University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA COLLEGE OF HUMANITIES DECENT WORK AND THE WORKING POOR: EVIDENCE FROM GHANA BY JOSHUA OPPONG (10416847) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF MPHIL ECONOMICS DEGREE. DEPARTMENT OF ECONOMICS DECEMBER, 2019. I University of Ghana http://ugspace.ug.edu.gh DECLARATION I, JOSHUA OPPONG, the author of this thesis, hereby declare that apart from the references made to other studies, which I have duly acknowledged, the work presented in this thesis “DECENT WORK AND THE WORKING POOR: EVIDENCE FROM GHANA”, was entirely done with my effort and was undertaken in the department of economics, University of Ghana, (Legon), under the supervision of Prof. William Baah-Boateng and Prof. Edward Nketiah-Amponsah. I bear sole responsibility for any shortcomings …………………………… …………………….. JOSHUA OPPONG DATE (10416748) …………………………… ……………………. PROF. WILLIAM BAAH-BOATENG DATE (SUPERVISOR) …………………………………… …………………… PROF. EDWARD NKETIAH-AMPONSAH DATE (SUPERVISOR) II University of Ghana http://ugspace.ug.edu.gh ABSTRACT The number of global working poor far exceeds the number of unemployed people. Thus, employment creation is not sufficient for poverty alleviation. For this reason, the concept of ‘decent work’, which seeks to measure the quality and quantity of work was proposed by the ILO in 1999 as a necessary and sufficient condition for poverty eradication. Achieving universal decent work is now one of the first eight SDG’s to be achieved by all countries in 2030. Previous studies on decent work have taken a macro perspective with most of them failing to reveal the individual level of decent work and what influences decent work at the micro-level. The relationship between decent work and the working poor have not been studied relative to its importance. This study utilizes the Round 7 of the Ghana Living Standards Survey (GLSS7), which contains decent work indicators and poverty dataset. The study creates decent work index using 11 indicators to elicit relevant information about the factors that determine decent work and analyze the relationship between decent work and the working poor in Ghana. The results from the descriptive statistics, binary and ordinal logistic regression indicate that individual personal characteristics (sex, age, education), employment status, industry, and employment sector are key determinants of decent work in Ghana. Women are more likely to experience decent work than their male counterparts. Education has a significantly positive influence on the probability of an individual engaging in decent work, with the returns being higher for women are than men. The study also found that decent work has a reducing effect on working poverty. III University of Ghana http://ugspace.ug.edu.gh The study concludes with a recommendation that for Ghana to achieve the first goal of the SDG of eliminating poverty of all forms, achieving universal decent work is key. In order to achieve universal decent work for Ghana, there is the need to invest in human capital through education, particularly that of women. Lastly, formalizing the informal economy in order to implement and enforce labour laws can help Ghana achieve universal decent work. IV University of Ghana http://ugspace.ug.edu.gh DEDICATION I dedicate this thesis to Almighty God and my family. V University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I am thankful to God for his guidance and protection over my life throughout the period of my study. I am also grateful for His mercies and unmerited favours bestowed upon me. My profound gratitude goes to my family for their collective support, both spiritual and physical and for solely sponsoring my education up to this level. I do not know how to express my gratitude to them. Again, to Professor Augustine Asaah, who have contributed a lot throughout my MPhil programme, many thanks to you. More so, this thesis would not have seen the daylight without the supervision of Professor William Baah-Boateng and Professor Edward Nketiah-Amponsah. For showing interest in this work and your constructive criticisms and corrections, many thanks to you and God richly bless you all. VI University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION..................................................................................................................... II ABSTRACT ............................................................................................................................. III DEDICATION .......................................................................................................................... V ACKNOWLEDGEMENT ....................................................................................................... VI TABLE OF CONTENTS ....................................................................................................... VII LIST OF TABLES ................................................................................................................... IX LIST OF FIGURES .................................................................................................................. X LIST OF ABBREVIATIONS .................................................................................................. XI CHAPTER ONE ........................................................................................................................ 1 INTRODUCTION ..................................................................................................................... 1 Background ...................................................................................................................... 1 Problem Statement ........................................................................................................... 4 The Study Objectives ....................................................................................................... 8 Significance of the study .................................................................................................. 8 Data Source and Methodology ......................................................................................... 9 Organization of The Study ............................................................................................... 9 CHAPTER TWO ..................................................................................................................... 10 DECENT WORK AND WORKING POVERTY IN GHANA .............................................. 10 Introduction ................................................................................................................ 10 Brief Overview of the Ghanaian Economy ................................................................ 10 Incidence of Poverty in Ghana ................................................................................... 12 Elements of Decent Work in Ghana ........................................................................... 18 2.3.1 Employment Opportunities ............................................................................ 18 2.3.2 Adequate Earnings and Productive Work. ..................................................... 23 2.3.3 Social Security and Social Dialogue. ............................................................. 25 CHAPTER THREE ................................................................................................................. 30 LITERATURE REVIEW ........................................................................................................ 30 3.0 Introduction ......................................................................................................... 30 3.1 Working Poverty ................................................................................................. 30 3.1.1 Conceptual Definitions of Poverty ............................................................. 31 3.1.2 Measurement of Poverty............................................................................. 32 3.1.3 Theory of Work .......................................................................................... 36 3.2 Decent Work ....................................................................................................... 39 VII University of Ghana http://ugspace.ug.edu.gh 3.3 Determinants of Decent Work............................................................................. 42 CHAPTER FOUR .................................................................................................................... 46 METHODOLOGY AND EMPIRICAL ANALYSIS ............................................................. 46 4.0 Introduction .................................................................................................. 46 4.1 Conceptual Framework ................................................................................ 46 4.1.1 Measurement of Decent Work .............................................................. 48 4.1.2 Construction of Composite Decent Work Index ................................... 49 4.1.3 Theoretical Model ................................................................................. 50 4.1.4 Empirical Model .................................................................................... 51 4.1.5 Explanation of Variables and their Expected Signs. ............................. 53 4.1.6 Econometric Approach and Estimation Procedure of the Logistic Model. ............................................................................................................ 57 4.1.7 Data Source ........................................................................................... 59 4.2 Presentation and Discussion of Empirical Results ....................................... 60 4.2.1 Descriptive Statistics of the Labour Force ............................................ 60 4.2.2 Empirical Analysis ................................................................................ 68 4.2.3 Descriptive Statistics of Composite Decent Work Index ...................... 68 4.2.4 Analysis of the Regression Results ....................................................... 81 4.2.5 Determinants of Decent Work............................................................... 81 4.2.6 Decent Work and Working Poverty ...................................................... 87 CHAPTER FIVE ..................................................................................................................... 90 SUMMARY, CONCLUSION AND POLICY RECOMMENDATION................................. 90 5.0 Introduction .......................................................................................... 90 5.1 Summary of Empirical Findings ........................................................... 90 5.2 Conclusion and Policy Recommendations ........................................... 92 REFERENCES ........................................................................................................................ 95 APPENDICES ....................................................................................................................... 104 VIII University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 2.1: Informal Employment (Percentage of Non-Agricultural Employment) ................. 25 Table 2. 2: Share Contribution to Social Security and Proportion of Population Covered by Social Security Systems in Ghana……………………………………………………… 29 Table 4. 1: Percentage of the Employed by Sex and Location ................................................ 61 Table 4. 2: Percentage of Occupational Employment by Sex and Locality ............................ 62 Table 4. 3: Percentage of Sector Employment by Sex and Locality ....................................... 63 Table 4. 4: Percentage of Employment Status by Sex and Locality ........................................ 64 Table 4. 5: Percentage of The Employment Type by Sex and Locality .................................. 65 Table 4. 6: Percentage of Employment by Educational Qualification ..................................... 65 Table 4. 7: Working Poverty Status by Sex and Locality ........................................................ 66 Table 4. 8: Summary Statistics of Decent Work Indicators..................................................... 67 Table 4. 9: Composite Index by 11 Indicators ......................................................................... 69 Table 4. 10: Ordered Composite Index of the Total Employment .......................................... 69 Table 4. 11: Decent Work Index by Sex (Percent) .................................................................. 70 Table 4. 12: Decent Work Index by Location and Sex (Percent) ............................................ 71 Table 4. 13: Decent Work Index by Occupation and Sex (Percent) ........................................ 73 Table 4. 14: Decent Work Index by Employment Sector and Sex (Percent) .......................... 75 Table 4. 15: Decent Work Index by Educational Qualification and Sex (Percent) ................. 76 Table 4. 16: Decent Work Index by Age and Sex (Percent) .................................................... 76 Table 4. 17: Decent Work Index by Employment Status and Sex (Percent) ........................... 78 Table 4. 18: Decent Work Index by Industry and Sex (Percent) ............................................. 80 Table 4. 19: Decent Work Index by Type of Employment and Sex (Percent) ........................ 81 Table 4. 20: Determinants of Decent Work ............................................................................. 86 IX University of Ghana http://ugspace.ug.edu.gh Table 4. 21: Marginal Effects from Logit Model 2 Regression ............................................... 89 LIST OF FIGURES Figure 2. 1: Sectoral Contributions to Ghana’s GDP (percentage) in 2017. ........................... 11 Figure 2. 2: Employment by Sector (percentage of total employment) from 1991-2017. ....... 11 Figure 2. 3: Population Living Below National Income Poverty Line (%) in 26 African Countries 2006-2017. ............................................................................................................... 13 Figure 2. 4: Incidence of Poverty Gap by Locality, 2005/06-2016/17(Poverty line=GHc1,314). ...................................................................................................................... 14 Figure 2. 5: Incidence of Poverty by Sex of Household Head from 2005/06-2016/17 (Poverty line=GHc1,314). ...................................................................................................................... 15 Figure 2. 6: Incidence of Poverty by Employment Status of Household Head (Poverty line=GHc1,314). ...................................................................................................................... 16 Figure 2. 7: Working Poverty among Occupation, Sector and Status in Employment ........... 17 Figure 2. 8: Employment to Population Ratio by Sex (Percentage) from 2000-2015. ............ 19 Figure 2. 9: Unemployment Rate by Sex and Youth (percentage of ages 15+, 15-24 of labour force), 1998-2015 ..................................................................................................................... 21 Figure 2. 10: Average monthly earnings of paid employees by sex and occupation above 15 years in 2015 ............................................................................................................................ 25 Figure 2. 11: Trade Unions Density and Collective Bargaining Coverage (Percentage) 2008- 2016, Ghana. ............................................................................................................................ 29 X University of Ghana http://ugspace.ug.edu.gh LIST OF ABBREVIATIONS CI Composite Index DWI Decent Work Index ERP Economic Recovery Programme GDP Gross Domestic Product GLSS Ghana Living Standard Survey GSS Ghana Statistical Service ILO International Labour Organization IMF International Monetary Fund LSMS Living Standards Measurement Study MCA Multiple Correspondence Analysis MDGs Millennium Development Goals NEET Youth Not in Education, Employment and Training OECD Organization for Economic Co-operation and Development PAMSCAD Programme of Action to Mitigate Social Cost of Adjustment SAP Structural Adjustment Programme SDGs Sustainable Development Goals TUC Trade Union Congress U. S United States UNDP United Nations Development Programme VIF Variance Inflation Factor WDI World Development Indicators WP Working Poor XI University of Ghana http://ugspace.ug.edu.gh XII University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION Background Traditionally, human welfare has been centered on the wealth of nations, societies, and individuals. This presumption holds because an individual’s ability to make consumption decision is purely based on income (or wealth) levels. The inability of an individual to make choices regarding their welfare levels or to meet the daily minimum standard of living renders them as poor. The concept of poverty has always been with humanity and is mostly attributed to a low level of consumption due to a low income from work. Though it is believed that income from work is the key antidote to poverty alleviation, some scholars hold the view that poverty is multi-dimensional which requires a multi-faceted approach to tackle it (Bell & Newitt, 2010). . The problem of inadequate labour income to meet the hopes and aspirations of the poor in Ghana worsened through the emergence of globalization and institutionalization. The economic reforms embarked upon in Ghana in the early 1980s led to the privatization of state institutions on grounds of the inefficacy of government businesses, the liberalization of financial and labour markets guided by the Breton Woods Institutions. In the advent of economic reforms and structural adjustment programme, economic growth became the traditional ‘measuring rod’ for countries’ development. On the contrary, this did not lead to a parallel growth in employment, let alone ‘decent work’. The effect of this has been the rising informalization in developing and emerging economies alike. In fact, evidence has shown that an increase in growth rate coupled with high inequality actually increase poverty (Fosu, 2017). 1 University of Ghana http://ugspace.ug.edu.gh Ghana is among few countries that subscribed to this programme (SAP/ERP) in 1983 under the IMF and World Bank, and the immediate results was the significant growth rate averaging 5.7% over 1984-1989 (Baah-Boateng, 2016). However, it triggered a severe widespread of poverty and economic turmoil. During the same period, it was acknowledged that the economic recovery programme (ERP) exacerbated economic conditions of Ghana and had a demeaning implication on certain vulnerable groups in the ensuing periods, which later led to the implementation of PAMSCAD (i.e. Programme of Actions to Mitigate the Social Costs of Adjustment) to cushion the poor households who were greatly affected (Anin, 2003) . One major shortfall of the market system is that it fails to provide certain essential commodities to the marginalized in societies for implementing such policies, and the negative effects on the poor in societies. This compelled Pope John Paul II to suggest the need to establish who is responsible for guaranteeing the global common good and the exercise of the economic and social right. The free market by itself cannot do it, because there are many human needs that have no place in the market (Pope John Paul II, 1999 p.6). Calling upon immediate interventions, the International Labour Organization (ILO) conceived the ‘decent work’ agenda as its ultimate goal thereafter (Somavia, 1999). The conception of decent work has remained the workhorse of the ILO in its technical cooperation programmes and has elevated the ILO’s profile as a development actor (Bell and Newitt, 2010). The concept of decent work is succinctly summarized in the address by Juan Somavia, the general secretary of the ILO during the International Labour Conference in 1999 in what he referred to as the four (4) ILO’s mandate. These include the fundamental principles and rights at work, the creation of full employment (more and better jobs for women and men), social protection, and tripartism and social dialogue, (Somavia, 1999). Thus, the ILO defines decent 2 University of Ghana http://ugspace.ug.edu.gh work as productive work (for all women and men) in which rights and security are upheld and command sufficient income, with adequate social protection and workers representation, (Somavia, 1999). Many scholars and institutions hold the view that income from work is the main catalyst in eradicating poverty. Contrary to this view, others assert that workers do not only consider pay in making employment decisions but also look at the entire attributes of work, which go beyond wages (Pautz & Stuart, 2016). The importance of ‘decent work’ as a global agenda was its acknowledgement and incorporation into the Millennium Development Goals (MDG’s) as a target to be achieved by 2015 by all countries. Decent work agenda was identified as a key player in achieving the sustainable development goals (SDGs) of the United Nations Development Programme (UNDP) in 2030. The reason is that, first, decent work puts money in peoples’ pockets, which enable them to spend on locally made goods. Second, an increase in the purchasing power triggers growth and development of new local medium and small firms, which in turn would enable them to expand and employ more workers to improve pay and conditions of service. Third, these have a trickledown effect of increasing the revenue base of the government that could be used to finance social protection measures to the marginalized in society. Since ILO introduced decent work agenda and was incorporated into the MDG’s, there has been a plethora of research in the area. Some scholars see decent work approach to development as a both ‘means’ and an ‘end’ and therefore applicable to both developed and developing countries alike. However, there is flexibility and variations for countries to choose the necessary elements available to them (Bell & Newitt, 2010; ILO, 2013). This is one major weakness of decent work measurement since it does not provide a common threshold for 3 University of Ghana http://ugspace.ug.edu.gh comparison (ECLAC/ILO, 2013). Ghai (2006) argues in favour of the ILO on the flexibility and variations regarding elements of decent work by stating that, countries are not in the same stages of development and should select or measure decent work element according to their level of development. It is therefore recommended that employing decent work paradigm can complement the traditional development indicators such as growth rate, unemployment, and inequality approach to development policy and identify alternative ways to foster development cooperation (Ghai, 2002, 2003, 2006). Studies on decent work have taken country profile dimension, which normally seeks to integrate decent work into countries’ local programmes. Studies on country’s profile of the ILO has been criticized based on the fact that such studies generally do not provide information on an individual decent work level, and the factors that influence decent work at the individual level. As a result, some studies have shown that individual economic and demographic factors such as sex, education, race, age and economic status are the factors that determine individuals decent work (Mackett, 2017; Moussa, 2017; Nizami, 2017). Problem Statement The concept of decent work from its inception has gained momentum and support from the global arena among development partners, donor countries and private institutions due to its underlying principles rooted in the global development agenda. The question is, why is decent work an important issue? According to Ghai (2006), whereas the MDG sought to measure the overall wellbeing of individuals, decent work, on the other hand, intends to measure the wellbeing of workers. The significant disproportionate change in the global employment emanating from the privatization and liberalization of state institutions and the labour market in the early 1980s has led to vulnerable employment in the informal sector (Mackett, 2017). 