COLLEGE OF HUMANITIES ECONOMIC GROWTH & UNEMPLOYMENT RELATIONSHIP: EVIDENCE FROM SUB-SAHARAN AFRICA. BY COSMOS OSEI ATTA JUNIOR 10558798 THIS THESIS IS SUBMITTED TO UNIVERSITY OF GHANA, LEGON, IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF MPHIL IN ECONOMICS DEGREE. DEPARTMENT OF ECONOMICS JANUARY, 2023 University of Ghana http://ugspace.ug.edu.gh ii University of Ghana http://ugspace.ug.edu.gh iii ABSTRACT Examining the economic growth-unemployment nexus in Sub-Saharan Africa from 2000 to 2021 was the main objective of the study. The study used both static panel data estimation methodologies (Random effect and fixed effect models) to estimate a balanced panel data. In the study, variables including GDP—a widely used proxy for economic growth—the unemployment rate, FDI, trade openness, inflation, the manufacturing sector, the agricultural sector, and the service sector were all used. AGRIC, MANU, and SERV variables are stationary at first difference, while GDP, UNEMP, FDI, TRADE, and INF variables are stationary at levels. The Granger causality test results showed a unidirectional relationship between unemployment (UNEMP) and gross domestic product (RGDP), with the economic causation running from GDP to UNEMP. The study found out that there is a negative linkage between economic growth and unemployment thus an increase in agricultural production will reduce unemployment in Sub-Saharan Africa, assuming all other variables remain constant. The estimated impact of the service industry on unemployment as a share of GDP is also shown to be positive. There is an inverse link between FDI and unemployment, suggesting that an increase in FDI has an impact on unemployment development. The paper recommends that the government should invest in agricultural modernization and the acquisition of cutting-edge machinery in order to increase participation in agriculture among all segments of the population. High rates of tariff and non-tariff obstacles continue to be a problem for many African countries. Efforts to increase trade should also focus on eliminating these obstacles. Policymakers should devise measures that support and encourage rapid and sustainable economic growth to foster a growth rate that is beneficial to the economy. This study did not estimate other significant variables including labour market institutions (unemployment benefits, employment protection legislation, etc.) due to a lack of data for several SSA nations. University of Ghana http://ugspace.ug.edu.gh iv DEDICATION Every daunting challenge calls for one's own efforts in addition to the counsel of more experienced individuals who played a vital role towards the attainment of this feet. My heartfelt appreciation goes to my loving and supportive parents for their relentless prayers and motivation which has helped me climb this academic ladder and to achieve these great feet. Your unwavering support and care are gratefully acknowledged, and may our Heavenly Father lavish on you His unceasing blessings now and throughout eternity. University of Ghana http://ugspace.ug.edu.gh v ACKNOWLEDGEMENT My sincere appreciation goes out to both Dr. Prince Adjei and Dr. Alfred Barimah, my supervisor and co-supervisor, respectively. Their advice, help, and encouragement are invaluable. I also appreciate and recognize the help and encouragement I received from loved ones and friends during my academic pursuits. Most importantly, I want to thank God, our Heavenly Father, for His providential grace and strength to carry out this study. University of Ghana http://ugspace.ug.edu.gh vi TABLE OF CONTENT DECLARATION ........................................................................................................................ ii ABSTRACT ............................................................................................................................... iii DEDICATION ........................................................................................................................... iv ACKNOWLEDGEMENT .......................................................................................................... v LIST OF FIGURES……………………………...…………………………………………viii LIST OF TABLES................................................................................................................... x LIST OF ABBREVIATIONS AND ACRONYMS ............................................................... xi CHAPTER ONE ........................................................................................................................... 1 INTRODUCTION......................................................................................................................... 1 1.1 Background of the study ....................................................................................................... 1 1.2 Problem Statement ................................................................................................................ 4 1.3 Research Objectives .............................................................................................................. 6 1.4 Research Questions ............................................................................................................... 6 1.5 Significance of the study ....................................................................................................... 7 1.6 Structure of the study ............................................................................................................ 8 CHAPTER TWO .......................................................................................................................... 9 REVIEW OF LITERATURE ...................................................................................................... 9 2.0 Introduction ........................................................................................................................... 9 2.1 Empirical Review .................................................................................................................. 9 2.2 Theoretical Review ............................................................................................................. 12 2.3 Unemployment theories ...................................................................................................... 13 2.4 Economic Growth theories .................................................................................................. 14 CHAPTER THREE .................................................................................................................... 16 OVERVIEW OF ECONOMIC GROWTH AND UNEMPLOYMENT IN SUB-SAHARAN AFRICA ....................................................................................................................................... 16 3.0 Introduction ......................................................................................................................... 16 3.02 Unemployment in Sub-Saharan Africa.......................................................................... 17 3.1 Economic Growth-Unemployment in Sub-Saharan Africa. ............................................... 20 University of Ghana http://ugspace.ug.edu.gh vii 3.11 Economic Growth and Unemployment Drivers. ............................................................... 21 3.2 Trend and features of unemployment in Sub-Saharan Africa. ............................................ 22 3.3 Nature of economic growth in Sub-Saharan Africa ............................................................ 25 3.4 The linkage between economic growth and unemployment ............................................... 26 CHAPTER FOUR ....................................................................................................................... 29 METHODOLOGY AND DATA ............................................................................................... 29 4.0 Introduction ......................................................................................................................... 29 4.1 Model Specification ............................................................................................................ 29 4.4 Source of Data and Variables Description .......................................................................... 31 4.2 Panel data Estimation Techniques....................................................................................... 34 Test for Over-identifying Restrictions ................................................................................... 36 Unit root test .......................................................................................................................... 36 Visual Inspection ................................................................................................................... 37 Augmented Dickey Fuller (ADF) Test .................................................................................. 37 Phillips-Perron ....................................................................................................................... 38 The choice of lag length ........................................................................................................ 39 Johansen cointegration test .................................................................................................... 40 Vector Error Correction Model ............................................................................................. 41 VECM granger causality ....................................................................................................... 43 Diagnostic test ....................................................................................................................... 43 CHAPTER FIVE ........................................................................................................................ 48 PRESENTATION AND DISCUSSION OF RESULTS .......................................................... 48 5.0 Introduction ......................................................................................................................... 48 5.1 Descriptive Analysis ........................................................................................................... 48 5.5. Presentation of Empirical Regression Results ................................................................... 53 CHAPTER SIX ........................................................................................................................... 60 SUMMARY, CONCLUSION AND POLICY RECOMMENDATION ................................ 60 6.1 Introduction ......................................................................................................................... 60 6.2 Summary and Conclusion ................................................................................................... 60 University of Ghana http://ugspace.ug.edu.gh viii 6.3 Policy Recommendations .................................................................................................... 62 6.4 Limitations of the Study ...................................................................................................... 63 REFERENCES ............................................................................................................................ 64 APENDICES ............................................................................................................................ 68 University of Ghana http://ugspace.ug.edu.gh ix LIST OF FIGURES Figure 3.1: Economic Growth and Unemployment in Sub-Saharan Africa. .............................................. 20 Figure 3.2: Unemployment Trend in Sub-Saharan Africa. ......................................................................... 22 Figure 3.3: Economic Growth Trend in Sub-Saharan Africa ..................................................................... 