University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA COLLEGE OF HUMANITIES GHANA’S POWER INDUSTRY AND MANUFACTURING SECTOR PERFORMANCE BY EDEM NY0AVOR (10347214) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF A MASTER OF PHILOSOPHY DEGREE IN ECONOMICS DEPARTMENT OF ECONOMICS OCTOBER, 2017 University of Ghana http://ugspace.ug.edu.gh DECLARATION I, EDEM NYAVOR hereby declare that except for references made to other studies, which have been duly acknowledged, this thesis is the original research undertaken entirely by myself under the guidance of my supervisors towards the award of the Master of Philosophy Degree in Economics in the Department of Economics, University of Ghana. ………………..………. EDEM NYAVOR (10347214) Date: ………………………………… …………………………..... …………………………..... PROF. AMOAH BAAH-NUAKOH DR. KWADWO TUTU (SUPERVISOR) (SUPERVISOR) Date: …………………………...... Date: …………………………........ i University of Ghana http://ugspace.ug.edu.gh ABSTRACT Manufacturing sector development in Ghana has undergone a lot of transformation from pre- independence to date, yet the sector is yet to reach its full potential of making Ghana an industrialized country. Ghana’s manufacturing sector has faced many impediments mainly in terms of infrastructural development especially in the provision of electricity as a cheap source of industrial energy. This study accounts for the status of the power industry in Ghana and how it affects the growth and development of the manufacturing sector in the country. The study therefore investigates the relationship between electricity supply and manufacturing sector performance in Ghana, given performance indicators to be value-added output, labour productivity, and employment. Six models were estimated in this study. The first three models dealt with the relationship between electricity consumption with manufacturing sector performance at the aggregate level employing the Autoregressive Distributive Lag (ARDL) method developed by Pesaran et al (2001) using time series data from 1990 to 2015. The other three models investigated the relationship between expenditure on electricity consumption and manufacturing performance at the firm level using a panel fixed-effects estimation model using panel data from 1992 to 2003. Results indicated that in the manufacturing sector, electricity had a long run relationship with all the manufacturing performance indicators and positively related to manufacturing value-added output growth and labour productivity in the long run. However, electricity was negatively related to employment growth in the long run. Results found for firm level manufacturing indicates that, expenditure on electricity consumption had a negative relationship with firm value-added output, productivity and employment. ii University of Ghana http://ugspace.ug.edu.gh DEDICATION I first and foremost dedicate this work to the almighty God for how far he has brought me. Secondly, to the memory of my late mother Janet Esinu Howusu, who was a great source of inspiration to me. You are gone but your belief in me has made this journey possible. Finally to my father Mr. Bernard Nyavor and all my siblings for their financial support, guidance and inspiration throughout the course of the work. iii University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENT I am eternally thankful to the Most High God for showering me with his abundant grace and love, for giving me good health and seeing me throughout the period of my study successfully. I am extremely grateful to my supervisors Prof. Amoah Baah-Nuakoh and Dr. Kwadwo Tutu for their guidance, useful criticisms and valuable comments that have aided me to put this work in good shape. I also express my appreciation to my friends and colleagues especially Manfred Kasapa, Evans Akpah, and Daisy Ofoley Annang who have helped me in various ways throughout my entire academic journey. To Ernest Agyekum, my manager at Options Microfinance Ltd, for providing me with the necessary support during period of the course. God bless you. Finally, I must express my very profound gratitude to my Father, Mr. Bernard Nyavor for being there for me and investing into my education throughout my years of study before and during my master’s degree programme. To my siblings Esenam Nyavor, Dela Nyavor, Mawuli Nyavor, Ewoenam Nyavor and my cousin Josephine Mantey, thank you for providing me with unfailing support and continuous encouragement through the process of researching and writing this thesis. This accomplishment would not have been possible without you. iv University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENT DECLARATION ........................................................................... Error! Bookmark not defined. ABSTRACT .................................................................................................................................... ii DEDICATION ............................................................................................................................... iii ACKNOWLEDGEMENT ............................................................................................................. iv TABLE OF CONTENT .................................................................................................................. v LIST OF TABLES ....................................................................................................................... viii LIST OF ABBREVIATIONS ......................................................................................................... x CHAPTER ONE ............................................................................................................................. 1 INTRODUCTION ....................................................................................................................... 1 1.1 Background ........................................................................................................................ 1 1.2 Statement of the Research Problem ................................................................................... 5 1.3 Objectives of the Study...................................................................................................... 8 1.4 Significance of the Study ................................................................................................... 8 1.5 Organization of the Study .................................................................................................. 9 CHAPTER TWO .......................................................................................................................... 10 OVERVIEW OF THE ELECTRICITY AND THE MANUFACTURING SECTOR OF GHANA. ................................................................................................................................... 10 2.1 Introduction ..................................................................................................................... 10 2.1.1 Electricity Generation In Ghana ............................................................................... 10 2.1.2 Electricity Transmission and Distribution in Ghana ................................................. 16 2.1.3 Electricity Access and Consumption in Ghana ......................................................... 20 2.2 Ghana’s Manufacturing Sector Development. ................................................................ 22 2.2.1 History of Ghana’s manufacturing sector ................................................................. 22 2.3 Conclusion. ...................................................................................................................... 30 CHAPTER THREE ...................................................................................................................... 32 LITERATURE REVIEW .......................................................................................................... 32 3.1 Introduction ..................................................................................................................... 32 3.2 Theoretical Review .......................................................................................................... 32 3.2.1 Output growth and productivity measurement ......................................................... 32 3.2.2 Productivity measures ............................................................................................... 37 3.3 Empirical Reviews ........................................................................................................... 44 v University of Ghana http://ugspace.ug.edu.gh 3.4 Conclusion ....................................................................................................................... 54 CHAPTER FOUR ......................................................................................................................... 56 METHODOLOGY .................................................................................................................... 56 4.1 Theoretical Methodology ................................................................................................ 56 4.2 Empirical Approach ......................................................................................................... 58 4.3 Model Specification for Manufacturing Aggregate level Analysis ................................. 58 4.4 Explanatory Variable Description and Interpretations (Aggregate level Data variables) 60 4.5 Model Specification for Manufacturing Firm Level Analysis ........................................ 64 4.6 Explanatory Variable Description and Interpretations (Firm level Data variables) ........ 65 4.7 Model Estimations ........................................................................................................... 67 4.8 Diagnostic Test ................................................................................................................ 68 4.8.1 Unit Root Test for Stationarity ................................................................................. 68 4.8.2 Test for Heteroscedasticity ....................................................................................... 70 4.8.3 Test for Autocorrelation ............................................................................................ 70 4.8.4 Stability Test ............................................................................................................. 70 4.9 Firm Level Analysis ........................................................................................................ 71 4.10 Data Sources .................................................................................................................. 72 CHAPTER FIVE .......................................................................................................................... 74 ESTIMATION AND DISCUSSION OF RESULTS ................................................................ 74 5.1 Introduction ..................................................................................................................... 74 5.2 Descriptive Statistics of Data .......................................................................................... 74 5.2.1 Time Series Dataset .................................................................................................. 74 5.3 Stationarity Tests ............................................................................................................. 76 5.4 Estimated Results for Model 1: The relationship between electricity consumed by manufacturing industries and manufacturing sector value-added output. ............................. 77 5.5 Estimated Results for Model 2: The relationship between electricity consumed by manufacturing industries and manufacturing sector labour productivity. ............................. 80 5.6 Estimated Results for Model 3: The relationship between electricity consumed by manufacturing industries and manufacturing sector employment. ........................................ 83 5.7 Diagnostic Test ................................................................................................................ 86 5.7.1 Test for Heteroskedasticity ....................................................................................... 86 5.7.2 Test for Serial Correlation ........................................................................................ 86 5.7.3 Stability Test ............................................................................................................. 86 vi University of Ghana http://ugspace.ug.edu.gh 5.8 Manufacturing Firm Level Analysis................................................................................ 87 5.8.1 Data Description ....................................................................................................... 88 5.9 Estimated Results for Model 4: The relationship between electricity expenditure consumption and firm level manufacturing value-added output. .......................................... 88 5.10 Estimated Results for Model 5: The relationship between electricity expenditure consumption and firm level manufacturing labour productivity growth. .............................. 89 5.11 Estimated Results for Model 6: The relationship between electricity expenditure consumption and firm level manufacturing employment growth. ........................................ 90 5.12 Conclusion ..................................................................................................................... 91 CHAPTER SIX ............................................................................................................................. 93 SUMMARY, CONCLUSION AND RECOMMENDATION ................................................. 93 6.1 Summary .......................................................................................................................... 93 6.2 Conclusion ....................................................................................................................... 95 6.3 Recommendations. .......................................................................................................... 96 6.4 Limitations of the Study. ................................................................................................. 96 REFERENCES ............................................................................................................................. 98 APPENDIX ................................................................................................................................. 106 vii University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 2.1: Components of Electricity Generation (% of total supply) ......................................... 12 Table 2.2: Electricity Generation by Plant (GWh) per installed capacity (MW) ......................... 13 Table 2.3: The electricity supply projections for three economic scenarios (2010-2020) ............ 14 Table 2.4: Transmission Losses in GWh and Percentage loses .................................................... 17 Table 2.5: Distribution System Losses (Technical and Commercial Losses) in GWh and Percentage loses ............................................................................................................................ 17 Table 2.6: Electricity Outages per consumer per year (in hours) ................................................. 18 Table 2.7:Total Domestic Energy and Energy Transmitted ......................................................... 19 Table 2.8Annual Peak Demand (MW) ......................................................................................... 20 Table 2.9: Electricity Consumption by Customer Class (GWh)................................................... 21 Table 2.10: Average growth rates of industrial subsectors (1981-2000) ..................................... 28 Table 2.11: Average growth rates of industrial subsectors (2001-2005) ..................................... 29 Table 2.12: Average growth rates of industrial subsectors (2006-2015) ...................................... 30 Table 5.1: Summarized Data Description (Time Series) .............................................................. 75 Table 5.2: Augmented Dickey-Fuller (at levels) ......................................................................... 76 Table 5.3: Augmented Dickey-Fuller (at First Difference) .......................................................... 76 Table 5.4: Phillips-Peron (at Levels) ............................................................................................ 