4 University of Ghana http://ugspace.ug.edu.gh Report from the SDG indicates that despite the success story told in pursuing the MDG’s for the past two decades, there exists an extra mile to cover (Kuhn, et al., 2018). The 2018 ILO Director-General report indicated that the inability of most African countries to achieve the MDGs goal (1) by halving extreme poverty is attributed to high uneven growth in decent work opportunities. As of 2019, 300 million workers were estimated to be extremely poor, which far exceeds 190 million unemployed people (ILO, 2019). However, poverty research focuses more on the unemployed relative to working poverty (Brady, et al., 2013). While the global unemployment rate stood at 5.6%, estimates show that about 250 million workers in Africa live in extreme or moderate poverty with an expected 4 million increases in every year. In the context of pervasive poor-quality employment, 66% of African workers estimated to be in vulnerable employment. Meanwhile, the extent of vulnerability and informality masks unemployment rate in sub- Saharan Africa (Kuhn, et al., 2018). Mostly, workers in this bracket often experience underemployment, low remuneration, least respected rights, inadequate social protection and a low voice in social dialogue (Kuhn, et al., 2018). Moreover, available statistics on health and safety at the workplace shows an alarming 2.78 million occupational death globally, 54,000 occupational death and 42 million work-related accidents in sub-Saharan Africa (GSS, 2015). The extent of poverty levels in Ghana is indisputably high though there has been a significant reduction from 51.7% in 1992 to 24.2% in 2013, leading to the achievement of MDG-1. Within the same period, Ghana attained lower-middle income status with per capita GDP of $1,820 in 2010 as well as registering 5.6% average growth rate over 31 years from 1983 to 2013. Empirically, it is found that Ghana’s growth has been less employment-oriented and that, on average, every 2% growth in GDP is associated with 1% growth in jobs (Baah-Boateng, 2016). 5 University of Ghana http://ugspace.ug.edu.gh Similarly, GSS (2018) indicated that every 1% change in growth is associated with 0.17 percentage decline in poverty between 2006 and 2013 but only resulted in 0.017 percentage change in poverty between 2013 to 2017. Many studies such as Selim (2006) focused on employment-oriented growth of countries and argue that countries should focus on employment dominating sectors such as agriculture in most developing countries. Nonetheless, the mere creation of employment is not enough since employment alone does not paint the true picture of the living conditions of people in a country. The use of unemployment to measure the extent of joblessness has been ineffective, particularly in developing countries since the unemployment rate does not depict the true picture of the working conditions of people. This is evidenced in Ghana, where the rural sector experiences the lowest unemployment rate of about 3.9% compared to 6.3% in the urban sector. However, GLSS6 report estimated that the rural population, which accounted for 50% of the country’s population contributed 78% of the country’s poverty rate. Although the agriculture sector is the highest employment sector contributing about 44.7% of total employment in 2013, it recorded the highest poverty rate of about 39.2% in the country which is higher than the unemployed poverty rate of 28.1% (GSS, 2014). It is estimated that about one out of every four workers is poor as of 2006, though working poverty in Ghana has drastically reduced to 11.54% as of 2017(Ackah & Baah-Boateng, 2012). In terms of social security, it is estimated that during old age, less than 30% of the employed persons benefit in their retirement. It is also estimated that 80.1% employed have no subsidized healthcare covering them in their workplace, the majority of which are engaged in the informal economy (GSS, 2014). Based on this, it is paradoxical for proponents to attribute mere employment creation as a robust tool in addressing the poverty situation of a country. It is 6 University of Ghana http://ugspace.ug.edu.gh obvious that decent work which is all-encompassing be used as both necessary and sufficient policy tool in eradicating poverty in a country. The ILO decent work agenda was first advocated for Ghana’s economic development and the country’s poverty reduction through Ghana Poverty Reduction Strategy (GPRS) programme in 2003. Ghana was used as a case study through the ILO’s pilot programme and cooperation to integrate decent work agenda into local economic development (Tijmstra, 2011). Nevertheless, there is scant research in the area of decent work in Ghana. Available studies have focused mostly on growth and employment creation, unemployment, inequality, poverty and gender. Baah-Boateng (2016) for instance, looked at the link between Ghana’s growth and job creation and how human capital development affects employment, while Abebrese (2014), on the other hand, argues for the need to enforce the adoption of decent work policy into the country’s laws. Heintz (2005) examined the linkages among employment, gender and poverty levels, while Baah-Boateng and Sparreboom (2011) accessed working poverty and trends of employment indicators and progress made in reducing decent work deficit in Ghana. There has been no quantitative and empirical study on Ghana that seeks to survey decent work at the individual level as a whole. In the literature, recent studies like Nizami (2013, 2017), Mackett (2017) and Moussa (2017) have explored determinants of decent work in developing countries. However, these studies failed to link decent work to the working poor. This study intends to close the gap in the existing literature and reinforce studies on decent work on Ghana, by looking at the determinants of decent work and investigate whether decent work has any implications for the working poor in Ghana. Thus, the study raises thought-provoking questions such as, (1) what are the determinants of decent work in the Ghanaian context? (2) does decent work have any implication on Ghanaian working poor? 7 University of Ghana http://ugspace.ug.edu.gh The Study Objectives The overall objective of the study is to empirically investigate factors that account for the probability of someone working decently in the Ghanaian economy and test whether the probability of working decently has any relationship with one’s poverty status. The study specifically seeks to 1. construct decent work index for Ghana; 2. identify the determinants of decent work in Ghana; and 3. investigate whether decent work is statistically significant in reducing working poverty in Ghana. Significance of the study The motivation in carrying out this study is first, to present Ghana in the global picture of the decent work agenda in the existing literature. Secondly, the study throws more light on elements of decent work in the Ghanaian labour market, “avoiding unemployment is not enough, an analysis of other forms of labour underutilization” (ILO, 2018 p.1). Labour underutilization is prevalent in the Ghanaian labour market. Estimates from GLSS VI in 2013 indicated that 4.2 million Ghanaians were time-related underemployed with 61.5% of agricultural workers being underemployed with about 47% of workers being underutilized. Out of this, females accounted for 51.5% with the male being 48.5% (GSS, 2014). The empirical results from the study will help to buttress existing knowledge to governments, development partners and policymakers in designing appropriate policy measures to curb indecent work within the country. Again, revealing the state of decent work in Ghana will help shape government policies towards employment creation that will enhance the well-being of citizens, eradicating vulnerability and poverty levels among the working poor. 8 University of Ghana http://ugspace.ug.edu.gh Data Source and Methodology The main data source for the study is the seventh round of the Ghana Living Standards Survey (GLSS VII). This is complemented by data from the Ghana Statistical Service, ILO, WDI of the World Bank. Ghana Living Standards Survey is a nationally representative survey collected by Ghana Statistical Services (GSS) periodically between 1987 and 2017. This dataset is used to track the impact of policies and programmes on an individual’s welfare in the country. The survey has the labour force and poverty modules, which contain elements of decent work that makes it suitable for this research. Composite indexes for decent work are constructed using 11 indicators of decent work. The logistic regression model is employed to analyze the determinants of decent work and investigate the relationship between decent work and the working poverty in Ghana. Organization of The Study The study consists of five chapters starting with the introduction where the problem statement, objectives of the study, brief methodology and data source, relevance and organization of the study are stated. Chapter two gives an overview of working poverty and decent work in Ghana, showing poverty dynamics in the country. This chapter also discusses elements of decent work in Ghana, taking gender into perspective. The theoretical underpinning of working poverty and empirical literature of decent work is reviewed in the third chapter, while the conceptual framework, methodology, empirical analysis and discussion of results are done in chapter four. Chapter five concludes with a summary of empirical findings, conclusion and policy implications from the study. 9 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO DECENT WORK AND WORKING POVERTY IN GHANA Introduction This chapter overviews the Ghanaian labour market with emphasis on decent work and working poverty, showing the trends in respect to gender and geographical status. Elements of decent work are well discussed, emphasizing employment-to-population ratio, unemployment, informality, youth not in education, employment or training, social security, participation in the labour through trade unions, earnings within occupation and industry. Brief Overview of the Ghanaian Economy Economic growth in Ghana has formally been spearheaded by the agricultural sector until the rebasing of the national account in 2010, which was necessitated by the structural changes in the Ghanaian economy. The exercise which took place in 2010 using 2006 as a base year was undertaken to improve the methodological and sources of data that is used in the computation of the national account. The new computation of the national account coupled with the discovery of oil in the country put Ghana on lower-middle income status in 2010 with per capita income of about $1,820 (Baah-Boateng, 2016). Since this exercise, sectoral contribution to GDP on the average has been led by the services sector, followed by the industrial sector and agricultural sector as of 2018. Ghana recorded non-oil and oil growth rate averaging 6.14% and 7.0% respectively, within the 11-year period from 2007 to 2017. Figure 2.1 shows the sectoral contributions to Ghana’s GDP, with the services sector leading with 52%, followed by the industrial sector recording 26% and lastly the agricultural sector accounting for 22% of the total GDP in 2017. 10 University of Ghana http://ugspace.ug.edu.gh Figure 2. 1: Sectoral Contributions to Ghana’s GDP (percentage) in 2017. 22% Agriculture 52% Services 26% Industry Source: Author’s construction from the National Account of Ghana Statistical Service (2018). The traditional dominance of agricultural contribution to total employment has also been overtaken by the services sector since 2015. As of 2017, the services sector contributed the largest to total employment of about 45.2% with the agriculture sector coming second with 40.6%, with manufacturing sector as third with 14.1%. Figure 2.2 presents the sectoral contributions to total employment from 1991 to 2017, showing upward movement for services and downward movement for agriculture sector even though the industrial sector contributions to total employment has averaged between 14% and 15% over the period (WDI, 2018). Figure 2. 2: Employment by Sector (percentage of total employment) from 1991-2017. Agriculture Industry Services 60 50 40 30 20 10 0 Source: Author’s Construction from the World Bank (WDI), 2018. 11 PERCENTAGE 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 University of Ghana http://ugspace.ug.edu.gh Incidence of Poverty in Ghana “Don’t ask me what poverty is because you have met it outside my house. Look at the house and count the number of holes. Look at my utensils and the clothes that am wearing. Look at everything and write what you see. What you see is poverty”. A poor man, Kenya 1997 (Narayan et al, 2000 p.26). According to Kofi Annan, the seventh Secretary-General of the United Nations, “extreme poverty anywhere is a threat to human security everywhere” the need to address this social canker has called for the attention of international bodies, and development partners to devote resources in helping to explore the changing dynamics, and to devise appropriate measurement and provide solutions to governments and international development agencies. According to ILO (2018) World Employment and Social Outlook report, though working poverty rate in Africa is reducing in terms of extreme poverty rate, as these are projected to have a downward movement to 31% in 2018. However, moderate poverty is expected to remain around 24% globally. United Nations through the implementation of the MDG1 has achieved a colossal feat by halving extreme poverty from 43% in 1990 to 21% in 2010 under the threshold $1.25 per day globally (ILO, 2018). Existing studies have revealed that poverty is a multifaceted social phenomenon, which emanates from many interlocking factors that need to be tackled under different approaches. Poverty can absolutely be defined as a situation where one is unable to meet his/her daily subsistence essentials of food, clothes and shelter (Todaro & Smith, 2015). The current threshold to determine extreme working poverty according to United Nations is when the share of workers living in a household with income or consumption less than $1.90 per day, and moderately poor when their consumption is above $1.90 per day but below $3.10 per day. That 12 University of Ghana http://ugspace.ug.edu.gh notwithstanding, individual countries have also come out with their own poverty lines deemed appropriate by the state authorities and this can be compared on parity levels against the global threshold. Figure 2.3 reports percentage of the population living below the income poverty line of 26 African countries including Ghana based on individual countries own poverty line estimates from 2006-2017. From the Figure, Ghana ranks second lowest having 24.2% of the country’s population living below its poverty line just behind Uganda which has the lowest of about 19.7% of its population living below the country’s income poverty line. Countries such as Zimbabwe, Madagascar, Burundi and DR Congo have the highest of their population living below their income poverty line of about 72.3%, 70.1%, 64.9% and 63.9% respectively. The Figure indicates that 16 countries out of the 26 countries have less than 50% of their population living below their poverty line and 10 of the countries living above their poverty line. Figure 2.3: Population Living Below National Income Poverty Line (%) in 26 African Countries 2006-2017. 80 60 40 64.9 63.9 70.7 72.3 36.640.1 46.746.5 27.829.848.4 57.157.1 24.236.1 50.746.128.744.539.146.752.9 55.1 54.4 20 28.2 19.7 0 Note: These estimates are based on countries own poverty line. All countries estimates are also computed from demographic and health surveys. Source: Computed from the Multidimensional Poverty Index of the Human Development Report 2018. In Ghana, the standard of living is measured based on consumption expenditure by an individual including food and nonfood (such as clothing and shelter) items, which then is used to estimate poverty levels in the country. Available statistics indicate a significant reduction of poverty levels in Ghana compared to most countries in the sub-region. Headcount poverty in 13 Angola Burki… Burundi Chad Congo DR… Egypt Ethiopia Gambia Ghana Kenya Lesotho Liberia Mada… Malawi Moza… Namibia Niger Rwanda Senegal Sierra… Tanzania Togo Uganda Zambia Zimba… University of Ghana http://ugspace.ug.edu.gh Ghana declined from 31.9% in 2005/06 to 24.2% in 2012/13 and slightly to 23.4% in 2016/17. Nonetheless, poverty gap declined from 11.0% in 2005/06 to 7.8% in 2012/13 but further rose to 12.4% in 2016/17 (see Figure 2.4). It means that poverty levels in the country somewhat increased by 400,000 people within 2012 and 2017according to GSS (2018) estimate. The rise in the total number of poor people living in the country could result from the increased in population growth and migration, and the widening in the poverty gap indicates more poor people in rural areas. Figure 2.4 presents incidence of poverty gap by locality. Figure 2. 4: Incidence of Poverty and Poverty Gap by Locality, 2005/06-2016/17(Poverty line=GHc1,314). 50 43.7 40 39.5 37.9 30 31.9 24.2 23.4 20 10 11 10.6 12.47.8 7.8 8.4 0 2005/06 2012/13 2016/17 Ghana(%) Urban(%) Rural(%) Poverty gap(%) Source: Author Construction from GLSS7 Poverty Report of Ghana Statistical Services. Incidence of poverty in Ghana is a rural phenomenon since the rural area's contribution to Ghana's poverty incidence is about 83.2%. This is explained by underemployment and the dominance of agricultural activities that prevail in the rural setting. At the regional level, the incidence of poverty is highly unevenly distributed across the country. The three northern regions (Upper West, Northern region and Upper East region) have the highest rate of poverty levels among the 10 regions in the country as they recorded 70.9%, 61.1% and 54.8% respectively. Contrary to this, the Greater Accra region recorded the lowest poverty incidence 14 University of Ghana http://ugspace.ug.edu.gh of about 2.5% and this is almost about 25 times the three regions average (GSS, 2018). This could be explained by the dominance of agriculture activities and the erratic rainfall pattern, and also the high concentration of most government businesses located in southern Ghana. One major worrying concern about the incidence of poverty in a country is how it is distributed between male and female-headed households. Statistics show that the incidence of poverty in Ghana is unevenly distributed among male-headed households as men constitute the bigger proportion of poverty in the country, that is, the proportion of male-headed household who are poor (living on less than GHc1,314 per year per adult equivalence) were 34.9%, 25.9% and 25.8% in 2005/06, 2012/13, and 2016/17 respectively which exceeds national levels and female-headed households in the same years as shown in figure 2.5 below. Figure 2. 5: Incidence of Poverty by Sex of Household Head from 2005/06-2016/17 (Poverty line=GHc1,314). Incidence of poverty by sex of the household head from 2005/06-2016/17 23.4 24.2 31.9 Female 17.619.1 22.1 Male 25.825.9 34.9 0 5 10 15 20 25 30 35 40 2016/17 2012/13 2005/06 Source: Author construction from GLSS7 report of Ghana Statistical Services (2018) Another crucial dimension where poverty can be scrutinized and which also forms the basis of this study is the individual employment status and how this influences their probability of being poor. Conventionally, taken the neo-classical theory of labour supply into consideration, one 15 Total Ghana(%) (%) University of Ghana http://ugspace.ug.edu.gh is likely to be poor if he/she has no job or work for fewer hours. However, all work does not pay the same remuneration and individuals’ income originates from many sources. Poverty by head-count is prevalent among self-employed agricultural workers in the country even more than the unemployed. Incidence of poverty is low among the retired, public employees and self-employed non-agriculture workers as depicted in Figure 2.6. The high poverty rate among self-employed agricultural workers is obvious in the sense that most people working in the sector experiences underemployment as indicated by a few hours of work, the lack of social and employment benefits. Again, the erratic nature of rainfall pattern in the northern part of Ghana as well as the seasonal nature of agricultural activities in the country reinforce the irregular income of those in the sector, hence contributing to high poverty levels in the sector. The low rate of poverty incidence among the retired and public employees explains the protection and enforcement of labour laws in the public sector. The availability of a retirement plan in the public sector employment provides support for employees in the public sector and thus accounting for their low rate of poverty incidence. Figure 2. 6: Incidence of Poverty by Employment Status of Household Head (Poverty line=GHc1,314). Poverty incidence by employment status(%) 60 42.7 40 29.4 26.8 20 11.44.8 8.9 1.6 0 Source: Author Construction from GLSS7 of Ghana Statistical Services 2018 report. 16 University of Ghana http://ugspace.ug.edu.gh The distribution of poverty among workers cut across all spectrum of work classifications. In respect to poverty distribution in employment status as indicated by Figure 2.7, working poverty is rampant among self-employed agriculture and non-agriculture workers without employees with 30.48% and 22.87% respectively. Meanwhile, poverty distribution is lowest within self-employed agriculture workers with employees and paid apprentice with 0.85% each. Figure 2. 7: Working Poverty by Status in Employment Other (Specify) Agric contributin family worker 16.81 Agric self-emp without emp 30.46 Agric-self-emp with emp 0.85 Non-agric contribut family worker 3.67 Non-agric self-emp without emp 22.87 Non-agric self-emp with emp 2.26 Paid apprentice 0.85 Casual worker 4.18 Paid employee 18.05 0 5 10 15 20 25 30 35 Source: Author’s construction from GLSS7 of GSS (2016/17) A look at poverty profile in Ghana clearly indicates that the incidence of poverty is geographically a rural phenomenon and highly concentrated in the northern part of Ghana. In terms of sex distribution, poverty is unevenly distributed towards male-headed households compare to female-headed households and similarly, the incidence of poverty is prevalent among agricultural headed households. Lastly, the incidence of poverty is lowest among household heads who have tertiary education level compared to 44.3% of no education which is the highest (GSS, 2018). 