects, and to defray public debts. Tax revenues are used to run government agencies and pay salaries of civil servants like police officers, judges, etc. and public servants like ministers, district chief executives etc. Governments invest majority of tax revenues into the provision of infrastructure in healthcare, education, housing, road construction, energy generation, etc. Often, taxes are imposed to pay public debts arising from budget deficits. 2.2.2 Meaning of Economic Growth and Related Concepts Governments aspire to achieve high economic growth; because, it drives prosperity (Myles, 2009b). Economic growth occurs when “the real market value of goods and services” manufactured in a particular economy appreciates over time. Generally, the numeric measure of economic growth is percentage increase in real GDP (IMF, 2012). To foster comparison of 7 University of Ghana http://ugspace.ug.edu.gh growth among countries, percentage increase in per capital real GDP is the preferred measure of economic growth. In this study, we measure economic growth with changes in real GDP. There are several factors that influence economic growth. Importantly, economic growth is influenced by advances in human and physical capital, improvement in productivity, positive innovation, and entrepreneurship (Lucas, 1988; Rosenberg, 1982; Roubini & Backus, 1998). Other factors responsible for growth include rule of law (Johnson & Koyama, 2017), labor force participation (Roubini & Backus, 1998), and demographic changes (Roubini & Backus, 1998). 2.2.3 Taxation in Ghana Every country has a tax system which is administered by a government agency. In Ghana, the agency legally mandated to administer tax on behalf of government is the Ghana Revenue Authority (GRA). Prior to the formation of the GRA, the function of tax administration was performed by four agencies - the Internal Revenue Service, the Value Added Tax Service, the Customs, Excise and Preventive Service and the Revenue Agencies Governing Board. In 2009, these institutions were merged to form the GRA which is functionally organised into three divisions – the Customs Division, the Domestic Tax Revenue Division and the Support Service Division. Apart from the support service division which performs administrative role, the other two divisions administer various taxes. The formation of the GRA represented an administrative reform of the tax system. Bekoe, Danquah, and Senahey (2016) classified tax reforms in Ghana under three issues – broadening the tax base (1983-84), increasing production incentives (1985-86) and improving tax efficiency and equity (1986-2000). They also analyzed the effects of tax reforms on revenue collection in Ghana and concluded that tax reforms increased government revenue. 8 University of Ghana http://ugspace.ug.edu.gh Even though tax revenue mobilization has improved considerably, Ghana’s tax to GDP ratio remains below the Africa average. Prior to the first tax reform in Ghana, the tax to GDP ratio was 5%. This improved significantly to 11% following the reintroduction of VAT in 1998. As at 2016, the tax to GDP ratio in Ghana is 17.6% below the Africa average of 18.2%. Figure 2.1 presents the tax to GDP ratio for Ghana, Africa and OECD countries from 2000 to 2016. The data is obtained from the 2018 OECD Revenue Statistics in Africa report. Figure 2.1: Tax to GDP Ratio for Ghana, Africa, and OECD Countries for 2000-2016 Ghana's tax to GDP ratio remains below the Africa average for 200-2016 0.4 0.35 0.3 0.25 0.2 Ghana 0.15 Africa OECD 0.1 0.05 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Indirect taxes remain the biggest source of tax revenue for government. Other taxes on goods and services contribute 35% to total tax revenue followed by VAT (29%). Figure 2.2 highlights the tax structure as at 2016. 9 tax-GDP ratio University of Ghana http://ugspace.ug.edu.gh Figure 2.2: Tax Structure of Ghana as at 2016 Tax revenue from indirect sources continous to dominate other sources of tax revenue 34% 29% 14% 13% 9% 1% Other taxes on goods and services Value added taxes Corporate income tax Personal income tax Social security contributions Other taxes 2.3 Theoretical Review Some researchers proposed theoretical models explaining the relationship between economic growth and derivers of economic growth. They are either exogenous growth theories or endogenous growth theories. 2.3.1 Exogenous Growth Theories Exogenous growth theories encompass early growth theories which posit that growth is consequence of capital accumulation or increase in productivity of capital stock due to technical progress (which is assumed to occur exogenously) (Kurz & Salvadori, 2014). Classical exogenous growth theories in the literature include theories by Alfred Marshall and Gustav Cassel (Cassel, 1932) and Knut Wicksell (Wicksell, 1934). Neo-classical exogenous growth theories in the literature include the Ricardian theory (Ricardo, 1951), and the Solow- Swan growth model (Solow, 1956; Swan, 1956). 10 University of Ghana http://ugspace.ug.edu.gh Because exogenous growth theories focused heavily on capital accumulation as a source of growth, it assumes there is a limit to growth mainly as a result of the assumption of diminishing marginal rate of capital. Per exogenous growth theories, taxes on capital impede capital accumulation which in the long-run decreases economic growth (Myles, 2009b). 2.3.2 Endogenous Growth Theories Endogenous growth models provide a framework for achieving sustainable growth over time. Sustainable growth under exogenous growth models is limited by the notion of diminishing marginal returns to capital. Also, exogenous growth models assume technical progress (technological advancement or innovation and increases in human capital or education) occur exogenously. Attempts by some researchers to investigate sources of technical progress lead to the promulgation of endogenous growth models. The AK model which is the earliest endogenous growth model was developed by Romer (1986). Under the AK model, the sole input for production is and constant returns to input (capital) prevail. This implies that continuous growth is possible through constant increase in capital stock (Myles, 2009b). In essence, tax policies that increase savings which in turn increase investment in capital stock may positively impact economic growth. Romer (1990) and Mankiw, Romer, and Weil (1992) modified the AK model by adding a human capital as a second input to the original production function. They referred to the new model as the human capital approach to achieving endogenous growth. Human capital refers to skillfulness of labour or the quality of labour. The model assumes that human capital just like physical capital can be developed and accumulated over time through training and education. By introducing human capital, quality of labour determines level of technical 11 University of Ghana http://ugspace.ug.edu.gh progress. Under the human capital model, tax policies that promote education are expected to increase economic growth. There are also endogenous growth models of innovation based on Schumpeter's (1934) idea of creative destruction. Creative destruction refers to the introduction of superior goods, services or processes to replace old ones. Segerstrom, Anant and Dinopoulos (1990) are credited with being the first researchers to model the growth-innovation process. In the innovation model, technical progress is captured in increased superior inputs arising from the replacement of old inputs by better products and processes. In essence, productivity increases overtime as better inputs replace old inputs. The tax policy implication is that taxes that incentivize innovation increase economic growth. 