77 Table 5.5: Phillips-Peron (at First Difference) ............................................................................. 77 Table 5.6: ARDL Bounds Test Model 1 ....................................................................................... 78 Table 5.7: Model 1Short-run Coefficients with Error Correction term ........................................ 80 Table 5.8: Model 1 Estimates of Long Run Coefficients ............................................................. 80 Table 5.9: Bounds test for Model 2 .............................................................................................. 81 Table 5.10: Model 2 Estimates of Short-run Coefficients with Error Correction Term ............... 83 viii University of Ghana http://ugspace.ug.edu.gh Table 5.11: Model 2 Estimates of long-run Coefficients .............................................................. 83 Table 5.12: Bounds test for Model 3 ............................................................................................ 84 Table 5.13: Model 3 Estimates of Short-run Coefficients with Error Correction Term ............... 85 Table 5.14: Model 3 Estimates of long-run Coefficients .............................................................. 86 Table 5.15: Summarized Data Description (Panel data) ............................................................... 88 Table 5.16: Model 4 Results from fixed effects estimations ........................................................ 89 Table 5.17: Model 5 Results from fixed effects estimations ........................................................ 90 Table 5.18: Model 6 Results from fixed effects estimations ........................................................ 91 ix University of Ghana http://ugspace.ug.edu.gh LIST OF ABBREVIATIONS ADF Augmented Dickey-Fuller AGI Association of Ghana Industries AIC Akaike Information Criterion ARDL Auto-Regressive Distributive Lag BAF Business Assisted Fund CEL Cenit Energy Limited ECG Electricity Company of Ghana ERP Economic Recovery Programme ERPP Emergency Reserve Power Plant FUSMED Fund for Small and Medium Scale Enterprises Development GDP Gross Domestic Product GIPC Ghana Investment Promotion Centre GLSS Ghana Living Standards Survey GRIDCO Ghana Grid Company Limited GSS Ghana Statistical Service GW Gigawatt GWh Gigawatt-hour HFO Heavy Fuel Oil IDC Industrial Development Corporation ISI Import Substitution Industrialization LCO Light Crude Oil MFP Multi-Factor Productivity MW Megawatt NDPC National Development Planning Commission NEDCO Northern Electricity Distribution Company OECD Organization for Economic Corporation and Development x University of Ghana http://ugspace.ug.edu.gh OLS Ordinary Least Squares PEED Private Enterprises and Export Development PNDC Provisional National Defense Council PP Phillips-Perron PV Photo-Voltaic SAIDI System Average Interruption Duration Index SAIFI System Average Interruption Frequency Index SAPP Sunon Asogli Power Plant SME Small and Medium Enterprises TAPCO Takoradi Power Company TICO Takoradi International Company TIP Trade and Investment Program TT1PP Tema Thermal 1 Power Plant TT2PP Tema Thermal 2 Power Plant TWh Terawatt-hour VALCO Volta Aluminum Company VRA Volta River Authority WAGP West African Gas Pipeline xi University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 Background Electricity is generally considered to be essential to industrialization and growth of any economy. Industrialization as explained by Kuznets (1973) is the development process that underdeveloped countries go through by changing the structure of their economy from an agrarian economy to an industrial one, including additional changes in the society that comes with the structural change. An industrial economic development according to Kuznets is based on progress and expansion of manufacturing industries. Cornwall (1977) described manufacturing as the “engine of growth” because it is the focal point of technological progress, which comes by either learning by doing or through entrepreneurial research activities. He added, manufacturing sector growth and productivity is characterized by economies of scale which reduces the unit cost of inputs and thereby leads to increasing output. The growth of manufacturing output in turn affects the growth rate of an economy through backward and forward linkages to non-manufacturing productivity. Developing countries that are in the process of industrialization are gaining more in terms of GDP from the development and growth of the manufacturing sector (Fagerberg and Verspagen, 1999). This is because whenever there is increased foreign direct investment in infrastructure especially in energy development, continuous technological improvement and investments in human capital to support manufacturing industries, GDP rates grow substantially. However, they argue that the effect of manufacturing on growth in highly industrialized countries is almost absent since the 1 University of Ghana http://ugspace.ug.edu.gh flexibility to shift towards manufacturing production is minute because opportunities to expand manufacturing has been exhausted. The main characteristic of industrial development over the years is that it is based on the growth in the amount of energy utilization per worker principally electricity utilization per worker (Rosenberg, 1998). The significance of electricity in manufacturing has largely been highlighted in most advanced economies who have built very big motor driven manufacturing factories, assembling and processing plants due to the presence of consistent electricity supply serving as a cheaper available source of power to these industries. However, manufacturing industries in developing economies especially in most African countries including Ghana have very low electricity utilization per worker even though they are endowed with several possible ranges of energy sources (Moyo, 2013). Several studies including Allcott et al. (2016), Cisshoko and Seck (2013) and Fisher-Vanden et al. (2015) have explored access to electricity use in developing countries especially in African countries and have found negative implications on their manufacturing industries as a result of inadequate electricity generation in these developing countries. African countries therefore experience unstable economic growth rates and are not able to take advantage of the potential of fast and sustainable economic growth and development. The manufacturing sector of Ghana has gone through several phases from pre-independence system of manufacturing through to post-independence Import Substitution Industrialization (ISI) strategy to the practice of private sector-led industrialization. Ghana’s manufacturing sector is mainly dominated by small and medium privately-owned firms, which are largely found within urban centres in the form of industrial clusters and interspersed with some large factories (Ackah, et al, 2014). The manufacturing sub-sector has been and still is the biggest component of Ghana’s 2 University of Ghana http://ugspace.ug.edu.gh industrial sector in terms of the number of firms, value addition and employment since independence. This is the case even though data from the Ghana Statistical Service (GSS) has shown that over the last few years, manufacturing growth has been declining. Construction and mining and quarrying, the other sub-sectors of Ghana’s Industrial sector, have gradually overtaken the manufacturing sub-sector in terms of their contribution to GDP as a result of the discovery and subsequent production of oil and gas. Data from GSS (2016) shows that economic activity within the Industrial sector saw Mining and Quarrying increase its share of GDP from 2.8% in 2006 to 6.4% in 2015 while the manufacturing sub-sector decreased its share of GDP from 10.2% in 2006 to 4.7% in 2015. Ghana’s economy seems to follow the “engine of growth” hypothesis since majority of the time, increases in manufacturing growth increases GDP growth rates and vise- versa as seen in the national statistics (GSS, 2016). One main reason for the continuous shrinking in the growth of the manufacturing sector is the poor nature of infrastructure across the country particularly electricity, viable source of water, good road and rail networks. Primarily, inadequate electricity supply leading to severe power outages and increased cost of electricity consumption for those who acquire power plants, hamper Ghanaian manufacturing sector activities (Amoako-Tuffour and Eshun, 2016). Industries within the manufacturing sub-sector consequently suffer from loss of revenue, employment and business failures. This is due to the fact that they are not able to produce output at the level that matches at least with their variable cost of production which involves paying wages, acquiring raw materials and maintenance of equipment. Ghana therefore has a large potential to transform and develop its economy through industrialization with the aim of creating employment and ensuring equitable distribution of wealth. Investing in infrastructure especially in electricity which is the biggest driver of 3 University of Ghana http://ugspace.ug.edu.gh manufacturing worldwide is seen to be the ideal catalyst for growing the economy (Ministry of Energy, 2010). Ghana’s manufacturing sector over the years has relied mainly on electricity in the production of goods and services even though some firms make use of wood fuel, residual fuel and petroleum. Electricity is mostly preferred because it serves as a better source of energy in terms of its cleanliness, efficiency, lower cost, its ease of use and how it promotes high worker productivity in relation to other energy sources. Considering the formal manufacturing sector, the major industrial energy used was electricity ranging between 55% and 56% followed by petroleum products which averaged between 39% to 42% and wood fuel which constituted about 5% as at 2006, (Ghana Energy Commission, 2006). Until 2003, the Volta Aluminum Company Ltd (VALCO) accounted for 50 to 60% of the manufacturing sector’s share of electricity consumed which represented 17% share of industrial energy, however fell below 2% due to suspension of smelter operations. VALCO mainly uses electricity needed for aluminum smelting (Ghana Energy Commission, 2006). Within Ghana’s industrial sector, manufacturing sub-sector’s share of electricity consumed was about 14%, whilst the mining & quarrying sub-sector increased its shares to 23% per annum in 2006 from 2002. Currently, the industrial sector of Ghana consumes about 48% of total electricity supplied to consumers. Ghana’s power sector is made up of a system controlled by separate establishments and power supply bodies which deal with the tasks of electricity generation, transmission and distribution. These establishments are the Ghana Energy Commission, Volta River Authority (VRA), Ghana Grid Company Limited (GRIDCo), and the Electricity Company of Ghana. The power sector has been bedeviled with inadequate power supply leading to power crisis for some time because of the over-reliance on hydro-electric power generation and lack of adequate investment in infrastructure 4 University of Ghana http://ugspace.ug.edu.gh especially in transmission and distribution systems (Amoaku-Tuffour and Eshun, 2016). Some other factors contributing to the power supply inefficiencies in Ghana are, poor supervisory and regulatory activities, electricity sector mismanagement, inability to enforce law on illegal utility consumers, and operational difficulties in the supply of electricity. The power insecurity problem has been detrimental to Ghana’s developmental ambitions as Ghana loses an amount of around US$ 2.2 million per day in revenues or US$ 686.4 million annually on the average according to ISSER (2015). It will take an investment of about $4 billion (2% of yearly GDP) into the electricity sector spread throughout the next decade to cover existing power deficits to bring an end to the power crisis (ISSER, 2015). It is therefore important to analyze the reaction of the manufacturing sector to electricity in line with total manufacturing activity in Ghana. In view of this, the study discusses the manufacturing sector highlighting the historical development of the sub-sector from Ghana’s independence to date in relation to the activities of the electricity sub-sector over the same period. The study also outlines the state of Ghana’s electricity sub-sector development including its current electricity supply and demand and how the electricity supply constraints may affect Ghana’s manufacturing growth and development. The study therefore investigates whether Ghana’s manufacturing performance and her electricity industry, given generation and supply over time, have a significant relationship existing between them or not and whether they move together in the same direction or not. 1.2 Statement of the Research Problem With the process of industrial expansion in Ghana, businesses tend to invest more in physical capital (machines and equipment) to improve their productivity. This makes electricity an important input in their production process as electricity is needed to power these capital inputs. 5 University of Ghana http://ugspace.ug.edu.gh In Ghana, power outages per consumer per year in terms of duration measured by System Average Interruption Duration Index (SAIDI) and in terms of number of outages measured by the System Average Interruption Frequency Index (SAIFI) on average is very high. The highest average duration of outages in Ghana recorded was 716 hours in 2012. Outages affect firms especially SMEs who tend to make a lot of losses more than large firms do when there are power outages (Scott et al., 2014). Apart from outages there is a high number of interruptions and fluctuations in voltage, which could lead to serious impairment to machinery and equipment, and a higher associated cost with frequent repairs and replacement. Inadequate Power Supply emerged as the leading problem restricting growth and start-up of businesses operating in Ghana in 2013, according to the AGI Business Barometer Survey result (2013). Most industrial firms complained of persistent power cuts which started in 2012 and increased in intensity over the first six months of 2013. These disruptions in power supply led to a short fall in production and revenue loses to firms especially those who were unable to procure generators to continue operations. Also, firms that could afford to invest in emergency plants and fuel to undertake normal operations incurred increased cost of doing business. The period between 2013 and 2015 saw manufacturing sub-sector grow at a negative rate from -0.5% to 2% and industrial sector growth plummeting severely from 16% to 3%. GRIDCo (2010) report attributed some of the reasons for the persistent power outages over the years to many years of prolonged underinvestment in generation and transmission infrastructure despite fast growth in demand and electricity consumption. Low tariffs charged over the years also contributed to limited capital for operations and investment and poor quality of service. In addition, the lack of competition characterized by a monopolized energy system played a big role in the poor performance of the sector. Furthermore, there were managerial decisions often influenced by 6 University of Ghana http://ugspace.ug.edu.gh government, usually for political gains, that were not in line with the utilities’ management and business objectives. Inadequate power supply affects production activities through various ways which negatively affects productivity in the long run. One of such ways is where interruptions in power supply disturbs the production process of firms causing productive resources to lie idle, resulting in lower output growth. Secondly, as a result of many power interruptions within a day, costs are incurred when machines malfunction and are needed to be fixed or replaced, or when finished products or inventory gets spoilt. Furthermore, power outages bring about extra input costs to firms, because they often depend on alternative source of power like rented or self-owned generators. Power outages in Ghana also tend to be abrupt and this brings about uncertainty in production where manufacturing firms cannot predict with accuracy the time of occurrence of power outages. Even though power is not supplied at a constant rate and producers have to contend with the uncertainty of production timelines, producers end up paying exorbitant charges placed on electricity sold to them by the distributor. Higher prices of electricity have come about in recent years due to high tariffs. This situation increases the likelihood of firms not meeting set targets in terms of output leading to reduced revenue, and production of low quality goods. Consequently, production cost further increases, and fewer workers are employed. It is evident from the above that electricity plays a role in the production processes of manufacturing industries in Ghana. However, most studies on the Ghanaian economy do not focus on the extent to which improvement in electricity supply or otherwise relates to the performance of the Ghanaian manufacturing sector. The focus of these studies mostly explain the impact of electricity consumption on total economic growth in Ghana, hence this study seeks to fill this research gap. It is therefore important to show how electricity affects Ghana’s economic growth 7 University of Ghana http://ugspace.ug.edu.gh at a sectoral or micro level, specifically on how electricity affects manufacturing performance in Ghana over the years since it has been shown that manufacturing growth is a catalyst for faster economic growth. 