25 University of Ghana http://ugspace.ug.edu.gh file:///C:/Users/MYF/Downloads/0-d186-44d7-8aac-49f09d7667e6_Cosmos_Mphil_Thesis-Main_Work.docx%23_Toc125065637 x LIST OF TABLES Table 5.1: Summary Statistics…………………………………………………………………...56 Table 5.5: Pooled OLS Regression………………………………………………………………61 Table 5.5.1: Fixed and Random Effects Estimation………………………………………….….62 Table 5.6: Hausman Test Results………………………………………………………………...63 University of Ghana http://ugspace.ug.edu.gh xi LIST OF ABBREVIATIONS AND ACRONYMS ILO International Labour Organization GDP Gross Domestic Product SSA Sub-Saharan Africa ECOWAS Economic Community of West African States VAR Vector Autoregressive VECM Vector Error Correction Model FAO Food and Agricultural Organization WESO World Employment and Social Outlook ADF Augmented Dickey-Fuller AIC Aikaike's Information Criterion PV Probability Value SIC Schward Information Criterion OLS Ordinary Least Squares UN United Nation HQ Hannann-Quin Criterion UNEMP Unemployment FDI Foreign Direct Investment INF Inflation University of Ghana http://ugspace.ug.edu.gh xii AGRIC Agriculture MANU Manufacturing SERV Service FE Fixed Effects RE Random Effects University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE INTRODUCTION 1.1 Background of the study Huge economic and financial problems currently facing the world include, among others, the issue of unemployment and unsustainable economic growth. (Murat, 2014). The reduction of unemployment and attainment of rapid economic growth are the most important economic objectives for both developed and developing countries. In order to get an accurate view of a nation's economic development, policymakers and the general public regularly monitor economic growth and unemployment, two essential macroeconomic indicators. One of the most prominent relationships in macroeconomic theory has been found to hold true for a variety of nations and regions, mostly in advanced economies rely on these indicators as essential components of their economic policy. (Lee, 2000; Farsio and Quade, 2003; Christopoulos, 2004). One of the primary questions that economic research aims to investigate is the underlying causes of economic growth. Economic growth is characterized as an increase in real GDP, GDP per capita, or the national product as measured in constant prices, according to Denison (1962). A measure of a country's welfare is economic growth, which quantifies the increase in the volume of goods and services produced there during a specific time period. Carree and Thurik (2000) affirmed that productivity, income distribution, and unemployment are the three aspects of an economy that matter most. According to the International Labour Organization (ILO, 2019), a person is considered unemployed if they are currently without a job, have the necessary skills to obtain a work, and are actively seeking employment. One of the most effective methods for obtaining high economic growth has been to reduce unemployment. This device is especially helpful for University of Ghana http://ugspace.ug.edu.gh 2 developing nations. As a result, governments can use unemployment as a key macroeconomic tool to shape the way their economy’s function. Unemployment is a multifaceted concept because it impacts both a country's economic activities and the social structure of society. As a result, these two dimensions add complexity to the problems faced by most SSA countries and necessitate substantial analysis to solve it since every policymaker's primary goal, whether fiscal or monetary, is to achieve strong economic growth. Many factors affect the stability of a country's growth rate and notable them is a high unemployment rate. Thus, Okun's Law serves as a theoretical statement that connects economic growth with unemployment which means that if unemployment declines by 1%, GDP also increases by 3%. The primary goal of economic policy is to promote strong economic growth, which leads to increased job demand through investment programs. Therefore, unemployment is a global issue with negative economic and social effects (Al-Habeas et al. 2012). Okun (1962) empirically showed and confirmed the relationship between growth and unemployment, and it is today recognized as one of the fundamental macroeconomic theories. It establishes the inverse relationship between economic growth and unemployment, which, for the US economy, means that a 1% decline in real production compared to its potential value results in a 0.5% increase in unemployment (Mankiw 2009). However, different applications of Okun's law have recently surfaced over the globe. In eight Eastern European countries, Soylu et al. (2018) discovered that a 1% increase in GDP growth rate led to a 0.08 percent decrease in the unemployment rate, while Abdul-Khaliq et al. (2014) discovered a 0.16 percent decrease in the unemployment rate in nine Arab countries. These results showed that, despite global geographical variations in the Okun's coefficient, literature consistently demonstrates a causally adverse relationship between GDP growth rate and University of Ghana http://ugspace.ug.edu.gh 3 unemployment rate. Furthermore, although Okun's coefficients vary somewhat globally, this rule is still applicable in the advanced world economy, according to later well-regarded research (Ball et al. 2019). As a result, Okun's law's rule of thumbs can be used as a benchmark for monitoring the development of long-term economic growth. Moreover, there is no clear-cut answer to the question of how economic growth and unemployment are related. Numerous findings were reached by researchers after examining unemployment and economic growth. It's been a highly contentious subject from both a theoretical and empirical standpoint. According to theories and earlier studies on the subject, there may be no relationship (Fuad, 2011), a negative relationship (Madito and Khumalo, 2014), or a positive relationship between unemployment and economic growth (Mahmoud, 2012). Once more, there has been debate over whether unemployment can forecast economic growth and vice versa. Granger causality determines whether one indication has any potential predictive power for the other. Unemployment, for instance, granger induces economic growth, which implies that unemployment holds information about future economic growth. Few studies and other pertinent publications also show the divergent viewpoints in the literature on how economic growth affects unemployment. Concerns today extend beyond the straightforward relationship to include the unemployment rate, which could have an effect on economic growth. Given the contentious issues surrounding economic growth and unemployment, the purpose of this paper is to examine the economic growth and unemployment relationships in Sub-Saharan African countries. University of Ghana http://ugspace.ug.edu.gh 4 1.2 Problem Statement The majority of countries desire rapid long-term economic growth. However, accomplishing such a goal has proven challenging due to a number of factors that affect economic growth. In addition to other factors, unemployment is one that might be viewed as a significant hindrance to economic growth, hence macroeconomic policy is extremely focused on these two relationships. (Barro, 1998). When the economy is in equilibrium, factors that support economic growth dynamics in classical economic growth models have no impact on the unemployment rate (the assumption of the full use of the production factors). The factors that affect the unemployment rate, on the other hand, have little consequence on economic growth. The only thing that can cause a short-term negative relationship between the variables is economic instability, especially volatility in the labour market (Gruchelski 2012). Despite abundant natural and human resources, Sub-Saharan Africa remains one of the world's poorest regions, due to a variety of factors including unpredictable regulatory systems and inadequate infrastructure. Unemployment remains a problem in many of the nations in the area and some SSA governments have made little or no effort to addressing the issue. Once more, a number of macroeconomic factors have been found to affect unemployment generally in the research. However, a more recent theory in the theoretical and empirical literature suggests that economic growth can affect the unemployment rate in a country. It is important to note that economic growth and unemployment are familiar topics in the areas of finance and economics. As seen by the conflicting results of earlier empirical research on the factors determining unemployment and economic growth, which is still an open question, the relationships between economic growth and unemployment have not yet been resolved or University of Ghana http://ugspace.ug.edu.gh 5 effectively addressed in the literature. For instance –, Gaber (2018), Maqbool et al (2013), Folawewo and Adeboje (2017), Riaz and Zafar (2018), Baah-Boateng (2014), Baah-Boateng (2016), and Ebaidalla (2016) found that economic growth reduced unemployment, whereas Kerckhoffs et al (1994) and Alrayes and Wadi (2018) found no clear relationship. Previous work has been inconclusive on the direction of causality, whether it is bi-directional or unidirectional (either from unemployment to economic growth or from economic growth to unemployment), or whether there is no relationship between unemployment and economic growth. Such ambiguity on the topic inspired the current study to contribute to the ongoing discussion economic growth and unemployment nexus in an African context. Also, despite the fact that Africa has high unemployment rates that translate into economic growth and faces similar unemployment issues, most empirical research has avoided looking at the economic growth and unemployment relationships across the continent in general. Single country studies are the foundation of the large majority of empirical research on the causes of unemployment and economic growth in Africa. (Kyei and Gyekye. 2011; Raifu. 2017; Khumalo and Eita. 2015; Baah- Boateng; 2014; Eita and Ashipala. 2010; Baah-Boateng. 2016; Dagume and Gyekye. 2016; Batu. 2016; Fila et al. 2016). Ebaidalla (2016) and Folawewo and Adeboje (2017) undertook some empirical study about the causes of unemployment in the Economic Community of West African States (ECOWAS). Nonetheless, these empirical studies were insufficiently representative of Africa and contained the following methodological flaws: they neglected the fact that unemployment data, and they failed to account for the endogeneity issue and thus this research helps to bridge these gaps. Finally, rather than simply lumping all sectors together in the form of GDP and studying its impact on unemployment, this study seeks to conduct a disaggregated sectorial analysis on growth to University of Ghana http://ugspace.ug.edu.gh 6 determine how the components (sectors of the economy) of economic growth (GDP) affect unemployment rates in Sub-Saharan Africa. As a result, this study investigates how value additions in the agricultural, service, and industrial sectors affect the unemployment rate in Sub-Saharan Africa. 1.3 Research Objectives Investigating the relationship between economic growth and unemployment in Sub-Saharan African economies is the main objective of this study. It uses a disaggregated sectorial approach to investigate the impact of sectorial growth on unemployment in Sub-Saharan Africa. However, to help comprehensively analyze these relationships, the following specific objectives are taken into consideration; 1. To explore the causal relationship between unemployment and economic growth in Sub-Saharan Africa. 2. To examine the impact of growth rate of agriculture, manufacturing and services sectors on unemployment rate in Sub-Saharan Africa. 