2.4 Empirical Review The lack of consensus in the theoretical literature on the impact of taxes on economic growth is reflected in the empirical literature as well. The empirical literature contain findings of negative relation, positive relation, and neutral relation between taxation and economic growth. Huang and Frentz (2014) observed that the mixed empirical results is partly explained by the fact that “many different studies examine different measures and types of taxes, across different political units, economic conditions, time periods, fiscal and monetary policies” For instance, Stoilova (2017) investigated the impact of taxes on economic growth across 28 European economies using cross-country pooled data from 1996-2014. The results of the study were mixed. Stoilova (2017) found that personal income taxes, retirement contributions, taxes on productions and imports positively impact economic growth while consumption taxes like VAT negatively impact economic growth. The results also indicated 12 University of Ghana http://ugspace.ug.edu.gh there is neutral relation between property taxes and economic growth. The use of cross- country data limits the policy application of these findings on individual country level. Lee and Gordon (2005) analyzed the effects of corporate tax rate and labour income tax rate on per capital GDP growth from 1970 – 1997 across 70 countries . They applied OLS to estimate the effects. Their conclusion was that corporate tax rates reduce growth in per capital GDP significantly while labour income taxes do not affect per capital GDP. According Lee and Gordon (2005), a 10 per cent reduction in corporate tax can spur growth by 1.1 per cent. A similar research by Acosta-Ormaechea, Sola and Yoo (2018) studied an updated data from 1970 to 2007 across 69 countries. The dataset was made up of 21 high- income economies, 23 middle-income economies and 25 low-income economies. Contrary to Lee and Gordon (2005), Acosta-Ormaechea et al. (2018) adopted Pooled Mean Group (PMG) estimation to study the relations between changes in tax composition and economic growth. The results suggested that for high and middle income economies, personal income taxes and social security contributions reduce economic growth while consumption taxes and property taxes boost economic growth. This result is contrary to Stoilova (2017) and to some extent Lee and Gordon (2005). Whereas Stoilova (2017) posited that personal income taxes enhance economic growth, Acosta-Ormaechea et al. (2018) argued otherwise. Finally, Lee and Gordon (2005) also concluded that personal income taxes influence economic growth insignificantly. The differences in empirical results can be attributed to differences in estimation methods and period of estimation as well as differences in tax systems across countries. Another study which used PMG estimation was carried by Ojede and Yamarik (2012). Ojede and Yamarik (2012) examined the relation between taxation and economic growth among 48 US states for 1967-2008. State-level economic growth was approximated using real personal 13 University of Ghana http://ugspace.ug.edu.gh income growth. The results showed that sales tax is reduce growth in the long-run while property tax reduce growth in the short-run. However, they found no relation between income taxes and short-run or long-run growth. Bujang, Hakim, and Ahmad (2013) investigated whether or not taxation impacts long-run economic growth among high-income OECD countries and developing countries. The dataset comprised of 24 high-income OECD countries and 24 developing countries spanning the period of 2000-2009. Unlike other studies which use GDP as the only estimator of economic growth, Bujang et al. (2013) included Saving, Foreign Direct Investment (FDI), Unemployment, and International trade as indicators of economic growth. Panel Cointegration Analysis was employed to estimate the relation. The results indicated an inconsistent relation between taxation and economic growth across the two groups. While there is significant association between taxation and GDP, saving, and FDI in high-income OECD countries, there is insignificant association between taxation and GDP and saving in developing countries. Also, while the study found no relation between taxation and International trade among high-income OECD countries, the study found taxation positively impacts international trade in developing countries. In essence, for a given tax policy in a high-income OECD country, there is opposite long-run effect in a developing country. The results of Bujang et al. (2013) demonstrated the impact of differences in tax systems on taxation and economic growth empirical studies. Therefore, recent empirical evaluations focused on country specific data to overcome challenges associated with different tax systems when using cross-country data. For instance, Benczúr et al. (2018) used a behavioural simulation model to investigate the outcomes of tax reforms implemented in Hungary from 2008 to 2013. They concluded that tax reforms which reduced employers’ contributions and increased VAT positively impacted long-run employment and GDP. Dladla and Khobai (2018) used Auto-Regressive Distribution Lag (ARDL) to analyze the impact of taxes on 14 University of Ghana http://ugspace.ug.edu.gh economic growth in South Africa from 1981 to 2016. They demonstrated that taxation reduced economic growth significantly over both the short-run and the long-run. In Pakistan, Azam and Shinwari (2015) used time series regression to study the impact of income tax, sales tax, custom duties, excise duties, and workers welfare tax on GDP. In exception of workers welfare tax which negatively impact GDP, they found that taxes positively affect GDP. In the case of Ghana, the closest empirical evaluation is carried out by Egbunike et al. (2018). Egbunike et al. (2018) used multiple regression to investigate the impact of taxation on GDP for the period 2000-2016. The results showed that tax revenue improves GDP in Ghana. The drawback to Egbunike et al. (2018) is they failed to decouple the relationship between taxation and GDP in Ghana. Precisely, Egbunike et al. (2018) did not demonstrate the relationship between the different types of taxes and GDP. This limits the policy implications of their findings. 2.5 Chapter Summary and Gaps in the Literature This chapter presented a conceptual review, a theoretical review, and an empirical review of extant literature on taxation and economic growth. The conceptual review highlighted tax reforms in Ghana and trends in tax to GDP ratio in Ghana, Africa, and OECD countries. Importantly, the trend showed that Ghana lagged behind the Africa average over the past 16 years. Exogenous growth theories and endogenous growth theories were discussed in the theoretical review. Exogenous growth theories represent early growth theories which explain growth as a function of capital accumulation or increase in productivity of capital stock. According to exogenous growth models, taxation on capital slow down economic growth in the long-run through reduced capital formation. Endogenous growth theories explain how to 15 University of Ghana http://ugspace.ug.edu.gh achieve sustained growth over time. Generally, endogenous growth theories predict relationship between taxation and growth; however, the direction of the relation is dependent on the type of taxation and the type of endogenous growth model. For instance, under the human capital model, tax policies that promote education are expected to increase economic growth The empirical review indicated lack of consensus in the empirical literature on the relation between taxation and economic growth. Whereas some studies suggested taxation positively impact economic growth, other studies found taxation negatively impacts economic growth. Amidst these extremes is another group of studies which indicated neutral impact of taxation on economic growth. Meta-researches which analyzed the various empirical studies attributed the mixed empirical results to differences in tax systems, social redistribution preferences, political administration, and economic conditions (Huang & Frentz, 2014; Shinohara, 2014; Stoilova, 2017). In attempt to avoid this drawback, recent empirical studies (Ahmad et al., 2016; Benczúr et al., 2018; Dladla & Khobai, 2018; Yahaya, n.d.) investigated country specific relations. As Myles, (2009a) pointed out, “individual country studies should be more informative about the causes of growth than cross-country aggregate analysis”. In Ghana, the closest empirical evaluation was performed by Egbunike et al. (2018). Egbunike et al. (2018) carried out an aggregate study and concluded that tax revenue increases GDP in Ghana and Nigeria. However, the results of Egbunike et al. (2018) raises several questions. For instance, which tax or tax structure positively impacts GDP? Which tax use case (government expenditure) affect GDP? The answers to these questions have significant policy implications. This study represents an attempt to examine the empirical data in Ghana and provide evidence which addresses these questions. 16 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE RESEARCH METHODOLOGY 3.1 Introduction This chapter presents and discusses the methodology employed to attend the research objectives. The first section which talks about the research design highlights the systematic approach to obtaining the research data to analysing the data. The next section describes the research data and the sources of the data. Subsequent sections presents the regression model, a descriptive statistics, and the method and tools adopted to estimate the regression model. The final section summarizes the chapter. 