17 University of Ghana http://ugspace.ug.edu.gh Elements of Decent Work in Ghana The concept of decent work as conceived by the ILO in 1999, is to consolidate the objectives of full-employment, rights at work, social protection and representation in a harmonized, gender-sensitive and geographical distribution that forms the foundation for all countries’ workers. To achieve this purpose, the ILO through its 18th International Conference of Labour Statisticians in December 2008 has come out with a measurement framework for decent work indicators (DWI)1. The conceptual structure on the measurement of decent work encompasses 10 substantive elements, which are closely related to the four-core mandate of the ILO. These are the creation of full employment, fundamental principles and rights at work, social protection and social dialogue and representation. These 10 elements include: “opportunities for work, adequate earnings and productive work, decent working time, combining work, family and personal life, work that should be abolished, stability and security of work, equal treatment and opportunity at work, safe working place, social security, social dialogue and representations” (ILO, 2013). Statistical indicators recommended in measuring decent work are sub-categorized into main, additional and future indicators and some are assessed for Ghana in this literature under the 10 substantive decent work elements. 2.3.1 Employment Opportunities The first indicator of employment opportunities is the employment-to-population ratio. It measures the extent to which a country generates adequate employment for its working population. The size of this ratio generally indicates employment opportunities available in the 1 ILO compilation on the basis of the Discussion paper for the Tripartite Meeting of Experts on the Measurement of Decent Work (Geneva, 8 -10 September 2008). *Wording modified by ILO in the pilot phase; **Indicator added by ILO in the pilot phase (2009-2013). 18 University of Ghana http://ugspace.ug.edu.gh country, in terms of the number of workers. However, a bigger ratio does not necessarily depict decent work. Absence of unemployment benefits, educational opportunities and economic hardship may force people to engage in any menial job available in the informal sector. The average employment-to-population ratio for about one-and-half decade in the Ghanaian economy is estimated to be 68.39% in total, 71% and 66.15% for male and female respectively. Meanwhile, youth employment-to-population ratio averaged 41.72% in total, 42.58% and 41.02% for male and female youth respectively as reported by Figure 2.8a for the total labour force and Figure2.8b for youth in employment. Figure 2. 8: Employment to Population Ratio by Sex (Percentage) from 2000-2015. Figure2.8a Employment-to-population ratio, ages 15+ 100 80 60 40 20 0 2000 2006 2010 2013 2015 Total Male Female Figure2.8b Employment-to-population ratio, ages 15-24 60 40 20 0 2006 2010 2013 2015 Total Male Female Source: Author’s Construction from the World Bank (WDI) data 2018. The trend analysis for employment-to-population ratio above 15 years as depicted by figure 2.8a, increased marginally from 2000 to 2006 and marginally decreased from 2006 to 2010 but further surged from 2010 to 2013 and a sharply declined to 2015 for both male and female in 19 University of Ghana http://ugspace.ug.edu.gh the respective years. The average employment-to-population ratio within this period means that employment opportunities in Ghana are inadequate in terms of quantity to the total population. On gender perspective, low rate of the female-to-male ratio is mostly as a result of cultural factors, low levels of education and labour market discrimination that hinders females in securing employment (Baah-Boateng, 2009). The low youth employment-to-population ratio can also be explained by the lack of job opportunities for youth in the country. However, low youth employment-to-population ratio could also mean the existence of educational opportunities available that intend to keep more youth in education. In order to ascertain the underlining cause of low youth employment-to-population ratio, information on youth not in education, employment or training (NEET) is needed to buttress this. Another dimension of assessing employment opportunities negatively in a country is by looking at the prevailing unemployment rate. The unemployment rate highlights the rate of labour underutilization that exists in an economy. Unemployment measures the situation where individual labour is willing to work, available for work at a given pay and is actively searching but cannot find paid-job to do. This reflects countries inability to provide productive employment for its working population. The upward and downward movement in the unemployment rate as depicted by Figure 2.9 generally reflects the business cycle, where upward movement often coincides with recessionary periods and vice-versa. Global youth unemployment is estimated to be 64.8 million as of 2018 (ILO, 2019). The unemployment rate in Ghana has had a general decline from 2000 to 2013, though the country saw a little rise in 2010 especially the youth unemployment rate. Unemployment in the country witnessed a sharp rise in 2015 with an 20 University of Ghana http://ugspace.ug.edu.gh unemployment rate of 6.81% and male, female and youth being 7.04%, 6.6%, and 15.18% respectively, and this is higher than 15 years average from 2000 to 2015. On gender terms, the unemployment rate among males has been relatively lower compared to females’ rate with the exception of 2015. The youth unemployment rate has been the highest over the period. The high youth unemployment rate mostly results from lack of job experience which makes the adult cohort more preferable by employers, and this reinforces the lower employment-to-population ratio for the youth in Ghana as a clear indication that, there are inadequate employment opportunities for the youth in the country. Figure 2. 9: Unemployment Rate by Sex and Youth (percentage of ages 15+, 15-24 of labour force), 1998-2015 Unemployment rate 20 15 10 5 0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Total Male Female Youth Source: Author Construction from the World Bank (WDI) data, 2018. Low unemployment rate though is an ideal target for a country, yet low rate, especially in developing countries, could mean high labour underutilization resulting from the fact that non- existence of unemployment benefits and social protection, usually pave the way for the 21 University of Ghana http://ugspace.ug.edu.gh unemployed to seek refuge in the informal economy as the last resort for survival, and hence resulting in low rate of unemployment (Baah-Boateng, 2015). The alarming youth unemployment rate is supported by the proportion of youth, not in education, employment and training (NEET) about; 24.42% in 2006, 11.08% in 2013 and 25.51% in 2015. The NEET also captures discouraged youth workers and inactive youth who engaged in household non-economic activities. Moreover, the unemployment rate in Ghana is rampant among those with intermediate education levels of 11.87% average from 2006-2015, followed by those with advanced and basic education of 4.91% and 4.58% respectively. Informality is a growing phenomenon among developing countries, and it has many repercussions on job quality. Normally, the informal economy is characterized by the lack of basic legal, social and employment protection which facilitates high poverty rate, inequality and large decent working deficit gap. Although informal employment may be found in the formal sector, however, nearly all categories of informal sector employment are classified under informal employment and the measurement of which is the proportion of informal employment in terms of total employment (ILO, 2013). There are about two billion people globally who make their living in the informal economy (ILO, 2019). An estimate shows that in some parts of sub-Saharan Africa, the gender gap in informal employment exceeds 20% (Kuhn, et al., 2018). Informal employment in Ghana, excluding agricultural workers, have witnessed a declining trend from 87.11% in 2006 to 83.18 in 2015, and this has a similar trend for both sexes. The bigger proportion of female in informal employment of about 88% compared to 75.8% as presented in Table 2.2 is attributed to low levels of education of females vis-à-vis their male 22 University of Ghana http://ugspace.ug.edu.gh cohort in the country. Generally, the size of the informality gives the extent of vulnerable employment since workers in this bracket are excluded from labour protection laws. Also, a declining informal employment rate from 2006 to 2015 could be explained by the expansion in the services sector in terms of employment relative to the agricultural sector. Table 2.1: Informal Employment (Percentage of Non-Agricultural Employment) Year Total Male Female Gender gap 2006 87.11 79.02 93.56 14.54 2013 85.94 78.72 91.85 13.13 2015 83.18 75.88 88.32 12.44 Source: Author Compilation from the World Bank (WDI) data, 2018. 2.3.2 Adequate Earnings and Productive Work. As enshrined in the preamble of the ILO constitution, adequate pay for productive work is the utmost desire of every worker who seeks employment. It is no doubt that the first characteristic of a job every worker desire is the pay rate even though workers do not work solely for the pay rate, however, it is the main incentive to work as posited by the neoclassical theory of labour supply. The availability of jobs alone in an economy is not enough in determining decent work but needs to be complemented by the rate of pay which defines the consumption level of a worker (ILO, 2013). Some proposed ILO statistical indicators to measure adequate earnings and productive work include: the in-work poor rate, employees with a low pay rate of pay, mean hourly earnings in selected occupations, minimum wage as a percentage of the median wage, manufacturing wage index and employees with recent job training. The scarcity of data in Ghana renders the analysis 23 University of Ghana http://ugspace.ug.edu.gh of all these indicators impossible. However, Ghana Statistical Services, through its labour force and living standards surveys, provide information on some of these indicators. Working poverty rate is the main indicator for adequate earnings and productive work. It measures the number of poor people who are economically active or someone who despite being employed, yet lives in a poor household (Baah-Boateng, 2013). GLSS7 reports 42.7% of self-employed agricultural workers, 11.4% private employees, 4.8% of public employees live below the upper poverty line of the country which suggests the spread of working poverty rate in the country. In addition, adequate earnings and productive employment could be determined by the average hourly, weekly or monthly earnings by occupation group and industry. The monthly average total occupational earnings for Ghana in 2015 was estimated to be GHc898.65 higher than service and sales workers of GHc499.96, skilled agricultural, forestry and fishery workers of GHc525.04 and elementary occupations of GHc504.99 indicating low average earnings in such occupations. In terms of gender distribution, with the exception of managerial level, where the earning gap is about 8.02 percentages points in favour of women. Men on average earn more than female across all other occupations with some even thrice, such as craft and related trade workers and the total wage gap estimated to be 17.32%, though less than the global gap of 20%. The highest paid occupation in Ghana on average is managers while the lowest paid is the service and sales workers as illustrated by Figure 2.10a. Analyzing average monthly earnings by industry shows that employees in 10 out of 20 industries in Ghana, earn below the total average earnings in the industry as depicted by the red bar in Figure 2.10b. The highest average paid industry is electricity, gas, steam and air conditioning supply of about GHc2,196 succeeded by mining and quarrying of GHc1,463 as 24 University of Ghana http://ugspace.ug.edu.gh the second highest paid industry. Contrary to this is the real estate sector whose average total earning is about GHc351, just about two-and-half times that of the total industry average earnings. Figure 2. 10: Average Monthly Earnings of Paid Employees 15 Years and Above by Sex and Occupation. FIGURE2.10A. Average Total Earnings of Paid Employees by Occupat ion and Sex Both sexes Male Female Source: Author’s construction from Ghana Statistical Services 2015 Labour Force Survey. 2.3.3 Social Security and Social Dialogue. ILO defines social security as a fundamental human right and a means to foster social cohesion, human dignity and social justice (Anker, et al., 2002). This definition forms the basis of the ILO existence and it seeks to measure cash and kind benefits and social protection among others, ranging from inadequate remuneration resulting from maternity, disability, illness, 25 898.65 1,011.13 715.22 1,378.76 1,305.61 1,533.40 1,017.31 1,122.90 905.59 1,201.19 1,427.13 658.86 1,163.24 1,328.63 985.82 499.96 590.62 380.74 525.04 658.48 259.4 1,071.77 1,327.07 409.64 714.71 732.62 348.82 504.99 684.42 276.01 1,162.50 1,162.50 0 University of Ghana http://ugspace.ug.edu.gh injury from work, inadequate support from family, lack of credit facility, old age, lack of access to health, general poverty, social exclusion among others (ILO, 2013). As stated by Anker et al. (2002) that, the three main coverage of social protection that should not be overlooked is the population coverage in terms of access, entitlement and contributions, benefits levels in relative terms and total expenditures devoted to social protection. Absence of pension support to workers on their retirement age is a contributing factor to poverty among the aged. Nursing mothers without maternity leave are more likely to have a fall in their standard of living. Social protection coverage in most cases is not widespread in developing countries, particularly in Africa. ILO 2017-2019 global social protection reports that, despite effort made to extend social protection throughout the world, there still remains 55% about 4 billion people who are uncovered by at least one social cash benefit which is higher compared to Ghana where at least one benefit coverage is about 18.3% in 2016 (see Table 2.3, contingency1), slightly above the African continent rate of 17.8%. Mothers with newborns receiving maternity benefits are about 41.7% (see Table 2.3, contingency4) slightly higher than global levels of 41.1%. Meanwhile, unemployment benefit coverage in Ghana is zero compared to global levels of about 21.8% coverage. A significant improvement has been made on persons above retirement age receiving a pension in Ghana of about 33.3% higher than the continent rate of 29.6% but falls far below the global rate of 68%. Table 2.3 gives a summary indicator of social protection for Ghana. Table 2. 2: Share Contribution to Social Security and Proportion of Population Covered by Social Security Systems in Ghana. Share Contribution to Social Security (percentage) Year Total (%) The share of population above pensionable age receiving an old age pension 2000 5.2 26 University of Ghana http://ugspace.ug.edu.gh 2011 7.6 Active contributors to an old-age contributory pension scheme as a 2011 9 percentage of labour force. Active contributors to an old-age contributory pension scheme as a 2011 6.7 percentage of working age population. Proportion of population covered by social protection systems (%) Contingency (1): Population covered by at least one social protection 2016 18.3 benefit Contingency (2): Persons above retirement age receiving a pension 2016 33.3 Contingency (3): Unemployed receiving unemployment benefits 2016 Contingency (4): Mothers with newborns receiving maternity benefits 2016 41.7 Contingency (5): Children covered by social protection benefits 2016 5.6 Contingency (6): Vulnerable persons covered by social assistance 2016 3.3 Source: Author compilation from ILO Social Security Database for Ghana 2019. ILO defines social dialogue “to covers various types of information exchange (e.g. negotiation and consultation), between representatives of governments, employers and workers, on any issue of common interest, and includes both bipartite and tripartite mechanisms” (Ghai, 2006 p.175). Recommended decent work indicators for social dialogue, employers and workers representation includes trade union density rate (measure the extent of the exercise of freedom of association and is computed as the percentage of workers in a given reference group who are trade union members), employers’ organization density rate, collective bargaining coverage rate (the proportion of workers in employment conditions and benefits are determined by one or more collective agreements directly or indirectly), and days and work lost due to strikes and lockouts ( ILO, 2013). 27 University of Ghana http://ugspace.ug.edu.gh Two types of trade unions exist in Ghana currently, 1) The Trade Union Congress which consists of 18 subsidiary national unions with an estimated membership of half a million throughout the economy, the second is Ghana Federation of Labour (GFL) which also comprises of 9 subsidiary groups with an estimated membership of 48,000 mainly concentrated in manufacturing and trade sectors (TUC, 2012). Figure 2.11 gives a pictorial view of both trade union density and collective bargaining rate portraying a dwindling trend from 2008 to 2016 similar to global levels. Even though trade union density does not give full accounts of union negotiation power. Nonetheless, it shows the strength, intensity and organization of the unions and tells us workers outside the union membership. Trade union density in Ghana has declined from 20.1% in 2008 to 14.7% in 2016. The strength of unions as determined to some extent is the bargaining potency of trade unions, however, it does not tell us the outcome of negotiations. Collective bargaining process, on the other hand, describes the nature of industrial relation that exists in a country in relation to the structure of bargaining process, number of agreements reached, labour law, regulation and administrative process. Figure 2.11 indicates that about one-fifth of the total employed persons pay rate and conditions of service are determined by collective agreement processes in the country as of 2016. The continuous fall in collective bargaining coverage rate within the eight years period could be explained by the fall in government share in total employment and since government employees form the larger proportion of the collective agreement process, hence its reduction. The implication here is that workers outside the trade union have no formal medium through which they can negotiate with their employers. As a result of that, the likelihood for employers to exploit them is high. 28 University of Ghana http://ugspace.ug.edu.gh Figure 2. 11: Trade Unions Density and Collective Bargaining Coverage (Percentage) 2008-2016, Ghana. 100% 80% 20.1 19.6 18.9 18.1 17.4 16.6 15.9 15.3 14.7 60% 40% 24.3 23.9 23.5 22.9 22.4 21.8 21.4 21 20.6 20% 0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 Trade union density rate (%) Collective bargaining coverage rate (%) Source: Constructed by the Author from World Bank (WDI) 2018. 29 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE LITERATURE REVIEW 3.0 Introduction The previous chapter gave an overview of poverty and decent work in Ghana. The prime focus of this chapter is to survey the theoretical and conceptual underpinnings of working poverty and decent work. Specifically, reviewing various definitional issues of poverty and some proposed measurements of poverty. The chapter reviews working definitions and how work and poverty interact in conceptualizing working poverty. An empirical review of working poverty and decent work are also discussed in this chapter, with an emphasis on methodologies and findings from various empirical studies. 3.1 Working Poverty The subject of working poverty has become a topical issue in recent times among poverty and development scholars, as well as policymakers. The study of working poverty has been necessitated by extreme deprivation and homelessness in some developed countries, the problems of informality in developing countries, which make workers fate indeterminate as to whether they would make it the next day (Crettaz, 2011), and the conundrum of achieving simultaneous employment quality and quantity for workers in developed and developing countries alike. The attempt to investigate the concept of working poverty raises two paramount concepts that need to be addressed. The first component is the definitional issues and measurement of poverty, and the second part is to explain what is work, and how these two concepts interact in other to set a threshold with regard to the amount of work performed (Crettaz, 2011, 2013; Crettaz & Bonoli, 2011; Joassart-Marcelli, 2005; Kusi, 2018). 30 University of Ghana http://ugspace.ug.edu.gh 3.1.1 Conceptual Definitions of Poverty The problem of poverty eradication has existed since time immemorial. The concept has evolved over time and space, dating as far back as the seventeenth century in reference to the “Poor Law Relief Act 1601” popularly known as Elizabethan Poor Law. Earlier scholars such as Adam Smith, Karl Marx and Joseph Rowntree, proposed some definitions to the concept. In spite of that, the adjectival description in terms of measurement in ‘relative’ and ‘absolute’ sense has created mixed feelings among contemporary economists unable to reconcile and settle on one standardized poverty definition. Contemporary economists such as Joseph Rowntree Foundation (JRF, 2013) define poverty as insufficiency of material resources of an individual, which place them below a daily minimum needs and prevent them from social participation (Sanchez-Martinez & Davis, 2014). Bradshaw (2006) postulates that poverty in general sense is simply lack of basic human needs ranging from food, shelter, medical care and safety that form the prime component of human dignity. Amartya-Sen, though blackened the biological approach to poverty developed by Rowntree (1901) who defined poverty as inadequate earnings to meet subsistence level of living, but emphasized that, it (the biological approach) should be redefined to incorporate the irreducible core of absolute deprivation to encompass the issues of starvation and hunger at the center of the concept. Nonetheless, Sen (1982) supported the sociological view of deprivation and acknowledged the shift in focus particularly from absolute to relative deprivation (Sen, 1982). A recent definition proposed by Ravallion (2017) said, “poverty exist in a given society when one or more persons do not attain the level of economic well-being deemed to constitute a reasonable minimum by the standards of that society”. This definition is in line with the World 31 University of Ghana http://ugspace.ug.edu.gh Bank poverty line definition, which describes a person as poor if his or her minimum consumption falls below a certain minimum threshold to meet basic needs. This definition, despite its wider usage, is still inadequate since what it refers to as “well-being deemed to constitute a reasonable minimum by the standards of that society” is vague and generally determined by a section of the society usually those in authorities which are influenced by their self-interest and personal understanding, but not the poor. Narayan et al., (2000) posit that poverty is multidimensional as their research “voice of the poor” findings indicated that, poverty emanates from several sources to make it multi-complex social phenomena, the inadequacy of material resources deprived the poor as demonstrated by their psychological behaviour. While many people were indifferent with regards to schooling, infrastructural development was unanimously a critical issue to the people, and lastly, poor people prefer asset and attributed their vulnerability and risk exposure to the lack of physical, human, social and environmental asset. They also found that while men in Ghana from both rural and urban areas associated poverty to lack of material wealth, women, on the other hand, linked poverty to lack of food security (Narayan, 2000). Though different scholars see the concept of poverty from different perspectives, it has many commonalities ranging from lack of material well-being, social participation, infrastructure, and education. 