3. To analyze the impact of economic growth on unemployment in Sub-Saharan Africa. 1.4 Research Questions To help achieve both the specific and general aims of the study, the following research questions are put forward; 1. What is the link between economic growth and unemployment in Sub-Saharan Africa? 2. What is the impact of growth rate of agriculture, manufacturing and services sectors on unemployment rate in Sub-Saharan Africa? 3. What is the impact of economic growth on unemployment in Sub-Saharan Africa? University of Ghana http://ugspace.ug.edu.gh 7 1.5 Significance of the study This study is significant because it seeks to uncover a wealth of evidence on the impact of economic growth on unemployment in Sub-Saharan Africa. Few research has been conducted on the link between economic growth and unemployment, and the results have been inconsistent. Some have found a negative relationship, while others have found a positive one, and only a few have found none. As a result of its empirical examination of SSA, this study will provide more insight into the issue and analyze whether economic growth has a significant impact on unemployment. The region has the fastest growing population, which is accompanied by rising unemployment rates in most countries. This study is important for SSA because it will inform countries about one of the major factors affecting unemployment, allowing them to seek solutions. Governments in SSA countries will be able to implement smart, practical policies that will allow the region to reap the full benefits of economic progress. Therefore, policymakers, such as various government agencies, will gain from this study in order to create long-term policies and recommendations to combat the issue of unemployment. Furthermore, the study will provide crucial insights into accomplishing the macroeconomic goal of African economies by examining the relationship between unemployment and economic growth. University of Ghana http://ugspace.ug.edu.gh 8 1.6 Structure of the study This study has been organized into six separate chapters as follows: Chapter one introduces the study by covering aspects like the study’s background, the problem being researched, questions that the study seeks to answer, objectives to be achieved at the end of the study as well as the importance of the study. The second chapter contains a review of the literature, which is divided into theoretical and empirical sections. It contains relevant and available literature on the topic investigation from other authors. Theories are also investigated and explained in this chapter. The third chapter offers an overview of the research in general. It expresses both the framework of unemployment and economic growth. This chapter discusses the links between economic growth and unemployment, as well as their respective determinants. The fourth chapter examines the study's research methodology. It considers the theoretical foundation of the methodology used. This chapter also captures the study's data and describes the data sources. The fifth chapter focuses on empirical findings, analysis, and results of the generated study based on the research questions and objectives. Chapter six summarizes and concludes the study while also offering policy recommendations and limitations. University of Ghana http://ugspace.ug.edu.gh 9 CHAPTER TWO REVIEW OF LITERATURE 2.0 Introduction For decades, the majority of theoretical and empirical research has focused on unemployment and general economic growth. However, research on the relationship between economic growth and unemployment appears to be scarce, particularly in Africa. Furthermore, most research on economic growth and other macroeconomic indicators use aggregated economic growth, which obscures the disparities in the effects of different sectors of the economy. This study will contribute to current research on economic growth and unemployment by revealing the full effects of various types of economic growth on unemployment in Sub-Saharan Africa using disaggregated forms of sectorial growth. 2.1 Empirical Review According to Andrei et al. (2009), in order to achieve long-term improvement in living standards, policymakers must examine the link between real GDP growth and unemployment. Mohr and Fourie (2008) looked into some of the primary factors that contribute to long-term economic growth. They claimed that these factors can be split into two main groups: supply factors (which result in an increase in production capacity) and demand factors (which refer to a sufficient and increasing demand for the produced goods and services), both of which are necessary for long- term economic growth. Fuad (2011) investigates the relationship between unemployment and economic growth in Jordan using Okun's law. Using annual data from 1970 to 2008, time series techniques are used to assess the link between unemployment and economic growth and to determine Okun's coefficient. The study employed Augmented Dickey-Fuller regression for unit root, cointegration test, and a simple University of Ghana http://ugspace.ug.edu.gh 10 regression between unemployment rate and economic growth (ADF). Okun's law does not applicable to Jordan, according to the empirical evidence. As a result, Jordan's unemployment problem cannot be attributed to a lack of economic growth. Mahmoud (2012) used Okun's law to examine the relationship between economic growth and unemployment in Arab nations. He found that while the two variables do have a positive relationship, high rates of economic growth and a decline in the unemployment rate do not necessarily indicate a strong correlation between GDP-growth and unemployment. Using the Hodrick-Prescott filter1 and the least square approach, Muhammad (2013) found that a one percent increase in unemployment was followed by a 0.3 percent decline in real GDP growth, demonstrating a substantially weaker link between the two variables, particularly in emerging nations. Ozel (2013) holds similar views to Mahmoud (2012), but this study was based on data collected from seven industrialized countries (G7) and used to examine the relationship between unemployment and GDP growth before and after the recession. Prior to the crises, productivity and GDP growth had a significant impact on unemployment reduction, according to the research. However, after the crises, the effects were insignificant because GDP expansion eliminated unemployment. Further research by Rosoiu (2014) investigating the link between GDP growth and unemployment in the United States from 1977 to 2011 indicated a substantial connection between the two variables. The research also indicated that the relationship between the two factors was significantly impacted by the financial crisis. However, Cashell (2009) emphasizes that there is a direct correlation between changes in the unemployment rate and the rate of economic growth. In order to evaluate the relationship between macroeconomic factors and unemployment, Dogan (2012) used the Vector Autoregressive (VAR) approach. The study's findings mainly addressed University of Ghana http://ugspace.ug.edu.gh 11 the effects of GDP growth and inflation on unemployment rates. The data showed that the variables' relationships adhered to both the Phillips curve theory and Okun's law. According to the VAR analysis's findings, unemployment and inflation have a positive link, while GDP growth and unemployment have a negative relationship. The strong positive coefficients of unemployment, labour, and capital revealed a long-run positive link with economic growth, according to Hussain (2012), who utilized a Vector Error Correction Model (VECM) to analyze the link between economic growth and unemployment. Contrary to the positive association between economic growth and employment, there is a correlation between economic growth and unemployment. In other words, it is expected that higher GDP or economic growth results in lower unemployment rates. Increased GDP growth results in higher employment levels as opposed to unemployment, indicating a positive correlation between GDP and employment. However, in reality, assuming a direct correlation between economic growth and unemployment is not always accurate because it is not always true that employment will increase solely if economic growth rates exceed productivity increases (potential output). Or to put it another way, GDP growth must exceed the growth rates of the labour force and productivity for the unemployment rate to decline over the long term (Levine, 2013). Due to the dynamic interrelationship between the variables used to gauge how quickly the economy was adapting to the unemployment crisis, Madito and Khumalo (2014) used the Error Correction Mechanism to study the relationship between unemployment and economic growth in South Africa from 1971Q1 to 2013Q4. Approximately 62 percent of economic growth is adjusted every quarter. Overall, the results showed that economic development and unemployment are negatively correlated in South Africa. University of Ghana http://ugspace.ug.edu.gh 12 Kemi and Dayo explore unemployment and economic growth in Nigeria (2014). The goal of this research is to determine the legality of Okun's law in Nigeria. The Error Correction Model (ECM) and Johansen cointegration test were used to determine both the short and long run relationships among the variables included in the study. According to empirical research, the unemployment rate and economic growth in Nigeria have a short- and long-term link. Yilmaz (2014) investigated the effects of general economic growth on unemployment in Turkey from the first quarter of 2010 to the third quarter of 2013. The model's variables include the unemployment rate, real GDP growth rate, real export growth rate, and real FDI inflows growth rate. In the inquiry, the bound testing technique based on the autoregressive distributed lag model was used. The empirical data in Turkey reveal a negative relationship between unemployment and economic growth. Ogbanga (2018) examined at the expansion of Nigeria's agricultural industry and its effects on employment creation. The study's variables included the employment rate, gross domestic product, foreign private investment, federal government spending, and industrial sector production. Using error correction and Granger Causality techniques, the impact of the agricultural sector on the creation of jobs was evaluated together with other explanatory variables. The main conclusion of the study is that the expansion of the agricultural and industrial sectors has a positive impact on employment creation. 2.2 Theoretical Review Numerous research have been conducted on the correlation between economic growth and other important macroeconomic variables in various nations. These studies produce a range of results and conclusions because of variations in the methodology and data employed. Limited but varied study has been done for years on developed, emerging, and developing nations to determine the University of Ghana http://ugspace.ug.edu.gh 13 impacts that economic growth has on unemployment and vice versa. The Okun's Law is primarily discussed in theoretical reviews. Therefore, the purpose of this study is to investigate theories and ideas about both economic growth and unemployment as a tool for fiscal policy. Prior to that, the study investigates unemployment and economic growth as a vehicle for fiscal policy in distinct and succinct reviews. 