3.2 Research Design We set out to estimate the relation between tax structure and economic growth in Ghana. We defined tax structure to be direct taxes and indirect taxes and we measured economic growth with change in real GDP. To achieve our research objective, we employed a quantitative research design. Specifically, we applied a time series analysis to estimate the magnitude and direction of effects of direct and indirect taxation on real GDP. We divided the data analysis stage into three sections: model identification, model prediction and model diagnostics. 3.3 Data The main data sets required for the research include quarterly real GDP figures and quarterly tax revenue aggregated into direct tax revenue and indirect tax revenue from 2009 to 2017. 17 University of Ghana http://ugspace.ug.edu.gh Data on quarterly real GDP was obtained from the Ghana Statistical Service (GSS) GDP bulletin available at http://www.statsghana.gov.gh/gdp_bulletin.html. Quarterly data on tax revenues from 2009 to 2017 was procured through the GRA data service unit. Government expenditure is included as an explanatory variable to enable us examine the transmission mechanism through which tax revenue impacts growth. The quarterly government expenditure data was retrieved from the Ministry of Finance fiscal data page available at https://www.mofep.gov.gh/fiscal-data. 3.4 Model Specification In chapter two, we established taxation impacts economic growth through endogenous growth models. Nevertheless, the direction of the relation is dependent on the type of taxation and the type of endogenous growth model (Myles, 2009b). For instance, under the human capital model, tax policies that promote education are expected to increase economic growth; under the AK model tax policies that increase savings are expected to increase economic growth; per the innovation model tax policies that encourage positive innovation are expected to boost productivity and economic growth. The objective of this research is to attempt to ascertain the magnitude and the direction of relation between taxes and economic growth in Ghana. Therefore, growth regression is most suitable (Myles, 2009b). Growth regression originated from (Barro, 1991) where the dependent variable “change in GDP” is explained by economic and non-economic variables capable of influencing GDP (Myles, 2009b). Thus, the general specification of a growth regression model is as follows; ∆𝐺𝐷𝑃𝑡 = 𝛼 + 𝛽𝑋𝑖𝑡 + 𝜀𝑖 (3.1) 18 University of Ghana http://ugspace.ug.edu.gh Where ∆𝐺𝐷𝑃𝑡 represents change in GDP in time 𝑡, 𝛼 and 𝛽 are coefficient estimates, 𝑋𝑖𝑡 represents potential economic and non-economic explanatory variables in time 𝑡, and 𝜀𝑖 is the error term. Growth regression is widely used in several growth and tax structure studies (Ahmad et al., 2016; Bonga-bonga & Ahiakpor, 2016; Dladla & Khobai, 2018; Easterly & Rebelo, 1993; Egbunike et al., 2018a; Kalaš et al., 2017; Lee & Gordon, 2005; Miller & Russek, 1997; Stoilova, 2017). For instance (Egbunike et al., 2018a) used growth regression to study the relation between GDP and tax revenue with GDP as dependent variable and tax revenue as independent variable. Likewise, Dladla and Khobai (2018) applied growth regression to examine the impact of taxes on economic growth in South Africa with GDP as dependent variables and several tax rates as independent variable. In order to estimate equation 3.1, we employed time series analysis. Time series analysis predicts future values based on past values. Using time series analysis enabled us to incorporate the effect of passage of time on growth into estimating the general growth model. There are several techniques for forecasting time series models. Each technique is suitable for a particular case. Considering the possibility that some independent variables may exhibit linear interdependencies across time, the Vector Autoregression (VAR) technique seemed the most appropriate approach for the research. However, the complexity of estimating a VAR model is beyond the scope of our research plan and design. The alternative approach which is the approach we adopted is an extension of the Autoregressive Integrated Moving Average (ARIMA) technique called the ARIMAX model (Peter & Silvia, 2013). ARIMAX stands for autoregressive integrated moving average with extra explanatory variables. As the name suggests, an ARIMAX model allows for consideration of extra explanatory variables which is not the case in ARIMA models. In order words, ARIMA is used for estimating univariate 19 University of Ghana http://ugspace.ug.edu.gh time series while ARIMAX is used for estimating multivariate time series with extra explanatory variables which are time independent. The extra explanatory variables are included in the model using a transfer function (Peter & Silvia, 2013). The general equation for estimating an ARIMAX model is given as; ?̂?𝑡 = ?̂? + 𝛽0̂𝑦𝑡−1 + 𝛽1̂𝑦𝑡−2 + ⋯ + 𝛽?̂?𝑦𝑡−𝑘 + ∅̂. 𝑋 + 𝜀?̂? + 𝜃1̂𝜀𝑡−1̂ + 𝜃2̂𝜀𝑡−2̂ + ⋯ + 𝜃?̂?𝜀𝑡−?̂? (3.2) where: 𝑌𝑡 is the dependent variable in time 𝑡 ∅̂𝑋 is the transfer function of any economic variable ?̂?, ?̂?, ∅̂,and 𝜃 are estimated coefficients 𝜀?̂? is the estimated error term in time 𝑡 We localized the general ARIMAX equation 3.2 to estimate the relationship between growth and tax structure as follows: 𝐺𝐷𝑃?̂? = ?̂? + 𝛽0̂𝐺𝐷𝑃𝑡−1 + 𝛽1̂𝐺𝐷𝑃𝑡−2 + ⋯ + 𝛽?̂?𝐺𝐷𝑃𝑡−𝑘 + ∅̂. 𝑋 + 𝜃?̂?𝜀𝑡−?̂? (3.3) where: 𝐺𝐷𝑃𝑡 is Ghana’s real gross domestic product in time 𝑡 ?̂?, ?̂?, ∅̂,and 𝜃 are estimated coefficients 𝜀?̂? is the estimated error term in time 𝑡 ∅̂𝑋 is the transfer function of any economic variable. The transfer term ∅̂𝑋 is further expressed as; 𝑑𝑖𝑟𝑒𝑐𝑡𝑡 + 𝑖𝑛𝑑𝑖𝑟𝑒𝑐𝑡𝑡 + 𝑐𝑎𝑝𝑒𝑥𝑡 + 𝑟𝑒𝑐𝑒𝑥𝑡 20 University of Ghana http://ugspace.ug.edu.gh where: 𝑑𝑖𝑟𝑒𝑐𝑡𝑡 is total value of direct taxes in time 𝑡 𝑖𝑛𝑑𝑖𝑟𝑒𝑐𝑡𝑡 is total value of indirect taxes in time 𝑡 𝑐𝑎𝑝𝑒𝑥𝑡 is total value of government capital expenditure in time 𝑡 𝑟𝑒𝑐𝑒𝑥𝑡 is total value of government recurrent expenditure in time 𝑡 Similarly, we specified equation 3.4 to estimate the relationship between GDP and some specific tax components. 𝐺𝐷𝑃?̂? = ?̂? + 𝛽0̂𝐺𝐷𝑃𝑡−1 + 𝛽1̂𝐺𝐷𝑃𝑡−2 + ⋯ + 𝛽?̂?𝐺𝐷𝑃𝑡−𝑘 + ∅̂. 𝑋 + 𝜃?̂?𝜀𝑡−?̂? (3.4) where: ∅̂. 𝑋 is defined as: 𝑃𝐴𝑌𝐸𝑡 + 𝑠𝑒𝑙𝑓_𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑑𝑡 + min_𝑟𝑜𝑦𝑎𝑙𝑡𝑖𝑒𝑠𝑡 + 𝑐𝑜𝑟𝑝𝑜𝑟𝑎𝑡𝑒𝑡 + 𝑉𝐴𝑇𝑡 + 𝑁𝐻𝐼𝐿𝑡 + 𝐶𝑆𝑇𝑡 + 𝑒𝑥𝑐𝑖𝑠𝑒𝑡 + 𝑐𝑎𝑝𝑒𝑥𝑡 + 𝑟𝑒𝑐𝑒𝑥𝑡 3.5 Stationarity Condition of Autoregressive Process The error term 𝜀?̂?, in an ARIMA model is assumed to follow a stationary univariate process (G. Box & Jenkins, 1970). In other words, the mean, covariance, and autocorrelation of the time series should remain constant over time. During the model identification stage, an autocorrelation plot of the time series indicated that GDP is non-stationary. A formal test of stationarity using Augmented Dickey-Fuller test also confirmed that the target variable is non-stationary. To convert the non-stationary time series into a stationary series, we used the method of differencing. Differencing is a widely applied method in time series analysis to achieve 21 University of Ghana http://ugspace.ug.edu.gh white-noise in a dependent variable (G. Box & Jenkins, 1970). It means, subtracting each data point in the series from its successor in the series. We used the “ndiffs” function in the “forecast” package within the R statistical programming environment to estimate the number of times of differencing necessary to achieve stationarity. 3.6 Dealing with Seasonality From the quarterly subseries plot of the target variable, GDP over the period 2009-2017 shown in figure 3.1, the series exhibited seasonality over each quarter. Often, seasonality in the target variable influences the relationship between the target variable and predictor variables. The general approach to dealing with this problem which we pursued is to deseasonalised the series in order to examine only the non-seasonal component. Figure 3.1: Seasonal subseries plot of GDP for 2009-2017 22 University of Ghana http://ugspace.ug.edu.gh 3.7 Measurement of Variables GDP The GSS publishes quarterly estimates of nominal GDP and price adjusted GDP periodically. These estimates are adjusted at the end of the year upon availability of new economic data. We used the price-adjusted figures in the research analysis. This enables us to eliminate price effects on growth as we try to examine effects of taxation on real output. Direct Tax Revenue (direct) Direct taxes as used in the data analysis constitutes quarterly aggregation of tax revenues from PAYE, self-employed assessment, corporate tax, and mineral royalties. Indirect Tax Revenue (indirect) Indirect taxes as used in the research is made up quarterly aggregation of tax revenues from VAT, NHIL, Excise duties, and CST. Government Capital Expenditure (capex) The Ministry of Finance publishes monthly fiscal data for government expenditures. The capital expenditure component is grouped two – capital expenditure financed by domestic resources and capital expenditure financed by foreign resources. Since we are interested in how tax revenues drive growth, we used domestic financed component of capital expenditure in the model estimation. 