1.3 Objectives of the Study The main objective of this study is to analyze the relationship between manufacturing electricity consumption and manufacturing sector performance in Ghana both at the aggregate and firm level using these specific performance indicators: value-added output growth, labour productivity growth and employment growth. The specific Objectives are:  To study the relationship between electricity consumption and manufacturing value-added output growth.  To investigate the relationship between electricity consumption and manufacturing labour productivity growth.  To examine the relationship between electricity consumption and manufacturing employment growth. 1.4 Significance of the Study There is enough evidence to suggest that there are enormous problems in the electricity sector including poor quality and inadequate electricity supply. Most manufacturing industries in Ghana use electricity in their production of goods and services. Power outages per hour in Ghana have been increasing significantly in the past decade and its intensity was mostly felt from 2013 to 2015. These outages imply that, firms had to procure other sources of energy to augment electricity 8 University of Ghana http://ugspace.ug.edu.gh supply which increases their cost of production. This study is therefore to understand the extent to which manufacturing firms and the entire sector in Ghana perform given the amount of available electricity that is supplied to them. 1.5 Organization of the Study This study is organized into six chapters. It is outlined as follows, chapter one provides background information on the topic, problem statement, and objectives. Chapter two discusses an overview of the electricity sector and the development of manufacturing sector in Ghana. Reviews of relevant literature done on the subject matter are presented in chapter three. Chapter four focuses on the methodology applied, models formulated, econometric estimation methods employed in the study, variables used and data sources. The results and discussions of the study are presented in Chapter five. Chapter six outlines the summary, conclusions and recommendations derived from the study. 9 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO OVERVIEW OF THE ELECTRICITY AND THE MANUFACTURING SECTOR OF GHANA. 2.1 Introduction Electricity is necessary in the day to day activities of people and economic agents in the world. Electricity is used in domestic activities from lighting to powering of technological equipment like phones and computers, used in offices and in industries including manufacturing. However, ensuring that electricity gets to the final consumer involves a multifaceted network of infrastructure that needs to be continuously managed and coordinated. This is why an insight into the development of the electricity sector in Ghana is ideal for this section of the overview. There are three steps that electricity undergoes before getting to the end user. They are the generation, transmission and the distribution processes. All three processes will be analyzed from the perspective of Ghana’s electricity sector. The demand side will also be discussed with emphasis on the industrial sector. Statistics on electricity production and consumption rates of different countries is looked at for comparative analysis. Finally, an analysis of the manufacturing sector will be undertaken. 2.1.1 Electricity Generation In Ghana The Volta River Authority (VRA) of Ghana indicates that total installed generation capacity as at October 2016 stood at 3,644 Megawatts (MW). This comprises VRA installed generation capacity of 2,434 MW and independent power producers’ capacity of 1,210 MW. The Akosombo Hydro plant which has been the main source of power generation since 1966 has a current generation capacity of 1,020 MW. Other hydro electricity generation plants are the Bui and Kpong hydro plants with generating capacities of 400MW and 160MW respectively. All the hydro plants 10 University of Ghana http://ugspace.ug.edu.gh contribute 50.86% share of the installed electricity generation capacity (Ghana Energy Commission, 2016). Apart from the hydro electricity generating plants there are the thermal and other renewable sources of electricity. The current thermal plants in Ghana are the Takoradi Power Company (TAPCO), Takoradi International Company (TICO), Takoradi T3, Tema Thermal 1 and 2 Power plants (TT1PP and TT2PP), Tema Reserve power plant (TRPP), Kumasi Reserve Power Plant (KRPP), Emergency Reserve Power Plant (ERPP), Sunon Asogli Power (Ghana) Ltd (SAPP) and Cenit Energy Ltd (CEL). All these thermal plants have installed capacity of 1,248 MW contributing 49.1% of the total electricity generation capacity. In terms of electricity generated from renewable resources, some solar PV systems have been installed over the years including systems for lighting health centres and schools, Solar streetlights, Rural and urban Solar home systems, and Solar Water Pumps. The major solar panels that contribute to the national electricity grid is the 2.5MW VRA Navrongo solar system and privately owned 20 MW BXC Chinese solar system. Wind energy source is not contributing to national electricity grid even though there are wind Power installations in places like Anloga, Anyanui, Lekpoguno and Akplabanya. Renewables excluding hydro, generates 2.5MW of electricity contributing 0.04% of total generation capacity. Electricity from renewables is electricity that is generated from renewable resources which can be replenished naturally, such as sunlight, water, wind, tides and geothermal heat. The relative importance of electricity from hydro source in the overall electricity supply has been declining, as more thermal plants have been added to the total generating capacity in recent years. For example, the share of power from hydro sources was estimated at 51.14% in 2015, compared 11 University of Ghana http://ugspace.ug.edu.gh with 67.5% in 2011 and 75% in 2009, while the share of power from thermal sources increased from 22.7% in 2009 to 48.83% in 2015 as can been seen in table 2.1 below. Table 2.1: Components of Electricity Generation (% of total supply) Sources 2009 2010 2011 2012 2013 2014 2015 Hydro 75 69 67.5 67.1 63.97 64.98 51.14 Thermal 22.7 31 32.5 32.9 36.01 34.99 48.83 Renewable 0.01 0.01 0.01 0.01 0.02 0.03 0.03 Table Source: Compiled by author from NDPC Annual Reports (2010-2016) The primary source of electricity generation had mainly been hydro until 1998 when more thermal plants were installed. As at 2016, the main fuel needed for powering thermal generating plants on the Ghana power system were the natural gas, Light crude oil (LCO), diesel and Heavy fuel oil (HFO). Natural gas is the fuel mostly used to power thermal plants and it is sometimes used as supplements to plants that use other fuels (Ghana Energy Commission, 2016). As at 2015, total gas needed for fueling the thermal plants was 46,912 mmscf, that is, almost twice that of 2014 of 23,631 mmscf. About 44% was from Nigeria via the WAGP and the remaining 56% coming from the Atuabo gas processing plant where the gas carried from the Jubilee field is processed by the Ghana National Gas Company in Atuabo. An estimated requirement of about 5.9 million barrels of LCO, 1.51 million barrels of Diesel oil and 2.8 million barrels of heavy fuel oil were needed to cut the deficits in natural gas supply. In all, about $1.18 billion was estimated to be used to procure fuel for electricity generation in 2016 (Ghana Energy Commission, 2016). From table 2.2 below, it is seen that electricity supply from hydro sources has decreased significantly from 8,387 Gigawatt-hour (GWh) in 2014 to 5,845 GWh in 2015 caused by decreased water levels in the Akosombo and Kpong Dam reservoirs. This affected overall electricity supplied from 12,963GWh in 2014 to 11,492GWh in 2015 even though electricity supply from thermal 12 University of Ghana http://ugspace.ug.edu.gh generating sources increased from 4,572GWh in 2014 to 5,845GWh in 2015. It is also seen from the table that electricity generation capacities of plants have more than doubled from the 2006 figure of 1,730 MW to 3,656 MW in 2015. However, due to increasing exponential growth in population and demand for electricity consumption, there has been the need to step up generation capacities to meet the increasing demand. Table 2.2: Electricity Generation by Plant (GWh) per installed capacity (MW) Source: (GRIDCo, 2016) The institution mandated to have the sole responsibility for the generation of electricity in Ghana is the Volta River Authority. The Authority operating under the Volta River Development Act, Act 46 of the Republic of Ghana in 1961 initially had the full responsibility to generate, transmit and distribute electricity (VRA, 2015). The Act was later on amended to its current status in 2005 which included the ability to draw Independent Power Producers (IPPs) onto the Ghana energy market. The VRA’s primary customers are the Electricity Company of Ghana (ECG) and Northern 13 University of Ghana http://ugspace.ug.edu.gh Electricity Distribution Company (NEDCo). Electricity is also made available in bulk to some mining companies such as; AngloGold Ashanti, Goldfields Ghana Ltd., Newmont Ghana Gold Ltd, and Golden Star Resources Group. Others are Aluworks, Akosombo Textiles Ltd., and Diamond Cement Ghana Ltd (VRA, 2015). Electricity supply projections were made to plan ahead of time to be able to help solve the problems of inadequate power supply. The projected time frame was between 2006 and 2020. The table below shows electricity supply projections made within this period after factoring in transmission and distribution losses. Table 2.3: The electricity supply projections for three economic scenarios (2010-2020) GROSS ELECTRICITY SUPPLY REQUIREMENT in (GWh) Year Business-as-usual, or Moderately high economic GPRS High economic Low growth growth economic growth Without Valco Without Valco Without Valco Valco 3–4potlines Valco 5 potlines Valco 5 potlines in in 2010 in2008-12 2008-12 5 potlines 6 potlines 6 potlines onwards onwards onwards 2010 8,502 12,846 13,848 17,484 19,194 22,500 2011 8,904 13,304 14,488 18,243 20,072 23,370 2012 9,325 13,786 14,600 18,500 20,144 24,650 2013 9,768 14,294 14,990 19,500 20,210 24,770 2014 10,233 14,828 15,676 20,200 21,120 25,620 2015 10,721 15,408 16,398 20,900 22,074 26,600 2016 11,234 16,000 17,155 21,660 23,077 27,600 2017 11,773 16,623 17,951 22,598 24,130 28,630 2018 12,340 17,280 18,787 23,613 25,235 29,948 2019 12,934 17,970 19,666 24,682 26,398 31,394 2020 13,560 18,714 20,590 25,815 27,620 32,915 Source: Ghana Energy Commission: Strategic National Energy Plan According to the Ghana Energy Commission 2016, projections of total electricity supply requisite for Ghana with VALCO operating at one potline for the year 2016 was revised to between 16,798 14 University of Ghana http://ugspace.ug.edu.gh to 16,900 GWh to achieve an expected marginal economic growth of 4.0-4.5% in 2016 over the previous year. Also projections with VALCO operating with at most two potlines was revised to between 18,185-18,737 GWh to raise the economic growth to over 4.5%. This shows the changing economic environment in the country through the expansive nature of the required electricity demand. Electricity supplied to household and industrial consumers for the past decade have fallen below the projected electricity requisite for the country and thus meeting the required target for 2020 is less likely to be achieved. Statistics from the Ghana Energy Commission’s 2016 energy outlook for Ghana showed that the aggregate electricity produced for gross transmission was only 12,927 GWh in 2013 which improved to 13,071 GWh in 2014 but fell by 12% to 11,692 GWh in 2015. The net grid electricity made available for transmission in the country amounted to 12,906 GWh in 2014 which was 10% more than that transmitted in 2015 that is 11,678 GWh. The net grid electricity available for that of 2015 was about 21 to 26% less than the projected electricity generation required for low economic growth status with VALCO operating at one potline. Comparatively, Ghana’s current electricity generation and supply capacity currently is 14 Terawatt-hour (TWh) which is in the same range as many other developing African countries if not better. According to Enerdata (2016), most African countries electricity generation fall below the 30 TWh threshold apart from Nigeria (33 TWh), South Africa (248 TWh), Egypt (185 TWh) and Algeria (74 TWh). Other African countries with lower electricity production are, Kenya (9 TWh), Cote d’Ivoire (6 TWh), Zambia (13 TWh), Angola (6 TWh), and Malawi (2 TWh). On the other hand, data shows highly industrialized countries have very huge generating capacities like United States of America, China, Russia, Japan, Germany and Canada have generating capacities of 4,324 TWh, 5,682 TWh, 1,062 TWh, 985 TWh, 638 TWh and 632 TWh respectively. Evidence 15 University of Ghana http://ugspace.ug.edu.gh shown here suggests that there is a wide gap between the developed and developing countries in terms of electricity generation. 