2.3 Unemployment theories The concepts of economic growth and unemployment are the most important factors that influence how all economies select and put into practise economic policies. Many theories attempt to explain the connection between unemployment and economic growth. Only one of the theories is, however, covered in this article. The main goal of Okun's Law was to clarify how unemployment and economic growth interacted in a particular economy. The idea is that, in any economy, unemployment has a bad impact on economic growth. The study found a correlation between a 3% drop in the unemployment rate and a 3% increase in economic growth. When Kwami (2015) evaluated the validity of Okun's theory of unemployment, he found that it did indeed suggest the existence of a negative relationship between unemployment and economic growth. The Keynesian unemployment theory, also referred to as the cyclical or deficient demand theory of unemployment, contends that insufficient demand in an economy is a major factor in unemployment, which occurs when people who are willing to work at the going rate of pay are unable to find employment at a particular time [Index Mundi 2017; FAO, 2019]. This theory contends that as demand for goods and services declines, so will output, resulting in a reduction in the number of workers needed for production. Even if full employment is reached, some employees will continue to be unemployed because there is an economic mismatch between the number of unemployed workers and the number of open positions. This theory contends that University of Ghana http://ugspace.ug.edu.gh 14 greater government spending is necessary for an economy to increase employment, increase aggregate demand, and reduce unemployment. Another popular explanation that explains unemployment is the Marxist theory, which was put forth in 1863 by the Marxist school headed by Karl Marx. The argument holds that because of the capitalism system's insatiable nature, unemployment is a given in any economy. Capitalists control the labour market by creating unemployment, which lowers the demand for labour and lowers wages. According to the idea, overthrowing capitalism and implementing a socialist economic system is the best way to minimize unemployment. 2.4 Economic Growth theories One of the theories that explain how an economy expands is the classical theory of economic growth. For instance, Adam Smith claimed in his classic work "An Inquiry into the Nature and Causes of the Wealth of Nations" that capital accumulation and labour productivity determine a country's wealth. This theory is in sharp contrast to Keynesian economic theory because it downplays the significance of the role that governments play in economies. According to Smith (1937), in order for economies to experience faster rates of growth, they must amass more capital, which is necessary for specialization and the division of labour. This model also emphasizes how crucial labour specialisation and division are to a nation's economic development. Another well-known theory is Solow's [1956] neoclassical growth model. This model demonstrates how an economy's capacity for capital accumulation affects its ability to grow. According to the model, labour and technology growth are determined exogenously. The model asserts that labour grows at a constant rate to support this assertion. As a result, the model contends that key variables influencing an economy's growth rate include population growth and the savings rate (growth rate per capita income). Countries with higher savings rates and lower capital costs University of Ghana http://ugspace.ug.edu.gh 15 are predicted to experience faster growth in per capita income because investments and savings add up to capital in an economy. The other growth model considered in this study is the endogenous growth model presented by Romer in 1986. This model differs from the neoclassical model in that it views technological advancement as an endogenous outcome, whereas the neoclassical model views it as an exogenous outcome. The endogenous growth model explains an economy's technological progress well by the country's knowledge acquisition. As a result, the basic idea of this model is that economic growth is powered by technological improvement, which is fueled by knowledge accumulation. Indeed, this model predicts that countries that spend a high proportion of their GDP on education will be more productive than countries that spend a low proportion of their GDP on education. As a result, this model emphasizes the critical significance of human capital investment in economic growth. Other growth theories include Keynesian growth theory, proposed by John Meynard Keynes in his important book "The General Theory of Employment, Interest, and Money" in 1936. The idea gained prominence after the classical school failed to explain the Great Depression in 1929. This model advocates for government intervention in the economy to enhance government spending, which leads to improved economic growth. University of Ghana http://ugspace.ug.edu.gh 16 CHAPTER THREE OVERVIEW OF ECONOMIC GROWTH AND UNEMPLOYMENT IN SUB-SAHARAN AFRICA 3.0 Introduction This chapter examines Sub-Saharan Africa's economic growth and unemployment. 3.01 Definition of Concepts Economic growth Gross domestic product (GDP) was employed as a proxy for economic growth in this analysis. According to Krugman (2000), GDP is the monetary value of all final goods and services produced inside a country's borders within a specified time period. Both real and nominal GDP can be used as indicators of economic growth. The rate of economic growth can be determined using both real and nominal measures of GDP. Real GDP refers to the quantity of economic output produced during a specific time period after adjusting for changes in the general price level. The nominal GDP is the total market value of a country's economic output produced inside its boundaries in a given year (Swan, 1956). Unemployment Unemployment is a phenomenon of job searching that occurs from joblessness. A person is deemed unemployed if they have attained the minimum age for employment, such as 15 years old, and were "without work," "available for work," and "actively seeking work" during a reference week (ILO, 1982). According to this definition, a jobless person who is available for work but for different reasons chooses not to look for a job cannot be categorized as being unemployed; instead, they are simply considered to be discouraged workers. Additionally, someone who is working full- time and chooses to look for (extra) work with another company is only moonlighting Baah- University of Ghana http://ugspace.ug.edu.gh 17 Boateng et al. (2013). A person is considered to be unemployed when they are actively seeking employment but are unsuccessful in doing so (Bean & Pissarides, 1993). This suggests that the unemployment rate is the percentage of working-age people over the age of 16 who lost their jobs or were unable to find work in the previous month and are now actively looking for work (Moosa, 2008). 3.02 Unemployment in Sub-Saharan Africa In 2019, there were over 34 million unemployed people in Africa. 12.2 million of them were young people between the ages of 15 and 24. (Figure 3.1). The number of young people without jobs increased by over 1.5 million, or 6.4 million, in 2010. Indicating that unemployment is a major labour market concern in Africa, the regional unemployment rate of 6.8% was much higher than the global average of 5.0%. Despite the continent of Africa having a low unemployment rate, most jobs there are informal. With approximately 34% of the labour force unemployed, South Africa has the highest unemployment rate in all of Africa in 2021. Djibouti and Eswatini were next, with unemployment rates of almost 28% and 26%, respectively. The lowest unemployment rates in Africa were found in Niger and Benin. The average for the continent that year was close to 8%. Due to a multitude of educational, socio-demographic, and economic factors, young people are more likely than older people to experience unemployment in the majority of countries around the world. In Africa, the average young unemployment rate in 2022 was 13%. The situation was especially bad in several countries. Some of the notable forms of unemployment in Sub-Saharan Africa are enumerated below: University of Ghana http://ugspace.ug.edu.gh 18 Seasonal unemployment When people experience seasonal unemployment, it's because they work in fields where their services aren't required all year round. According to Njoku and Inugba (2011), weather fluctuations, shifts in consumer preferences, or the enduring nature of such industries are the causes of seasonal variations in the operations of various industries. For instance, farm workers in Vineyards in the Western Cape are categorized as seasonal workers in South Africa's agricultural industry. They often experience high demand during the harvest and low demand throughout the off-seasons (Banda, et al., 2016). Frictional unemployment This type of unemployment occurs when employees leave their jobs but have not yet found new ones. The majority of workers leave on their own, either because they have to move or because they have enough money saved up to seek for decent opportunities (Sweezy, 1934). As parents are ready to enter the workforce or as students search for their first job, frictional unemployment can also happen. Additionally, it happens when workers are fired or, in some situations, laid off as a result of business-specific events like factory closures. Frictional unemployment is a frequent aspect of the employment search. Frictional unemployment does indeed help the economy by enabling workers to migrate to more fruitful positions (Chatterjee, 1995). Due to a lack of communication services like smartphones, the internet, and employment centers, this form of unemployment is prevalent in Sub-Saharan Africa, where unemployed unskilled laborers move from one location to another (Mafiri, 2002). University of Ghana http://ugspace.ug.edu.gh 19 Structural unemployment This type of unemployment happens when an industry or a nation's economic activities go through structural change, and it is rather widespread in Sub-Saharan Africa (Njoku & Inugba, 2011). Unexpected technical developments, inflation, recession, and changes in taste and preference are additional factors that contribute to rising unemployment rates. The South African economy had a period of rapid technical development, which increased the capital-intensiveness of many firms and led to structural unemployment because human labor was no longer necessary in many fields, according to Smit et al. (2006). According to Yager (2010), structural unemployment happens when economic developments lead to a mismatch between people's skills and what businesses need in terms of talent. Workers' abilities may get stale if they are unemployed for an extended period of time. If they are reluctant or unable to accept entry-level or unskilled jobs, they may remain unemployed for a long time after the economy has recovered (Pissarides, 1989). Natural unemployment rates rise in this situation as a result of structural unemployment. Cyclical unemployment During the business cycle, cyclical tendencies in growth and production are tied to cyclical variations in unemployment. Cyclical unemployment is a result of the business cycle's contraction period. When this occurs, demand for goods and services drops significantly, causing companies to lay off employees in an effort to reduce costs (Barro & Sal-Martin, 1995). Cyclical unemployment typically results in more unemployment due to the fact that fired employees have less money to spend on necessities which further diminishes demand (Sweezy, 1934). Because overall economic productivity is greatest during business cycle peaks, cyclical unemployment will be low. According to Mafiri (2002), cyclical unemployment in South Africa has a trait that makes it difficult to treat adequately. On top of that, there is a risk of widespread structural University of Ghana http://ugspace.ug.edu.gh 20 unemployment. As a result, the unemployment problem has gotten significantly worse, more complex, and difficult to resolve. 