23 University of Ghana http://ugspace.ug.edu.gh Government Recurrent Expenditure (RECEX) Recurrent government expenditure comprises expenditure line items associated with day-to- day administration of governance. The figures for recurrent expenditure used in the research is arrived at by deducting capital expenditure from total expenditure. PAYE PAYE as used is this research refers to quarterly amount of tax revenue raised by government of Ghana from pay-as-you-earn. Self-employed Self-employed as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from self-employed assessments. Corporate Corporate as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from corporate taxes. NHIL NHIL as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from NHIL levies. 24 University of Ghana http://ugspace.ug.edu.gh CST CST as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from CST levies. Excise Excise as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from excise duties. Mineral Royalties Mineral royalties as used in this research refers to quarterly amount of tax revenue raised by government of Ghana from mineral royalties. 3.8 Method of Data Analysis We employed a quantitative research design approach to investigate and obtain evidence to answer the research questions. The data analysis process was divided into three stages – model identification, model estimation, and model diagnostics. We started with model identification where we examined the descriptive properties and the distribution of each variables. We investigated whether the dependent series exhibits seasonality and requires seasonal differencing. Similarly, we examined whether the dependent series is stationary or requires stationary differencing. The next natural step in the model identification stage of a time series analysis is to determine order (that is 𝑝, 𝑞) of the model. However, the software 25 University of Ghana http://ugspace.ug.edu.gh procedure we applied in estimating the model is capable of iterating and discovering the appropriate 𝑝 and 𝑞 values to use. After model identification, we estimated the model using the 𝑎𝑢𝑡𝑜. 𝑎𝑟𝑖𝑚𝑎 function included in the 𝑓𝑜𝑟𝑒𝑐𝑎𝑠𝑡 statistical package within the R programming environment. 𝑎𝑢𝑡𝑜. 𝑎𝑟𝑖𝑚𝑎 is an automated procedure; therefore, we only specified the dependent variable and the transfer function as inputs for the function to use to estimate the model. Lastly, we examined whether the forecasted model is proper or not. To do this, we investigated whether a plot of the residuals is without any pattern, we examined whether the distribution of the residuals is normally distributed, and we performed portmanteau tests on the residuals to find out whether they exhibit autocorrelation. 3.9 Chapter Summary In this chapter, we presented and discussed the methodology we employed to gather evidence to answer the research questions raised. We began the chapter by stating that the research design is a quantitative research approach. After that, we presented how we obtained the research data and from which source. Subsequently, we described the procedure we used to specify the quantitative model to be used for the estimation of the relationships. We continued to discuss how the variables included in the regression estimation are measured and ended the chapter with a detailed discussion of the method of data analysis used to achieve the research objectives. 26 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR RESULTS AND DISCUSSIONS 4.1 Introduction This chapter presents and discusses the research findings. The first two sections describe the characteristics of the research data using descriptive statistics, Pearson’s correlation coefficient estimates, and scatterplot. The next section presents and discusses the results of the coefficient estimates for the tax structure and growth model (equation 3.3). This is followed by the fourth section which also presents and discusses the results of the coefficient estimates for the individual taxes and growth model (equation 3.4). The subsequent sections examine the robustness of the coefficient estimates while the final section provides a summary of the chapter. 4.2 Descriptive Statistics As indicated in chapter one, the theme of the research is to examine the relation between tax structure and growth where tax structure is represented as direct versus indirect taxation and growth is estimated using changes in GDP. Focusing on this theme, we looked at two characteristics of the data set using descriptive statistics. These characteristics are central tendency and variability. From Table 4.1, the mean, the median, and the maximum value as well as the minimum value portray central tendency of the research variables. From the same table, variance, standard deviation, and range indicate the extent of variability of each research variable. Over the period 2009-2017, total revenue from direct taxes is twice more than revenue from indirect taxes. Likewise, average revenue from direct taxes is greater than average revenue 27 University of Ghana http://ugspace.ug.edu.gh from indirect taxes. This is expected since the approach to taxation in Ghana is more income- based than consumption-based as mentioned by (Bekoe et al., 2016). In Figure 4.1, the linear scatterplot of direct taxes and indirect taxes suggests an upward rise in tax revenue over the period. The positive Pearson’s skewness estimates further support this observation. According Bekoe et al. (2016), the upward trend observed is mainly attributable to tax reforms aimed at boosting tax mobilization. On variability, the direct tax data appears more variable than the indirect tax data. This is potentially driven by variability in GDP as the Pearson correlation coefficient between direct taxes and GDP of 0.96 is higher than that of indirect taxes and GDP of 0.90. Table 4.1 Descriptive statistics All values are measured in GHȻ million. Capex and Recex standards for capital expenditure by government and recurrent expenditure of by government respectively. Indirect per capita Summary Statistics Direct Taxes Taxes GDP Capex Recex Mean 1646.28 635.68 7084.87 373.14 5790.34 Median 1482.28 472.92 7264.59 306.72 5157.54 Standard Deviation 944.46 438.30 1199.06 282.36 3732.03 Sample Variance 891999.56 192104.96 1437752.08 79728.79 13928023.80 Kurtosis -0.41 0.40 -0.95 1.57 -0.63 Skewness 0.54 1.06 -0.30 1.13 0.60 Range 3651.95 1,604.19 4223.06 1269.16 13492.07 Minimum 389.75 110.89 4889.35 0.32 1121.06 Maximum 4041.69 1715.08 9112.40 1269.48 14613.12 Sum 59265.90 22884.61 255055.27 13433.09 208452.22 28 University of Ghana http://ugspace.ug.edu.gh Figure 4.1 Graph of direct taxes and indirect taxes from 2009-2017 For 2009-2017, tax revenue from direct taxes is about twice tax revenue from indirect taxes 4,500.00 4,000.00 3,500.00 3,000.00 2,500.00 Direct taxes 2,000.00 Indirect taxes 1,500.00 1,000.00 500.00 - From Table 4.1, the difference between capital expenditure and recurrent expenditure for the period 2009 – 2017 is sizeable. On average, capital expenditure lags recurrent expenditure by about GHȻ 5.5 million. From Figure 4.2, the amount of expenditure peaks at the fourth quarter of any given year and drops in the first quarter of the following year. This pattern in expenditure may be attributable to the budget cycle of government and availability of funds for disbursements. In order words, funds tend to be more available towards the fourth quarter in the budget cycle than in the first quarter of the budget cycle. Analyzing Figure 4.2 with respect to election periods may also suggest that government spends more during election period (fourth quarter of election years 2012 and 2016). On variability, recurrent expenditure seemed more volatile compared to capital expenditure. 