2.1.2 Electricity Transmission and Distribution in Ghana Electricity transmission and distribution forms an important part of the energy sector since electricity is not generally utilized in the same place that it is generated and this is the means to which industrial users and households can have access to power. Energy is embedded in the fuel itself before it is transformed to electricity and once altered, it must be transferred onto the power lines or else will be lost. This suggests that very reliable long-distance transmission lines and distribution systems are necessary for the realization of a vibrant power sector. All over the world, countries whose energy sectors are adequately financed and structured have very good transmission terminals, cables and lines. They also have competitive power distribution centers. Transmission and distribution losses are major problems in Ghana. GRIDCo (2010) grouped transmission and distribution losses experienced in Ghana into two main types, namely technical and commercial losses. Commercial losses are mainly due to distribution activities, while technical losses occur in all phases of the supply chain that is generation, transmission, and distribution. Technical losses are the losses experienced through low efficiency transformers, conductors and cables. In Ghana, technical losses can be as a result of overburdened power lines. These losses can also be reduced if there is transfer of power from one line to another when there is detection of faults on a transmitting cable or equipment and using Circuit breakers to disconnect power to prevent damage from overloads. Commercial losses are dominant in Ghana’s electricity system and these losses are as a result of cable and wire theft, illegal connections, faulty meters, and inaccuracies in accounting for power 16 University of Ghana http://ugspace.ug.edu.gh utilization. A higher percentage of commercial losses can however be attributed to illegal connections and unpaid consumption. Total losses make up of an average of 24% of electricity demand in Ghana. Most of the losses are from distribution losses. The losses will keep increasing if there are inadequate maintenance of lines and cables by the utility body. The best utility practices around the world have the technical line loss during transmission and distribution at 7 to 8%. In comparison, electricity line wastages are most of the time restricted to only 6.5% of demand in the United States of America. Table 2.4: Transmission Losses in GWh and Percentage loses Year 2008 2009 2010 2011 2012 2013 2014 2015 Transmission losses 303 343 380 531 522 569.7 565 402 Losses (% of net generation) 3.5 3.8 3.7 4.7 4.3 4.8 4.3 3.8 Table Source: Compiled by author from Ghana Energy Commission, VRA and GRIDCo Table 2.5: Distribution System Losses (Technical and Commercial Losses) in GWh and Percentage loses Year 2008 2009 2010 2011 2012 2013 2014 2015 ECG (losses) 1,483 1,570 1,799 1,974 1,865 1,983 2,108 1,713 Losses (%) 25.6 25.9 26.6 27.2 23.5 23.4 25.2 22.7 NEDCo (losses) 138 153 124 139 165 200 240 273 Losses (%) 26.0 27.1 19.5 19.3 20.1 21.3 24 27.5 Table Source: Compiled by author from Ghana Energy Commission, ECG and NEDCo Customers on the distribution system are grouped into industrial, commercial and residential units. Since Industrial customers consume more electricity than residential units and industrial consumption is relatively stable throughout the day, industrial demand is considered to be a base load. Large scale industrial customers sometimes obtain electricity direct from the transmission system instead of the local distribution system. This makes industrial power less expensive load to serve. Household users and small scale businesses in turn have very low voltages and their power consumption is not very stable which varies from day to day depending on need of the individual, 17 University of Ghana http://ugspace.ug.edu.gh equipment use and seasonal economic changes (ISSER, 2005). An electricity distribution system is seen to be an efficient one when there is continuous or reliable power supply, voltage stability in the system and when it brings about lower cost in power sales. Supply of electricity in Ghana has been very inconsistent over the years indicated by the number of power outages experienced in the country. This is a result of power system shortages and inefficiencies and inability to manage transmission and distribution inefficiencies. Total outages per consumer per year have been very high, increasing from 413 hours/customer/year in 2010 to 1,619 hours/customer/year in 2014. The average number of hours of electricity outage per customer per year worsened significantly reaching an average of 623 and 653 hours/customer/year in the rural and urban areas respectively in 2014 in areas under ECG control. In NEDCo areas, electricity outages were 170 hours/customer/year in 2014 among urban customers, whilst in rural areas it was 173 hours/customer/year. This deteriorating situation was because of the power crises accompanied by the load shedding. Table 2.6 shows the data on hours of electricity outages per consumer per year in both rural and urban areas in Ghana. Table 2.6: Electricity Outages per consumer per year (in hours) Year ECG NEDCo Total 2009 78(51) 150(134) 413 2010 104(66) 97(89) 358 2011 185(68) 106(102) 461 2012 206(190) 141(160) 697 2013 143(174) 146(95) 558 2014 623(653) 173(170) 1,619 2015 203(136) n/a 339 Source: Compiled by author from NDPC Annual Reports (2010-2016) NB: urban statistics on outages in bracket and rural statistics on outages without bracket 18 University of Ghana http://ugspace.ug.edu.gh As at 2015, Ghana’s transmission network was made up of over 4,500 kilometers of 161 kV high voltage transmission lines, 429 kilometers of 330 kV transmission lines, and 138 kilometers of 69 kV transmission lines. According to GRIDCo, these transmission lines are connected to 53 major transformer substations and switching stations with a total installed transformer capacity of 3,870.86 MVA. With regards to sharing power across borders, electric power from Ghana is connected to Côte d’Ivoire through a single circuit 225 kV transmission interconnection and connected to Togo and Benin using a double circuit 161 kV transmission interconnection. Ghana also provides electric power to Burkina Faso with a high voltage transmission system using a low voltage distribution network. In all, Ghana is a net exporter to Togo and Benin, and a net importer from Côte d’Ivoire. Total electricity transmitted by GRIDCo in 2015 was 11,216.546 GWh out of which 10,652.346 GWh was consumed within Ghana. Overall electricity exported to neigbouring countries totaled 564.200 GWh. There was an increase in electricity transmitted for consumption within Ghana by 40.6% from 7,577.358 GWh in 2008 to 10,652.346 in 2015. Maximum peak demand in 2015 was 1,933 MW indicating a 10.54% rise over the 2008 peak demand of 1,367MW. Below is a table that shows figures of transmission activities in Ghana since 2007. Table 2.7: Total Domestic Energy and Energy Transmitted 2008 2009 2010 2011 2012 2013 2014 2015 Total Domestic 7,577.358 8,017.466 8,811.141 10,027.607 10,027.607 11,687.658 11.870.00 10,652.346 Energy Total Export 538.021 766.611 1,036.289 774.991 774.991 602.016 744.51 564.200 Energy Total Energy 8,115.379 8,784.077 9,847.430 10,802.598 10,802.598 12,289.674 13,070.00 11,216.546 transmitted Table Source: GRIDCo 2015 19 University of Ghana http://ugspace.ug.edu.gh Table 2.8: Annual Peak Demand (MW) 2008 2009 2010 2011 2012 2013 2014 2015 Ghana Load at Peak 1,208 1,263 1,391 1,520 1,658 1,791 1,853 1,757 System Peak 1,367 1,423 1,506 1,665 1,729 1,943 1,970 1,933 Source: VRA & GRIDCo The Ghana Load at Peak is made up of maximum electricity demand for Ghana including (ECG, NEDCo, Direct Customers of VRA and mines) and System Peak is made up of Ghana Load at Peak in addition to Export Load. Load at Peak describes the period where electricity is expected to be supplied for continuous period at a significantly higher than average supply level. Peak demand describes a period of simultaneous, strong consumer demand. System at Peak is the highest electricity that is supplied within the year. In 2015, the System Average Availability recorded was 99.46%, which was above the PURC’s stipulated target of 95%. The system availability recorded is a drop in the 2014 figure of 99.79%. The drop was due to interruptions in power supply to customers because of a shortfall in available generation. The short fall in power supply with its consequent load management regime also resulted in a reduced feeder availability for the year. System Average Availability shows the probability that a supply system will work as and when required during the period of transmission and distribution recorded normally within the year. 2.1.3 Electricity Access and Consumption in Ghana Access to electricity in Ghana has been increasing over the years. The proportion of the population with access to electricity increased from to 67% in 2010 to 80.5% in 2015 but didn’t reach the medium term target of 85%. In terms of access to electricity, the main aim is to attain 100% universal electrification nationwide by 2020. This is to be achieved through a decentralized 20 University of Ghana http://ugspace.ug.edu.gh renewable energy production mainly solar and off-grid electricity services (Ghana Energy Commission, 2006). The Energy Plan put in place is aimed at improving the quality and reliability reaching 95% uninterrupted power supply to households and industrial sector per annum. Data already seen from table 2.6 shows that as of early 2014 there were a lot of power cuts that affected industries especially the small and medium industries. From the perspective of electricity consumption in Ghana, consumption is put into four consumer classes namely Residential, Non-Residential, Industrial and Street Lighting. Below is a table that shows the various classes and the amount of electricity they consume in GWh per year Table 2.9: Electricity Consumption by Customer Class (GWh) Year 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Residential 2,130 2,095 2,269 2,418 2,738 2,761 2,803 3,228 3,223 2,437 Non-Residential 790 802 927 884 966 1,041 1,153 1,525 1,522 1,532 Industrial 3,593 2,687 2,963 2,921 3,156 3,900 4,153 4,224 4,055 4,144 Street Lighting 144 137 171 184 264 274 315 377 382 534 Total in GWh 6,657 5,721 6,330 6,407 7,124 7,976 8,424 9,354 10,182 8,646 Source: Energy Commission of Ghana (2015) From table 2.9 above, it is seen that electricity consumption by industrial users increased from 3,593 GWh in 2006 to 4,144 GWh in 2015 given a difference of 551 GWh representing only 13.3% increase from 2006. From this observation, it is realized that even though value-added industries are growing rapidly and their demand for electricity consumption is also increasing, the rate of electricity supply to industries over the decade is not encouraging. Electricity consumption by residential users on the other hand increased by 65% from 1,956 GWh in 2005 to 3,223 in 2014 but fell by 24% in 2015. The non-residential electric consumption more than doubled from 676 21 University of Ghana http://ugspace.ug.edu.gh GWh in 2005 to 1,522 GWh in 2014 while street lighting consumed a small rate of 534GWh of electricity in 2015. Developed nations like the United States of America, United Kingdom, Canada, and Finland according to Enerdata (2016) had an electricity consumption per capita of 12,988.3 kilowatt-hour (kWh), 5,407.3kWh, 15,519.3kWh and 15,509.7kWh respectively. Ghana’s electricity consumption per capita at 382.3 kWh is much higher than that of Cote d’Ivoire standing at (278kWh), Kenya (167.7kWh) and Nigeria (141.87 kWh). The highest electricity consumption per capita realized for Ghana was 425.94 kWh in 1980 when the population was 10.8million. Brazil, a developing South American country has seen an improvement in its electricity consumption rate per capita from 455.6 kWh in 1971 and 2,529 kWh in 2013. One can notice that countries who have attained energy sufficiency are far advanced in terms of industrial and economic development than countries that have barely met their energy consumption needs. 2.2 Ghana’s Manufacturing Sector Development. 2.2.1 History of Ghana’s manufacturing sector Industrial development entails activities of consistent expansion of new industries with the view to increasing economic growth, creating jobs and reducing poverty. Ghana’s industrialization process has however been stagnating due to the inability of the state to execute efficiently programs for sustainable growth of industries over the years. Ghana’s manufacturing sector before independence was underdeveloped and did not contribute substantially to economic growth since the government then concentrated solely on extraction of raw materials for export to Britain while importing most of the manufactured goods (Ackah et al., 2014). Most of the firms established before 1950 were in the wood and cork industries, exporting 22 University of Ghana http://ugspace.ug.edu.gh mainly logs of timber. According to Baah-Nuakoh (1997), the enactment of the Gold Coast Industrial Act of 1947 marked the first conscious effort on the part of the Government to affect the course of industrialization. This Act, inter alia, established the Industrial Development Corporation (IDC) and charged the Corporation with the task of "securing the investigation, formulation and carrying out of projects for development industries in the Gold Coast”. Also the IDC was tasked to operate as a loans agency where they were to lend small loans to individual private industries. In 1951 the Ghana government allocated 4.6% of total planned expenditure to industrial development connected with mainly manufacturing which later was increased to 7.6% of the total planned expenditure (1959 to 1964). Despite this, the growth in manufacturing was not substantial (Ewusi, 1981). In the late 1950’s, it was discovered that the development of the economy and of industries, in particular could not be left entirely with the private sector. After Ghana gained independence in 1957, the government adopted the creation of import substitution industries dominated by state-owned capital intensive manufacturing industry. In order to make significant improvement in the development of manufacturing in Ghana, basic services such as communication, power and water were considered important. This led to the establishment of the Volta River Project where the Akosombo hydro power plant was built in 1966 to provide electricity to back the industrialization processes. The Volta River Project was established as a fundament tool to developing an integrated aluminum industry using the vast bauxite reserves exploited and its hydroelectric potential. The construction of the aluminum smelter, the Volta Aluminum Company (VALCO) in 1964 at Tema was thus the largest investment made in the manufacturing industry. VALCO became the main consumer of hydroelectricity from the Volta River Project, consuming 60% of electricity from Akosombo hydropower plant to produce about 200,000 tons of aluminum yearly during the 1970’s. However, 23 University of Ghana http://ugspace.ug.edu.gh VALCO’s operations have experienced many shutdowns and restarts largely due to unavailability of sufficient electricity to run at full capacity at all times. The plant has been shut down and restarted thirteen times between 1984 and 2004 (VALCO, 2017). Currently, only one of the five operating potlines is in use. The Import Substitution Industrialization (ISI) strategy ensured the exploitation of Ghana’s natural resources to be processed into finished goods for local consumption and exports. Manufacturing firms were mostly light non-durable consumer goods made up of food, beverages, tobacco, textiles and clothing (Ewusi, 1981). There was also the introduction of the chemical, electronics, machine and electrical equipment industries. Furthermore, the Government provided some appreciable level of industrial development in about twenty district and local administrative centres by building manufacturing enterprises where raw materials were readily available. Some of the factories built under the ISI strategy were the Akosombo Textiles Limited for the purpose of reducing the importation of foreign textiles into the country and the GIHOC Fibre Products Company to manufacture sacks to produce shopping bags, money sacks and most importantly sacks for export of agricultural products like cocoa and maize. The following Food processing, canning and food packing industries were established; the tomato and mango canning factories at Wenchi, meat processing factory at Bolgatanga, cocoa processing factory at Takoradi, and Asuatuare sugar factory. The Kumasi shoe factory, the Glass factory at Aboso and Tema steel works were some other well-known manufacturing factories that were established. Lastly, some assembling plants were put up to supply tools and equipment to larger factories such as VALCO. Apart from VALCO, all the firms established under the ISI were state-owned and state-run Substantial manufacturing growth was experienced after the implementation of the ISI strategy. The manufacturing sector grew rapidly and contributed 9% share of GDP in 1969 from 2% in 24 University of Ghana http://ugspace.ug.edu.gh 1957. During the 1960s, manufacturing output grew at an average rate of 13% per annum, while its share in overall industrial output grew from 10% in 1960 to 14% in 1970. This also led to a fast growth of manufacturing sector employment by about 90% between 1962 and 1970. Due to this, industrial sector employment averaged 8% per annum within that period (Ackah et al., 2014). The direct intervention of government in the early 1960’s led to the growth in contribution to gross manufacturing output by the fully state owned firms from 11.8% in 1962 to 24.1% in 1967. The growth rate in manufacturing by joint public-private sector also increased from 7.1% in 1962 to 17.5% in 1967. However private-owned firms fell from 80.9% in 1963 to 58.2%.in 1966. The number of establishments rose from 167 in 1962 to 382 in 1974. In 1972, value added in the medium and large-scale establishments accounted for 8.7% of GDP. The emphasis on the state as the engine of growth was changed as a result of the military coup d’etat in 1966. The Private sector was given the responsibility between 1968 and 1970 to spearhead the development of industries because many public enterprises were considered to be unproductive and were either abandoned or sold over to private entrepreneurs. The Military coup that overthrew the Busia Regime of 1969-72 changed the strategy of development of the nation. The new government put emphasis on empowering Ghanaians by assisting them to own manufacturing enterprises. All foreign enterprises in certain specified industrial sectors especially those in value- addition and food