3.1.2 Measurement of Poverty Defining the concept of poverty though necessary but not sufficient in addressing the poverty situation in a country. Poverty eradication requires one to figure out the root cause of the phenomenon, and design credible measurement to ascertain the intensity so that policymakers and development actors can address the situation. According to Ringen (1988), poverty can be measured directly in terms of consumption and indirectly in terms of income and revealed that 32 University of Ghana http://ugspace.ug.edu.gh recent studies on poverty combine direct definition and indirect measurement and this, therefore, makes statistics produced invalid. Ravallion (2017) argued that the essence of measuring poverty is not merely computing a number for a place and date rather enhance comparability within and across countries (Ravallion, 2017) . Goedhart et al. (1977) proposed a poverty line in their research where family heads were quizzed on how much their family considers as a minimum level of income. They found that the higher the amount specified by the respondents, the greater the actual income and family size, thus the poverty line is defined to vary with family size. In settling on a particular poverty line definition, it was found to be problematic in situations where politicians, representative citizens, and experts called to determine the poverty line. All because these people earn far above the minimum subsistence income, and their judgement and perception may misinform the true status of the poor, but rather, reflect their own social background (Goedhart, et al., 1977). Another widely used welfare measurement is the poverty line propounded by the World Bank (Ravallion, 1988) in living standard measurement study (LSMS). This poverty line is defined as “the monetary cost to a given person, at a given place and time, of a reference level of welfare” and considers individuals who do not meet this criterion as poor and non-poor for those above it. According to Ravallion (1988), this ‘line’ helps to create the awareness of governments and civil societies to implement policies that serve to fine-tune living standards of the poor and more importantly serving dichotomous purpose of 1) determining the minimum level of living needed to restore the poor to non-poor status, 2) to ensure interpersonal comparison. Atkinson (1987) contended that the choice of measuring poverty has remained contentious on the subject of an extensive literature and diverse measurement has been 33 University of Ghana http://ugspace.ug.edu.gh proposed. In terms of measuring poverty, absolute poverty can be distinguished from relative poverty, whereby the former is fixed over time and preferable for the purpose of informing anti-poverty policies. However, the latter measures poverty based on average expenditure or rest on mean or median income of households and also has the advantage of international comparison (Crettaz, 2011; Ravallion, 1998). The most renowned and widely acceptable measure of the incidence, depth and severity of poverty is the headcount ratio and poverty gap indicator of absolute poverty. The headcount ratio measures poverty incidence by dividing the number of poor persons over the size of the population. The headcount poverty index is estimated by equation (3.1) below: 𝑁𝑝 𝑃0 = (3.1) 𝑁 Where 𝑁𝑝 measures the number of poor and N is the sample total population. This indicator of poverty incidence is simple in computation and easily interpreted, it suffers from major shortcomings in respect of the fact that the index does not tell how intense poverty levels are. Moreover, the headcount ratio is insensitive to changes in the number of poor people and therefore does not tell how severe poverty levels are among the poor. Lastly, this indicator is also chastised for the fact that it should be applied to individuals and not households. The problem of intensity among the poor is addressed by the poverty gap index which measures the depth of poverty by showing how disadvantaged households are, and also how much should in case, be transferred from the rich or by the government to uplift such households above the poverty line and this is given by equation (3.2) 1 𝑁 pg = ∑ 𝑃 (𝑝𝑙−𝑥𝑖) 𝑖=1 (3.2) 𝑁𝑝 𝑝𝑙 34 University of Ghana http://ugspace.ug.edu.gh Where 𝑁𝑝 represents the number of individuals who are poor, pl being the poverty line and 𝑥𝑖 denoting the i-th observation of disposable income among the population who are poor. This indicator though does improve the headcount ratio, it fails to reveal the severity or inequality among the poor and the squared poverty gap given by equation (3.3) is used to rectify this. 1 𝑁𝑝 (𝑝𝑙−𝑥 spg= ∑ [ 𝑖] ^2 (3.3) 𝑁𝑝 𝑖=1 𝑝𝑙 This index is also referred to as one of the family measures of Foster, Greer and Thorbecke (FGT) used in meta-analysis and allows for various aspect of poverty specified as: 1 𝑁𝑝 (𝑝𝑙−𝑥𝑖) FGT= ∑ [ ] ^𝛼 (3.4) 𝑁𝑝 𝑖=1 𝑝𝑙 Where 𝛼 measures the elasticity of index to poverty, pl denotes poverty line and other variables defined as in (3.2). When α=0, the formulae are reduced to headcount ratio, when α=1 and 2, it makes FGT the poverty gap and squared poverty gap respectively, and this makes the FGT a measure of severity of poverty. The advantage of FGT is that it allows for disaggregation of the population into sub-groups so that the contribution of each sub-group to national poverty can be estimated. However, the choice of α value is arbitrary (Crettaz, 2011; Haughton & Khandker, 2005). Sen (1976) proposed a poverty measurement that captures the effects of the number of poor, the depth of poverty, and how they are distributed within a group known as Sen index. The index is expressed as the average of the headcount and poverty gap measures weighted by the Gini coefficient of the poor, giving as: 𝑃𝑠 = 𝑃 𝑃 𝑃 0𝐺 + 𝑃1(1 − 𝐺 ) (3.5) Where 𝑃0 measures the headcount index, 𝐺 𝑃 is the Gini coefficient of inequality among the poor and lies between 0 and 1. This index though widely discussed, is limited in the application 35 University of Ghana http://ugspace.ug.edu.gh all because it lacks the intuitive appeal and simplicity compared to other measures (Khandker and Haughton, 2005). Sen index has been modified by some scholars and the most appealing among them is the Sen-Shorrocks-Thon (SST) index which is the product of the headcount index, the poverty gap index and a term with the Gini coefficient of the poverty gap ratios defined as: 𝑃𝑆𝑆𝑇 = 𝑃0𝑃 𝑃(1 + ?̂?𝑃1 ) (3.6) Where 𝑃0 measures the headcount ratio, 𝑃 𝑃 1 is the poverty gap for the poor only, and the ?̂? 𝑃 represent the Gini index for poverty gaps of the entire population. This indicator allows one to disaggregate poverty into three components that raises the question: how many poor are they? How is the poor poorer? And is there exist inequality among the poor? (Khandker and Haughton, 2005). 3.1.3 Theory of Work In a sociological context, work can either be paid or unpaid, public or private, and formal or informal (Eichler & Matthews, 2004; Taylor, 2004). The concept of work in economics perspective theoretically can be traced from the neoclassical theory of labour supply, and labour demand from production theory. Thus, employment is technically defined as the interaction between demand and supply of labour. Examples of demand-side factors include the wage rate and productivity of the worker. Supply-side factors are factors which enables the worker more desirable to be employed. These include the worker’s education, age, experience and other related personal characteristics (Ackah & Baah-Boateng, 2012; Islam, 2004; Rodgers, 2002). 36 University of Ghana http://ugspace.ug.edu.gh According to ILO (2013), an individual is said to be employed if during a reference period, normally a week, was in paid employment, self-employment or temporarily unavailable due to reasons such as maternity or paternity leave, illness, holidays, training or/and industrial disputes. This definition expatiates the concept of work to be an activity performed by an individual during a reference period (at least for an hour within a week) for pay either in cash or in-kind, profit or family gains (ILO, 2013). The central theme of this definition fixed on payment is its weakness, since it does not regard housework activities such as cooking for one’s own family or caring for children or the aged including voluntary services as work for lack of payment. Crettaz (2013) put together official ‘work’ definitions from various sources by different countries. In the E.U., an individual is in ‘work’ if he is employed at least 7 months in the reference year, 6 months employment in a reference year (either working or searching for a job) is referred to as work in France and Belgium. Meanwhile, in the U.S. and Switzerland, total family hours are put together to determine whether the family is in work or not, and in the U.S., family in work should have more than 1750 hours or 44 weeks a year. While in Switzerland, total family hours per week should be at least 36 hours. In Canada, all workers should have at least 50% of their income coming from wages, salary or self-employment, while in Australia, any active person is considered a worker. These work definitions raise the issue of full-time and part-time work, and the definitions above reflect full-time work based on countries definitions, otherwise, individuals are said to be part- time workers. The ILO specify 40-48 hours per week to be full-time work, at least 31 hours per week in the U.K. (Cooke & Lawton, 2008). Although working below this threshold could mean either the individual is time-related underemployed or in part-time work, however, working 37 University of Ghana http://ugspace.ug.edu.gh beyond 48 hours per week is considered by the ILO as indecent work as it may affect an individual’s health (ILO, 2013). A person or mostly family (since poverty is measured at the household level) is said to be ‘working poor, or ‘in-work poor’ if that individual has a work or lives in a household where the main earner is working but lives below the poverty line deemed appropriate by that society (Cooke & Lawton, 2008; Crettaz, 2011, 2013; Heintz, 2005; Joassart-Marcelli, 2005; Kusi, 2018; Maleček & Čermáková, 2015; Slack, 2010). Many scholars have emphatically, stressed that the lack of specific theoretic and conceptual models, and normally the type of definition used, could affect the findings (Joassart-Marcelli, 2005; Crettaz, 2011, 2013). Working poverty evolves from several factors such as economic, socio-demographic, and cultural factors. The premier among them, and unanimously agreed by all scholars is inadequate income from work as the main cause of working poverty (Cooke & Lawton, 2008; Crettaz, 2013; Danziger & Gottschalk, 1986; Fleury, 2007; Joassart-Marcelli, 2005). Another closely related work characteristics that were also found to cause working poverty are workers in services, workers not in a union, workers in small companies, and self-employment (Joassart-Marcelli, 2005) in southern California, Brady et al. (2013) in the U.S., and Crettaz (2013) in Europe and the United States. Scholars such as Slack (2010), and Lawton and Cooke (2008), have argued that increasing the minimum wage will suffice in curbing working poverty rates. On the contrary, evidence indicated that increased in the federal minimum wage in the U.S. from the period 2003 and 2007 had no significant impact on the working poor, but rather exacerbated unemployment levels among the working poor (Ray, 2010; Sabia & Burkhauser, 2010). 38 University of Ghana http://ugspace.ug.edu.gh Socio-demographic factors such as education, sex, age, race, marital status, family size, women with children, and parental status also influence working poverty. Evidence shows that working families with younger children, single mothers with children and women generally are more likely to be working poor. Thus, women with younger children are unable to participate in the labour market. Moreover, women generally have low education and normally faces labour market discrimination that could also affect their choice of getting decent jobs. Joassart- Marcelli, (2005) in California, Ferguson (2005) and Danziger & Gottschalk (1986) in the U.S.A., Malecek & Cermakova (2015) in the Czech Republic and Heintz (2005) and Kusi (2018) in Ghana. In addition, education, skills and training are also other individuals’ characteristics that were found to influence working poverty. That is, people with higher education are more likely to earn enough income that places them above the poverty line compared to those with low education (Malecek and Cermakova, 2015; Joassart-Marcelli, 2005). With regard to age, race and geographical locations, papers such as JRF (2014), Crettaz (2013) Joassart-Marcelli (2005) found young workers likely to be poor due to inexperience in the job market, rural dwellers are more likely to be working poor (Kusi, 2018), while Damaske et al. (2016) observed a dwindling racial working poverty disparities between the whites and non- whites in the U.S., in terms of geographical distributions, Slack (2010) found working poverty to be high in non-metro households, though observe a convergence of working poverty rate between metro and non-metro households due to the rising informality in the metro areas. 3.2 Decent Work As described by Somovia (1999), the decent work agenda is to ensure productive employment for men and women alike, and serves to secure the rights of workers, allow workers 39 University of Ghana http://ugspace.ug.edu.gh participation in social dialogue and provides appropriate social security at the time individual worker is not in employment either on retirement, leave, injury and/or even unemployed. Ban Ki-moon, described the sustainable development as one of the most comprehensive development agenda ever existed; and decent work has been incorporated into the Millennium Development Goals to be achieved by all countries by 2030 (UNDP, 2015). The concept of decent work seeks to achieve workers’ well-being before, during and after work (retirement) as perceived by some scholars as to the restatement of the ILO’s traditional objective in a more comprehensive manner which also defines the multidimensional nature of the concept of decent work (Nizami, 2017). There are numerous synonyms that seem to express decent work concepts such as good work, employment quality and job satisfaction and therefore referred to them as “family of names”. Some scholars believed that decent work concept lacks theoretical foundation since ILO did not include the opinions of academic scholars in the conceptualization of the concepts due to its political foundation. Moreover, the title ‘employment quality’ is extensively highlighted in the European literature instead of ‘decent work’, (Burchell, et al., 2013). Nevertheless, Nizami (2017) argued that the concept of ‘decent work’ is more extensive than ‘employment quality’ since decent work ensures the aspiration of the worker. The concepts of decent work, models, indicators and measurement are highlighted by Ghai (2002, 2006), Anker et al. (2002), Nizami (2017), and ILO (1999, 2013). According to Ghai (2002), the conceptual framework of decent work as stipulated by the Director-General Juan Somovia is based on four key elements. These four elements include productive employment for all men and women, protection of rights of all workers, social protection, and social dialogue, on these four substantive elements are further broken down into eleven major 40 University of Ghana http://ugspace.ug.edu.gh indicators which form the conceptual framework of decent work. These are made up of opportunities for employment creation, work that provides enough income and enhance productivity, adequate working hours which do not negatively hamper workers health and enables individual worker to have enough time for their family and personal life, stable and secured work, unwanted work such as forced and child labour, adhering to zero employment discrimination, occupational safety, social protection as well as social dialogue and workers representation (Anker, et al., 2003; Ghai, 2003; ILO, 2013). Decent work can be studied at the macro level by considering factors such as laws and regulatory institutions that serve to create a conducive and security atmosphere for businesses to thrive, individual workers, to have productive work and adequate earnings. At the firm level, it describes the workplace safety and protective mechanisms for workers, while the individual level of decent work ensures productive work for all forms of security-related issues that enhance individual wellbeing (Bonnet, et al., 2003; Nizami, 2017). Most studies in Europe have focused on job quality since the launching of ‘Lisbon Strategy’ in 2000. This aspect of job quality also measures the dimensions of job characteristics, which ensures workers wellbeing and economic growth. Job characteristics such as earnings, the security of work, social protection, participation in decision-making and others (Findlay, et al., 2017; Green, et al., 2013; Said, 2012; Leschke and Watt, 2008). This has also been emphasized by Piasna et al. (2013) as the term ‘job quality’ being researched than the concept ‘decent work’, albeit CIPD (2013) report suggested that job quality and its dimensions are better described by the term decent work (ILO, 2013) . 41 University of Ghana http://ugspace.ug.edu.gh ILO decent work agenda to reduce poverty primarily hinges on equitable distribution of resources between men and women mostly in the areas of quantity and quality of work. Research by the ILO has focused on country profile studies of developing countries such as Sri Lanka, Philippines, Brazil, South Africa, Ghana, Ethiopia, Moldova, Vietnam, and India (ILO, 2017 and 2018)2. These studies largely emphasized the implementation and integration of decent work agenda into the national plan, and do not spell out clearly, what really, are the factors that determine an individual’s possibility of working decently. Moreover, the ILO country profile studies also highlight decent work on the national levels, that is, country profile studies do not give a vivid picture of individuals and firms level decent work, which makes it difficult for one to tell the intensity of decent work among individual workers and within firms in a particular country. 3.3 Determinants of Decent Work Earlier studies such as Ghai (2003, 2006) undertook studies of decent work at the macro level among 22 developed countries. However, few empirical studies in the literature have attempted to undertake studies related to determinants of decent work at the individual levels such as Moussa (2017) in Ivory Coast, Mackett (2017) in Gauteng City-Region in South Africa, while Pautz and Stuart (2016) have investigated what individuals in Scotland perceived as decent work. At the firm (meso) level, Nizami (2017) looked at decent work levels in the IT industry in India. These studies were made possible by computing decent work indices, although the methodological approach adopted by these scholars in the computation of decent work indices vary across studies. Ghai (2003) for instance, developed decent work indexes for 22 OECD countries with homogeneous features using a simple methodology, where countries were given 2 See also ILO (2006, 2011, 2013a, 2013b, 2013c, 2014, 2016) 42 University of Ghana http://ugspace.ug.edu.gh a score of 1 as being the best and 22 as the worst for each indicator3 present. The scores of each individual component were then averaged with equal weight, and normalized between 0 to 1 to determine countries with highest decent work index. This approach of studying decent work had been maligned due to the fact that it does not reveal the state of decent work among individual workers in a particular country. Secondly, this approach can only be replicated for countries with similar characteristics and therefore not appropriate for global comparison. Due to the defects associated with the macro level studies, Mackett (2017) used quality of life survey in the Gauteng region, which has in it the decent work index. In this survey, questions on eleven elements of decent work were asked, and the respondent was coded 1 if the element is present to him/her otherwise zero. The index was computed by summing up the response of the respondent of the 11 indicators, where 0 meant that the individual has no decent work, and 11 meant that the individual has full decent work. Again, the scores were subcategorized into three levels of decent work, that is, an individual has low, medium and high decent work if he/she scored between 0 and 4, 5 and 8, and 9 to 11 respectively. Ordinal logistic regression was employed to analyze the determinants of decent work. This approach of computing decent work indexes at the individual level though is an improvement on the macro level studies. The choice of weight and decomposition of the index into low, medium and high decent work is based on the researcher’s judgement. Hence, there is no reference point for international comparison unless the same indicators and range are used. Moreover, the geographical location of the study (Gauteng City-Region) raises doubt in generalizing the results to the entire South African country due to the fact that, Gauteng is the 3 The indicators are; employment opportunities, remunerative employment, social security, forced and child labour, discrimination at work, freedom of association, collective bargaining, economic freedom, and participation at the national level (Ghai, 2003). 