29 tax revenue (GHȻ million) University of Ghana http://ugspace.ug.edu.gh Figure 4.2: Capital Expenditure and Recurrent Expenditure by Government of Ghana for 2009 – 2017 For 2009 - 2017, Government spends more than three-quarter of tax revenue on recurrent expenditure than capital expenditure 2017 2017 2017 2017 2016 2016 2016 2016 2015 2015 2015 2015 2014 2014 2014 2014 2013 2013 2013 2013 2012 2012 2012 2012 2011 2011 2011 2011 2010 2010 2010 2010 2009 2009 2009 2009 CAPEX RECEX 30 University of Ghana http://ugspace.ug.edu.gh 4.3 Degree of Association Among Variables In the previous section, we examined central tendency and variability properties of each variable using some descriptive statistics. In this section, we look at how each variable is related to the other variable using Pearson correlation coefficients and multiple scatterplot or correlation plot diagram. The correlation matrix shown in Figure 4.3 consists of Pearson’s pairwise correlation coefficient estimates at 0.01(***), 0.05(**), and 0.1(*) significance levels. The multiple correlation plot in Figure 4.4 provides a diagrammatic illustration of the association contained in the pairwise correlation matrix. From the correlation matrix, direct taxes and indirect taxes are positively significantly associated with recurrent expenditure and insignificantly associated with capital expenditure. This suggests that tax revenues are applied to finance recurrent expenditure more than capital expenditure. The correlation matrix also shows that apart from capital expenditure, direct taxes, indirect taxes, and recurrent expenditure have positive significant association with GDP. This gives an idea of the transmission mechanism through which taxes relate to growth. Roughly, the association between direct taxes, indirect taxes, and recurrent expenditure implies that, recurrent expenditure financed through tax revenues impact growth positively. Figure 4.3 Correlation Matrix GDP Direct taxes Indirect taxes Capex Recex P er capita GDP 1 Direct taxes 0.9551*** 1 Indirect taxes 0.9045*** 0.9607*** 1 Capex 0.1618 0.1493 0.0177 1 Recex 0.9272*** 0.9493*** 0.9131*** 0.2563 1 31 University of Ghana http://ugspace.ug.edu.gh Figure 4.4 Multiple Correlation Plot 4.4 Tax Structure and Growth The main objective of this study is to examine the relation between tax structure and growth. In the previous chapter, we indicated that the study uses the direct and indirect classification of tax as tax structure and real GDP as a measure of growth. The results of the effect of direct taxes and indirect taxes on real GDP are presented in Table 4.2. 32 University of Ghana http://ugspace.ug.edu.gh Table 4.2 Coefficient Estimates for Tax structure versus Growth (Equation 3.3) Variable Estimate Std.Error z value Pr ( > |z|) intercept 17.6678 0.3184 55.4854 0.0000 *** GDTt-1 0.3645 0.1963 1.8568 0.0633 * GDTt-2 0.3532 0.1686 2.0954 0.0361 ** direct 0.0636 0.0350 1.8147 0.0696 * indirect 0.1106 0.0254 4.3474 0.0000 *** Capex -0.0008 0.0024 -0.3256 0.7447 Recex 0.0656 0.0252 2.5979 0.0094 *** Regression with ARIMA(2,0,0) errors Sigma^2 estimated as 0.0004338: log likelihood = 91.86 AIC = -167.72 AICc = -162.38 BIC = -155.05 ***, **, and * indicate 1%, 5%, and 10% levels of significance respectively. The coefficient estimates in Table 4.2 shows that GDPt-1, GDPt-2, direct taxes and indirect taxes significantly impact GDP in a positive direction. On average, one percentage increase in direct tax revenue increases GDP by 0.06% while one percentage increase in indirect tax revenue increases GDP by 0.11%. Just as Egbunike et al. (2018) concluded on the aggregate level that there is a positive relation between taxation and GDP in Ghana, these results also suggest that on the tax structure level, there is a positive relation between taxation and GDP in Ghana. Importantly, the results in Table 4.2 supports the endogenous growth theory canvased in chapter three which underpins this research. Particularly, we noted that endogenous growth theories predict that taxation impacts growth unlike exogenous growth theories which predict otherwise. 33 University of Ghana http://ugspace.ug.edu.gh The regression estimates for capital expenditure by government and recurrent expenditure by government give an indication of the mechanism through which taxes impact GDP. This was not the case in Egbunike et al. (2018). From Table 4.2, there is a significant positive relationship between recurrent expenditure and GDP, but insignificant negative relationship between capital expenditure and GDP. In other words, tax revenue drive growth through recurrent expenditure. On average one percentage increase in recurrent expenditure by government increases GDP by 0.07% while one percentage increase in capital expenditure by government decreases GDP by 0.0008%. We expected a significant relationship between GDP and recurrent expenditure since we observed from Figure 4.2 that government allocates more than three-quarter of annual tax revenue to recurrent expenditure. However, the direction of the relationship is unexpected, since the major recurrent expenditure items financed by government is deemed unproductive. For instance, every year government spends close to 70% of tax revenue on salaries and debt servicing. The probable explanation for the predicted relationship is that, individuals spend salaries received from government on goods and services leading to increased economic activity and output. Unfortunately, this realm of conjecture is beyond the scope of our results. Even though the magnitude of the relationship between GDP and capital expenditure is not significant, the direction of the relationship is contrary to what we expected. We expected capital expenditure to positively impact GDP. Nevertheless, the negative relationship is probably due to fact that we used a narrow measurement of capital expenditure, which is capital expenditure financed from tax revenues. In fact, the general evidence is that, a greater proportion of government capital expenditure is either financed by debt or private equity or a combination. 34 University of Ghana http://ugspace.ug.edu.gh 4.5 Individual Taxes and Growth The second objective requires we examine the effects of specific individual taxes on GDP. Table 4.3 contains the regression outputs to that effect. Eight individual taxes were investigated. Out of these, PAYE, self-employed tax, corporate tax, mineral royalties, and NHIL are of direct nature, and VAT, CST, and excise duty are of indirect nature. The results in Table 4.3 present two themes – taxes of direct nature positively impact GDP and in exception of NHIL, taxes of indirect nature negatively impact GDP. For indirect taxes, a percentage increase in tax revenue from PAYE increases GDP insignificantly by 0.06%; a percentage increase in tax revenue from self-employed tax increases GDP significantly by 0.13%; a percentage increase in tax revenue from corporate tax increases GDP insignificantly by 0.02%; a percentage increase in tax revenue from mineral royalties increases GDP insignificantly by 0.01%. The positive net effect of these taxes supports the finding in the previous section that direct taxes positively impact GDP. With regard to indirect taxes, a percentage increase in tax revenue from VAT decreases GDP insignificantly by 0.01%; a percentage increase in tax revenue from NHIL increases GDP insignificantly by 0.13%; a percentage increase in tax revenue from CST decreases GDP insignificantly by 0.07%; a percentage increase in tax revenue from excise duties decreases GDP insignificantly by 0.005. Notably, the positive effect of tax revenue from NHIL on GDP dominates the gross negative effect of tax revenue from VAT, CST, and excise duty. 35 University of Ghana http://ugspace.ug.edu.gh Table 4.3 Coefficient Estimates for Individual Taxes versus Growth (Equation 3.4) Variable Estimate Std.Error z value Pr ( > |z|) GDPt-1 0.1563 0.2430 0.6433 0.5200 GDPt-1 0.1730 0.2405 0.7192 0.4720 Intercept 17.6905 0.3123 56.6511 0.0000 *** PAYE 0.0620 0.0379 1.6352 0.1020 self-employed 0.1314 0.0617 2.1303 0.0332 ** corporate 0.0195 0.0204 0.9536 0.3403 mineral royalties 0.0086 0.0179 0.4803 0.6310 VAT -0.0122 0.0859 -0.1420 0.8871 NHIL 0.1298 0.0874 1.4844 0.1377 Excise -0.0201 0.0217 -0.9262 0.3543 CST -0.0656 0.0477 -1.3746 0.1693 Capex -0.0052 0.0024 -2.1520 0.0314 ** Recex 0.0175 0.0254 0.6890 0.4908 Regression with ARIMA(2,0,0) errors Sigma^2 estimated as 0.0004251: log likelihood = 95.22 AIC = -166.44 AICc = -152.87 BIC = -147.43 ***, **, and * indicate 1%, 5%, and 10% levels of significance respectively. The foregoing observations collaborate findings of some extant literature (Ojede & Yamarik, 2012; Stoilova, 2017) and contradict findings of other studies (Azam & Shinwari, 2015; Benczúr et al., 2018). After studying 28 European economies, Stoilova (2017) concluded that personal income taxes positively impact GDP while consumption taxes like VAT negatively impact GDP. Although Ojede and Yamarik (2012) found that income taxes negatively impact 36 University of Ghana http://ugspace.ug.edu.gh growth in the USA, their finding that sales tax negatively impact growth is supported by Stoilova (2017). Results from other studies contradict the observations summarized in Table 4.3. For instance, Benczúr et al. (2018) in there study of the impact of taxes on growth in Hungary, found that VAT positively impact on employment and GDP. Likewise, Azam and Shinwari (2015) concluded that in Pakistan, sales taxes and excise duties positively affect GDP. 4.6 Model Diagnostics In the previous two sections, we presented and discussed the results of predicting equation 3.3 and equation 3.4. In this section, we examined the robustness of the model predictions. For each model, we examine the distribution and autocorrelation of forecast errors using various plots and numerical tests. 