processing were transferred to indigenous Ghanaians and all dormant state industries were revamped. The period starting from the mid-1970’s to 1983 saw Ghana’s economic and financial situation declining throughout the period. Severe drought reduced the electricity generating capacity of the Akosombo Dam which led to a temporary shutdown of the VALCO aluminum smelter. As noted earlier, aluminum production did not get better after the temporary shutdown and this continued 25 University of Ghana http://ugspace.ug.edu.gh through early 1990s till date. This was compounded by external shocks from international markets and inadequate internal policies such as huge fiscal deficits which were supported predominantly by borrowing from the domestic financial sector leading to high levels of inflation at the same time crowding out of private investment. There was also an existence of an over-valued exchange rate, increasing lending rates during the 1970’s and the after effects of the problems that the ISI strategies brought. The ISI strategy by design did not encourage the growth of agriculture and exports. Also, the import-dependent industries became ineffective at exploiting domestic resources because of the effective protection that was given to them under the ISI strategy (Ackah et al., 2014). All these led to balance of payment problems as export earnings declined, capital flows and official grants significantly fell, in addition to the loss of creditworthiness which hit Ghana’s manufacturing industries severely. Since there was inadequate foreign exchange to import essential primary materials and spare parts, capacity utilization in most of the large scale state owned import substituting industries declined averagely from between 43-52% during the mid- 1970s to a low of 21% by the end of 1982 (Ackah et al., 2014). Manufacturing in turn recorded an average negative growth rate of 8.11% between 1980 and 1983. In 1983, the Economic Recovery Programme (ERP) was put in place. Some measures were put in place for the resuscitation and growth of manufacturing and the industrial sector which included, restructuring of the manufacturing sector and its associated sectors especially the agricultural sector. This involved upgrading the small/medium-scale manufacturing industries at the same time promoting the expansion of agro and resource-based industries. This was aimed to create economically viable linkages between local industries and strategic economic sectors. Some 26 University of Ghana http://ugspace.ug.edu.gh measures were also put in place to find solutions to the limitations the manufacturing sector faced under the implementation of the ISI strategy Increasing output and improving upon manufacturing production efficiency through greater use of existing capacity was of great importance. Strengthening financial organizations that gave assistance to the manufacturing firms was deemed very important. Some steps were taken under the ERP to restructure the manufacturing sector such as privatization of some State Owned Manufacturing Enterprises, removal of barriers such as price and distribution controls, eliminating import licensing and dealing with market-determined prices After the implementation of the Economic Recovery Programme, manufacturing industry averaged 12% growth in output between 1984 and 1988 from the negative growth average of 8% between 1981 and 1983. This translated into the average growth of Ghana’s GDP by 6% per year between 1984 and 1988 from the negative growth average of 5% per year between 1981 and 1983 as seen in table 2.10 below. Nevertheless, there was a fall in manufacturing growth rate from 1989 to 1994, averaging a growth rate of only 2% per year. This was coupled with a fall in the average growth rate of the electricity sector from above 22% to 9% due to the over dependence on hydro- electric sources of energy to enable the generation of electricity. Ackah et al, (2014) also attributed liberalization as one of the causes of dipping industrial growth as manufacturing firms faced “intense competition from productive-efficient economies like China”, fast increasing depreciation of the domestic currency leading to high production costs making many domestic firms ‘economically inefficient’. 27 University of Ghana http://ugspace.ug.edu.gh Table 2.10: Average growth rates of industrial subsectors (1981-2000) Period GDP Industry growth Manufacturing Electricity & water 1981-83 -5 -12.49 -8.11 -10.36 1984-88 5.87 11.18 12.66 22.64 1989-94 4.29 4.1 2.33 9.07 1995-2000 5.07 4.3 4.12 3.27 Table source: Compiled by Ackah et al., (2014) from Ghana Statistical Service The share of manufacturing in the industrial sector had fallen considerably, even though it was still the leading industrial subsector in the 1990’s. Midway through 1990’s, there were some programs put in place to support growing businesses that could not withstand the harsh conditions of the liberalized economic system. These included: tariff policy reforms; institutional and regulatory reforms; Trade and Investment Programme (TIP), Private Enterprises and Export Development (PEED) policy and the Investment Policy Fund for Small and Medium Scale Enterprises Development (FUSMED). The Ghana Investment Centre; the Business Assistance Fund (BAF) and the Ghana Investment Promotion Centre (GIPC) Act of 1994 to PNDC Law 116 were put in place. These interventions led to a growth in manufacturing output of 1.8% from 1995 to 5% in 2005 and thereby increasing growth rates of the industrial sector from 3.3% to 7.6%. GDP growth increased from 4% to 5.9% within the same period as seen from table 2.11 below. Throughout the period from 2001 to 2005, there were consistent increases in the provision of electricity and water. 28 University of Ghana http://ugspace.ug.edu.gh Table 2.11: Average growth rates of industrial subsectors (2001-2005) Year GDP Industrial Manufacturing Electricity and Sector Water 2001 4 2.9 3.7 4 2002 4.5 4.7 4.8 4.1 2003 5.2 5.1 4.6 4.2 2004 5.6 5.1 4.7 3.7 2005 5.9 7.6 5 12.3 Source: Compiled by Ackah, Adjasi and Turkson, (2014) from Ghana Statistical Service The growth rate of the industrial sector declined to 6.1% in 2007 due to negative growth rates in the manufacturing sector (-1.2%) and electricity and water sectors (-17.2%). The industrial sector however picked up the following year by growing at a rate of 15.1% mainly due to substantial growth in electricity and water by a substantial rate of 19.4%. There was an increase in manufacturing output growth from 7.6% in 2010 to 17% in 2011 the largest output growth in recent period. However, manufacturing growth fell drastically to 2% the following year and further down to -0.8% in 2014. The energy sector experienced some difficulties in electricity supply in 2011 which led to an increase in outage duration per consumer by 77.5% from the previous year (ECG, 2011). This inadequate power supply began to have adverse effect on manufacturing companies in Ghana from 2012 to 2014 (AGI, 2014). 29 University of Ghana http://ugspace.ug.edu.gh Table 2.12: Average growth rates of industrial subsectors (2006-2015) Year GDP Industrial Manufacturing Electricity Sector 2006 6.4 9.5 4.2 24.2 2007 4.4 6.1 -1.2 -17.2 2008 9.1 15.1 3.7 19.4 2009 4.8 4.5 -1.3 7.5 2010 7.9 6.9 7.6 12.3 2011 14 41.6 17 -0.8 2012 9.3 11 2 11.1 2013 7.3 6.6 -0.5 16.3 2014 4 0.8 -0.8 0.3 2015 3.9 9.1 -2 3.2 Source: Compiled by author from Ghana Statistical Service (Annual GDP 2015 edition) Ghana’s manufacturing sector comprises mainly of micro establishments constituting 55% of manufacturing firms, small firms (39.3%), medium firms (3.6%) and large firms (2.1%), (GSS).In terms of employment growth, manufacturing sector’s share of industrial employment increased from 69% in year 2000 to 80% in 2006 while in mining and quarrying subsector decreased from 9 to 5.2%, electricity and water (2.6 to 1.5%) and construction decreased from 19% to 13% according to labour statistics from GLSS IV and V reports, (GSS, 1998) and (GSS, 2008). 2.3 Conclusion. This Chapter discussed the electricity generation trends in the power sector, and the statistics on electricity transmission, distribution and consumption. It looked at the differences in the generation and supply capabilities of Ghana’s power sector and that of other countries. It was realized that Ghana’s power industry needs a lot of investment made into it in terms of expanding generation and supply capacities to be able to become energy self-reliant. This Chapter also discussed the contribution of the manufacturing sector to economic growth which has not been consistent through the years due to continuous changing policies with every new government. It has been 30 University of Ghana http://ugspace.ug.edu.gh realized that the electricity sector growth tends go hand in hand with the trend in manufacturing sector growth and that in order for the manufacturing sector and economy to develop there is the need secure adequate electricity needs. 31 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE LITERATURE REVIEW 3.1 Introduction This Chapter examines the relevant literature on electricity and manufacturing sector activities. The first section looks at the theoretical frameworks of output and productivity measurements. It also analyses the concepts of electricity as a direct input in a firm’s production processes given the traditional production inputs. This section finally examines the theory of how electricity serves as a measure of capital utilization. The second section would cover a review of the empirical literature which considers the different approaches that have been used to estimate the impact that electricity has on industrial production. 3.2 Theoretical Review 3.2.1 Output growth and productivity measurement Industrial firms or business are set up to produce goods and services. For the purposes of this study, it is important to understand the processes of deriving the measurement of output growth and the quality of that output growth at the firm level and at industry or sectoral level. A defined production function model with “neutral” or “output-augmenting” technical change is a method of acquiring gross output for a manufacturing firm or industry, in relation with both direct and intermediate inputs. This technical change is described as “the currently known ways of converting resources into outputs desired by the economy” (Griliches, 1987). Multifactor productivity (MFP) growth accounts for the rate of neutral, disembodied technical change when MFP measures are built on the concept of gross-output. 32 University of Ghana http://ugspace.ug.edu.gh On the other hand, MFP measures could be based on the value-addition theory in the case where value addition is taken as the firm’s final output and only primary inputs are considered to be the firm’s input. Productivity measures based on value addition mirrors an industry’s ability to expand income on a macroeconomic scale and match a wider varied economic demand. From this perspective, we can say value-addition based measures are complementary to gross-output based measures. Any good or service that is produced within a production firm and becomes accessible outside the firm is called gross output. In the calculation of gross output, the value of sales and net additions to raw materials or work-in-process products as well as purchases of intermediate inputs are recorded. Value added measure of output is obtained when intermediate inputs bought are not added to gross output. This makes value added a net measure. In order to get the various ways in which to measure growth and productivity of firms, we have to make use of a production function. With respect to gross output, the production function shows how total output (𝑌) that can be produced by both primary inputs such as capital and labour (𝑋), and intermediate inputs(𝑍). Intermediate inputs are factor inputs such as energy, raw materials, semi-finished goods, and services purchased by the firm that are used in production rather than for final consumption. This production function also captures disembodied technological shifts which is represented by the parameter 𝐴(𝑡). The technical change is referred to as “disembodied” because it is not based on or affects a particular factor of production, instead it affects inputs evenly. This type of technical change was first introduced by Hicks (1932) and was named “Hicks-neutral”. It is also termed as “output augmenting” technical change. It is defined as the technical change which increases the maximum output where the effects of the marginal rate of substitution between each pair of inputs on output is independent of technical change. 33 University of Ghana http://ugspace.ug.edu.gh With the above description of the production function Schreyer (2001) uses the theory by Hicks to determine how technology determines gross output and value added productivity growth. Therefore, the production function is; Y = H (A, X, Z) = A(t) ∙ F(X, Z) ………………………….. (3.1) From equation 1 it was noticed that A(t) = …………...…………………….. (3.2) ( , ) Where Y is gross output, A is rate of technological improvements, X is direct inputs and Z is intermediate inputs. Rate of technological improvements for a production unit can be determined by equation 3.2. Multifactor productivity is positive when gross output is greater than the rate of change in all combined measured inputs. When technology changes given the prices of all inputs, the production frontier shifts. Since technological changes cannot be observed directly, an appropriate way of observing technical change is through the rate at which the production function shifts over time. The shift in production function equals the rate of change of the technology whenever technology is Hicks-neutral. Thus from equations 3.1 and 3.2, the change in H with respect to time is equal to the change in A with respect to time = ………..…………………… (3.3) Multifactor productivity growth is therefore growth rate of divisia index of output and divisia index of inputs. Combining equations 3.2 and 3.3, we find the logarithmic rates of both output and inputs with respect to time. Divisia index of inputs is made up of the logarithmic rates of change of primary and intermediate inputs, weighted with their respective shares(𝑠 , 𝑠 ) 34 University of Ghana http://ugspace.ug.edu.gh = = − 𝑠 − 𝑠 ...………. (3.4) 𝑠 is the share of primary inputs in total inputs, while 𝑠 is the share of intermediate inputs in total inputs. Hence if total inputs equals 1, then 𝑠 + 𝑠 = 1. Both shares are therefore fractions of total input such that 𝑠 = 1 − 𝑠 and 𝑠 = 1 − 𝑠 In terms of measuring value added output, a value-added function is established. The value-added output represented by the maximum amount of current-price value added that can be produced is a function of a set of given primary inputs, given prices of intermediate inputs and output and a level of technical change. Denoting the value added function as 𝐺 = 𝐺(𝐴(𝑡), 𝑋 , 𝑃 , 𝑃) ……………….….... (3.5) With this function, it is realized that producers are affected by the use of intermediate inputs when there are changes in relative prices. Similarly to that of gross output, the measure of output growth associated with technical change for the value-added function is defined as the shift of the value- added function over time. Given the definition of value-added above, MFP growth is: = − …………...……..... (3.6) Where = − s ……………........ (3.7) Where VA is value-added output, Y is gross output, A is rate of technological improvements, X is direct inputs and Z is intermediate inputs. Equation 3.7 is volume index of value added where the value of intermediate goods is taken out of final output (to solve the problem of double counting) multiplied by the inverse share of value added in gross output. This led us to the equation of productivity based value added output similar 35 University of Ghana http://ugspace.ug.edu.gh to that of gross output where changes in technological change (in this case G) is determined where changes in direct input is deducted from changes in output (in this case value added output VA). Inputs in this scenario are only direct inputs. Only direct inputs are deducted from value-added output since intermediate goods are already taken care of in the computation of value-added output and are not included in the value of the final output as is shown in equation 3.7. The relationship between the gross-output and the value-added productivity measure according to Bruno (1978) is a direct one. He described the rate of change of value-added based MFP as the multiplication of the rate of change of gross-output based MFP by the inverse of the nominal share of value added in gross output: = ∙ ………………………….. (3.8) 𝐺 𝑤ℎ𝑒𝑟𝑒 𝑠 = , 𝑖. 