3.1 Economic Growth-Unemployment in Sub-Saharan Africa. All African states that are largely or completely south of the Saharan Desert are classified as Sub- Saharan Africa (SSA) (United Nations, 2011). The SSA population is currently projected to be 936.1 million people, with forty-eight nations making up the region (World Bank, 2018). According to the International Labour Organization (ILO) forecast for 2017, there will be over 201 million unemployed people globally in 2017, with an extra 2.7 million projected in 2018. According to the report, third-world countries, particularly Africa, will be the hardest hit, with high unemployment and poverty rates. The issues of high unemployment rates and slow output growth are not unique to emerging countries; but industrialized countries have employed strong economic and political policies to reduce unemployment over time. There is empirical support for Okun's (1962) assertion that a high unemployment rate has a negative effect on economic growth. The unemployment rate of a nation ultimately determines its economic stability. Figure 3.1: Economic Growth and Unemployment in Sub-Saharan Africa. -6 -4 -2 0 2 4 6 8 10 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 Sub-Saharan Africa GDP per capita growth (annual %) [NY.GDP.PCAP.KD.ZG] Unemployment, total (% of total labor force) University of Ghana http://ugspace.ug.edu.gh 21 Source: World Development Indicator (2022) 3.11 Economic Growth and Unemployment Drivers. Numerous research on how employment and unemployment affect GDP growth rate show that employment has a positive effect on GDP and that unemployment has a negative effect. The Sub- Saharan African region has one of the lowest growth rates in the world, which can be attributed to the effects of unemployment and potential resource mismanagement. Employment and economic growth do not appear to go hand-in-hand in Africa over time, despite the supposed benefit of pro-employment growth in the fight against poverty (Martins, 2013). The World Bank states that between 2016 and 2017, Sub-Saharan Africa's GDP growth rate increased. According to the Bank's most recent report, economic growth reached 2.6% in 2017 and was projected to reach 3.2% in 2018 and 3.5% in 2019. (2019). Sub-Saharan Africa has the problem of low-quality employment as opposed to unemployment and occasionally fails to experience improvement. According to estimates from WESO, the unavailability of opportunities for working-class youth and adults in Sub-Saharan Africa suggested that at least 247 million people were in vulnerable employment in 2016, which is comparable to 68 percent of all people who had jobs. While a little decline in the proportion of risky employment is expected in the next two years due to an increase in the working-class population, 14.6 million additional persons are expected to enter the population of vulnerable employment. According to the WESO research, the Sub-Saharan area has experienced its worst economic development in 20 years. The supply and demand of workers generally have an impact on the labour market. It is a factor in the labour force's availability to meet market needs for the creation of goods and services. The employment rate, unemployment rate, and the number of working-class job openings make up the fundamental characteristics of a University of Ghana http://ugspace.ug.edu.gh 22 labour market. The fundamental concept therefore revolves around salary, description of employment class size, and employee-to-employer match, which is determined by the human capital's talents, working experience, and educational level. The labour market's efficiency is thus a result of the interaction between the workforce's trend and the needs of employers. Therefore, the question that arises is whether the job-seeking population meets the requirements demanded by businesses in the labor market. This is because technology is replacing human labor. As a result, the question is whether the hungry working class possesses the required skills and/or knowledge to replace the newly introduced technology in the labor market. And how fast and efficiently are humans able to compete with technology and machinery? Considering all of this, it should come as no surprise that occasionally people lose their jobs as a result of companies firing employees. Human power is being replaced by technology, so it is up to the population looking for work to better equip themselves with technology in order to keep up with developments in the labour market. 3.2 Trend and features of unemployment in Sub-Saharan Africa. Figure 3.2: Unemployment Trend in Sub-Saharan Africa. -6 -4 -2 0 2 4 6 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 Sub-Saharan Africa GDP per capita growth (annual %) University of Ghana http://ugspace.ug.edu.gh 23 Source: World Development Indicator (2022) According to the graph, Sub-Saharan Africa's unemployment rate rose from a record high of 23.8% in 2004 to 22.6% in 2006, before continuing to decline until 2008. When the economy entered a recession in 2008, the unemployment rate was lower than it had been in prior years. As can be seen from the graph, unemployment began to increase slightly during the recession in 2009, and it has continued to rise since then. The rate reached a high of 26.7% in 2016 from a low of 25.3% in 2015. After the 2008 global recession, which resulted in the loss of over one million jobs in South Africa and disproportionately affected young people, the need for employment growth was highlighted negatively. Unemployment has been a huge issue around the world, particularly among the youth. Statistics from the World Bank, World Employment and Social Outlook (WESO), International Labor Organization (ILO), United Nations (UN), and other organizations indicate that unemployment worldwide is increasing and does not appear to be going down, and Sub-Saharan Africa is no exception. There are more people joining the workforce than leaving it, and little to no effort is being made to increase employment to keep up with the expanding labour force. The worldwide unemployment rate was 197.1 million in 2015, and the World Employment and Social Outlook predicted that it would increase by 2.3 million to 199.4 million in 2016. Additionally, it anticipated that in 2017, the global jobless rate would increase by 1.1 million individuals (WESO 2017). Sub-Saharan Africa is characterized by its large young population (15-24 years of age), which accounts for 20% of the overall population, 40% of the labor force (employed), and 60% of the unemployed thus this emphasizes the severity of the problem. As a result, the working force is made up of persons ranging in age from 15 to 65. Only youth aged 15 to 24 years account for 40% of the total workforce, implying that those aged 25 to 65 years account for a larger proportion of University of Ghana http://ugspace.ug.edu.gh 24 the overall population. This means that the unemployed account for a bigger percentage of the population than those who are employed. According to the International Labor Organization, about 75 million young people aged 15 to 24 were unemployed worldwide in 2011. This represented an increase of almost 4 million people since the start of the global financial and economic crisis in 2007, with nearly 20% of this rise occurring in Africa (ILO, 2012). In 2011, the youth unemployment rate in Sub-Saharan Africa was slightly over the global average of 12.8%, but North Africa had the highest rate with 27.1% globally (Anyanwu, 2014). Employment opportunities for young people are currently a major concern for international policymakers. This is due to the fact that increasing employment opportunities for young people has the potential to stimulate economic growth, foster political and social stability, and have a positive impact on overall progress toward the Millennium Development Goals and overall poverty reduction. In addition to this, it helps to achieve greater social equality among the children of any given region, as well as more effective resource allocation, higher potential for productivity, and a lower dependency ratio (Ncube and Anyawu, 2012). University of Ghana http://ugspace.ug.edu.gh 25 3.3 Nature of economic growth in Sub-Saharan Africa Source: World Development Indicator (2022) Figure 3.3 depicts the trend of the real GDP from 2000 to 2021 in Sub-Saharan Africa and it’s evident from the graph that Sub-Saharan Africa's economy has been expanding, albeit very slowly. From 2003 to 2007, it exhibits an upward trend, which corresponds to an increase in GDP, until the economic downturn in 2009, which exhibits a decline. Though it is developing extremely slowly, the economy appears to be doing well following the recession after slowly recovering after 2010 In recent years, African countries have had rather robust economic growth. SSA countries' growth lagged behind East Asia's 7.2 percent and South Asia's 4.3 percent yearly averages. Angola, Chad, Equatorial Guinea, Ethiopia, Mozambique, and Sierra Leone are among the most populous. In recent years, there has been a movement in the structure and composition of GDP in some SSA nations, particularly Mauritius, away from agriculture and towards services and industry. In 2007, services made for 44.3 percent of SSA's GDP, followed by industry (41.7 percent), and agriculture (41.7 percent) (14.0 percent). Agriculture, manufacturing, and services all had lower relative Figure 3.3: Economic Growth Trend in Sub-Saharan Africa -6 -4 -2 0 2 4 6 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 2 0 2 0 2 0 2 1 GDP per capita growth (annual %) GDP per capita growth (annual %) [NY.GDP.PCAP.KD.ZG] University of Ghana http://ugspace.ug.edu.gh 26 proportions in 2007 than in 2000. The overall picture of African economic growth reflects erratic and unsustainable growth patterns. 3.4 The linkage between economic growth and unemployment When we look at the functional link between growth and unemployment, we can detect two-way causality. Employment has an impact on growth. This means that growth is influenced by the unemployment rate. If the unemployment rate rises, all other factors being equal, the growth rate will fall. The unemployment rate will increase if businesses fire a lot of workers. All other things being equal, a higher unemployment rate will lead to a lower overall output, which will stifle growth. On the other side, expansion leads to employment. As a result, unemployment is a result of economic expansion. All other things being equal, if the rate of growth improves, more earnings and increased demand will follow, and the unemployment rate will decline. According to Obadan (1997) and Sagbamah (1997), employment and growth move in perfect harmony. The employment rate increases with a quicker rate of growth when all other parameters are held constant. Because of the aforementioned, growth and unemployment go in different ways and hence unemployment rate will fall if the rate of growth rises. Among other things, output affects employment. Overall, more employment will lead to higher output and consequent economic growth. Instead, a decline in employment (unemployment) will have a negative impact on output and, consequently, economic growth (all things being equal). The preceding indicates that there is a negative association between growth and unemployment. Growth-promoting policies should be designed and executed in order to reduce unemployment. However, it is critical to remember that for growth to reduce unemployment, it must be accompanied with higher labour force participation. According to the classical school of thought, this raises demand for goods and services, which raises demand for labour services, leading in an University of Ghana http://ugspace.ug.edu.gh 27 increase in employment and, as a result, a decrease in unemployment. However, it is