4.6.1 Residual Plot To evaluate the fitness of the two models, we first examined whether the plot of residuals overtime exhibit any discernible pattern or not. When the residuals exhibit a pattern, it is indicative of improper fitting. The plot of residuals shown in Figure 4.5 and 4. 6 suggests there is no discernible pattern in the distribution of the residuals overtime. The lack of pattern in the residuals indicate equation 3.3 and equation 3.4 are somewhat proper fit. Thus, the explanatory variables account for a greater proportion of variation in GDP over the research period. 37 University of Ghana http://ugspace.ug.edu.gh Figure 4.5: Plot of Forecast Errors for Equation 3.3 Figure 4.6: Plot of Forecast Errors for Equation 3.4 4.6.2 Distribution of Residuals To investigate whether the residuals follow a normally distribution, we overlaid a normal density plot on a histogram of the forecast errors for the two models. From Figure 4.7 and 4.8, the distribution of the forecast errors is centered on zero and assumes the bell shape of the normal density plot. 38 University of Ghana http://ugspace.ug.edu.gh Figure 4.7: Normal Density Plot Overlaid on Histogram of Forecast Errors for Equation 3.3 Figure 4.8: Normal Density Plot Overlaid on Histogram of Forecast Errors for Equation 3.4 A formal test of normality using the Shapiro-Wilk test of normality supports the foregoing observation that the distribution of the forecast errors are normally distributed. Specifically, the Shapiro-Wilk test statistic (W) for forecast errors from equation 3.3 is 0.975 and a p-value of 0.58 and the Shapiro-Wilk test statistic for forecast errors from equation 3.4 is 0.977 with p-value of 0.67. Since the p-values of for the two equations are greater than the 5% 39 University of Ghana http://ugspace.ug.edu.gh significance level, the distribution of the forecast errors for the two equations are not significantly different from normal distribution. 4.6.3 Correlogram of Forecast Errors From lag 1 – 10 Correlogram is a plot of correlations at different time lags. It is used to assess randomness in data (Friendly, 2012). When there is randomness in the data, the Correlogram is near zero. Figure 4.9 and 4.10 shows the correlograms for forecast errors for equation 3.3 and 3.4 for lags 1 – 10. From Figure 4.9, the sample autocorrelations of forecast errors from the first model falls within the significance bounds for lags 1 – 10, indicating that there is little evidence for non- zero autocorrelations of residuals from the first prediction at lags 1 – 10. The Ljung-Box test of autocorrelation in the next section confirms the foregoing observation. For the second prediction, the sample autocorrelations of forecast errors exceed the lower significance bound at lag 1.25 (see Figure 4.10). Apart from that the sample autocorrelations for other lags 1 – 10 remain within the significance bounds. The formal test of autocorrelation in the next section shows that the sample autocorrelation exceed the lower significance bound at lag 1.25 per chance, and that there is little evidence for non-zero autocorrelations of residuals at lags 1 – 10. 40 University of Ghana http://ugspace.ug.edu.gh Figure 4.9: Correlogram of Forecast Errors at Lags 1-10 for Equation 3.3 Figure 4.10: Correlogram of Forecast Errors at Lags 1-10 for Equation 3.4 4.6.4 Portmanteau Test (Ljung-Box Test) Portmanteau test provides a lot of flexibility in specifying the alternative hypothesis to a well- defined null hypothesis (Castle & Hendry, 2010). In time series analysis, there are two common portmanteau tests for assessing autocorrelations of variables. These are the Ljung- 41 University of Ghana http://ugspace.ug.edu.gh Box test (Ljung & Box, 1978) and the Box-Pierce test (G. E. P. Box & Pierce, 1970). Ljung- Box is an improvement on the Box-Pierce test (Castle & Hendry, 2010). In this study, we adopted the Ljung-Box test to examine whether or not there is evidence of autocorrelation in the forecasts errors for each prediction model. Evidence of non-zero autocorrelation suggests the prediction model is not properly fitted. For the tax structure and growth model, the test statistics (x-squared) for the Ljung-Box test for 20 lags is 11.328 with p-value of 0.939. For the individual taxes and growth model, the test x-squared for the Ljung-Box test for 20 lags is 20.889 with p-value of 0.403. Since the p- values for the Ljung-Box test for either prediction model is greater than the 5% significance level, there is little evidence of non-zero autocorrelations in the forecasts errors. In other words, both prediction models somewhat explains all the variations observed in the outcome variable. 4.7 Chapter Summary In this chapter, we presented and discussed the results and findings of the study. We started by examining some of the relevant characteristics of the dataset. Thus, we looked at measures of central tendency, variability, and degree of association among the variables. Next, we examined the coefficient estimates for equation 3.3, the tax structure and growth model as well as the coefficient estimates for equation 3.4, the individual taxes and growth model. Afterwards, we assessed the robustness of the model estimations using residual plots, test of normality, correlogram, and Ljung-Box test. From the descriptive statistics, we observed that over the research period, revenue from direct taxes is more than revenue from indirect taxes; direct taxes are more volatile than indirect taxes; tax revenue were on an upward trajectory; capital expenditure lags recurrent 42 University of Ghana http://ugspace.ug.edu.gh expenditure by about GHȻ 5.5 million every year. Thus, tax revenues are applied to finance recurrent expenditure more than capital expenditure. The results from the model predictions suggests that there exist a significant positive relationship between tax structure and growth. Particularly, we observed that one percentage increase in direct tax revenue increases real GDP significantly by 0.06% and one percentage increase in indirect tax revenue increases real GDP significantly by 0.11%. This finding collaborates Egbunike et al. (2018) finding that taxes generate growth at the aggregate level in Ghana. The results also showed that growth in two successive previous periods positively impact current growth significantly. To a large extent the results from the second model estimation support the foregoing observations. Thus, we observed that on net effect basis, specific taxes of direct nature positively impact real GDP and specific taxes of indirect nature positively impact real GDP. Particularly, the results show that a percentage increase in tax revenue from PAYE increases real GDP insignificantly by 0.06%; a percentage increase in tax revenue self-employed tax increases real GDP significantly by 0.13%; a percentage increase in tax revenue from corporate tax increases real GDP insignificantly by 0.02%; a percentage increase in tax revenue from mineral royalties increases real GDP insignificantly by 0.01%. On taxes with indirect nature, the results show that a percentage increase in tax revenue from VAT decreases real GDP insignificantly by 0.01%; a percentage increase in tax revenue from NHIL increases real GDP insignificantly by 0.13%; a percentage increase in tax revenue from CST decreases real GDP insignificantly by 0.07%; a percentage increase in tax revenue from excise duty decreases real GDP insignificantly by 0.005%. Lastly, the results give a fair indication of the mechanism through which taxes affect growth. Particularly, we observed that there is a significant positive relationship between recurrent 43 University of Ghana http://ugspace.ug.edu.gh expenditure and real GDP, but insignificant negative relationship between capital expenditure and real GDP. 44 University of Ghana http://ugspace.ug.edu.gh CHAPTER 5 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 5.1 Introduction This chapter summarizes the major research findings, highlights contributions of the research to extant literature, and makes recommendations for future research. The first section of the chapter summarizes the observed relation between tax structure and growth as well as the observed relation between individual taxes and growth. The next section highlights key contributions of the research to extant literature. Subsequent sections discuss policy recommendations and research limitations. 