𝑒, 𝑠ℎ𝑎𝑟𝑒 𝑜𝑓 𝑣𝑎𝑙𝑢𝑒 𝑎𝑑𝑑𝑒𝑑 𝑖𝑛 𝑔𝑟𝑜𝑠𝑠 𝑜𝑢𝑡𝑝𝑢𝑡 𝑃 ∙ 𝑄 Productivity growth in value-addition for a specific industry especially in manufacturing normally exhibits higher values in terms of output due to value added output’s percentage in gross output being less than or equal to one. Even though gross-output and value-added based MFP measures complement each other, gross output measures are considered when technical change affects all production inputs proportionally. However, the value-added based measure reflects an industry’s ability to transform technical change into income and into a contribution to final demand. Gross output based MFP measures are not highly responsive to intra-industry flows of products or to when the degree of vertical integration within industry changes. Value-added based MFP measures however tend to differ with changes in the amount of intermediate inputs calculated over the period and it gives an indication of how improvement in productivity for the entire economy 36 University of Ghana http://ugspace.ug.edu.gh can be achieved. It shows how much added quality to final demand per unit of primary inputs created in an industry. Information regarding value added figures are most of the time readily available than gross output data even though in principle, gross-output measures first and foremost are essential in deriving value-added figures. These are some of the reasons several researchers most of the time employ the use of aggregate value-added based measures in their studies. It is also important to note that the difference in choosing between gross output and value added at the firm level is insignificant however becomes bigger at the detailed industry level. 3.2.2 Productivity measures Productivity measurement theories date back to the work of Tinbergen (1942) and in modern times developed by Griliches (1998) and Jorgenson et al. (2014). Productivity measures are commonly defined as a ratio of a volume measure of output to a volume measure of input use. Most literature on productivity including its applications reveal that there is no single measure of productivity. Productivity measures entail ways of dealing with efficiency, real cost savings, and can be used as a tool to compare performances between industries and assessing living standards. The use of a productivity measure depends on the objective the particular measure of productivity has to accomplish and on data availability. Productivity measures are mainly categorized under either a single factor productivity (that is measuring value of output per a single input) or a multifactor productivity measure (measuring value of output per a weighted average of all the relevant inputs). Also, differences in measurement at the industry or firm level depends on whether measuring productivity relates gross output to one or several inputs or measures that relates value- added to one or several inputs. 37 University of Ghana http://ugspace.ug.edu.gh This study makes use of the single factor productivity measure which is labour productivity. It is a measure that uses only labour or labour-hours as the input measure that is, output per labour, or output per labour-hour. With the study employing value-added as the output measure, changes in labour productivity can be explained or disintegrated into three measurement scenarios. The first one relates to the quality of labour. Whenever there is improvement in the quality of labour, output is expected to grow even if the number of labour hours in production were to remain the same. The second scenario of labour productivity changes is that explained by changes in the amount of capital per labour-hour. This means as capital increases holding labour constant, capital per labour will also increase which will subsequently increase output without any increase in labour effort. Last but not least measure of change in labour productivity is through the multi-factor productivity growth. This deals with all other factors including the efficiency of organization such as structured and co-ordinated workflows within a firm. Therefore, in the case where labour quality is rising and capital equipment that is used per labour hour also rises, then labour productivity growth will be greater than multi-factor productivity growth. In terms of labour productivity, value-added based measures are less sensitive to variations in the degree of vertical integration than gross-output based measures. 3.2.3 Electricity as an input and a measure of Capital Utilization. Unlike statistics on labour, employment and wage variables, it is more difficult to access information on capital stock formation and utilization at the firm and industry level. Productiveness of capital according to Robinson (1954) consists in the fact that “a unit of labour that was expended at a certain time in the past is more valuable today than a unit expended to-day, because its fruits are already ripe”. This means that gains accrued from output today is as a result of past investment into labour unlike labour employed today which produces for current 38 University of Ghana http://ugspace.ug.edu.gh consumption. Productiveness of capital is thought of in terms of labour units because an increment of capital depends upon how much work is done in terms of final product of an hour's labour, not upon the cost (Robinson, 1954). In order to show how capital is used efficiently to promote productive output there must be the discussion of capital utilization in the analysis of the sources of firm growth (Kwon, 1986). Theoretical models of capital utilization analyze and explain the termination of production activities below 100% of technical maximum productive capacity, caused by regularly changing factor prices (Bossworth and Dawkins, 1983). Capital utilization measures are intended to tell how much time on an average the capital stock of a manufacturing firm, sector or economy is used and how much of the time it is idle. The central issue in the study of capital utilization is to find out, how much of the time a firm desires to operate and how much of the time it wants to be idle. It is argued that the best way to quantify the extent of capital utilization is by valuing its ability and the capacity to which it can produce output. Capital utilization measures have been derived using electricity consumption data which is the focus of this study. This measure was derived by Foss (1964) and later used by Jorgenson and Griliches (1967) for U.S., Heathfield (1972) for U.K. and Kim and Kwon (1977) for South Korea. The basis for this measure is that plants, machinery, equipment or tools used in factories or manufacturing firms do most of the mechanical work in these factories using electricity. These machines have inbuilt motors and engines designed to fire and propel the machines to be able to do the work they were built to do adequately. Therefore, finding the extent to which electric motors are used to operate the machinery can inform the level of intensity of machine use within a period. However, without electrical power these plants become least useful. The utilization rate here is defined as the ratio of the actual number of hours per period 39 University of Ghana http://ugspace.ug.edu.gh during which a given electric motor driven plant operates to the total number of hours available during the same period (Kim and Kwon, 1977). Quantifying work done by plants as an input in terms of their value during the plants’ lifespan is difficult to compute. The closest solution to this problem is to put a value on the quantum of electricity consumed by these plants during their operation. Therefore in this case, for purposes of computation, the rate of capital utilization is computed by the ratio of actual consumption of electricity (in kwh) to the maximum possible consumption by installed electric motor plants; where the maximum possible consumption is derived with the assumption that all installed electric motor plants are used continuously without any cessation of operation within a given period beyond which it can no longer operate. Heathfield (1972) justified reasons why electricity consumption should be used as a method of measuring capital usage. He initially distinguished between direct and indirect measures of capital usage. Direct measures according to Heathfield are the sample surveys conducted by statistical institutions and take the form of questionnaires filled in by businessmen who inevitably rely on subjective judgments as to what is meant by "normal'" or "maximum"' capital usage. The direct methods of estimating capital usage all rely on some presumed relationship between capital usage and other, more easily measured, variables like output, labour stock, labour hours or investments. Indirect measures of capital on the other hand encounter problems such as effect of Quality change and index number measurements applied to output, capital stock and labour. Even if these measures are assumed away the link between labour inputs and capital usage may be lost when labour is hoarded in times of a fall in demand. Once the reduction in the amount of labour used does not match with existing capital it causes capital usage to be overestimated when output is low. Technological changes may also cause the relationships between labour and capital (or output 40 University of Ghana http://ugspace.ug.edu.gh and capital) to change in the long run and this technological advancement would be incorporated/absorbed into the definition of capital usage. In order to separate technological progress and capital usage, identifying technical progress as a simple or quadratic time trend so that deviations of output/capital ratios about a trend are regarded as over-use and underuse of capital. Electricity consumption relates to capital utilization via the production function. In this respect it is an addition to the group of indirect measures discussed above and is free from the subjective nature of the "direct" measures. The objections to the indirect measures can, however, be overcome in the case of electricity consumption. According to Heathfield (1972), electricity is a perfectly homogeneous input of invariate quality and hence presents no aggregation or measurement problems. Secondly, electricity cannot easily be stored and so the flow into a process corresponds exactly with what is currently used up by the process. Thus there is no "hoarding" problem as there is with labour inputs. There are however some arguments against the use of electricity consumption as a measure of capital use especially in developing countries where plants are operated by electric power as well as non-electric power, such as steam engines or gasoline engines or generators especially when there are issues of very inadequate power supply. For electricity, there is also a measurement problem where the relationship between capital services and electricity consumption may change with time, so that one unit of electricity corresponds to one capital hour in time t1 but two capital hours in time t2. Energy variable which is a composite of all fuels is considered as a measure of capital to account for the use of non-electric motors. According to Heathfield (1972), even though there are reasons for not regarding the electricity consumption as an absolute measure of capital 41 University of Ghana http://ugspace.ug.edu.gh usage, they may nevertheless be useful for inter-regional and international comparisons of similar industry groups. It can be argued however that most firms in less developed countries are not large enough to absorb the cost that comes with acquiring plants and fueling it daily so they either reduce the number of hours they operate during times there is unavailable cheap electricity supply or make losses from fueling generators. This makes these small and medium firms from similar competitive industries more dependent on electricity from the national grid. Hence, the usage of electricity consumption as a measure of capital is a suitable measure for this study in particular. Another theory that backs electricity to be used as a measure of capital input use is that developed by Stern (2004). He explains energy to be an innovation and the driver of new technology. According to Stern, classical economists typically assume capital, labor, and land to be the primary inputs in production, and goods such as fuels and materials as the intermediate inputs. Therefore, in this application of classical approach to production, much focus has been directed towards primary inputs, especially on capital and labour, and a limited attention paid to exploring the role energy plays in the production process which has been relegated to the background as an indirect factor and not the main facilitator of growth. The classical theory of production and growth assumed diminishing returns to both capital and labour and it did not explain how to lessen the effect of diminishing returns to achieve long run growth in output. The neoclassical economists then came out with the argument that the only drive of continued growth is through technological progress. However, the neoclassical model fails to explain how improvements in technology will come about. They also assume technology to be fixed. 42 University of Ghana http://ugspace.ug.edu.gh Stern (2004) explained reproducibility as an important notion in the economics of production. This means, some production inputs cannot be reproduced, while some inputs can be reproduced at a cost within the economic production system. Physical capital and natural resources are examples of reproducible factors of production. Electricity is a typical example of inputs that are non- reproducible since its production and use follows the ‘first law of thermodynamics’. The law states that “When energy passes, as work, as heat, or with matter, into or out from a system, the system's internal energy changes in accord with the law of conservation of energy. Equivalently, perpetual motion machines of the first kind are impossible.” This simply means electricity once produced cannot be stored and that electricity would have to be converted to energy in order for machines and equipment to be operational. This energy used by machines cannot be regenerated and reused, even though, energy vectors such as fuels, wind, hydro, solar are reproducible factors. Modern scientists, energy and ecological economists have therefore placed an emphasis on the role of energy and its availability in the economic growth and development processes. It is argued that instead of making energy an indirect input, it is made a direct input in the production process. The argument for making energy as a direct input is primarily based on the idea of the new growth theorists stipulating that technological advancement leads to exponential growth in an economy through increasing level of savings and capital formation. Stern explained how technological change as a solution to growth comes about which he describes as not so different from knowledge substitution. The theory put forward in Stern (2004) assumes there are several efficient technologies that exist at a particular time and each technology can be substituted for one another. Whenever there are new efficient techniques developed, there would be changes in the technology within a production 43 University of Ghana http://ugspace.ug.edu.gh unit for the best outcomes. Stern further adds that applying new techniques represent the “substitution of knowledge for the other factors of production”. The knowledge is embedded in better-quality physical capital goods and improved human capital in the form of trained labour, all of which require some kind of energy, materials, and technical training to produce and maintain. Thus, no matter how sophisticated and technical workers and machinery become, there is still the need for energy for them to work. The use of existing capital stock and future addition to capital from this notion make use of some sort of energy to operate as Heathfield (1972) indicated in his work in terms of using electricity in production. Therefore, the concept of production used in this study examines the linkage between energy use as the primary input and economic activity within an industry over time. 3.3 Empirical Reviews Several authors and economic researchers have done some extensive work on the impact of electricity on industrial growth and productivity in different economies. The empirical review seeks to understand how results from different studies based on theory are similar and differ from each other in different scenarios and contexts. Beaudreau, (1995) in identifying different reasons for growth in US manufacturing from 1950 to 1984, showed that the biggest factor that propelled growth in manufacturing value-added in this period was industrial electricity consumption. Increased electricity consumption accounted for 79% of growth in manufacturing value-added within the period increasing at an average annual rate of 4.455%. Per worker consumption of electricity also increased from 12,534 kilowatt hours in 1950 to 41,688 kilowatt hours in 1984, a total increase of 232% over the entire period. Therefore, the slowdown in productivity which began in 1973-1974 in the US was seen by most authors as the immediate impact of the energy crisis as prices of electricity went up at the same period. 