vital to note that the growth that leads to an increase in employment (or a decrease in unemployment) is labor- intensive and accompanied by higher labor-force participation. The relationship between economic growth and the unemployment rate may not be positive in the short run. It is common for the unemployment rate to keep decreasing even after broad indications of economic activity have improved (Levine, 2013). Because some companies may have employees who aren't being used to their full potential, unemployment may not decline significantly when the economy emerges from the recession. Costs are associated with firing employees during periods of low product demand and hiring them back during periods of high demand. By increasing the productivity of their current personnel, employers may be able to raise output to meet increased demand without adding new staff at the start of the recovery. Long term, this speeds up the rise of labor productivity momentarily over the average rate. Once labor is fully employed, businesses can only increase output at the rate of productivity growth going forward. The amount of production produced will depend on the rates of growth in the labor supply and labor productivity as an economic expansion continues. The unemployment rate will decrease if employment growth outpaces labor force growth (McCarthy & Potter, 2012). Real GDP growth rates that fluctuate and unemployment have a protracted negative association. The economist Arthur Okun is credited with identifying this long-term connection between these two variables in the early 1960s. In the field of economics, Okun's law is generally acknowledged (Alan, 1997). It argues that real GDP growth approximately equal to the rate of potential output growth is typically necessary to achieve a stable unemployment rate (Knotek, 2011). The rate of growth in potential output is crucial to comprehending the link between GDP growth rate fluctuations and high unemployment over the long run. If all current labor and capital University of Ghana http://ugspace.ug.edu.gh 28 resources are put to use, an economy's potential output is an abstract measure of its future productivity (Gordon, 2008). Potential output growth at full employment is set by the rate of increase in both potential productivity and the available labor force (Mitra, 2002). Real GDP falls short of potential GDP when unemployment is high and the term "production gap" describes this condition. So long as every new member of the labor force is able to find a job, production growth will equal the rise in the number of people looking for work, even if productivity growth remains stagnant. There won't be enough openings to go around if GDP growth lags behind the increase in the labor force (Basu & Fernald, 2009). The number of people employed will consequently decrease while the unemployment rate will increase. When firms increase hiring to meet rising demand for goods and services, the unemployed may end up filling some of those positions if production growth outpaces the growth in the labor force. Consequently, the unemployment rate will decrease (Weidner & Williams, 2010). More people will enter the labor force than are necessary to produce a certain amount of goods and services if GDP growth keeps pace with labor force expansion in the presence of productivity growth. The percentage of people who are employed will decline (Levine, 2013). Alternatively, the unemployment rate will increase. As long as GDP growth outpaces the combined growth rates of the labor force and productivity, unemployment will eventually decline. The GDP rate may be helpful to policymakers who are interested in adopting stimulus plans to lower unemployment. It is obvious that a tendency of economic growth is necessary to lower the unemployment rate (Bernanke, 2012). University of Ghana http://ugspace.ug.edu.gh 29 CHAPTER FOUR METHODOLOGY AND DATA 4.0 Introduction This chapter discusses the methodologies and procedures utilized to achieve the study's objectives, as described in section 1.3. It consists of the theoretical framework, model specification, data source, and estimations. 25 Sub-Saharan African countries were studied using the panel data estimation method between 2000 and 2021. Panel data models assess group (individual-specific) effects, time effects, or both to address heterogeneity or individual effects that may or may not be observed (Park, 2011). These impacts may be random or fixed. We used pooled OLS regression, fixed effects, and random effects modeling as our three panel data estimation methods. To be more precise, an initial visual inspection was carried out using a graphical method. A test for variable stationarity will be carried out to ensure that the variables are stationary in order to prevent any incorrect analysis in the first place. The Fisher type Panel unit root test, developed by Choi (2001), will be used in this inquiry. The Johansen cointegration test and the Vector Error Correction Model (VECM) were then used. The VECM investigates both short-term dynamics and the long-term relationship, in contrast to the cointegration test which only looks at the long-term equilibrium relationship between variables. On the other side, the granger causality test, variance decomposition, and impulse response were used. We then examined the impact of economic growth on unemployment using a pooled OLS model. 4.1 Model Specification This study models economic growth as a function of the explanatory variables made up of unemployment and the relevant control variables, which are foreign direct investment (FDI), trade openness, and inflation. This study aims to analyze the relationship between economic University of Ghana http://ugspace.ug.edu.gh 30 growth and unemployment. This study employs regression analysis to examine the relationship between the dependent variable and the independent factors discussed later in the paper. The panel data analysis baseline model is described below; GDP= f (UNEMP, FDI, TRO, INF) The function is transformed into the generalized equation below in order to estimate the β parameters; Yit =α + β1 Unempit + βj+1 Xit +δt + ϕj+𝜇it ……………………………………… (1) Where: I= 1……10 t= 1……., T Yit = Annual GDP growth for country I at period t. Unempit = Unemployment rate for country I at period t. Xit = control variables including FDI, trade openness, population and agric. δt = Time variant factors affecting GDP (Time -Fixed Effects) ϕj = Country Fixed Effect. β1 = the coefficient on unemployment. βj+1 = the coefficient on control variables. 𝜇𝑡 = Error term i = index for countries. t = index for time period which is in years. Moreover, Makun and Nnanna (2015) and Yilmaz (2014) specified an unemployment-economic growth model relationship as thus, University of Ghana http://ugspace.ug.edu.gh 31 UNEM = f (GDP)……... (Model 1) Where: UNEM = Unemployment rate and GDP = Gross domestic product The current research alters equation 1 and uses a disaggregated method, which means that some of the sectors that contribute to the Gross Domestic Product (GDP) in Sub-Saharan Africa are incorporated into the model as explanatory variables alongside a few control variables. This enables us to determine how the unemployment rate affects the various economic sectors. Agriculture, manufacturing, and services make up these sectors, and the essential control variables include foreign direct investment (FDI), trade openness, and inflation. As a result, the relationship between the unemployment rate and economic growth forms the basis of this study. UNEMP = f (GRAGR, GRMAN, GRSER, FDI, TRO, INF) ……... (Model 2) UNEMP = β0 + β 1GRAGR + β2GRMAN + β3 GRSER + β4FDI + β5TRO + β6INF+ εt … (1) UNEMR = f (GDPGR, FDI, TRO, INF) UNEMP = α0 + α1GDPGR + α2FDI + α3TRO + α4INF+ εt …… (2) 4.4 Source of Data and Variables Description In order to quantify the impact of unemployment on the growth (GDP annual growth) of a few selected emerging Sub-Saharan African nations, the study used macro level yearly secondary panel data. The World Bank's African Development Indicators (2022) edition provided the data for this study, which was conducted between 2000 and 2021 on an annual basis. All required analysis for the study was performed using the statistical software package EViews 12. These 25 nations were chosen based on their economies' high unemployment rates and the availability of data; as a result, the analysis of the study shed greater light on the significant fluctuations in these core macroeconomic variables. The aforementioned countries include Burkina Faso, Burundi, Cote University of Ghana http://ugspace.ug.edu.gh 32 d'Ivoire, Cameroon, Niger, Nigeria, Rwanda, Senegal, Gambia, Mauritania, Tanzania, Ghana, Namibia, Guinea-Bissau, Lesotho, Madagascar, Malawi, Sierra Leone, Mauritius, Uganda, Zimbabwe, South Africa, Mozambique, Guinea, and Zambia. To identify evidence of causation between variables in higher middle-income countries, low-income countries, and lower middle- income countries, respectively, the study will divide SSA nations into three groups using the World Atlas technique in order to conduct the analysis. As indicated in Chapter 3, one of the reasons this study focuses on Sub-Saharan Africa is because it has one of the highest unemployment rates in the entire globe. Real GDP Growth: This serves as the study's primary explanatory variable. To account for the impact of economic growth on unemployment, we take into account the real GDP growth lag. Most empirical investigations have shown the Okun's law relationship between unemployment and GDP growth, which broadly represents the notion that as economic growth rises, unemployment falls. According to Anyanwu (2014), a nation's economic growth should boost employment through incentivizing people to join the labor force. Because the majority of the nations in the region have had consistent economic growth, we anticipate a negative correlation between GDP growth and unemployment in SSA, as predicted by Anyanwu (2014). This control variable's data comes from the World Development Indicators database. Unemployment: According to the WDI, unemployment is defined as the proportion of the labour force that is unemployed yet willing and eager to work. The data is modelled in accordance with International Labor Organization (ILO) standards and is expressed as a proportion of the entire labour force. Foreign Direct Investment (net inflows): According to WDI, foreign direct investment (FDI) refers to net investment inflows that are utilized to acquire long-term management ownership in a University of Ghana http://ugspace.ug.edu.gh 33 business that operates in a market other than the investor's own. According to the balance of payments, it is made up of short-term capital, other long-term capital, reinvested earnings, and equity capital. "Foreign direct investment" refers to an investment that involves a long-term partnership and shows a persistent interest and control by a resident entity in one country (foreign direct investor or parent firm) over a company with its headquarters in another economy. FDI is crucial to the expansion of the labor market in the majority of African nations where access to capital is a key barrier. Additionally, FDI-related employment generally tends to increase the country's overall labor productivity (Coniglio et al., 2015). Trade Openness: Researchers typically quantify the degree to which nations are open to international trade by looking at how much they import and how much they export. This is referred to as "trade openness." On the other hand, the economic performance of these nations is typically evaluated based on their gross domestic product (GDP) or their productivity in a variety of forms. Agriculture: The agriculture sector was represented by agriculture value-added and this factor is expressed as a percentage share of