5.2 Summary of Findings Taxation is a major economic policy tool. It is also a means to promote positive social behaviour. Unfortunately, in Ghana taxation is almost reduced to a tool for winning political power. Majority of the time, political discourse and economic policy discourse on taxation is at best anecdotal. The purpose of this research is to investigate and provide empirical evidence on the relation between tax structure and growth in Ghana. First of all, we investigated the effects of direct taxes and indirect taxes on real GDP over the research period 2009-2017 and observed that both direct taxes and indirect taxes significantly impact real GDP in the positive direction. Specifically, we found that on average one percentage increase in direct tax revenue increases real GDP significantly by 0.06% and one percentage increase in indirect tax revenue increases real GDP significantly by 0.11%. 45 University of Ghana http://ugspace.ug.edu.gh We also examined the transmission mechanism through which taxes impact growth. From the research results, we observed that taxes turn to impact growth through recurrent expenditure more than capital expenditure. Particularly, we noted that on average, a one percent increase in recurrent expenditure increases real GDP significantly by 0.07% while the same percentage increase in capital expenditure decreases real GDP insignificantly by 0.0008%. Finally, we examined the granular relation between individual taxes and real GDP in order to obtain actionable insight for economic policy recommendations. We found that all individual taxes of direct nature positively impact real GDP. Conversely, we found that all individual taxes of indirect nature in exception of National Health Insurance Levy negatively impacts real GDP. 5.3 Contribution of the Study The relation between taxation and growth is a widely researched topic; however, results in the empirical literature is mixed. Some studies found positive relation between taxation and growth while other studies concluded on negative relation or insignificant association between taxation and growth. Huang and Frentz (2014), Shinohara (2014), and Stoilova (2017) attributed the mixed empirical results to differences in tax systems, social redistribution preferences, political administration, and economic conditions. This implies that there is need for more country-specific empirical evaluations examining the relation between taxation and growth, and this study is a response to that need. It examines the relation between tax structure and growth in Ghana. Moreover, this study expands on extant literature in the Ghanaian empirical corpus by looking at the relation between tax structure (direct and indirect taxes) and growth, instead of the relation between aggregate tax revenue and growth. Also, the study investigate the 46 University of Ghana http://ugspace.ug.edu.gh transmission mechanism through which taxes impact growth, which to the best of our knowledge has not been examined by any researcher. Finally, the study employs time series analysis to estimate the relation of interest unlike majority of extant literature in Ghanaian corpus which used multiple linear regression to estimate the relation of interest. The upside to using time series is that autocorrelations in the forecast errors due to past values of the predicted variable is minimized. 5.4 Conclusions Based on the results obtained over the research period, we make the following conclusions; i. There exits significant positive relation between tax structure (direct taxes and indirect taxes) and growth. ii. Indirect taxes impact growth more than direct taxes mainly as a result of National Health Insurance Levy. iii. Taxes impact growth significantly through recurrent expenditures more than through capital expenditure. 5.5 Policy Recommendations The importance of taxation to economic policy cannot be overemphasized enough and this study has done exactly that. The study establishes that even though both direct taxes and indirect taxes improves growth, a percentage increase in indirect taxes impacts growth more than direct taxes solely due to the positive effect of National Health Insurance Levy. In fact, the results suggests that other forms of indirect taxes like Value Added Tax, Excise Tax, and Communication Service Tax reduces growth. 47 University of Ghana http://ugspace.ug.edu.gh Based on the foregoing observation, we recommend that future taxation policy is more direct tax orientated than indirect tax oriented. The current tax structure is such that more than three-quarter of every one cedi of tax revenue comes from indirect taxes. This is highly expected because indirect taxes are easy to administer considering the highly informal nature of economic activities. Unfortunately, the evidence that indirect taxes in exception of the National Health Insurance Levy positively impact growth is weak. Hence, it is imperative that future tax policy aggravate more towards direct taxation than indirect taxation. 5.6 Limitations and Suggestions for Further Research First of all, the period of the study is somewhat unextended. As a result, we were unable to investigate the long-term versus short-term dynamics of the relationship between tax structure and growth. Possibly, the observed relationship that tax structure positively impacts growth may not hold in the long-term, but we cannot tell from the current study. Knowing the short-term relationship between tax structure and growth, as well as the long-term relationship between tax structure and growth would give policy makers a complete picture about the subject matter contributing to more effective policy considerations. Therefore, we suggest future researchers investigate a longer time period to examine the short-term and long-term dynamics of tax structure and growth. Secondly, the study does not address the transmission mechanism through which taxes impact growth comprehensively. The current study only tells us that taxes impact growth through recurrent expenditure more than capital expenditure. This raises the question of what nature of recurrent expenditure. We recommend that future studies explore other classifications of expenditure in order to comprehensively examine the transmission 48 University of Ghana http://ugspace.ug.edu.gh mechanism by which taxes impact growth. For instance, future researchers can categorize government expenditure into productive expenditure and unproductive expenditure. Lastly, the observations from the current study is not targeted enough. We suggest future researchers look at optimization models in order to find an optimal tax structure mix that achieve sustainable growth. For instance, the current study only tells us that there exists positive association between tax structure and growth, but it does not tell us what mix of tax structure achieve a certain growth target. Any future study that can develop an optimization model to estimate the optimal combinations of direct taxes and indirect taxes to attend a specified growth target would revolutionize the tax policy debate considerably. 5.7 Chapter Summary We started the chapter with a summary of the major research findings. We noted that there is a positive relationship between tax structure and growth over the research period. We proceeded to discuss the contribution of the study to extant literature. Afterwards, we made some tax policy suggestions based on the research observations that going forward, government should focus more on increasing direct taxes than indirect taxes. We concluded the chapter with a discussion of limitations of the research and offer suggestions for future studies. 49 University of Ghana http://ugspace.ug.edu.gh REFERENCES Acosta-Ormaechea, S., Sola, S., & Yoo, J. (2018). Tax Composition and Growth: A Broad Cross-country Perspective. German Economic Review. https://doi.org/10.1111/geer.12156 Ahmad, S., Sial, M. H., & Ahmad, N. (2016). Taxes and Economic Growth: an Empirical Analysis of Pakistan. European Journal of Business and Social Sciences, 5(02), 16–29. Retrieved from http://www.ejbss.com/recent.aspx-/ Alesina, A., & Ardagna, S. (2010). Large changes in fiscal policy: taxes versus spending. Tax Policy and the Economy, 24. Arnold, J. (2008). Do tax structures affect aggregate economic growth? (No. 643). Atems, B. (2015). Another look at tax policy and state economic growth: the long and short run of it. Economics Letters, 127(1), 64–67. Azam, M. F., & Shinwari, S. (2015). Empirical Analysis of Tax Revenues and Its Impact on Economic Growth of Pakistan, 6(1), 110–118. Barro, R. J. (1991). Economic Growth in a Cross Section of Countries". Quarterly Journal of Economics, 106, 407–444. Barro, R., & Redlick, C. J. (2011). Macroeconomic Effexts of Government Purchases and Taxes. Journal of Economics, 126, 51–102. Bekoe, W., Danquah, M., & Senahey, S. K. (2016). Tax reforms and revenue mobilization in Ghana. Journal of Economic Studies, 43(4), 522–534. https://doi.org/10.1108/JES-01- 2015-0007 Benczúr, P., Kátay, G., & Kiss, Á. (2018). Assessing the economic and social impact of tax 50 University of Ghana http://ugspace.ug.edu.gh and benefit reforms: A general-equilibrium microsimulation approach applied to Hungary. Economic Modelling, 75(April), 441–457. https://doi.org/10.1016/j.econmod.2018.06.016 Blanchard, O., & Perotti, R. (2002). An Empirical Characterization Of The Dynamic Effects of Changes in Government Spending and Taxes on Output. Quarterly Journal of Economics, 107, 1329–1368. Bleaney, M. F., & Kneller, R. (2001). Testing the endogenous growth model: Public expenditure, taxation, and growth over the long run. Canadian Journal of Economics, 34(1), 36–57. Bonga-bonga, L., & Ahiakpor, F. (2016). Assessing the Detreminants of Economic Growth in Ghana, 50(4). Box, G. E. P., & Pierce, D. A. (1970). Distribution of Residual Autocorrelations in Autoregressive-Integrated Moving Average Time Series Models. Journal of American Statistical Association, 65(332), 1509–1526. Box, G., & Jenkins, G. (1970). Time Series Analysis: Forecasting and Control. San Francisco: Holden-Day. Bujang, I., Hakim, T. A., & Ahmad, I. (2013). Tax Structure and Economic Indicators in Developing and High-income OECD Countries: Panel Cointegration Analysis. Procedia Economics and Finance, 7(13), 164–173. https://doi.org/10.1016/S2212- 5671(13)00231-1 Cassel, G. (1932). The Theory of Social Economy (1st ed.). New York: Harcourt Brace. Castle, J., & Hendry, D. (2010). A Low-Dimension Portmanteau Test for Non-linearity. Journal of Economics, 158(2), 231–245. 