44 University of Ghana http://ugspace.ug.edu.gh However, Beaudreau (1995) in his article said, some growth accountants such as (Denison, Gullickson and Harper) did not agree with the hypothesis that electric power played an enormous part in productivity growth. This was due to the weight given to electricity input in computing the total factor productivity output. He found that the percentage of electric energy in total factor income is largely insignificant. Beaudreau (1995) with this fore knowledge reexamined the role of electricity on productivity in the United States. While several literature use factor shares rather than actual estimates of Bi, the relevant output-factor input elasticities, he estimated the various input elasticities directly using output and input data. This is because implicit in the factor shares assumption is that factor markets are perfectly competitive (Christensen and Jorgenson, Gollop and Jorgenson). However, since the electricity market in the USA was largely controlled, he assumed the marginal value product of electric power would be unequal to its price. Hence, he estimated electricity consumption, employment and capital on manufacturing value- added for the period 1950-84 using both direct and indirect techniques. Results showed that the estimation using indirect techniques, by assuming perfectly competitive factor markets, produced estimates that were far less biased of the value-added electricity elasticity and increasing biased estimates of the value-added employment and value-added capital elasticities. Results from this technique showed the output elasticity for electricity was in the range of 0.015 and 0.06. Using actual elasticity values, results showed the value-added electricity elasticity, of 0.533043. This means the period between 1950 and 1984 showed, a 1% rise in electricity consumption in manufacturing led to a rise in manufacturing value-added by 0.53%. The energy crisis in Beaudreau’s estimations was the main cause of fall in productivity. 45 University of Ghana http://ugspace.ug.edu.gh Soytas and Sari, (2006) investigated the relationship between electricity consumption and industry level production in Turkey using a multivariate framework. The relationship between electricity consumption and value added output was studied within the Turkish manufacturing industry by applying multivariate co-integration tests to indicate if there exists a long term relationship between the two variables, and also taking into consideration labor and fixed investment. It was established that there was a causality running from electricity consumption to manufacturing value added. Also in the long run electricity granger caused labour and fixed investment. This therefore indicates that electricity consumption is a key input for the Turkish manufacturing industry. This means that any form of electricity shortage or conservation may unfavourably affect manufacturing output. Conversely, increase in electricity supply, improved electrical energy saving technologies and increased energy efficiency possible might contribute positively to manufacturing value added growth. Soytas and Sari (2006) also found that there was a positively significant effect of positive electricity consumption shocks on manufacturing output. It was realized that the impact that electricity had on value added was slightly more than labour and investment initially. Using the generalized impulse response and variance decomposition tests it was noted that throughout the first five years of observation, electricity shocks explained more of the differences in value added than the shocks in labor and fixed investment. Also After the fifth year, fixed investment accounted for about 50% of the variation in value added, whereas electricity’s percentage in value added variation fell to 26%. This means that increases in electricity affected value added significantly in the first five years but did not have a significant impact in the longer term. In the long run labor and fixed investment shocks accounted for majority of the difference in value added. 46 University of Ghana http://ugspace.ug.edu.gh Dike et al., (2016) studied the impact of electricity supply on industrial output in Nigeria between the period 1980 and 2014 using a double-log linear formulation. The estimated model showed that electricity consumption had a negative and insignificant impact on industrial production which is not in line with expected results and findings from other studies. This result depicts the situation where increases in industrial output was not brought about solely by increases in electricity supply from the national grid but rather came about largely through consumption of private and self- generated electricity. In analyzing the extent of influence of other factors vis a vis electricity supply on industrial output, trade openness as another explanatory variable was found to have an inverse relationship with industrial output and also found to be insignificant. Dike et al., (2016) recommended that the government of Nigeria provide tax reductions for all firms that privately generate power for industrial production. This effectively will reduce the excessive cost manufacturers have to incur in production. The study also recommended that it is important for future trade treaties entered into by the government should take into account the actual state of Nigeria’s industrial sector. This is to prevent creation of platform in Nigeria for dumping of cheaper products from other countries in order to motivate industrial firms to produce and export more. Abeberese (2013) analyzed the effect of electricity price on firm output and firm productivity. The analysis was done using firm-level panel data from the Indian Annual Survey of Industries (ASI) collected from 2001 to 2008 covering about 30,000 manufacturing firms each year. Abeberese (2013) estimated the equation where output of firms was regressed on electricity prices paid by a firm in rupees per kilowatt-hour of electricity, price of coal used by a firm, GDP per capita and on state population. Firm fixed effects was used to account for differences in firm characteristics over time which may affect individual firm outputs. There was a negative relationship between 47 University of Ghana http://ugspace.ug.edu.gh electricity prices and firm output from results obtained. A decrease in electricity prices by a percentage point results in a two percent increase in firm output and vice versa. In terms of effects of electricity prices on labor productivity, it was found that labor productivity decreases with an upsurge in electricity price. This relates to the finding of a positive relationship between an industry’s electricity intensity and its labor productivity. This is because as price rises, firms reduce the intensity to which they consume electricity or switch to less electricity intensive production activities that tends to reduce the speed of production which eventually brings down productivity. Even though the results showed the coefficients on the log of electricity price for the productivity regressions were negative, they were not statistically significant. High electricity prices consequently reduces the rate at which the manufacturing sector of India grows. Jayalath & Wijayatunga (2004) studied the economic impact of power interruptions on manufacturing industry in Sri Lanka. The study revealed that power interruptions is the main cause of loss of revenue in Sri Lanka’s manufacturing industry. Averagely, the national proportions of outage costs in different industry for planned power outages was US$ 0.36 million per hr and the costs for unplanned power outages amounted to US$ 0.58 million per hr. In terms of both planned and unplanned interruptions, most of the national economic losses came from industries that deal in food, beverages, tobacco, textiles, leather, chemical, petroleum, plastic, metal products, and machinery and transport equipment In terms of national economic losses in Sri Lanka, the study valued the loss that arose as result of 300 hours on average of planned power interruptions to be in the region of US$ 47–117 million (90% confidence level), representing 0.4–0.9% of Sri Lanka’s 2001 GDP estimates, which was US$ 12.5 billion. Generally, unplanned power interruptions within that period were much lower compared to planned outages since these unplanned power cuts did not last long. The duration for 48 University of Ghana http://ugspace.ug.edu.gh unplanned interruptions was estimated to average about 100 hours in the year 2000/2001. About US$ 30–58 million per year was the estimated loss associated with unplanned interruptions which represented approximately 0.25–0.46% of GDP. Another form in which Jayalath & Wijayatunga, (2004) interpreted economic losses within the Sri Lankan industry is the economic loss per unit of supply loss (US$ per kW h), also referred to as the cost of unserved energy. Unserved energy is where demand for electricity exceeds supply of electricity as result of interruptions as power is lost through transmission. The study indicates the average cost of unserved energy for the Sri Lanka system is US$ 0.66 per kWh in the case of planned interruptions, while that of unplanned interruptions is US$ 1.08 per kWh. It was revealed that losses from unserved energy due to unplanned interruptions were about 1.6 times more than losses from planned outages. This gives an indication that improvements in the country’s electricity power supply must be made to lessen the occurrence of unplanned outages. It was recommended that in addition to increasing Sri Lanka’s generation capacity, improving system reliability and establishing processes that enables the relay of any planned outages in advance to all customers is of utmost importance. Darko, et al, (2014) analyzed the statistical effect of electricity insecurity in some low and middle income countries on their firms’ total factor and labour productivity. Data was gathered from the World Bank Enterprise Surveys from six selected countries (Bangladesh, Nepal, Nigeria, Pakistan, Tanzania and Uganda), and qualitative information collected from four countries (Bangladesh, Nepal, Nigeria, and Uganda). Their study focused on manufacturing firms, which make up 56% of the total firms surveyed by the Enterprise Surveys. These SME’s provide major employment avenues in these countries and higher per capita GDP through increased incomes. Empirical 49 University of Ghana http://ugspace.ug.edu.gh findings from this study indicated that inadequate electricity supply affects total factor productivity and labour productivity of manufacturing SMEs negatively. However, the negative effect did not always show statistically significant outcomes, and results could be directed in a particular direction by the way electricity insecurity is measured by the researcher. For instance, some manufacturing firms in two out of the six countries dealing with the problems of inadequate power supply have higher productivity. The estimation showed that making use of the length or duration of outages was a better measure of electricity insecurity than using the number of outages. According to Darko, et al, (2014), the differences in the way firms reacted to electricity insecurity might be because of the dissimilarities between the surveyed countries in terms of both structural and behavioural influences in each economy. For instance, firms experiencing outages may be high productivity firms in one country and low productivity firms in another. Also the impact of electricity insecurity on SMEs’ productivity at the firm level clearly differs, and the effect depends on factors that relates to both the environment within which the firm is operating and to the firms’ internal capabilities. Mensah (2016) studied the effects of power outages in Sub-Saharan Africa, by estimating the impact electricity shortages has on the performance of firms and labor demand. He used a panel data of 2,144 firms in 15 Sub-Saharan Countries surveyed at different time periods between 2003 and 2014. In analyzing the data acquired using the Instrumental Variable (IV) approach, his results showed a strong negative relationship between intensity of power outages and firm revenue, and consequently productivity of factor inputs. The results indicated that a 1% rise in the intensity of outage led to a decline in firm level productivity by between 0.6% and 1.1%. Fixed effects model for firm attributes and industry was included in the models estimated to control for heterogeneity. This is to account for differences in the ability of firms to cope with the negative effects of the 50 University of Ghana http://ugspace.ug.edu.gh power outages. After accounting for fixed effects in the estimations, results revealed significant differences in the way firms across different countries responded to outages. It showed that power outages had the greatest impact on Nigerian firms (that is increases in the intensity of power outages by a percent firm productivity in Nigeria falls by about 3.5%) while South African (1.5%) and Malian (1.4%) firms experienced the lowest impact. Countries such as Nigeria, Cameroon and Ghana who had the highest predicted negative impacts had higher outage intensities. The overall effect power outages on firms’ total employment was also found to be negative although the effect it had on temporary workers was greater than the effect it had on permanent workers. This is because cutbacks of permanent full time workers would mean payment of compensation and might be restricted by labor market regulations and opposition from trade unions. For policy recommendations, Mensah (2016) advised that, it is important to improve provision of electricity since it leads to growth in output, productivity, and employment. Reducing the uncertainties regarding outage periods through planned and full information disclosure of outage schedules can also be extremely beneficial to firms for plan and organize their production activities efficiently. Cissokho and Seck (2013) made use of cost efficiency scores and allocative efficiency scores to examine the relationship between electricity outages and firms’ productivity in Senegal. Based on the number of employees, data was collected on a final sample of 528 businesses stratified on regional and sectorial basis. Out of this sample were 411 small firms, 104 medium ones, and 13 large businesses, representing 77.8%, 19.7%, and 2.5% respectively of the entire sample surveyed. In Senegal, the industrial sector contributes averagely 20% to GDP yearly and gives employment to about 12% of the labor force. Small and Medium Enterprises (SMEs) interviewed in the survey, which constitute 95 percent of total businesses appeared to be affected much more by economic 51 University of Ghana http://ugspace.ug.edu.gh shocks especially from the electricity sector than large businesses, largely because of their constrained financial abilities and insufficient human and physical capital resources. In the survey done by Cissokho and Seck (2013), 57% of businesses among which 97% were SMEs indicated that electricity was a big problem. 