GDP. According to the WDI measure, agriculture value-added includes animal production, crop cultivation, hunting, and fishing in addition to these other activities. Value added is the term used to describe the net output of a sector after adding up all outputs and subtracting intermediate inputs (WDI, 2019). In Africa, agriculture continues to provide employment for a sizable section of the working population (ILO, 2018). In the sub- Saharan region, particularly in rural regions, the agricultural sector employs a bigger number of employees. Manufacturing: It is the production of goods for sale or consumption through the use of tools, machinery, human labor, chemical and biological formulations. It is measured as a percentage share of GDP and it incorporates both human handiwork and advanced technology in the University of Ghana http://ugspace.ug.edu.gh 34 transformation of raw materials into final commodities. The development of industries in the modern economy relies heavily on the technological advancement of productive techniques. Service: The service sector includes a variety of tertiary economic activities that are further split into the following categories: transportation, storage, and communications; wholesale and retail business; restaurant and hotel facilities; finance, insurance, real estate; and business services. It is measured as a percentage share of GDP. Inflation: This suggests an increase in the nation's average price level that would probably last a year. According to several empirical investigations, the well-known Phillips curve connects unemployment and inflation. Anyanwu (2014) finds evidence in favor of the Phillips curve in his empirical study of young unemployment and intra-African trade. Similar to this, Chaudhry et al. (2012) explain that inflation could lower unemployment in a nation when real wages are declining more rapidly than expected. This is if actual price levels exceed expectations. Salary discussions increase employment, which also reduces unemployment. A contradictory theoretical prediction states that unemployment may increase during periods of high inflation due to increased manufacturing costs when employees adjust their real wage. Due to the increased cost of production and low wages of their employees, employers would likewise respond by cutting employment. As a result, the inflation variable's sign is uncertain. The change in the Consumer Price Index (CPI), which is based on data from the World Development Indicators database, calculates the cost of purchasing a typical consumer's normal basket of goods and services. 4.2 Panel data Estimation Techniques The relationship between economic growth and unemployment was examined using the panel data econometric technique. First, in this study, methods for fixed and random effects were employed. A fixed effect model analyzes the correlation between the predictor and outcome variables within University of Ghana http://ugspace.ug.edu.gh 35 an entity (i.e., Country, firm, etc.). The predictor variable may be influenced by the individual, which must be considered, according to the fixed effect model. Calculating the overall contribution of the predictors to the dependent variable is made simple by the fixed effect, which removes the time-invariant characteristics from the equation. To do this, either the within transformation or the first difference transformation is applied. On the other hand, we use the random effect model if there is any reason to believe that differences across units influence the outcome variable. The random effect model assumes that each individual effect is random and independent to the regressors. However, a decision must be taken regarding which of the two models, fixed effect or random effect, provides accurate and dependable parameter estimates. The Hausman test will be conducted to decide which model should be utilized. It involves comparing the alternative hypothesis, that the fixed effect model is the most appropriate and effective model, with the null hypothesis, that there is no relationship between regressors and individual effects (the random effect is appropriate). If the results are significant, that is, if the prob>chi2 is greater than 0.05 and we cannot reject the null hypothesis that the random effect model is appropriate, we favor the random effect model over the fixed effect model; otherwise, we opt for the fixed effect model. Additionally, the Granger causality test, Johansen's (1988) cointegration approach, the Vector Error Correction Model (VECM), and panel data analysis were employed to examine the relationship between economic growth and unemployment in SSA nations. In this investigation, the VECM was employed because Vector Autoregression (VAR) models incorrectly represent cointegrated variables. In general, VAR models do not imply long-term relationships; rather, they eliminate them when they diverge. However, the VECM can differentiate between long-term and short-term correlations and can reveal causation between variables that normal Granger causality University of Ghana http://ugspace.ug.edu.gh 36 cannot (Oh & Lee, 2004). Before undertaking cointegration analysis, it is vital to determine whether the series is stationary. The Augmented Dickey-Fuller (1979) test was used to determine whether or not the series was stationary. There is the possibility of a cointegration relationship between the series if their differences are stationary but their levels are not, revealing their long- term link. Johansen's cointegration test was utilized to assess the long-term relationship between two variables. Test for Over-identifying Restrictions Over-fitting of endogenous variables is caused by the tendency of panel data estimations for the number of instruments to expand exponentially with the number of time periods (Heid and Larch, 2012). However, the instruments of the endogenous variables must be legitimate for the estimator to give accurate and dependable findings. Thus, a correlation must exist between the instruments and endogenous explanatory variables, but not between the residuals and the explanatory variables. Consequently, we evaluate the validity of instruments using the Sargan tests for over-identifying constraints. If the employed instruments are actually valid and therefore exogenous under the null hypothesis that they are valid against the alternative that they are invalid, and are also connected with the error term, then the null hypothesis is not rejected. This is indicated by a p-value > 0.05. Unit root test The stationary time series theory is used to estimate the unit root test. The empirical study begins with a look at the statistical properties of the economic variables generated by the model, which are time series in nature. If a series' mean and auto covariance are not dependent on time, it is said to be stationary. The goal of examining these qualities is to determine whether the model's variables are stationary in order to prevent false regression that could produce inaccurate findings (Austeriou & Hall, 2011). The inability of non-stationary time series results to be generalized to subsequent periods provides yet another argument for doing stationarity tests (Banda, et al., 2014). University of Ghana http://ugspace.ug.edu.gh 37 Phillips Perron unit root test and Augmented Dickey-Fuller unit root test are the two most frequently employed stationary tests (Said and Dickey, 1984). If the variables are found to be stationary in the model, it is frequently anticipated that they will have a constant variance and specific components of autocorrelation over time (Mosikari, 2013). Visual Inspection Graphs or a correlogram test can be utilized to demonstrate this technique for stationarity testing. According to Mah (2013), the strategy is effective since it provides an initial indication of the nature of the time series. It indicates whether the tracked variables are increasing, decreasing, or remaining constant. According to Gujarati and Porter, when the logged variable rises or falls, it indicates that the logged variable's mean is changing with time, making the recorded variable non- stationary (2009). The variable is stationary and oscillates around the trend line when it is constant. Due to the fact that it is unaffected by time, more robust techniques, such as the Augmented Dickey-Fuller and Phillips Perron unit root tests, should be utilized to confirm these conclusions. Augmented Dickey Fuller (ADF) Test The Dickey Fuller test presupposes that each error term is uncorrelated. Dickey and Fuller (1976) revised the fundamental Dickey-Fuller test to develop and utilize the already Augmented Dickey- Fuller (ADF) unit root test to account for the chance that the error components are linked. On models containing lagged values of the dependent variables, the Augment Dickey Fuller unit root test is utilized to reduce serial correlation in the error term. In accordance with Mosikari (2013) and Madito & Khumalo, the following regression is generated for each model variable for the ADF test (2014). University of Ghana http://ugspace.ug.edu.gh 38 Where 𝛼o is a constant, 𝛽t is the coefficient on a time trend and 𝜎1 is the parameter to be estimated and 𝜀t is the error term which is assumed to be normally distributed. The ADF test uses the presence or absence of a unit root in the time series as the null hypothesis. To be considered stationary, a variable must have a critical value greater than the test statistics and a probability value (PV) less than the 1%, 5%, and 10% levels of significance. Because of this, the Augmented Dickey-Fuller statistic has a negative value. Since the unit root grows with a larger negative, no significance can be accepted (Dickey, 1988). Phillips-Perron A different approach to determining if a unit root exists in a general time series context is the Phillips-Perron test. The linear trend was part of Phillips' and Perron's (1988) definition. To account for the possibility of serial correlation in the error terms, the ADF unit root test modifies the Dickey-Fuller test by including the lagged differenced terms of the independent variables. Additionally, Phillips and Perron (1988) use non-parametric statistical methods (Gujarati & Porter, 2008) to analyze serial correlation in the error terms without including the delayed differenced terms. The PP test uses the following regression estimation procedure for all model variables: Where 𝜇0 and 𝜇1 are parameters estimates and 𝜀t is error term. Time series are considered stationary when the critical value exceeds the test statistics and the probability value (PV) is less than the 1%, 5%, and 10% levels of significance. To determine if a unit root exists in a time series, one can apply the PP test. University of Ghana http://ugspace.ug.edu.gh 39 The choice of lag length Selecting a suitable lag for our variables is crucial if we want to get desirable outcomes. The lag time is determined by completely at random variables. In order to ensure that the error components are normally distributed without autocorrelation or heteroskedasticity, the lag length criteria is essential. Variables that are eliminated have an effect on the lag time. Therefore, it has an instantaneous impact on how the model operates. The best lag time can be found using any of a number of different metrics, including Aikaike's Information Criterion (AIC), Schward's Information Criterion (SC), Hannann-Criterion Quin's (HQ), sequential modified Likelihood Ratio (LR), and Final Prediction Error (FPE) (FPE). According to Mah (2013) and Davidson & MacKinnon (2011), the following formula is used to determine these criteria for information (2004): Where l is the log likelihood l is computed as: Mah (2013) claims that the SC is an AIC substitute that penalizes heavily for having additional coefficients. In addition, HQ has a