51 University of Ghana http://ugspace.ug.edu.gh Dladla, K., & Khobai, H. (2018). The impact of taxation on economic growth in South Africa (No. 86219). Munich Personal RePEc Archive. Retrieved from https://mpra.ub.uni- muenchen.de/86219/ Easterly, W., & Rebelo, S. (1993). Fiscal Policty and Economic Growth: An Empirical Investigation. Journal of Monetary Economics, 32, 417–458. Egbunike, F. C., Emudainohwo, O. B., & Gunardi, A. (2018a). Tax Revenue and Economic Growth : A Study of Nigeria and Ghana, 7(2), 213–220. Egbunike, F. C., Emudainohwo, O. B., & Gunardi, A. (2018b). Tax Revenue and Economic Growth: A Study of Nigeria and Ghana. Signifikan: Jurnal Ilmu Ekonomi, 7(2), 213– 220. https://doi.org/10.15408/sjie.v7i2.7341 Ferede, E., & Dahlby, B. (2012). The impact of tax cuts on economic growth: Evidences from the Canadian Provinces. National Tax Journal, 65, 563–594. Friendly, M. (2012). Corrrams: Exploratory displays for correlation matrices. The American Statistician, 56(4), 316–324. Gale, W. G., Krupkin, A., & Rueben, K. (2015). The relationship between taxes and growth at the state level: new evidence, 68(December), 919–942. Gemmell, N., Kneller, R., & Sant, I. (2011). The timing and persistence of fiscal policy impacts on growth: evidence from OECD countries. The Economic Journal, 121, 33– 558. Goode, R. (1984). Government finance in developing countries. Washington DC: Brookings Institute. Huang, B. C., & Frentz, N. (2014). What Really Is the Evidence on Taxes and Growth ? Washington DC. Retrieved from https://www.cbpp.org/sites/default/files/atoms/files/2- 52 University of Ghana http://ugspace.ug.edu.gh 18-14tax.pdf International Monetary Fund. (2010). Will it hurt? Macroeconomic effects of fiscal consolidation. World Economic Outlook: Recovery, Risk and Rebalancing. International Monetary Fund. (2012). Statistics on the Growth of the Global Gross Domestic Product (GDP) from 2003 to 2013. Retrieved from http://www.statista.com/statistics/197039/growth-of-the-global-gross-domestic-product- gdp/ Johnson, N. D., & Koyama, M. (2017). States and Economic Growth: Capacity and Constraints. Explorations in Economic History, 64, 1–20. Kalaš, B., Mirović, V., & Andrašić, J. (2017). Estimating the Impact of Taxes on the Economic Growth in the United States. Economic Themes, 55(4), 481–499. https://doi.org/10.1515/ethemes-2017-0027 Katz, C. J., Mahler, V. A., & Franz, M. G. (1983). The impact of taxes on growth and distribution in developed capitalist countries: a cross-national study. American Political Science Review, 77, 871–886. Kizito, E. U. (2014). The Nexus between Tax Structure and Economic Growth in Nigeria: A Prognosis. Journal of Economic and Social Studies, 4(1), 107–131. https://doi.org/10.14706/JECOSS11417 Kneller, R., Bleaney, M. F., & Gemmell, N. (1999). Fiscal policy and growth: Evidence from OECD countries. Journal of Public Economics, 74, 171–190. Koester, R., & Kormendi, R. (1989). Taxation, Aggregate Activity and Economic Growth: Cross-Country Evidence on Some Supply-Side Hypotheses. Economic Inquiry, 27, 367– 386. 53 University of Ghana http://ugspace.ug.edu.gh Kurz, H. D., & Salvadori, N. (2014). Theories of Economic Growth – Old and New, (May). Lee, Y., & Gordon, R. (2005). Tax Structure and Economic Growth. Journal of Public Economics, 89, 1027–1043. Ljung, G. M., & Box, G. E. P. (1978). On a Measure of a Lack of Fit in Time Series Models. Biometrika1, 65(2), 297–303. https://doi.org/10.1093/biomet/65.2.297 Lucas, R. E. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3–42. Mankiw, N. G., Romer, D., & Weil, D. (1992). A contribution to the Empirics of Economic Growth. Quarterly Journal of Economics, 107, 407–437. Mcbride, W. (2012). What Is the Evidence on Taxes and, 1945(207). Mendoza, E. G., Milesi-Ferretti, G. M., & Asea, P. (1997). Do taxes matter for long-run growth?: Harberger’s superneutrality conjecture. Journal of Public Economics, 66, 99– 126. Mendoza, E. G., Razin, A., & Tesar, L.-L. (1994). Effective Tax Rates in Macroeconomics: Cross-Country Estimates of Tax Rates on Factor Incomes and Consumption. Journal of Monetary Economics, 34(297–323). Mercedes, G. E., & Mehrez, G. (2004). The impact of government size and the composition of revenue and expenditure on growth. Austria. Mertens, K., & Ravn, M. (2012). The dynamic effects of personal and corporate income tax changes in the United States. American Economic Review. Miller, S., & Russek, F. (1997). Fiscal Structures and Economic Growth: International Evidence. Economic Inquiry, 35, 603–613. Retrieved from 54 University of Ghana http://ugspace.ug.edu.gh http://econpapaers.repec.org/paper/wpawuwpma/9309001.htm Myles, G. D. (2009a). Economic Growth and the Role of Taxation-Aggregate Data. OECD Economics Department Working Papers, (No 714), 64. https://doi.org/10.1787/222800633678 Myles, G. D. (2009b). Economic Growth and the Role of Taxation-Theory. OECD Economics Department Working Papers, (No 714), 64. https://doi.org/10.1787/222800633678 Ojede, A., & Yamarik, S. (2012). Tax policy and state economic growth: The long-run and short-run of it. Economics Letters, 116(2), 161–165. https://doi.org/10.1016/j.econlet.2012.02.023 Padovano, F., & Galli, E. E. (2001). Tax rates and economic growth in the OECD countries (1950-1990). Economic Inquiry, 29, 44–57. Peter, Ď., & Silvia, P. (2013). ARIMA vs . ARIMAX – which approach is better to analyze and forecast macroeconomic time series ? In International Conference of Mathematical Methods in Economics (pp. 136–140). Ricardo, D. (1951). On the Principles of Political Economy and Taxation. In P. Sraffa & M. H. Dobb (Eds.), The works and Correspondence of David Ricardo (3rd ed.). Cambridge: Cambridge University Press. Romer, C., & Romer, D. (2010). The macroeconomic effects of tax changes: estimates based on a new measure of fiscal shocks. American Economic Review2, 100, 763–801. Romer, P. (1986). Increasing Returns and Long-Run Growth. Journal of Political Economy, 94(5), 1002–1037. Romer, P. (1990). Human Capital and Growth: Theory and Evidence. Carnegie-Rochester 55 University of Ghana http://ugspace.ug.edu.gh Conference Series on Public Policy, 32, 251–286. Rosenberg, N. (1982). Inside the Black Box: Technology and Economics. New York: Cambridge University Press. Roubini, N., & Backus, D. (1998). Productivity and Growth. Lectures in Macroeconomics. Schumpeter, J. A. (1934). The Theory of Economic Development. Cambridge: Harvard University Press. Segerstrom, P. S., Anant, T. C. A., & Dinopoulos, E. (1990). A Schumperterian Model of the Product Life Cycle. American Economic Review, 80, 1077–1091. Shinohara, M. (2014). Tax Structure and Economic Growth Faculty of Economics Tax Structure and Economic Growth – A Survey of Empirical Analyzes – Faculty of Economics, (217). Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70(1), 65–94. Stoilova, D. (2017). Sistema fiscal y el crecimiento económico: evidencia de la Unión Europea. Contaduria y Administracion, 62(3), 1041–1057. https://doi.org/10.1016/j.cya.2017.04.006 Swan, T. W. (1956). Economics of Growth and Capital Accumulation. Economic Record, 32, 334–361. Wicksell, K. (1934). No Title. Lectures on Political Economy. Yahaya, K. A. (n.d.). EFFECT OF PETROLEUM PROFIT TAX AND COMPANIES INCOME TAX ON ECONOMIC GROWTH IN NIGERIA, (13), 100–122. Zhou, S., Shi, M., Li, N., & Yuan, Y. (2011). Impacts of Carbon Tax Policy on CO2 56 University of Ghana http://ugspace.ug.edu.gh Mitigation and Economic Growth in China. Advances in Climate Change Research, 2(3), 124–133. https://doi.org/10.3724/SP.J.1248.2011.00124 57