43 University of Ghana http://ugspace.ug.edu.gh most developed region in the country which has different working conditions compared to other less developed regions in the country. The dimensional nature of decent work has made some researchers developed a composite index at the firm (meso) and individual (micro) levels. Nizami (2017) developed a composite index for the I.T industry in India, where decent work index was created for 9 indicators using the UNDP normalization principle. Multinomial logistic regression, correspondence and cluster analyses were used to identify variables that affect decent work. However, the development of the composite index here again also suffers from the researcher’s judgment of the weighting choice. For this reason, Moussa (2017) adopted a statistical approach (multiple correspondence analysis), which has the ability to control for the researcher’s judgment of weighting choice, and a hierarchical classification to group workers into decent and non-decent working groups in Ivory Coast. Logistic regression was then used to investigate the determinants of decent work. There have been different approaches in the computation of decent work indices among different scholars, that notwithstanding, studies have shown similar findings indicating low levels of decent work among developing countries. Major findings evolved from these empirical studies revealed that, about 5.08% workers in Ivory Coast, work decently (Moussa, 2017), 6% workers in the I.T industry in India were found to have decent work (Nizami, 2017), while 10.88% workers in Gauteng region in South Africa have a high decent work (Moussa, 2017). The main determinants of decent work were individuals social and job characteristics such as age, sex, education background, social class, marital status, migration, race, geographical 44 University of Ghana http://ugspace.ug.edu.gh location, occupation, employment type and institutional sector. Nizami (2017) and Mackett (2017) found that males were more likely to have high decent work relative to females in India and Gauteng region, even though Moussa (2017) observed no statistical significance in terms of decent work between men and women in Ivory Coast. Job quality also increases as one move from rural to urban areas, indicating a high decent work in cities found by Nizami (2017) and Moussa (2017). There was also a direct association between decent work and age (Moussa, 2017; Nizami, 2017, Mackett, 2017), except that in India, it was observed that decent work increases with age up to 30 years and then falls as the individual worker ages. The race variable was found to be significant in South Africa and non-significant in Ivory Coast, and this could possibly be the number of white people who reside in the two countries. With regards to the institutional sector, public administration, informal and formal private sector had high decent work respectively. The ability to migrate, full-time employment, and working in the mining sector also influenced high decent work (Moussa, 2017; Mackett, 2017). The above discussions clearly show that the proposition that, decent work as a tool to eradicate poverty, specifically the working poor, has not been investigated in the existing literature. This study adopts the methodology employed by Mackett (2017), to compute composite decent work indexes for Ghana and investigate whether there exists a statistically significant relationship between decent work and working poverty, and investigate factors that influence the probability of working decently in Ghana. 45 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR METHODOLOGY AND EMPIRICAL ANALYSIS 4.0 Introduction This section consists of two major sub-sections. Section one describes the conceptual framework and methodology employed in the analysis of this work. The main methodology consists of using addictive binary response coding to construct composite decent work indexes at the individual level using GLSS7 dataset. The logistic regression is used to investigate factors that determine decent work in Ghana and analyze the relationship between decent work and working poverty. Section two uses descriptive and inferential statistics to analyze and discusses the empirical results. 4.1 Conceptual Framework Theoretical underpinnings for work (employment) and poverty have been studied through the theory of growth-employment-poverty linkage, in which income from work allows workers to obtain basic goods and services for themselves and their families (Ackah & Baah-Boateng, 2012; Aryeetey & Baah-Boateng, 2015; Islam, 2004; Osmani, 2003). The concept of decent work, unlike the ‘human development’ which has its basis rooted in theoretical literature such as “Sen’s theory of capability approach”. Decent work, on the other hand, emerged from the ILO's institutional literature and do not have any theoretical grounding in any of the social sciences. Decent work did not challenge any existing theoretical models in labour studies that form the basis of development thinking, and issues grounded in ethical philosophy (Ramos and Acosta, 2006; Standing, 2008) cited in (Sehnbruch, et al., 2015). Standing (2008) referred decent work to has been borne out of a vacuum, and hence, its limitation in academic literature (Sehnbruch, et al., 2015). However, some studies have 46 University of Ghana http://ugspace.ug.edu.gh conceptualized decent work and poverty reduction (Bell & Newitt, 2010; ILO, 1999; Rodgers, 2002; Saget, 2001). As noted by Rodgers (2002), there is a need to study the link between decent work and poverty- reducing effects. Decent work, like poverty, has a threshold and relate to economic and social factors with multidimensions. The association between these two concepts expatiate below. Firstly, the poverty level is determined against a threshold set by society of a given minimum consumption below which an individual should not fall. Similarly, decent work is set against a floor which encompasses universal rights and principles. The value of this universality is that decent work is meant for all, particularly ensuring gender equality in the labour market. Decent work deficit, therefore, means that there is a lack of social deprivation of living conditions, access to education and public services Rodgers (2002). The linkage between decent work and working poverty can be analyzed based on the four fundamental pillars of decent work (Rodgers, 2002). The first step in eradicating working poverty through decent work is full employment for all men and women. Ensuring full and productive employment means there are jobs for all who want to work in a country. The job should be able to provide training and skills which will enable the individual to be more productive and receive an adequate payment that will enable them to meet the basic standard of living. Providing employment, though necessary but not sufficient in addressing the poverty situation among workers in a country. That is, after ensuring the availability of employment, individuals should have the right to choose the type of work they deemed appropriate and serves to fulfil their aspirations. Also, individual workers should be free from discrimination on grounds of 47 University of Ghana http://ugspace.ug.edu.gh sex in entering the labour market (Ghai, 2006). There should be equal treatment for both men and women in the job market with respect to earnings and opportunity for career development. This will enable a substantial reduction of poverty, particularly among the females since they are the largest victims of global poverty (Rodgers, 2002). Moreso, guaranteeing rights at work means individual workers especially those at the margin (the poor mostly women) will have the liberty to form associations that will empower them to articulate their grievances and interests in collective decision-making at the national level which affect their lives and enhances their welfare. Also, within the working environment, workers participation in decision making give them a sense of belongingness which uplifts their psychological well-being and influence productivity. Lastly, decent work agenda through social protection is an effective approach in curbing poverty among the workers. Social protection to reduce poverty takes the form of providing social amenities such as education, health, insurance, pension schemes, and compensation to the individual workers before, during, and after participation in the labour market (ILO, 2008). 4.1.1 Measurement of Decent Work Decent work encompasses 10 elements in measuring the concept based on the four pillars of decent work. The multi-dimensional nature of the concepts of decent work renders it appropriate to employ a composite index in its measurement. As it has been highlighted in the previous chapters, some researchers have utilized composite indices to quantify decent work (Greyling, 2013; Mackett, 2017; Moussa, 2017; Nizami, 2017). The use of the composite index developed by the OECD (2008) was to enhance computation of the composite indexes for the purpose of comparison within and among countries. The use of a composite index is to 48 University of Ghana http://ugspace.ug.edu.gh aggregate dimensions of a complex variable into a single indicator, which makes it simple to interpret as compared to individual indicators with multiple trends. Moreover, the composite index allows benchmarking for inter-countries comparison and to access the progress of countries on a particular variable. Although the composite index has many strengths, it could be misleading for public policy purposes, if it is poorly constructed (Bonnet, et al., 2003; OECD, 2008). 4.1.2 Construction of Composite Decent Work Index This study uses inductive logic to construct composite decent work indices for Ghana. The GLSS7 has decent work index built into the survey, where individual workers were quizzed on some decent work elements. A respondent who gave a positive response was coded 1 and classified as having decent work and 0 as indecent work. For instance, ‘yes’ response to social security means decent work and is coded 1, however, yes response to suffered job injury means indecent work and is coded 0. Equal weight was given to each indicator since there is no criterion to determine which of the indicator is superior to others. Eleven (11) indicators of decent work based on the four pillars of decent work were adopted, and they are: 1) seek to change job, 2) sick/maternity leave, 3) regular pay, 4) working hours, 5) paid holidays, 6) contractual agreement, 7) medical care, 8) occupational injury, 9) occupational hazard, 10) health insurance coverage, and 11) social security. The response to these indicators is then added to give the individual decent work. Thus, individual scores 0 if he/she gave a negative response to all the indicators and scored 11 if he/she gave a positive response to all the indicators. The composite indices are then created by arbitrarily ordering the scores from (0-4) as low decent work, (5-8) as medium and (9-11) as high decent work. 49 University of Ghana http://ugspace.ug.edu.gh 4.1.3 Theoretical Model The determinants of decent work are analyzed by modelling decent work as a function of individual demographic and economic factors that influence decent work following Mackett (2017) and Moussa (2017). The equation is specified as: 𝐷𝑊𝐼 = 𝑋′𝑖 𝑖𝛽 + 𝜀 ………………….………………………………………………. (4.1) Where 𝐷𝑊𝐼𝑖 measures the individual decent work status, 𝑋 ′ 𝑖 measure the vector of individual economic and demographic characteristics (sector of employment, education, age, sex, industry, staus in employment, type of employment, location), and 𝛽 is their corresponding coefficients, while 𝜀 captures the error term. Secondly, it is argued that there is a lack of theoretical underpinnings of decent work and working poverty (Sehnbruch et al. 2015). The study adopts growth-employment-poverty linkages approach used by Islam (2004), to model the relationship between poverty and employment. Poverty is a function of employment-related factors and individual characteristics given by equation (4.2) 𝑃𝑂𝑉 = 𝑓(𝐸𝑀𝑃𝑖, 𝑍𝑖)…………………………………………………………..…………. (4.2) Where 𝑃𝑂𝑉 measures headcount poverty, 𝐸𝑀𝑃𝑖 measure employment-related factors and 𝑍𝑖 measure the individual’s related characteristics (Islam, 2004). Equation (4.2) is modified by replacing employment-related factors with decent work, which is a comprehensive measure of employment and its related characteristics (see Burchell et al. 2013, Rodger, 2002; ILO, 1999) to give equation (4.2). 𝑊𝑃𝑖 = 𝑓(𝐷𝑊𝐼𝑖, 𝑍𝑖)…………………………………………………….………………… (4.3) Where 𝑊𝑃𝑖 measure the working poverty status 50 University of Ghana http://ugspace.ug.edu.gh 𝐷𝑊𝐼𝑖 measure individual decent work status 𝑍𝑖 measure individual supply-side characteristics such as education, sex, location, marital status, and factors that enable them to secure decent work. 4.1.4 Empirical Model The analysis of the determinants of decent work has generally taken the form of discrete choice modelling (Mackett, 2017; Moussa, 2017; Nizami, 2017). The construction of the composite decent work indices, which is the dependent variable produces three categorical ordered outcomes. For this reason, the appropriate econometric technique is the ordered logistic regression as used by Mackett (2017). The estimation equation is given as: 𝐷𝑊𝐼𝑖 = 𝜔 +  𝐸𝑑𝑢𝑐1 𝑖 +  𝐹𝑒 +  𝑅𝑢𝑟 +  𝐴𝑔𝑒 +  𝐽𝑠𝑡 +  𝑇𝑒𝑚 +  𝐸𝑠𝑒𝑐 +  𝐼𝑛𝑑 +2 3 4 5 𝑖 6 7 𝑖 8 𝑖 𝜎𝑖 …………………………………………………………………………………...…….. (4.4) The second empirical estimation seeks to investigate the relationship between working poverty and decent work following from equation (4.3). The dependent variables include a demand- side variable i.e. decent work and controlling for supply-side factors such as the marital status, and occupation of the worker. Other variables such as education, age, sex, location and status of employment of the worker, are equally important to be included in the model. However, Islam (2004) caution that estimating such a model generally encounter high probability of multicollinearity among the independent variables. Again, the individual characteristics have already be included in the first equation (4.4), which determines decent work. This underscores the justification for being excluded from the second empirical estimation, which is equation (4.5). 51 University of Ghana http://ugspace.ug.edu.gh The dependent variable is the working poor, and it measures the poverty status of the worker. It involves two outcomes either the worker is ‘poor’ or ‘nonpoor’ and therefore it is captured by a dummy variable. The appropriate econometric technique to employ is a binary logistic regression. 𝑊𝑃 = 𝜔 +  𝐷𝑊𝐼 +  𝐷𝑊𝐼 2𝑖 𝑖 𝑖 +  𝐻𝐻𝑧 +  𝑀𝑎𝑟𝑟 + 𝜎𝑖……………………………. (4.5) 1 2 2 3 Definitions of variables: 𝑫𝑾𝑰𝒊 measures the individual decent work index in an ordered category and takes on 0 as low, 1 as medium, and 2 as high decent work. 𝑫𝑾𝑰𝒊 measures decent work index in its continuous form, from 0-11 𝑫𝑾𝑰 𝟐𝒊 measures the squared of decent work index. 𝑾𝑷𝒊 is the measure of individual workers’ poverty status and it is observed as 1 if a worker is poor and 0 as non-poor. 𝑯𝑯𝒛 measures household size and is a continuous variable 𝑬𝒅𝒖𝒄𝒊 is the educational qualification of the worker measured in a form of four categorical dummies as no educational qualification (reference dummy), BECE, High, POSEC, and Tertiary respectively. 𝑱𝒔𝒕𝒊 represents an individual’s working status proxied by six categorical dummies. 𝑻𝒆𝒎 represents the working type of the individual and is coded 0 for permanent, and 1 for temporary work. 𝑬𝒔𝒆𝒄𝒊 is the employment sector measured into 6 categorical dummies namely government, parastatals, NGO’s, cooperatives, international organization, and private. 𝐴𝑔𝑒 measures the age of the worker and is a continuous variable starting from 15 and above. 𝑭𝒆 is the gender variable which takes on 1 for female and 0 for male. 𝑹𝒖𝒓 is location variable which takes on 0 for urban and 1 for rural. 52 University of Ghana http://ugspace.ug.edu.gh 𝑴𝒂𝒓𝒓 tells the marital status of the worker either married or unmarried, where married is coded 0 and unmarried is coded 1. 𝑰𝒏𝒅𝒊 describes the industry of the worker are categorized into five as agriculture, mining and quarrying, manufacturing, wholesale and retail trade, and others. 4.1.5 Explanation of Variables and their Expected Signs. Empirical Model 1: Decent work (DW) Decent work is the response variable in equation (4.4) and has been defined in this work to encompass full employment, protection of rights, benefiting from social protection and social inclusion. Construction of composite decent work indexes (DWI) at the individual level is an ordered one, ranging from ‘low’ as the base outcome and is coded 0, ‘medium’ is coded 1 and ‘high’ is coded 2. Regressors in Model 1: The study seeks to analyze the factors that determine an individual’s probability of working decently in the Ghanaian labour market and hence, individual economic and demographic characteristics analyzed. Age of the Worker According to ILO definition, the working age ranges from 15 years and above. The age variable is a key characteristic that influences individual’s probability of decent work as most studies have found a positive association between age and decent work (Nizami, 2017; Moussa, 2017; Mackett, 2017). Age is a continuous variable and since age comes with experience, adult workers are more likely to secure decent work. Mackett (2017) and Moussa (2017) found that 53 University of Ghana http://ugspace.ug.edu.gh decent work increases with age. However, Nizami (2017) found that beyond a certain threshold of age, decent work decreases. Sex of the Worker The sex variable is dichotomous, which indicate either the individual worker is a male or female, where a male is coded 0 and female coded 1. Generally, the labour market does not favour women in terms of employment, particularly formal sector jobs, which are likely to affect their decent work levels. While Moussa (2017) found no statistical significance of the sex variable in Cote d’Ivoire, Mackett (2017) found a positive effect for men. Therefore, the a priori expectation of the sex variable in this study is uncertain. Location of the Worker The location variable also takes on two responses, that is, either the work is performed in the urban area and is coded 1, otherwise 0 in the rural area. Though evidence has shown that unemployment is low in the rural areas in Ghana, on the contrary, employment in the rural areas are characterized by high informality and indecent work (Kuhn, et al., 2018). Education of the Worker Education variable describes an individual worker’s level of educational qualification measured in the form of five categorical dummies namely no educational qualification (as reference dummy), BECE, High, POSEC and Tertiary. Education. Education is found to influence an individual’s probability of working decently in that educated folks are likely to find formal jobs which mostly provide necessary conditions for decent work (Baah-Boateng, 2009; Mackett, 2017). 54 University of Ghana http://ugspace.ug.edu.gh Job Status of the Worker Job status describes individual workers position on the job, and it is found to influence individuals decent work. Job status is measured by a categorical dummy as paid employees (as reference dummies), casual worker, Non-Agric self-employed, Agric self-employed, contributing family worker and others). Individual workers in paid employees are likely to have higher decent work, given the fact that they are more likely to have a sustained income and be protected by labour laws. Sector of Employment Employment sector is measured by categorical dummies as government sector (reference dummy), parastatals, NGOs, cooperatives, international organization/diploma, private sector and others). Most studies have found that the private sector mostly in developing countries are largely informal and therefore characterized by indecent work, though Moussa (2017) found the private sector to be among the leading contributors to high decent work in Côte d'Ivoire. Form of Employment Form of employment type defines the permanence or temporary nature of the individual’s work and is found to influence individuals decent work. It is dichotomous and is coded 1=temporal and 0=permanent. Generally, temporal workers do have access to employment benefits. As a result, workers in permanent employment are more likely to have higher decent work compared to those in temporal work. Industry of the Worker The industry describes the goods or services the worker produces. The industry of the worker are categorized into five as agriculture=1, mining and quarrying=2, manufacturing=3, 55 University of Ghana http://ugspace.ug.edu.gh wholesale and retail trade=4, and other=5. Mackett (2017) observed that workers in mining industry have higher decent work compared to other industry. Model 2: Working poor (WP) The study adopted monetary poverty measurement and define an individual to be working poor if he/she is working but live in a household whose income is insufficient to place him/her above the national poverty line of that country. Poverty estimation in Ghana uses the FGT poverty measurement approach to classify individuals into three poverty status consisting of ‘very poor’ ‘poor’ and ‘nonpoor’. Thus, an individual is said to be ‘very poor’ if his annual consumption falls below GHc 792.05 and ‘poor’ if it is below GHc1,314, otherwise, he/she is ‘nonpoor’. We reclassified ‘very poor’ and ‘poor’ as working poor. The working poor is the response variable and is coded 0 as working poor and 1 as non-working poor. Regressors in Model 2: The second model investigates the statistically significant association between working poverty and decent work. Decent work is an explanatory variable, and other factors that affect working poverty based on findings in the literature are also analyzed. Composite Decent Work Index (CDWI) The scores of decent work index from 0-11 is used as a continuous variable. The squared decent scores are included to see the effect that as decent work scores increase beyond a certain threshold, what happens to the worker poverty status. In equation (4.5), decent work index is the independent variable and is expected to have an inverse relationship with working poor. As an individual worker decent work increases, his/her probability of being poor must reduce. 56 University of Ghana http://ugspace.ug.edu.gh Marital Status Marital status tells either the individual is married or not married. Marital status is dummy variable and coded 0 for married and 1 for unmarried. Married workers are more likely to be poor because of the likelihood of the presence of children that they have to feed. Kusi (2018) found poverty to be lower among married workers than unmarried ones. Household Size Household size measures the number of people in a household and is a continuous variable. The chances that individual lives in a bigger household should increase their chances of being poor. Anyanwu (2014) found a positive relationship between household size and poverty status. 4.1.6 Econometric Approach and Estimation Procedure of the Logistic Model. The first question involves, what determines decent work? And since the computation of composite decent work index (CDWI) results in a trichotomous outcome (low, medium and high) which is an ordered one, the ordered logistic regression is more appropriate in this case. The second question is to investigate whether decent work has any statistical relationship with the working poor? And because the dependent variable 'working poor' has a binary outcome, the binary logistic regression is used. Studies that have investigated determinants of decent work such as Mackett (2017) and Nizami (2017) used the ordered and multinomial logistic regression respectively, while Moussa (2017) also employed the binary logistic model to elicit determinants of decent work. The choice of the technique is based on the categorization of the dependent variable. 57 University of Ghana http://ugspace.ug.edu.gh From the cross-sectional model given by equation (4.6) below, y takes on 1 if the individual is a working poor and 0 as non-working poor, 𝜎𝑖 captures the unobservable factors and is assumed to be normally distributed in a large sample, 𝑋𝑖 represents the vector of independent variables and  is the vector of the corresponding coefficients of the independent variables. 𝑖 𝑦 ′𝑖 = 𝜔 + 𝑋𝑖 + 𝜎𝑖, 𝑖 = 1,2, … . . 𝑛 ……………………………………………... (4.6) 𝑖 The cumulative logistic distribution function is given by equation (4.6) as: 𝑒𝑋 ′ 𝑖𝑖 𝑃𝑟𝑜𝑏(𝑦𝑖 = 1|𝑋 ′ 𝑖 ) = ′ ……………………………………………….