punishment feature. M is the number of parameters in the equation for the alternative in the sequential modified likelihood ratio (LR) test. We start with the University of Ghana http://ugspace.ug.edu.gh 40 greatest latency, reduce it incrementally by one lag, then compare the altered LR statistics to the crucial value until we get a rejection. Each criterion is marked with an asterisk next to the selected lag, which minimizes the criterion. When looking for a good lag length criterion, Liew (2004) asserts that the AIC and FPE findings outperform the SC and HQ results when the number of observations is sixty or less. Johansen cointegration test Unlike the Engle Granger technique, the Johansen cointegration test allows for a large number of cointegrating interactions. This test considers large samples, or asymptotic properties (Phillips & Ouliaris, 1990). The Auto Regressive Distributed Lags should be used since the results would be invalid if the sample size were too small (Giles, 2014). The Johansen cointegration test represents a linear combination of the data with the highest correlation and is based on the eigenvalues of data transformations. It is guaranteed that the eigenvalues are either nonnegative or real. According to Johansen (1988), the Johansen methodology starts with the Vector Autoregression (VAR) of order p, as shown below: Where Y is a n ×1 vector variables that are integrated of the order one I (1) and εt is a n ×1 vector innovations. This VAR can be written as: Where ∏ = ∑ 𝐴𝑝 𝑡=1 t-1 and Γ = − ∑ 𝐴𝑝 𝑗=𝑡+1 j University of Ghana http://ugspace.ug.edu.gh 41 The 𝑛 × 𝑟 matrices 𝛼 and 𝛽 exist if the coefficient matrix ∏ has a reduced rank 𝑟 ×𝑛. ∏ = 𝛼𝛽1 and 𝛽1Yt are stationary, r is the number of cointegration relationship, 𝛼 is the adjustment parameter in the vector error correction model and each column of 𝛽 is the cointegrating vector. The maximum eigenvalue test and the trace test are names for the Johansen tests. Let r represent the rank of. Probability ratio tests include the Johansen tests. The initial Johansen test pits the null hypothesis of no cointegration against the alternative of cointegration for both test statistics (Gregory & Hansen, 1996). The alternative hypothesis is different for each test. The maximum and trace eigenvalue tests are as follows: Where T represents the sample size and 𝜆 represents the ith greatest canonical correlation. r cointegrating vectors is the null hypothesis of the trace test, while 𝑛 cointegrating vectors is the alternative hypothesis. The maximum eigenvalue test evaluates the null hypothesis of r cointegrating vectors veres the alternative hypothesis of r plus one cointegrating vectors (Thomas, 1985). Vector Error Correction Model For use with nonstationary series that are known to be integrated, the vector error correction model (VECM) is a constrained VAR with cointegration restrictions imposed in the specification (Amassoma & Nwosa, 2013). The short run dynamics and cointegrating equation of the series are investigated using the conventional vector error correcting model. When the series fails to cointegrate, the short run model is the next estimation technique since the error correction term for University of Ghana http://ugspace.ug.edu.gh 42 the coefficient is estimated in this fashion. The link between short run dynamics and long run equilibrium relationships among data series is explained by the VECM concept. Use of VECM is crucial because it is utilized to rectify transient short-run series aberrations from the long-run equilibrium relationship (Eze, Atuma, Egbeoma, 2016). Here is how the VECM model is presented: Where: Yt= Yt- Yt-1, 𝛼1 and 𝛼2 represent the dynamic adjustment coefficients of the variables, while 𝜇t-1 is the residual lag, the term, denoted by t, is the error term that is estimated to rectify the long-term equilibrium error and shows the short-run divergence from equilibrium. Because the inquiry included more than one explanatory variable, it was chosen to apply VECM based on conventional least squares (Asoluka & Okezie, 2011). As a result, applying the procedure in the study is required. The model is depicted below: ΔGDPt = β0 + β1 ΔGDPt-1 + β2UNt-1 + β3 FDI t-1+ β4 TROt-1 + β5INFt-1 +𝜇t Where GDP denotes the gross domestic product growth rate, β0 is a constant, β1, β2, β3 β4 β5 are the vaues of the explanatory variables and 𝜇t is the error term of the long run equilibrium error. The dynamic behavior of the study's important variables is analyzed via a vector error correction model approach after the long run equilibrium relationship has been confirmed. By using the appropriate differenced variables to build the model's short-term adjustment, it circumvents the issues of spurious regression (Mah, 2013). University of Ghana http://ugspace.ug.edu.gh 43 VECM granger causality Within a Vector Error Correction Model (VECM) framework, the Granger causality test determines the direction of the causal relationship between variables. Effectiveness of the VECM- based causality test can be attributed to its ability to model interconnected but non-stationary variables (Austeriou & Hall, 2011). The following regressions can be used to implement the Granger causality test in a VECM framework (Ageli, 2013; Odhiambo, 2009). Where, in the preceding equations: • n represents the number of lag variables. • 𝛼1, 𝛼2, β1 and β2 are a list of the parameters that need to be estimated. • 𝛼0 and β0 are constant terms that represent the equation intercepts • 𝜇t and εt are the error terms or the mutually uncorrelated white noise residual • ECTt is the error correction term lagged one period Diagnostic test Diagnostic tests are performed to validate the vector error correction model's precision and predictability. Below is a list of diagnostic tests that may be performed: The Breusch-Pagan, White, ARCH, Breusch-Godfrey, Jarque-Bera, Ljung-Box Q, and Ramsey RESET tests are only a few examples. University of Ghana http://ugspace.ug.edu.gh 44 Normality test There is a special Lagrange multiplier test for normalcy known as the Jarque-Bera test (Jarque & Bera, 1987). Many statistical tests, such as the T test and the F test, rely on the assumption of normality. The Jarque-Bera test is commonly used to establish normalcy before these kinds of tests. Since the normal distribution has a kurtosis of three and a skew of zero, it is perfectly symmetrical about the mean. The distribution's peak height and the amount of data in the tails are both indicated by kurtosis (Jarque & Bera, 1987). We put the assumption that residuals follow a normal distribution to the test. Autocorrelation tests The Ljung-Box Q is a diagnostic tool that was developed by Box and Pierce in 1970. Its purpose is to evaluate the robustness of a time series model, or the lack thereof. Following the application of an ARMA model to the data, the test is carried out on the residuals of the time series. The test investigates the residuals' innate relationships with one another (Ljung & Box, 1978). The Breusch-Pagan Test, which was devised by Breusch and Pagan in 1980, is utilized in the process of determining whether or not the model contains serial correlation. The alternative null hypothesis for the Breusch-Pagan test is given by: Where h is an unknown, continuously differentiable function (that do not depend on i) such that h(0) > 0 and h(0) = 1 . A test for: H0: 𝛼 =0 versus H0: 𝛼 ≠0 can be derived independently of the function h. The simple Breusch-Pagan test can be computed by multiplying the number of observations by the R2 of and of 𝑒𝑡 2 (the squared OLS residuals) on zi together with a constant. This allows for the calculation of the test. A Langrage multiplier test for heteroskedasticity, the University of Ghana http://ugspace.ug.edu.gh 45 Breusch-Pagan test is sometimes known as the Breusch-Pagan test. Lagrange multiplier tests' main advantage is that they may be calculated using the R2 of an auxiliary regression without requiring the model to be estimated under the alternative. The Breusch Godfrey test, which was expanded by Breusch (1978) and Godfrey, is applied in order to assess whether or not a time series contains serial correlation (1978). The test makes use of the residuals that were generated by the regression analysis model, and a test statistic is generated from the data that they provide. The absence of any kind of serial relationship, regardless of the order, is referred to as the null hypothesis. Heteroskedasticity tests One of the most important steps in the estimating process for panel data is the test for heteroskedasticity, a cross-sectional data estimation problem. Because cross-section is covered by panel data, this estimation problem is probably present in our estimations. Heteroskedasticity is present when the error term's variance fluctuates between observations and renders it non-constant (that is Var (𝜀𝑖𝑡) ≠σ2). Wooldridge (2008) argues that the T and F tests produce misleading findings because heteroskedasticity renders parameters meaningless. When the Hausman test is consistent with the fixed effects model, we will employ the modified Wald test for heteroskedasticity. To account for the possibility of heteroskedasticity, a powerful command will also be executed. This method provides standard errors of regression coefficients that are robust against heteroskedasticity. The white test (White, 1980) employs the idea of a heteroskedasticity consistent covariance matrix for the OLS estimator and does not necessitate any supplementary structure for the alternative hypothesis. The conventional estimator is: University of Ghana http://ugspace.ug.edu.gh 46 If there is no heteroskedasticity, the equation above will produce a consistent estimate of V(b), but it will not if there is. The Chi-squared asymptotic distribution of the test statistic has P degrees of freedom, where P is the number of auxiliary regressions regressors (regressors in the auxiliary regression that do not include the intercept). Engle's Autoregressive Conditional Heteroskedasticity (ARCH) model is the most commonly used heteroskedasticity model in time series (1982). An ARCH model starts from the premise that we have a static regression model: y =α + β0+ β1+ εt Ramsey RESET Ramsey RESET is used in addition to the diagnostic test to assess whether the error correction model is appropriately stated. Incorrect equation specification, according to Austeriou & Hall (2011), can result in improper functional forms and misspecification bias, which can lead to high R-squared and deceptive results. Stability test The reliability of the model is examined using a series of tests predicated on recursive residuals. The CUSUM test and the CUSUM of squares test are the two most essential tests; they use data ordered by time rather than by the value of the explanatory variable (Steward, 2005). A visualization showing the sum of the recursive residuals serves as the foundation for the CUSUM test. If this total deviates from a critical bound, it is assumed that a structural break occurred at the moment the sum started moving in that direction. The CUSUM of squares test is a proportional representation of the cumulative sum of squared recursive residuals, which is similar to the CUSUM test (Marno, 2004). University of Ghana http://ugspace.ug.edu.gh 47 Impulse response function An impulse response function is a mathematical model that describes how the value of a particular variable shifts over a predetermined amount of time after being subjected to a shock. It's possible that Granger causality can't provide the complete story of how the variables in a model interact with one another. It is essential to take into account the impulse response relationship in a system with a higher dimensionality (Rossi, 2010). Variance Decomposition After a vector auto regression (VAR) model has been fitted, the understanding of