………….. (4.6) 1+𝑒𝑋𝑖𝑖 If 𝑃𝑖 is the probability that the individual worker is poor, then the probability that a worker is nonpoor is (1-𝑃𝑖). Estimating the odds-ratio in favour of working poor by taking the natural log of a simplified equation (4.6) gave the logit model as: 𝑃 𝐿𝑖 = ln ( 𝑖 ) = 𝜔 + 𝑋′𝑖 + 𝜎𝑖 (4.7) 1−𝑃𝑖 𝑖 Where 𝑋𝑖 and  are already defined in equation (4.6). Equation (4.7) is estimated using the 𝑖 maximum likelihood method which ensures that the parameters are consistent and normally distributed especially in a large sample (Damodar, 2004; Greene, 2003; Stock & Watson, 2003). To allow within variables comparison, marginal effects of the ’s are computed from the binary logit model. The study employs the ordered logistic model to investigate the determinants of decent work since decent work which is the dependent variable has three ordered outcomes; ‘low’, ‘medium’, and ‘high’. The ordered logistic model is also estimated with the maximum likelihood and reports the logit coefficients. Diagnostic checks are conducted for suspected multicollinearity among the regressors to enhance the model predictive power. 58 University of Ghana http://ugspace.ug.edu.gh 4.1.7 Data Source The data source used in carrying out this research is the Ghana Living Standards Survey round seven (7) of the Ghana Statistical Services (GSS). Ghana Statistical Services conducts periodic household living standard survey dated back from 1987 when the first round was conducted. Since then, second, third, fourth, fifth, sixth and seventh rounds have been conducted in the years 1987/88, 1988/89, 1991/92, 1998/99, 2005/06, 2012/13 and 2016/17 respectively. The ultimate goal of this exercise is to record information on the living standards of individual citizens in the country. This is to ascertain whether people living in the country have experienced better or worse living standards within the period. It also reveals the poverty levels in the country by estimating absolute and relative poverty threshold and to inform policy to improve conditions of life, and most importantly, to eradicate poverty in the country. More so, Ghana Living Standards Surveys (GLSS) dataset helps to revise national account estimates of the country, and most importantly, to set up a nationwide database on decent work indicators, as well as labour force and child labour statistics. The seventh round of the Ghana Living Standard Survey is a regional and nationally representative all-encompassing survey that collect data on households’ demographic characteristics, education, skills/training, health and fertility, housing conditions, employment and time use, including child labour, income and expenditure of households, financial services and assets of households, and prices of consumer items. The study made use of employment and time use section, which is decent work in-built and a subset of the entire Ghana Living Standard Survey dataset. This is a cross-sectional nationwide representative survey consisting of 51,295 observations. This consists of 48.07% males and 59 University of Ghana http://ugspace.ug.edu.gh 51.93% of females, out of which 36.82% were in urban areas and 63.18% in rural areas across the ten (10) regions of Ghana. The survey also collected data on individual household characteristics such as individual working status, position, working conditions and job characteristics. Questions on individual decent work indicators were asked whether individual worker received or experienced the following; maternity/sick leave, had permanent or temporal work, number of hours work per day, received medical care, had a contractual agreement, received paid vacation or social security, was willing to change job, whether in trade union and other related variables, including individual working, social and demographic characteristics such as education, age, sex, location, sector of employment, occupational and industrial status. Respondent was coded 1, if he/she responded yes to the above questions otherwise 0, to reveal information on individual’s decent work. This means that the study focuses on the employed, that is those individuals aged 15 years and above who during the reference period performed at least an hour of work for economic gain. Again, GLSS7 has poverty module, which makes it a good data set for analyzing the association between decent work and the working poor as well as investigating determinants of decent work in Ghana. 4.2 Presentation and Discussion of Empirical Results 4.2.1 Descriptive Statistics of the Labour Force Estimates from GLSS VII datasets of 2016/17 indicate that employment to population is about 84.67% while the labour force participation rate is 74.29%. Though this figure is quite high, the extent of labour underutilization is enormous. The composite index for labour underutilization measures labour underutilization in the form of unemployment, time-related underemployment, and the potential labour force is estimated to be 32.91% of the total working age. This means that three out of ten Ghanaian workers are underutilized either through 60 University of Ghana http://ugspace.ug.edu.gh unemployment or time underemployed or through potential labour force. Out of the total number of people employed, men accounted for 47.96% and the remaining 52.04% were women. Also, 37.53% of the employed are located in the urban areas, while 62.47% are located in the rural areas as indicated by Table 4.1. Table 4. 1: Percentage of the Employed by Sex and Location Sample Sex Frequency Percent Male 10,728 47.96 Female 11,641 52.04 Total 22,369 100 Location Urban 8,396 37.53 Rural 13,973 62.47 Total 22,369 100 Source: Author’s computation from GLSS7 data (2016/2017) Occupational distribution of the employed according to sex and locality is given by table 4.2. Employment is dominated by the skilled agriculture and fishery workers, accounting for 47.83% of the total employment, followed by services/sales workers, and craft and related trades workers with 19.03% and 14.72% respectively. The lowest occupational employment occurred among the clerical support workers, legislators/managers, and technicians and associate professionals with 1.32%, 1.42% and 1.66% respectively. In terms of sex distribution, males dominate in all occupations except for services/sales workers and craft related and trade workers. Also, rural dominance by agricultural activities remains as it employs 68.71%, followed by craft and related trades workers with 9.95%. The concentration 61 University of Ghana http://ugspace.ug.edu.gh of government work in urban areas is also clear in the areas of managerial and legislative workers, professionals, and clerical support workers. This gives a hint of how informalized the rural areas are, and the level of indecent work that is likely to exist among the rural folks. Table 4. 2: Percentage of Occupational Employment by Sex and Locality Total Male Female Urban Rural Main Occupation (%) (%) (%) (%) (%) Legislators/managers 1.42 1.96 0.93 2.75 0.62 Professionals 5.42 6.88 4.07 9.2 3.13 Technicians and associate professionals 1.66 2.41 0.97 3.06 0.81 Clerical support workers 1.32 1.77 0.91 2.61 0.54 Service/sales workers 19.03 10.28 27.08 34.01 9.95 Skilled agric/fishery workers 47.83 51.19 44.73 13.34 68.71 Craft and related trades workers 14.76 12.41 16.93 23.37 9.55 Plant machine operators and assemblers 4.43 8.82 0.37 6.21 3.35 Elementary occupations 4.11 4.22 4.01 5.41 3.32 Other Occupations 0.03 0.06 0.04 0.02 Total 100 100 100 100 100 Source: Author’s computation from GLSS7 data (2016/2017). The private sector (both formal and informal) continued to provide the greatest employment in Ghana as indicated by Table 4.3, where the private sector comprising both formal and informal contributes to 92.89% of the total employment, while the government sector employs as low 62 University of Ghana http://ugspace.ug.edu.gh as 6.38%. Out of the employed women, 94.86% of them engaged in the private sector, while 92.89% out of total employed men engaged in the private sector. However, men dominance in government employment is almost twice as women. Also, in terms of locality, government employment in urban areas is 11.09% and rural areas with 3.54%. This also confirms government employment concentration in the urban areas compared to rural areas in Ghana. About 95.83% of workers in the rural areas partake in the private sector, while 88% of the urban workers also in the private sector. Table 4. 3: Percentage of Sector Employment by Sex and Locality Total Male Female urban Rural Sector of Employment (%) (%) (%) (%) (%) Government Sector 6.38 8.26 4.64 11.09 3.54 Parastatals 0.12 0.19 0.05 0.21 0.06 NGOs (Local & International) 0.23 0.32 0.15 0.33 0.16 Cooperatives 0.23 0.28 0.19 0.14 0.29 International Organization / Diploma 0.08 0.12 0.04 0.1 0.07 Private Sector 92.89 90.75 94.86 88 95.83 Other 0.08 0.08 0.07 0.12 0.05 Total 100 100 100 100 100 Source: Author’s computation from GLSS7 data (2016/2017) Table 4.4 depicts status in employment by sex and locality. Out of the total employment, 19% are paid employees, out of which 26.9% are men, and 11.71% are women. In terms of locality, 33.48% from urban and 10.29% are in rural areas. Agriculture self-employed workers are the highest with 29.13% in total, out of which 37.35% are males, and 21.56% are females and also 63 University of Ghana http://ugspace.ug.edu.gh 8.27% located in the urban, and 41.67% in the rural areas. In terms of locality, it is the reverse for non-Agriculture self-employed workers, where 42.13% reside in the urban and 15.69% in the rural areas. Again, contributing family workers which includes both non-Agriculture self- employed and Agriculture self-employed is high of about 21.79% of the total employed, out of which 14.78% are men and 28.24% are women and also 10.03% of them located in the urban and 28.85% in the rural areas. Table 4. 4: Percentage of Employment Status by Sex and Locality Job Status Total (%) Male (%) Female (%) Urban (%) Rural (%) Paid employees 19 26.9 11.71 33.48 10.29 Casual worker 4.43 6.47 2.55 6.05 3.46 Non-Agric self-employed 25.61 14.42 35.93 42.13 15.69 Agric self-employed 29.13 37.35 21.56 8.27 41.67 Contributing family worker 21.79 14.78 28.24 10.03 28.85 Other (Specify) 0.04 0.07 0.02 0.04 0.05 Total 100 100 100 100 100 Source: Author’s computation from GLSS7 data (2016/2017) Table 4.5 shows the percentage of the employed who have permanent work and part-time work. From the Table, 84.26% of the total employed have permanent work, while 15.74% of the employed worked as part-time workers. In terms of sex distribution, males in permanent employment are 85.35% higher than that of females of 83.24%. Permanent employment is also high among those in the urban areas compared to rural areas and this could be explained by more formal jobs in the urban areas. 64 University of Ghana http://ugspace.ug.edu.gh Table 4. 5: Percentage of Form of Employment by Sex and Locality Type Total (%) Male (%) Female (%) Urban (%) Rural (%) Permanent 84.26 85.35 83.24 87.04 82.58 Part-Time 15.74 14.64 16.75 12.97 17.42 Total 100 100 100 100 100 Source: Author’s computation from GLSS7 data (2016/2017) With regard to education qualification, total employment is high in ascending order of educational qualification as those with no education have the highest employment rate of 39.79%, followed by those with basic education of 21.81% and those with Secondary and above in that order. This is true since those with higher education are more likely to be employed in the government sector while those with no or less education, to be employed by the private sector. With the exception of no education where females dominate men in employment, however, men dominate in all other levels of employment as displayed by Table 4.6. Table 4. 6: Percentage of Employment by Educational Qualification educational qualification Total (%) Male (%) Female (%) None 39.79 33.47 46.9 Basic 21.81 21.66 21.97 High 25.41 29.2 21.15 POSEC 8.79 10.1 7.32 Tertiary 4.2 5.58 2.66 Total 100 100 100 Source: Author’s computation from GLSS7 data (2016/2017) 65 University of Ghana http://ugspace.ug.edu.gh Working poverty definition employed by this study is someone who is employed or lives in a household whose consumption falls short of the estimated national poverty line of GHc1314 (higher poverty line). Based on GLSS7 survey estimates indicated by Table 4.7, poverty among the employed is 11.54%, while 88.46% of the employed are non-working poor. The rate of working poverty between men and women is almost the same. In terms of location, the rate of working poverty is higher for urban workers than rural workers with 14.67% and 9.66% respectively. Table 4. 7: Working Poverty Status by Sex and Locality Poverty Status Total (%) Male (%) Women (%) Urban (%) Rural (%) Working poor 11.54 11.51 11.57 14.67 9.66 Non-working poor 88.46 88.49 88.43 85.33 90.334 Total 100 100 100 100 100 Source: Author’s computation from GLSS7 data (GLSS7 2016/2017) After compiling the employed dataset, eleven (11) indicators under six (6) dimensions of decent work elements were extracted to construct composite decent work indexes for Ghana. Since decent work index is constructed using binary response, indicators such as hours of work, which is a continuous variable were dichotomized to enhance easy construction of the composite index. The hours of work were coded 1 as normal for individuals with 48 working hours per week or below, and above 48 hours per week was coded 0 as excessive working hours according to ILO (2013). Table 4.8 presents summary statistics and indicators of decent work for Ghana. 66 University of Ghana http://ugspace.ug.edu.gh Table 4. 8: Summary Statistics of Decent Work Indicators Dimension Indicators Categories Percent Employment No 96.15 opportunities Seek to change job To find a better job 3.85 Adequate pay and With regular pay 22.97 productive job Regular pay No regular pay 77.03 Normal working hours 76.08 Work hours/week(48hrs) Excessive working hours 23.92 Paid holidays/leave 9.16 Decent working time Paid holidays/leave No paid holidays/leave 90.84 Stability and security of With contract 18.55 work Contractual agreement No contract 81.45 Medical care 4.14 Medical care No medical care 95.86 No injury 96.15 Occupational injury Suffered injury 3.85 No exposure to hazard 65.23 Safe work environment Occupational hazards Exposure to hazard 34.77 With sick/maternity leave 10.47 Social security Sick/maternity leave No sick/maternity leave 89.53 With insurance 3.59 Health insurance No insurance 96.41 In social security 8.24 Social security coverage Not in social security 91.76 Source: Author’s computation from GLSS7 data (GLSS7 2016/2017). 67 University of Ghana http://ugspace.ug.edu.gh The extent at which people in the Ghanaian labour market are willing to change their jobs for better ones is quite low, about 3.85%, while 22.97% of the employed receive regular pay for their work, 77.03% do not receive regular payment. With regards to decent working hours, 23.92% of Ghanaians work beyond 48 hours per week. Meanwhile, only 18.55% of Ghanaian workers have formal and informal contract leaving the rest with no contractual agreement for their work. Workers who have access to health insurance, social security, medical care, and sick/maternity leave are 3.59%, 8.24%, 4.14%, and 10.47% respectively, as showed in Table 4.8. 4.2.2 Empirical Analysis 4.2.3 Descriptive Statistics of Composite Decent Work Index Composite decent index based on the 11 indicators selected is present in Table 4.9, where 0 indicates a complete deficit of decent work, and 11 indicate zero deficit decent work. From Table 4.9, it is observed that 0.04% of the total employed population had no decent work based on the 11 indicators shortlisted, meaning a complete deficit of decent work. Again, only 1.16% of the total employed population in Ghana had zero decent work deficit. Despite these two extremes of a complete deficit of decent work and zero decent work deficit, the distribution of decent work from the scores of the 11 indicators, wherein decent work distribution is highest among workers who scored 4, 3 and 2 respectively. The individual composite decent work index was recoded into ranges from (0-4) as low, (5-8) as medium and (9-11) as high. The results from Table 4.10 show that only 6.08% of the total employed population have high decent work in Ghana, while 13.64% and 80.27% with medium and low decent work. This is a reflection of high informality accounting for over 80% of the total employed population (Ackah and Baah-Boateng, 2012). This result is consistent with 68 University of Ghana http://ugspace.ug.edu.gh Moussa (2017), who found that only 5.08% of the total employed workers in Côte d'Ivoire have high decent work. Table 4. 9: Composite Index by 11 Indicators DWI (scores) Frequency Percent of the employed Cum. 0 8 0.04 0.04 1 361 1.61 1.65 2 2,384 10.66 12.31 3 7,281 32.55 44.86 4 7,922 35.42 80.27 5 1,142 5.11 85.38 6 908 4.06 89.44 7 479 2.14 91.58 8 523 2.34 93.92 9 678 3.03 96.95 10 424 1.9 98.84 11 259 1.16 100 Total 22,369 100 Source: Author’s computation from GLSS7 (GLSS7, 2016/2017) Table 4. 10: Ordered Composite Index of the Total Employment DWI Frequency Percent of employed Cum. Low (0-4) 17,956 80.27 80.27 Medium (5-8) 3,052 13.64 93.92 High (9-11) 1,361 6.08 100 Total 22,369 100 69 University of Ghana http://ugspace.ug.edu.gh Source: Author’s computation from GLSS7 (GLSS7, 2016/2017). The sex classification of decent work displayed by Table 4.11 revealed 75.29% of Ghanaians have low decent work, wherein women accounted for 83.96% and men for 66.02%. However, in terms of high and medium decent work, the proportion of men in medium and high decent work are 23.82% and 10.16% compared to 10.57% and 5.47% of women respectively, and this indicates high decent work deficit for women relative to men in the Ghanaian labour market. Table 4. 11: Decent Work Index by Sex (Percent) DWI Low Medium High Total 66.02 23.82 10.16 100 Male (0.61) (0.57) (0.41) 83.96 10.57 5.47 100 Female (0.48) (0.42) (0.29) Total 75.29 16.97 7.74 100 (0.38) (0.35) (0.24) N 22369 Note: data are weighted and standard errors in parenthesis Source: Author’s computations from GLSS7 (GLSS7, 2016/2017). There is a higher probability of individual workers in rural areas of Ghana to have higher decent work deficit relative to their counterparts in the urban areas, explained by the higher informality in the rural areas. As indicated by Table 4.12, 86.38% of rural workers have low decent work and 64.29% of the urban workers have low decent work. On the contrary, the proportion of 70 University of Ghana http://ugspace.ug.edu.gh urban workers with high decent work are more than three times that of rural workers with 12.1% for urban and 3.33% for rural. In terms of sex distribution by location, 81.26% of men in rural areas have low decent work compared to 50.56% of their fellow men in the urban areas, while an alarming rate of 91.28% of women in the rural areas have low decent work compared to their fellow women in the urban rate of 76.87%. High decent work favour workers in the urban than rural areas both for men and women. However, decent work disparity between both sexes still exist in both locations. In summary, being a man and working in an urban area increases one’s chances of working decently in the Ghanaian labour market. This could result from government jobs concentration and more formalized jobs in urban areas relative to rural areas. Table 4. 12: Decent Work Index by Location and Sex (Percent) ALL MEN WOMEN DWI Low Medium High Low Medium High Low medium High 64.29 23.6 12.1 50.56 33.28 16.16 76.87 14.74 8.39 Urban (0.66) (0.6) (0.43) 86.38 10.28 3.33 81.26 14.49 4.25 91.28 6.26 2.46 Rural (0.39) (0.34) (0.23) Total 75.29 16.97 7.74 66.02 23.82 10.16 83.96 10.57 5.47 (0.38) (0.35) (0.24) N 22369 10728 11641 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) 71 University of Ghana http://ugspace.ug.edu.gh Table 4.13. presents decent work index by main occupation and sex distribution. From the Table, with the exception of other occupation where low decent work is zero. Low decent work is predominant among skilled agriculture and forestry workers with a similar rate of 98.39% of the total agricultural workers in the country. Following agriculture and forestry workers are craft and trade-related workers, services and sales workers and elementary occupations with 81.17%, 78.35% and 51.23% respectively. However, low decent work is lowest among professionals with 7.23%, clerical support workers with 22.39%, technicians/professionals with 27.28% and legislators/managers with 29.04%. The reverse in terms of high decent work is true for other occupation with the highest decent work rate of 73.3%, followed by professionals with 54.2%, legislators/managers with 46.9% and technicians/professionals with 38.2% in the highest order of decent work. Sex distribution of decent work among all the occupations also gives an interesting trend. Among the formalized occupations, excluding legislators/managers. The proportion of decent work for men and women in high decent work are similar in occupations such as professionals, technicians/professionals, and clerical support workers. Women have a lower proportion of low decent work among the other three formal occupations, and a higher proportion of high decent work in the same occupation. The opposite occurs in other informal occupation such as; agriculture, service and sales, craft and related trade, plant machine/operators, and elementary occupation, wherein the proportion of women in low decent work precede that of men, and proportion of women in high decent work falls below men. The foregoing discussion indicates that individual workers in occupations such as legislators/managers, professionals, technicians, clerical support workers, and other occupation are more likely to have higher decent work. Nizami (2017) found similar results where decent work was high in a large corporation which is normally formalized relative to small firms that are mostly informal. 72 University of Ghana http://ugspace.ug.edu.gh Table 4. 13: Decent Work Index by Occupation and Sex (Percent) ALL MEN WOMEN DWI Low medium High Low Medium High low medium high 29.04 24.04 46.9 24.31 28.22 47.47 38.29 15.85 45.86 Legislators (3.1) (2.75) (3.4) Professionals 7.23 38.57 54.2 9.96 37.26 52.78 3.19 40.51 56.3 (0.95) (1.85) (1.9) 27.78 34.02 38.2 32.22 34.95 32.84 15.92 31.55 52.53 Technicians (3.55) (3.68) (3.5) Clerical 22.39 43.93 33.7 24.84 45.16 30 18.57 42.02 39.42 workers (3.17) (3.48) (3.3) 78.35 17.95 3.7 55.66 35.24 9.1 86.24 11.94 1.82 Sales workers (0.86) (0.81) (0.4) 98.39 1.45 0.15 97.94 1.88 0.18 98.89 0.99 0.13 Agric workers (0.16) (0.15) (0.1) 81.17 17.19 1.64 65.67 30.79 3.55 93.38 6.49 0.13 Craft workers (0.94) (0.92) (0.3) 39.38 52.98 7.64 39.41 52.72 7.87 38.78 57.69 3.53 Plant workers (2.02) (2.09) (1.1) 51.23 41.32 7.45 44.94 46.92 8.14 57.75 35.5 6.74 Elementary occ (2.2) (2.17) (1.1) 0 26.72 73.3 0 26.72 73.28 Other occ 0 (21.87) (22) Total 75.22 17.02 7.76 65.93 23.87 10.19 83.92 10.6 5.49 (0.38) (0.35) (0.2) N 22259 10676 11583 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% 73 University of Ghana http://ugspace.ug.edu.gh Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) As reported by Table 4.14, overall low decent work is lowest among parastatals with 0.42%, government sector with 2.26% but highest among the private sector of 81.14%, other sector with 55.9%, and cooperatives with 29.84%. Nonetheless, high decent work prevails more in the government sector of 64.04%, followed by parastatals with 59.35% as the second with the highest decent work, but lowest in the private sector of just 3.34% of the total private sector workers in high decent work. The distribution of decent work in the government sector and international organizations for both sexes are closely related with marginal difference in favour of men. In spite of this, women in parastatals and NGOs dominate in high decent work over men, and have a low rate of low decent work relative to men. The private sector decent work distribution favours men in high decent work with 4.62% compared to 2.2% for women and 72.75% low decent work to women rate of 88.64%. This is also an evidence to show that workers in the public sector have a higher probability of working decently than the private sector as a result of more formalized public sector where workers are protected by laws that ensure decent work. Educational qualification plays a key role in determining decent work as indicated by 87.32%,76.31%, 66.63%, 41.17%, and 13.37% of workers with no educational qualification, BECE, High, POSEC, and Degree with low decent work respectively. Contrary to this trend, high decent work is highest among workers with degree qualification of 66.06%, followed by POSEC with 32.34% in high decent work and other qualifications in descending order as presented by table 4.15. Thus, decent work increases with higher educational qualification. 74 University of Ghana http://ugspace.ug.edu.gh Table 4. 14: Decent Work Index by Employment Sector and Sex (Percent) ALL MEN WOMEN DWI Low Medium High Low Medium High Low medium high 2.26 33.71 64.04 2.1 33.69 64.2 2.5 33.73 63.77 Government Sector (0.56) (1.57) (1.61) 0.42 40.23 59.35 0.56 42.85 56.59 0 32.61 67.39 Parastatals (0.43) (10.58) (10.59) NGOs (Local & 11.66 54.93 33.41 11.31 62.45 26.23 12.08 45.69 42.23 Inter) (4.77) (8.73) (8.31) 29.84 39.77 30.39 23.24 36.61 40.15 44.13 46.6 9.27 Cooperatives (7.47) (11.52) (13.01) International 6.41 42.45 51.14 7.1 41.89 51.02 5.18 43.46 51.36 Org./diploma (4.01) (14.71) (15.51) 81.14 15.52 3.34 72.75 22.63 4.62 88.64 9.16 2.2 Private Sector (0.37) (0.36) (0.18) 55.9 37.42 6.68 52.87 34.74 12.38 59.44 40.56 0 Other (12.5) (12.27) (6.36) 75.29 16.98 7.74 66.02 23.82 10.16 83.96 10.57 5.47 Total (0.38) (0.35) (0.24) N 22368 10728 11640 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017). While men with no educational qualification, BECE and High qualification are lower in low decent work and higher in high decent relative to women, women workers with Matric and 75 University of Ghana http://ugspace.ug.edu.gh Degree qualification, on the other hand, dominate in high decent work compared to men with the same educational qualification. This shows that higher educational qualification increases women’s decent work relative to men, while no education reduces women’s decent work. Table 4. 15: Decent Work Index by Educational Qualification and Sex (Percent) ALL MEN WOMEN DWI low medium high Low Medium High low medium High None 87.32 11.87 0.82 79.75 18.84 1.41 92.85 6.77 0.38 BECE 76.31 22.1 1.58 67.01 31.18 1.81 86.44 12.22 1.34 High 66.63 24.54 8.83 62.38 28.05 9.57 72.81 19.44 7.75 Matric 41.17 26.49 32.34 38.53 31.23 30.23 45.13 19.36 35.51 Degree 13.37 20.57 66.06 14.45 19.99 65.56 10.9 21.9 67.2 Total 70.84 19.63 9.54 62.41 26.06 11.53 80.1 12.55 7.34 N 16060 8501 7559 Note: data are weighted, row percent add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017). Table 4.16 reports age dimension of decent work such that low decent work is highest among the age group above 44 years and lowest among age group between the ages of 24-35 years old. High decent work as well occurred among the age group between 24-35 years old and lowest for the age group below 25 years old. High and low decent work for men and women are similar among workers below 25 years and between 24-35 years, even though medium decent work between both sexes vary significantly. 76 University of Ghana http://ugspace.ug.edu.gh Table 4. 16: Decent Work Index by Age and Sex (Percent) ALL MEN WOMEN DWI low medium High low Medium High Low medium high Below 25 77.22 20.38 2.4 74.81 22. 47 2.72 79.71 18.22 2.07 years (0.92) (0.9) (0.35) 25-34 65.73 22.7 11.57 55.07 32.18 12.75 75.79 13.75 10.46 years (0.87) (0.8) (0.58) 35-44 74.98 16.55 8.47 62.6 24.96 12.43 86.23 8.9 4.86 years (0.86) (0.74) (0.58) Above 44 82.44 10.65 6.9 72.76 16.62 10.62 91.16 5.28 3.56 years (0.63) (0.54) (0.4) Total 75.29 16.97 7.74 66.02 23.82 10.16 83.96 10.57 5.47 (0.38) (0.35) (0.24) N 22369 10728 11641 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) Almost all the workers in contributing family work (both Agriculture and non-Agriculture), agriculture workers (both those with employees and without), and non-agriculture activities (both those with employees and without) have complete medium and high decent work deficit for both sexes. The above working status is purely informal and absence of regulations that could trigger a complete deficit of decent work among such workers. The situation is quite opposite among paid employees who are more likely to be protected by the laws that ensure their decent work probability. Also, the percentage of women in high decent work both in paid employee and casual worker exceed that of men as shown by table 4.17. 77 University of Ghana http://ugspace.ug.edu.gh Table 4. 17: Decent Work Index by Employment Status and Sex (Percent) ALL MEN WOMEN DWI low medium High Low Medium high Low medium high Paid 11.65 55.26 33.09 13.32 55.45 31.23 8.21 54.88 36.91 employee (0.67) (1.01) (0.93) Casual 25.28 73.26 1.47 28.75 70.14 1.11 17.06 80.63 2.31 worker (1.7) (1.74) (0.47) Paid 50.07 49.41 0.52 47.79 51.58 0.63 61.58 38.42 0 apprentice (6.1) (6.11) (0.52) Non-Agric 100 0 0 100 0 0 100 0 0 employed Agric 99.96 0.04 0 99.94 0.06 0 100 0 0 employed (0.04) (0.04) Contributi 99.95 0.05 0 99.94 0.06 0 99.95 0.05 0 ng (0.04) (0.04) workers Total 75.29 16.98 7.74 66.02 23.82 10.16 83.96 10.57 5.47 (0.38) (0.35) (0.24) N 22368 10728 11640 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) The industry of work dimension of decent work is reported in Table 4.18 and show that low decent work is highest among agriculture, forestry and fishing workers with 97.13% of the total in the industry, leaving only 2.71% and 0.17% in medium and high decent work respectively. 78 University of Ghana http://ugspace.ug.edu.gh Though there is a marginal improvement for men compared to the overall, the case for women worsened. High decent work occurred most in industry such as public administration and defence with 64.47%, health and social workers with 63.58%, information and communication with 57.25%, education and financial institutions with 51.91% and 51.83% respectively. In terms of low decent work, education recorded the lowest of its workers being in low decent work about 2.9%. Women dominate in industries that provide for high decent work such as health and social work, information and communication, and have the same proportion as men in the financial institutions. While men mostly dominate in industries which usually do not provide for high decent work such as agriculture. Theoretically, being in permanent employment ensures regular income and hence improve decent work probability. This assertion is confirmed in the Ghanaian labour market by results displayed in Table 4.19. High decent work among permanent workers is 8.69% compared to 1.55% of temporal workers. Low decent work is also low for permanent workers with 74.89% relative to 77.65% of temporal workers. The trend is similar for both sexes. 79 University of Ghana http://ugspace.ug.edu.gh Table 4. 18: Decent Work Index by Industry or Economic Sector and Sex (Percent) ALL MEN WOMEN low medium high low medium high low medi high um Agriculture, 97.13 2.71 0.17 96.29 3.51 0.2 98.06 1.81 0.13 forestry and fishing Mining and 27.65 58.49 13.86 28.62 54.81 16.57 23.04 76.12 0.84 quarrying Manufacturing 82.22 13.81 3.97 63.54 26.24 10.22 90.42 8.35 1.23 Electricity, gas, 5.24 54.01 40.74 6.67 50.94 42.39 0 65.31 34.69 stream and air condit Construction 47.71 47.43 4.86 48.05 47.57 4.38 34.12 42.01 23.87 Wholesale and 85.83 12.67 1.5 69.85 26.42 3.72 92.14 7.24 0.62 retail trade Information and 16.49 26.27 57.25 17.29 27.56 55.16 10.17 16.11 73.73 communication Financial and 12.85 35.32 51.83 11.11 37.05 51.84 15.61 32.58 51.81 insurance activities Professional, 41.37 24.74 33.89 40.2 22.73 37.07 44.73 30.49 24.78 scientific and technical Public 4.66 30.87 64.47 4.47 30.16 65.37 5.07 32.36 62.57 administration and defence Education 2.9 45.19 51.91 2.58 41.01 56.41 3.28 50.26 46.47 Human health 5.84 30.58 63.58 11.63 35.9 52.47 2.15 27.19 70.66 and social work activities Total 75.22 17.02 7.76 65.93 23.87 10.19 83.92 10.6 5.49 N 22259 1067 11583 6 Note: data are weighted, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) 80 University of Ghana http://ugspace.ug.edu.gh Table 4. 19: Decent Work Index by Type of Employment and Sex (Percent) ALL MEN WOMEN DWI low Medium High Low Medium High Low medium High 74.89 16.42 8.69 65.02 23.49 11.49 83.98 9.91 6.11 Permanent (0.42) (0.38) (0.28) 77.65 20.8 1.55 71.93 26.03 2.05 83.65 15.31 1.04 Temporal (0.98) (0.95) (0.31) 75.26 17 7.75 65.98 23.84 10.17 83.94 10.58 5.48 Total (0.38) (0.35) (0.240 N 22328 10715 11613 Note: data are weighted, standard errors in parenthesis for ‘ALL’, rows add up to 100% Source: Author’s computations from GLSS7 (GLSS7, 2016/2017) 4.2.4 Analysis of the Regression Results This section discusses the empirical estimations of the regression results. The first estimation deals with determinants of decent work, while the second estimation deals with the relationship between decent work and the working poor in Ghana. 4.2.5 Determinants of Decent Work Determinants of decent work are analyzed through the pooled and gender-disaggregated ordered logistic regression results displayed by Table 4.20. To ensure that the model has good predictive power, diagnostics checks for the presence of multicollinearity and heteroskedasticity are carried out. The test for multicollinearity showed very low collinearity among the regressors, indicated by the variance inflation factor (VIF) being less than 5 for all regressors (see Table IA in appendix). According to Gujarati (2004), the rule of thumb for 81 University of Ghana http://ugspace.ug.edu.gh using the VIF is that, when the VIF for any of the regressors is less than 5, then multicollinearity is not a problem. Also, the White test results indicated the presence of heteroskedasticity (see Table IB in appendix). The robust standard error was used to correct the presence of heteroskedasticity and this show a significant improvement on some of the regressors. The pseudo-R-square of 0.7021 shows the overall model goodness of fit at 1% significance level and the Wald chi (2) =218884.59 at Prob>chi2 =0.0000 means the joint significance of all the regressors in the model. The likelihood ratio, log-likelihood ratio, 1%, 5% and 10% significance level of the regressors are reported below Table 4.20. Table 4.20 reports only the regression coefficients for the pooled and gender-disaggregated data from the ordered logit model. This is because the focus of the study is to look at the factors that affect an individual’s decent work level in Ghana and therefore, variable significance, sign and magnitude of coefficients are paid attention to. The results from Table 4.20 show that being female increases the probability of having higher decent work compared to male. Though this result contradicts the descriptive statistics shown in Table 4.11, it is clearly observed that women seem to dominate in high decent work in areas that provide higher decent work such as employees, educational qualification above High, information and communication, human health and social work, professionals, parastatals and others. This result contradicts Mackett (2017) who found a negative effect for women in Gauteng City Region in South Africa, though Moussa (2017) found no statistically significant relationship between men and women in Cote d'Ivoire. This could be that gender inequality debate, laws and policies implemented in Ghana both in the education system and the labour 82 University of Ghana http://ugspace.ug.edu.gh market is having a positive impact on females. Baah-Boateng (2009) observed this changing trend for women in the Ghanaian labour market. The prevalence of indecent work in rural areas is not a new phenomenon. The negative coefficient for rural means that rural workers have low decent work even the coefficient is not statistically significant. Also, temporary workers are significantly less likely to have decent work compared to workers with permanent jobs as indicated by the pooled and gender- disaggregated regressions with negative coefficients 0.727, 0.609 and 1.146 respectively. This is theoretical evidence that shows that permanent workers are more likely to have sustained income and enjoy work-related benefits such as medical care, social benefits that increase an individual’s probability of having higher decent work. The age variable for the pooled and gender-disaggregated regression (all, men and women) has positive coefficients of 0.009, 0.011 and 0.004 respectively, and statistically significant form for the pooled and men regressions. This means that as an individual grows older, his/her probability of having higher decent work increases, which could result from experience in the labour market and higher education.. Age has been found to increase individuals decent work as indicated by Moussa (2017), Mackett (2017) and Nizami (2017) in Cote d'Ivoire, South Africa and India respectively. With reference to no educational qualification, BECE is not significant for the pooled and men regression but is significant for women. Educational qualification has a robust increase in individuals higher decent work probability. Despite of the general increase in higher decent work through education, the effect is bigger for women than both the pooled and men regressions indicated by the size of the coefficients. This also confirms the descriptive statistics 83 University of Ghana http://ugspace.ug.edu.gh for high decent work for women in higher education, and from the industrial level where women dominate in high decent work in industries that have high decent work probabilities. The finding in this study is in line with Mackett (2017) and Moussa (2017) for the pooled result but contradict the gender disaggregated results of Mackett (2017). Thus, men have higher decent work than women at all levels of educational qualification in the Gauteng City Region in South Africa. The mixed findings here explain the puzzle in these two countries decent work effect on sex. Thus, the affirmative actions that have been put in place in Ghana, specifically in the education systems are showing a positive effect for women labour market participation and decent work. Studies have found that higher education usually discourages people from the agricultural sector or informal low paid jobs (Baah-Boateng, 2009). There is less probability of decent work among other sectors relative to the government sector. Table 4.20 shows that indecent work is higher in sectors such as the private, international organizations, and the NGO’s, with the coefficients 1.547, 1.492, and 1.070 respectively, where a bigger value indicates less decent work. With the exception of NGO’s, indecent work is higher among men than women in all other sectors. Parastatals and cooperatives are not significant. Moussa (2017) found similar results in Cote d'Ivoire, where workers in public administration had higher decent work compared to other sectors. As noted by Moussa (2017), this is as a result of controls in the government sector in enforcing laws that protect workers from indecent work. For instance, trade unions that seek to address workers grievances in Ghana, basically involve government sector workers. Turning to employment status and with reference to paid employees, all the categories are statistically significant in determining decent work. The negative coefficients for the pooled regression results in Table 4.20 reveal that indecent work is highest among Non-agriculture 84 University of Ghana http://ugspace.ug.edu.gh employed with 23.715, slightly lower than agriculture employed with 23.291. While indecent work is higher for men than women in agriculture status, women dominate indecent work in Non-agriculture status. Compared to Non-agriculture and agriculture status, decent work is higher among casual worker, paid apprentice and contributing family worker respectively, with negative coefficients 0.609, 1.303 and 8.643 in the pooled regression. Smaller value indicates more decent work and higher value indicates less decent work. The results mean that decent work is higher for employees who are more likely to be in formal employment than other status. This evidence is supported by GSS (2013), who found that workers in the informal economy have low decent work. The last determinant of decent work variable discussed here is the industry in which the individual is engaged. With agriculture, forestry and fishing as a reference industry, decent work is highest among workers in the mining and quarrying, manufacturing, others and wholesale and retail trade respectively, with positive coefficients 0.554, 0.434, 0.316 and 0.009 for the pooled regression, where a bigger value indicates higher decent work and smaller value indicates indecent work. The pooled regression result is similar to male regression in terms of magnitude, sign and significance of the coefficients. However, female regression showed no significant results. Comparing the decent work of other industries to agriculture, Mackett (2017) found similar results. Workers in the mining industry had the highest decent work in the Gauteng City Region in South Africa, similar to Ghana in this study. The mining industry being one of the formalized industries in Ghana, is more likely to provide decent work compared to other industries. 85 University of Ghana http://ugspace.ug.edu.gh Table 4. 20: Determinants of Decent Work Variables All Men Women Female 0.387*** Rural -0.106 -0.112 -0.060 Temporary -0.727*** -0.609*** -1.146*** Age 0.009*** 0.011*** 0.004 Educational Qual. (ref. None) BECE 0.155 0.023 0.603*** High 0.837*** 0.641*** 1.425*** POSEC 1.680*** 1.388*** 2.458*** Tertiary 2.570*** 2.400*** 3.040*** Sector of Employment (ref. government) Parastatals -0.119 -0.285 0.338 NGO'S -1.070*** -1.044*** -1.036** Cooperatives -0.494 -0.354 -1.167 International org. & diploma -1.492*** -1.583*** -1.227 Private -1.547*** -1.661*** -1.295*** Others -1.369** -1.629* -0.668 Employment Status (ref. Employees) Casual worker -0.609*** -0.703*** -0.165 Paid apprentice -1.303*** -1.182*** -2.639** Non-Agric self-employed -23.715*** -23.535*** -22.981*** Agric self-employed -23.291*** -22.984*** -22.641*** Contributing family worker -8.643*** -8.084*** -9.243*** Industry (ref. Agric, forestry & fishing) Mining and quarrying 0.554** 0.677*** 0.089 86 University of Ghana http://ugspace.ug.edu.gh Manufacturing 0.434** 0.593*** 0.137 Wholesale and retail trade 0.009 0.156 -0.318 Others 0.316** 0.360** 0.325 cut1 -2.257*** -2.285*** -2.614*** cut2 1.639*** 1.462*** 1.860*** N 16019 8470 7549 Note: *** significance at 1% level, ** significance at 5% level, * significance at 10%, Wald chi2(25) =218884.59, Prob>chi2=0.000, Pseudo 𝑅2 =0.7021, Log likelihood= -3478.6107 4.2.6 Decent Work and Working Poverty The second question the study attempts to answer is to investigate the empirical link between decent work and working poverty status in Ghana. Diagnostics checks were conducted to ensure the model predictive power. The ‘conditional number’ for multicollinearity test was 8.5, which according to Joshi et al. (2012) implies light collinearity, and is not a problem (Joshi, et al., 2012). The presence of heteroskedasticity revealed by the White test was addressed with the robust standard error. Marginal effects of the results from the pooled and gender- disaggregated regressions after the estimation of the logit model is presented in Table 4.21 (see Table II in appendix for logit coefficients). The overall significance of the model is indicated by the chi-square at 1% significance level and the pseudo 𝑅2 of 0.1004 for the pooled regression. After controlling for some individual characteristics such as the marital status, household size and the squared of decent work score (𝐷𝑊𝐼2), an increase in individual decent work level (scores) is more likely to reduce their probability of being poor. From Table 4.21, an increase 87 University of Ghana http://ugspace.ug.edu.gh in decent work scores by an additional indicator of decent work is statistically significant in reducing an individual likelihood of being poor by 1.1 percentage points holding other variables constant. The gender-disaggregated results showed that decent-work-poverty reduction for women is higher relative to men, given by 2.1 and 0.2 percentage points respectively. This result confirms the ILO’s proposition that decent work could be used to reduce poverty. Some related studies on poverty and employment-related factors have shown a negative relationship that confirms this result. Saget (2001) observed a negative relationship between decent work and poverty, through increasing the minimum wage. Islam (2004) find that more decent occupation such as manufacturing reduces poverty more compared to agriculture, which generally provides less decent work. Control variables that were included in the model showed statistically significant results except for the coefficients of decent work scores and the squared of it, for the men regression. The squared of decent work score was positive and statistically significant, which means that increasing the decent work indicators beyond certain threshold could increase poverty. Decent work is a demand-side variable, which means that increasing the number of indicators will lead to an increased cost of production. The theoretical implication is that workers in decent employment could be displaced into the informal economy where decent work deficit is high and poverty predominant. Married workers are less likely to be poor compared to unmarried by 3.3 percentage points. Female workers are less likely to be poor than male workers indicated by 4.2 and 2.3 percentage points respectively. This results is supported by Kusi (2018). However, Osei-Amponsh et al. (2010) found that unmarried people are less likely to be poor compared to married, though not significant. The last control variable is the household size and is positive and statistically significant. That is, an additional person to a household is more likely to increase their poverty status by 3.5 percentage points. This result is supported by Anyanwu (2014) and Kusi (2018), who found a positive relationship between poverty and 88 University of Ghana http://ugspace.ug.edu.gh household size. However, Osei-Amponsh et al. (2010) found a conflicting result, where a large household size is needed to maintain and expand their business and hence, reduces poverty. Table 4. 21: Marginal Effects from Logit Model 2 Regression Variables Total Men Women DWI -0.011** -0.002 -0.021*** DWI^2 0.001*** 0.001 0.002*** Married -0.033*** -0.023*** -0.042*** HHSIZE 0.035*** 0.035*** 0.034*** N 22,369 10,728 11,641 Wald chi2(2) =863.17 Wald chi2(2) =384.2 Wald chi2(2) =499.58 Prob>chi2=0.000 Prob>chi2=0.000 Prob>chi2=0.000 Pseudo R2=0.1004 Pseudo R2=0.0973 Pseudo R2=0.1045 Note: dy/dx for factor levels is the discrete change from the base level, *** significance at 1% level, ** significance at 5% level, * significance at 10%. 89 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE SUMMARY, CONCLUSION AND POLICY RECOMMENDATION 5.0 Introduction The preceding chapters have explored the concept of decent work and working poverty in the literature and looked at the construction of individual decent work index for the respondents. The empirical analysis of the determinants of decent work and the relationship between decent work and working poverty status in Ghana was carried out. This chapter summarizes the findings from the study, conclude and make some policy recommendations. 5.1 Summary of Empirical Findings The study sought to investigate two main issues. First, the study sought to ascertain the factors that influence an individual’s probability of working decently in Ghana and second, does the probability of working decently have any implication on ones working poverty status? In answering these questions, the study is motivated to by the following objectives: 1) To construct individual composite decent work indexes for Ghana, 2) To elicit factors that determine the probability of having decent work in Ghana, and 3) to look at the statistically significant association between decent work and working poverty. Based on the 11 indicators used for the construction of individual composite decent work indices, we found that there is 6.08% probability of an individual worker in Ghanaian labour market to have high decent work, 13.64% workers likely to have medium decent work and 80.27% chances that a worker in Ghanaian labour market have low decent work (see Table 4.10). This is a reflection of high informality in the Ghanaian economy and lack of controls to implement laws that protect individual workers. 90 University of Ghana http://ugspace.ug.edu.gh The descriptive statistics and the results from the logistic regression confirm that individual characteristics such as sex, age, educational qualification, sector of employment, status of the worker, employment type, and industry in which the individual is, are statistically significant in determining their probability of working decently in Ghana. Examining these factors, we saw that being female increases one’s decent work level than being a male in the Ghanaian labour market. Also, workers in permanent employment have higher chances of decent work than those in temporary employment. An ageing worker increases his/her decent work probability. However, decent work increasing effect on men is higher than women among women. The educational qualification has a positive change in decent work, but education contribution to decent work has a bigger impact for women relative to men. In reference to the government sector, workers in the private sector, NGOs, international organization and diplomacy are less likely to have decent work. There is also a higher deficit of decent work in the private sector for men than women. More so, individual worker status was found to significantly influence their decent work level. With employees as the reference variable, decent work deficit is highest among Non-agriculture employed, agriculture employed and contributing family worker respectively. While decent work deficit in Non- agriculture and the contributing-family worker is higher for women than men, the deficit in agriculture is higher for men compared to women. Among the industries, agriculture, forestry and fishing had the lowest probability of decent work compared to the mining and quarrying, manufacturing, wholesale and retail trade, and other industries. Decent work in the nature of industries selected was favourable to men than women. 91 University of Ghana http://ugspace.ug.edu.gh The third objective which sought to investigate the association between decent work working poverty was found to be statistically significant. Thus, decent work is statistically significant in reducing working poverty in Ghana. However, beyond a certain threshold of the number of decent work variables, it might lead to poverty. Also, the gender-disaggregated results showed that decent work-poverty reduction is high for women than men. Married workers are less likely to be poor compared to unmarried. While being in a larger family increases an individual’s likelihood of being poor. 5.2 Conclusion and Policy Recommendations Conclusions, drawn from the empirical findings, from this study certainly have policy relevance to policymakers both locally and internationally. Our results showed that 80.27% Ghanaians have low decent work which partly emanates from high informality confirmed by the prevalence of low decent work among the private sector workers, casual, paid apprentice, contributing family work, agriculture and non-agriculture as well as workers in temporary employment due to lack of institutional controls and regulations. This calls for institutional interventions in implementing policies geared towards formalizing the informal economy so that labour laws could be enforced to enhance improvement in decent work among Ghanaian workers. Given that females have higher decent work index than males, is a positive signal to implementation of affirmative actions and gender policies to address gender inequalities that hinder women from labour market participation and even in career advancement jobs. This calls for the enforcement of gender policies to promote women in the job market. However, in pursuing such policies to promote women should not be a cost to men in the job market participation and decent work. 92 University of Ghana http://ugspace.ug.edu.gh Educational qualification as a positive and robust determinant of decent work reveals the importance of human capital investment in promoting decent work among Ghanaian workers. The interesting part of this is its implication on female labour market participation and decent work improvement. This requires private and national concern with regards to investing in one’s education to improve their probability of decent work levels, and also government providing free education at the basic and secondary level as well as subsidizing higher education, particularly for women to promote national development. There exists a strong inverse relationship between decent work and working poverty. Which means decent work is a significant tool that could be used to address working poverty. This evidence provides theoretical support to the ILO claims of addressing global working poverty canker through the decent work agenda. This finding has a heterogeneous effect in addressing a wide range of global issues pertaining to the SDGs. Thus, policymakers such as governments should ensure quality and accessible education, formalizing the informal sector or putting in place mechanisms to ensure that workers in the informal sector are protected. Achieving decent work has national and global relevance through the reduction of poverty and hunger, gender inequality, and inequality at large. The findings provide three key lessons: To employers, in the advent of high unemployment, the possibility of employers in the quest of cutting down the cost of production, intend to avoid paying benefits due employees. This normally occurs in the form of putting workers on contract to avoid payment of social security and other benefits. To the government, employment creation should not be over-emphasized in meeting political promises but rather use decent work as a ‘means to an end’ to reduce poverty in the country. To employees, demanding too 93 University of Ghana http://ugspace.ug.edu.gh much employment benefits as a way of improving their decent work levels could lead to their displacement into the informal economy and increase their poverty likelihood. This work has provided a significant contribution to decent work literature in two major ways. The study has provided the first empirical evidence demonstrating that decent work is statistically significant in reducing working poverty. Secondly, the work is also the first attempt to study decent work at the micro level in Ghana. Despite a significant contribution made by this work, it has some shortfalls. The key among them is the number of decent work indicators used are non-exhaustive, the addition of which may improve the index construction. Again, the consumption poverty measurement has been criticized, since it does not capture non-monetary factors of poverty that could affect the validity of the results. Notwithstanding, the results are still valid and useful for policy implications, especially in developing countries. This work also opens the way to discussions on decent work in the country and invite criticisms to improve future studies. The study recommends future investigations into the relationship between decent work and economic growth, and also access the contributions of human capital to decent work. 94 University of Ghana http://ugspace.ug.edu.gh REFERENCES Abebrese, J. (2014). " We Need to Back that Dialogue with Some Action": Programme and Practice of Decent Work in Ghana. Friedrich-Ebert-Stiftung, Ghana Office. Ackah, C., & Aryeetey, E. (Eds.). (2012). Globalization, trade and poverty in Ghana. IDRC. Ackah, C., & Baah-Boateng, W. (2012). Trends in growth, employment and poverty in Ghana. Globalization, trade and poverty in Ghana, 33. Anin T.E. (2003). An economic blueprint for Ghana. Woeli Publishing Services Anyanwu, J. C. (2014). Marital status, household size and poverty in Nigeria: evidence from the 2009/2010 survey data. African Development Review, 26(1), 118-137. Aryeetey, E., & Baah-Boateng, W. (2015). Understanding Ghana's growth success story and job creation challenges (No. 2015/140). WIDER Working Paper. Asselin, L. M. (2009). Analysis of multidimensional poverty: Theory and case studies (Vol. Springer Science & Business Media. Atkinson, A. B. (1987). On the measurement of poverty. Econometrica: Journal of the Econometric Society, 749-764. Baah-Boateng W and Sparreboom Theo (2011) “Ghana: MDG Employment Indicators” Sparreboom (ed.) Towards Decent Work: Monitoring MDG1B employment Indicators in Sun-Saharan Africa, published by ILO, Geneva Switzerland, forthcoming Baah-Boateng, W. (2009). Gender perspective of labour market discrimination in Ghana (Doctoral dissertation, University of Ghana). Baah-Boateng, W. (2015). Unemployment in Africa: how appropriate is the global definition and measurement for policy purpose. International Journal of Manpower, 36(5), 650- 667. 95 University of Ghana http://ugspace.ug.edu.gh Baah-Boateng, W. (2016). Economic growth and employment generation nexus: Insight from Ghana. Volume 16 of ICDD Working Papers, Kassel University Press GmbH Publication, ISBN 3737600686, 9783737600682. Bell, S., & Newitt, K. (2010). Decent work and poverty eradication: Literature review and two-country study. London: Ergon Associates. Bonnet, F., Figueiredo, J. B., & Standing, G. (2003). A family of decent work indexes. Int'l Lab. Rev., 142, 213. Brady, D., Baker, R. S., & Finnigan, R. (2013). When unionization disappears: State-level unionization and working poverty in the United States. American Sociological Review, 78(5), 872-896. Cattan, P. (1998). The effect of working wives on the incidence of poverty. Monthly Labour Review, 121, 22. Chambers, R. (2006). What is poverty? Who asks? Who answers? Crettaz, E. (2011). Fighting working poverty in post-industrial economies: causes, trade-offs and policy solutions. Edward Elgar Publishing. Crettaz, E. (2013). A state-of-the-art review of working poverty in advanced economies: theoretical models, measurement issues and risk groups. Journal of European Social Policy, 23(4), 347-362. Crettaz, E., & Bonoli, G. (2011). Worlds of working poverty: National variations in mechanisms. In Working Poverty in Europe (pp. 46-69). Palgrave Macmillan, London. Damaske, S., Bratter, J. L., & Frech, A. (2017). Single mother families and employment, race, and poverty in changing economic times. Social science research, 62, 120-133. Damodar, N. (2004). Basic econometrics. The Mc-Graw Hill. 96 University of Ghana http://ugspace.ug.edu.gh Danziger, S., & Gottschalk, P. (1986). Work, poverty, and the working poor: a multifaceted problem. Monthly Labour Review, 109, 17. Danziger, S., & Gottschalk, P. (1986). Work, poverty, and the working poor: a multifaceted problem. Monthly Labour Review, 109, 17. ECLAC/ILO (2013). The employment situation in Latin America and the Caribbean: Advances and challenges in measuring decent work. United Nations Publication LC/L.3630 Ehrenpreis, D. (2008). Jobs, Jobs, Jobs? The Policy Challenge (No. 16). International Policy Centre for Inclusive Growth. Eichler, M., & Matthews, A. (2004). What is work? Looking at all work through the lens of unpaid housework. Manuscript). Retrieved from http://wall. oise. utoronto. ca/events/What is Work. pdf. Post-retirement practices, 131. Ferguson, R. F. (2005). The working-poverty trap. Public Interest, (158), 71. Ferraro, T., Pais, L., & Dos Santos, N. R. (2015). Decent work: An aim for all made by all. International Journal of Social Sciences, IV (3), 30-42. Fleury, D. (2007). A study of poverty and working poverty among recent immigrants to Canada. Ottawa: Human Resources and Social Development Canada. Fosu, A. K. (2017). Growth, inequality, and poverty reduction in developing countries: recent global evidence. Research in Economics, 71(2), 306-336. Fremstad, S. (2012). Married… without Means: Poverty and Economic Hardship Among Married Americans (No. 2012-26). Center for Economic and Policy Research (CEPR). Ghai, D. (2003). Decent work: Concept and indicators. International Labour Review, 142(2), 113-145. 97 University of Ghana http://ugspace.ug.edu.gh Ghai, D. P. (2002). Decent work: Concepts, models and indicators (Vol. 139). International Institute for Labour Studies. Ghai, D.P. (2006). Decent work: objectives and strategies”, International Labour Organization Ghana Statistical Service (2014). Ghana Living Standards Survey Round 7 (GLSS7): Labour Force Report. Ghana Statistical Service (2015). Integrated Business Establishment Survey: National Employment Report. Ghana Statistical Services (2018). Ghana Living Standards Survey Round 7 (GLSS7): Poverty trends in Ghana (2005-2017) Ghana Trades Union Congress (TUC)/Rosa Luxemburg Stiftung (RLS) 2012: trade unions and industrial relations in ghana Goedhart, T., Halberstadt, V., Kapteyn, A., & Van Praag, B. (1977). The poverty line: concept and measurement. Journal of Human Resources, 503-520. Goulden, C. (2010). Cycles of poverty, unemployment and low pay. Policy and Research, Joseph Rowntree Foundation publication. Greene, W. H. (2003). Econometric analysis. Pearson Education India. Greyling, T. (2015). A composite index of quality of life for the Gauteng city-region: A principal component analysis approach. Harvey, A. S., & Mukhopadhyay, A. K. (2007). When twenty-four hours is not enough: Time poverty of working parents. Social indicators research, 82(1), 57-77. Heintz, J. (2005). Employment, gender, and poverty in Ghana (No. 92). Working Paper Series. Herman, E. (2014). Working poverty in the European Union and its main determinants: An empirical analysis. Engineering Economics, 25(4), 427-436. 98 University of Ghana http://ugspace.ug.edu.gh ILO (2019). Work for a brighter future – Global Commission on the Future of Work, International Labour Office – Geneva: ILO, 2019 ILO (2017). Universal social protection to achieve the Sustainable Development Goals, World Social Protection Report, 2017–19, International Labour Office – Geneva: ILO (2013). Decent work indicators: Guidelines for producers and users of statistical and legal framework indicators. ILO (2012). Up-scaling Local Economic Development: Lessons from the Ghana Decent Work Programme. International Training Centre of the ILO Viale Maestri del Lavoro, 10 - 10127 Turin, Italy, Publications, ISBN: 978-92-9049-661-8 Islam, R. (2004). The nexus of economic growth, employment and poverty reduction: An empirical analysis. Joassart-Marcelli, P. (2005). Working poverty in southern California: Towards an operational measure. Social Science Research, 34(1), 20-43. Joint Research Centre-European Commission. (2008). Handbook on constructing composite indicators: methodology and user guide. OECD publishing. Joshi, H., Kulkarni, H., & Deshpande, S. (2012). Multicollinearity Diagnostics in Statistical Modeling & Remedies to deal with it using SAS. Pharmaceutical Users Software Exchange, (1), 1-34. Khandker, S. R., & Haughton, J. (2005). Introduction to poverty analysis. World Bank Institute. Kuhn, S., Milasi, S., & Yoon, S. (2018). World employment social outlook: Trends 2018. Geneva: ILO. Kusi, G. (2018). Moonlighting to Survive: The Effect of Multiple Job Holding on the Working Poor in Ghana. University of Ghana,- MPhil Desertation 99 University of Ghana http://ugspace.ug.edu.gh Lawton, K., & Cooke, G. (2008). Working Out of Poverty: A Study of the Low Paid and the Working Poor. Institute for Public Policy Research. MacInnes, T., Bushe, S., Tinson, A., Born, T. B., & Aldridge, H. (2014). Monitoring poverty and social exclusion 2014. York: Joseph Rowntree Foundation. MacInnes, T., Tinson, A., Hughes, C., Born, T. B., & Aldridge, H. (2015). Monitoring poverty and social exclusion 2015. Children, 3, 3-7. Mackett, O. Measuring the determinants of decent work: Evidence from the Gauteng City- Region. Maleček, P., & Čermáková, K. (2015). In-work Poverty in the Czech Republic: Identification of the Most Vulnerable Groups. Procedia Economics and Finance, 30, 566-572. Mehran, F., Anker, R., Chernyshev, I., Egger, P., & Ritter, J. (2002). Measuring decent work with statistical indicators. International Labour Office, Policy Integration Working Paper, (2). Moore, P., Dannreuther, C., & Möllmann, C. (2015). Guest editors’ introduction: The future and praxis of decent work. Global Labour Journal, 6(2), 127-137. Moussa, R. K. (2017). Elicitation of the determinants of decent work in developing countries: Evidence from Côte d’Ivoire. Economics Bulletin, 37(1), 494-507. Narayan, D., Patel, R., Schafft, K. and Rademacher, A. (2000). Voices of the poor: can Anyone hear us? Oxford University press, New York. Newitt, K. and Bell S. (2010). Decent work and poverty eradication: Literature review and two-country study, A study for the Decent Work and Labour Standards Forum. Ergon, London Nizami, N. (2017). Changing Work Organizations and Implications for Decent Work-A Case Study of Information Technology Industry in India. 100 University of Ghana http://ugspace.ug.edu.gh Nizami, N. and Prasad, N. (2013). Decent work in India: Insights from IT industry. Sage Publications India Private Limited, New Delhi Nizami, N., & Prasad, N. (2017). Characteristics of IT Industry’s Labour Market. Decent Work: Concept, Theory and Measurement (pp. 125-139). Palgrave Macmillan, Singapore. OECD. (2008). Handbook on constructing composite indicators: methodology and user guide: OECD publishing. Osei-Amponsah, C., Anaman, K. A., & Sarpong, D. B. (2010). Marital Status and Household Size as Determinants of Poverty among Fishmongers in Tema, Ghana”, In Marriage: Roles, Stability and Conflict, (ed. Kenneth S. Pearlman). Nova Science Publishers. Osmani, S. R. (2003). Exploring the employment nexus: topics in employment and poverty. UNDP New York. Pautz, H., & Stuart, F. (2016). What does ‘decent work’mean to low-paid workers? Working practices and how to improve them. British Politics and Policy at LSE. Ravallion, M. (1998). Poverty lines in theory and practice. The World Bank. Ravallion, M. (2017). Poverty comparisons. Routledge. Ray, K., Hoggart, L., Vegeris, S., & Taylor, R. (2010). Better off working? Work, poverty and benefit cycling. Ringen, S. (1988). Direct and indirect measures of poverty. Journal of social policy, 17(3), 351-365. Rodgers, G. (2002). Decent work as a goal for the global economy. unpublished mimeograph. Rose M.E. (1986). The Causes of Poverty in The Relief of Poverty. 1834–1914. Studies in Economic and Social History. Palgrave, London 101 University of Ghana http://ugspace.ug.edu.gh Sabia, J. J., & Burkhauser, R. V. (2010). Minimum wages and poverty: will a $9.50 Federal minimum wage really help the working poor? Southern Economic Journal, 76(3), 592-623. Saget, C. (2001). Poverty reduction and decent work in developing countries: Do minimum wages help. International Labour Review, 140(3), 237-270. Chicago 7th ed. Sanchez-Martinez, M., & Davis, P. (2014). A review of the economic theories of poverty (No. 435). National Institute of Economic and Social Research. Sehnbruch, K., Burchell, B., Agloni, N., & Piasna, A. (2015). Human development and decent work: why some concepts succeed and others fail to make an impact. Development and Change, 46(2), 197-224. Selim, R. (2006). Employment-poverty linkages and pro-poor growth: A synthesis paper based on country studies of Bangladesh, Bolivia and Ethiopia. ILO. Sen, A. (1982). Poverty and famines: an essay on entitlement and deprivation. Oxford university press. Slack, T. (2010). Working Poverty across the Metro‐Nonmetro Divide: A Quarter Century in Perspective, 1979–2003. Rural sociology, 75(3), 363-387. Somavia, J., & General, I. D. (1999). Decent work. In Report of the Director-General to the 87th Session of the International Labour Conference.(Geneva: ILO). Sparreboom, T., & Albee, A. (Eds.). (2011). Towards Decent Work in a Sub-Saharan Africa: Monitoring MDG Employment Indicators. ILO. Stock, J. H., & Watson, M. W. (2003). Introduction to econometrics (Vol. 104). Boston: Addison Wesley. Stuart, F., Pautz, H., and Wright, S. (2016). Decent work for Scotland’s low-paid workers: a job to be done. 102 University of Ghana http://ugspace.ug.edu.gh Taylor, R. F. (2004). Extending conceptual boundaries: work, voluntary work and employment. Work, employment and society, 18(1), 29-49. Tijmstra, S. A. (2011). Up-scaling local economic development: lessons from the Ghana decent work programme. International Training Centre of the ILO, Turin. Todaro, M. P., & Smith, S. C. (2012). Economic development. Pearson Education Publication, 12th edition. ISBN 978-0-13-340678-8 -- ISBN 0-13-340678-4 103 University of Ghana http://ugspace.ug.edu.gh APPENDICES Table IA: Multicollinearity Test in Model 1 Collinearity diagnostics SQRT R- Variable VIF VIF Tolerance Squared loc2 1.38 1.17 0.7247 0.2753 Sex 1.1 1.05 0.9109 0.0891 sector 1.42 1.19 0.7048 0.2952 status 2.34 1.53 0.427 0.573 industry 2.3 1.52 0.4347 0.5653 EDUC 1.45 1.2 0.6909 0.3091 agebrac 1.18 1.08 0.85 0.15 Emp type 1.17 1.08 0.8566 0.1434 Mean VIF 1.54 Source: Author’s computation from Stata 15.1 Table IB: Test for Heteroskedasticity in Model 1 White's test for Ho: homoskedasticity against Ha: unrestricted heteroskedasticity chi2(41) = 3231.67 Prob > chi2 = 0.0000 Cameron & Trivedi's decomposition of IM-test Source chi2 df p Heteroskedasticity 3231.67 41 0 Skewness 1974.45 8 0 Kurtosis 20.2 1 0 Total 5226.32 50 0 Source: Author’s computation from Stata 15.1 104 University of Ghana http://ugspace.ug.edu.gh Table II: Logit Coefficients of Decent Work and The Working Poverty Variables All Men Women DWI -0.114** -0.019 -0.232*** -0.05 -0.07 -0.07 DWI^2 0.015*** 0.008 0.024*** 0 -0.01 -0.01 Marital status (ref. Unmarried) Married -0.355*** -0.250*** -0.449*** -0.04 -0.06 -0.06 Household size 0.376*** 0.379*** 0.374*** -0.01 -0.02 -0.02 _cons -3.638*** -3.967*** -3.263*** -0.15 -0.2 -0.21 N 22369 10728 11641 All** Wald chi2(2) =863.17, Prob>chi2=0.000, Pseudo R2=0.1004 Source: Author’s computation from Stata 15.1 105