91% of firms enduring hardship due to scarcity of electricity owned or shared a generator. Firm’s losses amounted to 5.1% of their total sales, which was greater than the world average loss. The outcome of the study showed a significant positive effect on cost and technical efficiencies which may indicate that the power outages did not become a hindrance to businesses, however it may just reflect survivor bias. The number and severity of power outages, however, had a negative impact on scale efficiency. A firm is technically efficient if it produces the maximum output from a minimum quantity of inputs while scale efficiency expresses whether a firm is operating at its optimal size given long run cost. Cissokho and Seck (2013) advised that policy should be directed towards improving the quality of electricity service. Resolving problems with electricity outages enables businesses to save and invest more and not use resources for coping purposes. Firms will therefore gain more on cost efficiency and technical efficiency if resources are reallocated to improve production of goods and services. Ketelhodt and Wöcke, (2008) surveyed 250 SMEs in Cape Town and discovered the impact of the electricity crisis in 2006 severely hit these industries. Most of these industries experienced fall in productivity due to incurring of costs in production while not trading. Out of the majority of SMEs (89%) who depended on a reliable supply of electricity, 69% indicated they were seriously affected by the outages. Also, 68% of the respondents said they were not given enough information about the magnitude of the electricity problems in Cape Town and were unable to plan adequately. 52 University of Ghana http://ugspace.ug.edu.gh Ketelhodt and Wöcke (2008) found that the existing policies directed at changing electricity consumption behaviour such as information campaigns, increasing prices, and providing rebates for energy savings, provided little or no results and were unsustainable when applied to SMEs. Generally, SMEs in South Africa are relied upon as source of growth linking the informal economy to the formal economy and bringing women and youth into the mainstream of the South African economy. However, in the study’s final analysis it was noted that SMEs were one of the sectors that were susceptible to an unstable economic environment and policy changes. This is because unlike large scale firms, SMEs most of the time do not have resources necessary to invest in alternative sources of energy. Hence, there is the need to ensure a stable electricity supply which is very essential to the growth of small and medium industries. Sing’andu (2009) in a case study of ten Zambia Association of Manufacturers affiliated Lusaka based firms determined how power rationing by the main power producer of electricity in Zambia, Zambia Electricity Supply Corporation (ZESCO), affects firm growth and productivity. The research showed that load shedding had led to a reduction in productivity in the selected firms, ultimately leading to a reduction in firm profits. Productivity had reduced by a monthly average margin of 11.8% for six firms in the food, beverage and tobacco sub-sector, 5% in the non-metallic mineral products sub-sector, 15% in the chemical, rubber and plastic products sub-sector and 30% in the paper and paper products sub-sector. This is because firms failed to meet targeted sales volume as a result of a drop in production and the increase in production costs as a result of using alternative sources of energy and measures to mitigate the impact of load shedding like the use of overtime to minimize the drop in production. Monthly Production ranged from 5 % to 15% compared to firms that did not use any alternative energy sources which recorded declines in production ranging from 6% to 30%. Alternative sources of energy and work overtime led to an 53 University of Ghana http://ugspace.ug.edu.gh increase in production costs by an average of 15% per month impacting negatively on productions which consequently leads to reduced profits. According to Sing’andu (2009) it was established that (ZESCO) could still have managed to ease the impact of the power shortages even though they are able to generate enough power to meet demand. It was seen that ZESCO’s failure to follow load shedding schedules had worsened the situation because firms were not able to plan for power outages to avoid costs associated with unanticipated disruptions in production process. Findings further revealed that firms thought increased investment in electricity generation was a long term solution to electricity deficit because it implied opening up newer electricity generation plants therefore increasing the electricity generation capacities. None of the firms thought that privatization of ZESCO and increasing electricity tariffs to generate more revenue for ZESCO were long term solutions to the power deficit. This is because increase tariffs the firms believed could lead to an increase in the firm’s production costs. 3.4 Conclusion The theoretical literature looked at the differences in measuring gross output and value added output based on multi-factor productivity. It looked at how technical change brings about changes in productivity. It also highlights the various types of productivity measures with emphasis on labour productivity due to its ease of measurement and how it forms the basis of determining living standards. It analyzed why electricity is a good measure of capital utilization and how it impacts on firm growth and productivity directly as an input. Most of the empirical literature reviewed discussed how insufficient electricity supply affects growth, productivity, revenues and employment. Mensah (2016), Darko et al (2014), Ketelhodt & 54 University of Ghana http://ugspace.ug.edu.gh Wöcke (2008), Cissokho & Seck (2013) all found that electricity outages as a result of inadequate electricity supply had a negative impact on firm productivity and employment. Beaudeau, (1995) and Soytas & Sari (2006) on the other found positive effect on manufacturing growth in the US and Turkey respectively. In relation to these reviews, this study seeks to improve on literature pertaining to the Ghanaian economy, where existing studies are mainly restricted to relationships between electricity consumption and total economic growth or industrial output growth. This study seeks to contribute to existing literature by exploring the relationship between electricity consumption and manufacturing sector performance in Ghana focusing on performance indicators such as output, productivity and employment growth. 55 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR METHODOLOGY 4.1 Introduction This chapter looks at the theoretical framework and methodology for the study which deals with the relationship between electricity and manufacturing sector performance in Ghana. This chapter is divided into two parts, the first part deals with manufacturing performance at the aggregate level and the second part deals with manufacturing performance at the firm level. 4.1 Theoretical Methodology This study is adopted from the model developed by Stern (2004) in his study “Economic growth and Energy”. The study thus uses the empirical model of output function derived in Stern’s work given as; (𝑄 , … . , 𝑄 ) = 𝑓(𝐴, 𝑋 … . 𝑋 , 𝐸 , … . 𝐸 ) ……………..…. (4.1) Where, the 𝑄 are the various outputs (such as manufactured goods and services), the 𝑋 are various inputs (such as capital, labor, etc.), the 𝐸 are different energy inputs (such as coal, oil, hydro- electric etc.), and A is the state of technology as defined by the total factor productivity indicator. According to Stern (2004), the neoclassical theorists only introduced energy as an intermediate good, with land, labour and capital being the primary inputs. Stern then introduces the concept of substitutability of knowledge which is based on learning, research and development to be the measure of technological improvement in his output function model. He added that energy is the driver of technology change. Therefore in the computation of output, the concept of production function used in his study examines the linkage between energy use as a primary input having a bigger weight and economic activity within an industry over time. 56 University of Ghana http://ugspace.ug.edu.gh Since Electricity is the most common source of energy and for the purposes of this study, energy inputs (𝐸 , … 𝐸 ) is restricted to electricity input (𝐸𝐶). Hence industrial output is a function of electricity and other inputs of production that impacts on production of industries. The amount of electricity available to the economy in any period is endogenous, though restricted by biophysical and economic constraints such as generating capacity and supply constraints. Given the introduction of electricity variable, Sterns model given as equation 4.1 above is modified to give; 𝑄 = 𝑓(𝐴, 𝐸𝐶, 𝑋 … 𝑋 ) …………….……………. (4.2) Where EC is the Electricity consumption factor. Applying the Cobb-Douglas Production function we get the production function to be 𝑄 = 𝐴𝐸𝐶 𝑋 . . 𝑋 …………………………. (4.3) The above equation 4.3 shows Output as a direct function of Electricity consumed by a manufacturing firm, technological knowledge accumulated through research and development and other inputs (labour capital, raw materials etc). In testing his production function in equation 4.1 using a multivariate approach, Stern found the existence of bidirectional causation between energy (electricity) and output and that energy was significant in explaining output. His findings also revealed a co-integration in a relation including energy, output, capital, and labour. With this analysis, an estimation framework which shows how electricity as an endogenous and direct input affects production in manufacturing. 57 University of Ghana http://ugspace.ug.edu.gh 4.2 Empirical Approach For purposes of econometric analysis, the empirical models for this study in all cases are of the form. ∗ 𝑌 = 𝑋 + 𝜀 …………………………... (4.4) Where Y is the dependent variable dependent on X the vector of explanatory variables, B* represents the corresponding coefficients which is an unbiased estimator and is the white noise stochastic term. The Log Transformation is: 𝑙𝑛𝑌 = 𝛽∗𝑙𝑛𝑋 + 𝜀 ……………………….... (4.5) Using this log transformation, the study applies Stern’s theory of energy consumption and economic growth to produce empirical models which captures electricity and other factors that affect manufacturing sector performance both at the aggregate level and firm level manufacturing. 4.3 Model Specification for Manufacturing Aggregate level Analysis At the aggregate manufacturing level the study employs an augmented form of the model by Stern (2004) where he analyzed the relationship between energy and industrial output growth. However, in analyzing the time series models the study will use co-integration and error correction techniques to estimate the short-run and long-run dynamics of the models as Olufemi (2013) used in the case of Nigeria. Olufemi’s model and estimation technique is used as a basis for estimating the time series models in this study because Olufemi (2013) made use of electricity as the only energy input whereas in the case of Stern (2004) other energy sources were considered. Olufemi (2013) based his model on Sterns theory of growth and energy where energy is an endogenous variable even though the estimation techniques are different. Olufemi’s model is given as; 58 University of Ghana http://ugspace.ug.edu.gh 𝑖𝑑𝑔 = 𝛽 + 𝛽 𝑘𝑖 + 𝛽 𝑙𝑒 + 𝛽 𝑒𝑐 + 𝛽 𝑒𝑔 + 𝛽 𝑓𝑒𝑟 + 𝜇 …………………… (4.6) Where idg is industrial output growth, ki capital, le is employed labour, ec is electricity consumption, eg is electricity generation and fer is foreign exchange rate. This study extends the outcome variable in Olufemi’s model from just manufacturing industrial output growth to include labour productivity and employment growth. Labour employed (LE) in this study is used as an outcome variable instead of an explanatory variable. Electricity generation variable is incorporated with Electricity consumption variable in this study since electricity which is generated is that which is supplied to manufacturing industries. This study also includes other explanatory variables which are Company taxes and GDP per capita in addition to Foreign Exchange variable. These explanatory variables are variables which are deemed key factors that affect the above mentioned manufacturing performance indicators in the context of the Ghanaian manufacturing sector. These models are then estimated econometrically using the Autoregressive Distributive Lag (ARDL) model. The models to be estimated are: Model 1: The relationship between electricity consumed by manufacturing industries and manufacturing sector value-added output. 𝑙𝑛 𝑚𝑣𝑎 = 𝛼 + 𝛼 𝑙𝑛𝑒𝑐𝑖 + 𝛼 𝑙𝑛𝑔𝑑𝑝𝑝𝑐 + 𝛼 𝑙𝑛𝑓𝑥 + 𝛼 𝑙𝑛𝑐𝑡 + 𝜀 …………….. (4.7) Where lnmva is the log of manufacturing value-added output, lneci is the log of electricity consumed by manufacturing industries, lnfx is the log of foreign exchange rate, lnct is the log of company taxes, lngdppc is the log of Gross Domestic Product per capita. 59 University of Ghana http://ugspace.ug.edu.gh Model 2: The relationship between electricity consumed by manufacturing industries and manufacturing sector labour productivity. 𝑙𝑛𝑝𝑟 = 𝛾 + 𝛾 𝑙𝑛𝑒𝑐𝑖 + 𝛾 𝑙𝑛𝑔𝑑𝑝𝑝𝑐 + 𝛾 𝑙𝑛𝑐𝑡 + 𝛾 𝑙𝑛𝑓𝑥 + µ …………….. (4.8) Where 𝑙𝑛𝑝𝑟 is the log of manufacturing labour productivity (manufacturing value added per labour employed), lneci is the log of electricity consumed by manufacturing industries, lnfx is the log of foreign exchange rate, lnct is the log of company taxes, lngdppc is the log of Gross Domestic Product per capita. Model 3: The relationship between electricity consumed by manufacturing industries and manufacturing sector employment. 𝑙𝑛𝑚𝑙 = 𝛿 + 𝛿 𝑙𝑛𝑒𝑐𝑖 + 𝛿 𝑙𝑛𝑔𝑑𝑝𝑝𝑐 + 𝛿 𝑙𝑛𝑐𝑡 + 𝛿 𝑙𝑛𝑓𝑥 + µ …………….. (4.9) Where, 𝑙𝑛𝑚𝑙 is the log of labour employed in manufacturing, lneci is the log of electricity consumed by manufacturing industries, lnfx is the log of foreign exchange rate, lnct is the log of company taxes, lngdppc is the log of Gross Domestic Product per capita. 4.4 Explanatory Variable Description and Interpretations (Aggregate level Data variables) Manufacturing Electricity Consumption: Electricity consumption variable is derived by electricity supply from public grid sold to manufacturing industries. Electricity consumption from theory is expected to have a direct effect on manufacturing growth and productivity. This is because increasing power supply to manufacturing industries increases their output growth and productivity (Soytas and Sari, 2006) and (Beaudreau, 1995). On the other hand, in the case where 60 University of Ghana http://ugspace.ug.edu.gh there is a reduction in the amount of power consumed due to electricity shortages or outages, output growth and productivity is expected to fall, (Sisshoko and Seck, 2013). Electricity use or its unavailability affects employment directly mostly when firms’ production are capital intensive in nature and hence when there is an increase in electricity consumption, it is cost effective to reduce employment. This is because less manpower is needed for mechanical work done by equipment and machines. On the other hand, constraints in electricity supply can lead to substitution through increased labour use and decreased capital use, thus employing more workers. If production processes in some industries are already labour intensive in nature, there is the expectation that labour employed is not affected when there is inadequate power for manufacturing use. Thus the manufacturing electricity consumption variable is expected to be either positive or negative. Gross Domestic Product (GDP) per capita: GDP per capita sometimes referred to as per capita income expresses the average income for citizens of a particular country. Production of goods from theory are an interactive cyclic result of effective demand, production capabilities and income. Thus higher GDP per capita means higher incomes arising from payments distributed to factors of production particularly labour. As income increases, the propensity to consume also becomes high and thus demand for goods and services also increases. This leads to an increase in investment in the production of more goods and services leading to higher output growth and productivity (measured by output per labour unit). The GDP per capita variable is thus expected to have positive relationship with output growth and labour productivity. That is as GDP per capita increases, output growth and labour productivity is expected to increase. On the other hand, if GDP per cap decreases, output growth and labour productivity is also expected to fall. Similarly, the GDP per capita variable is expected to have 61 University of Ghana http://ugspace.ug.edu.gh positive relationship with manufacturing employment rate, that is, an increase in GDP per capita increases employment and vice versa. This is inferred from Okun’s law which states that a 2% increase in GDP, reduces the unemployment rate by 1%. Analytically, whenever unemployment rate decreases, it means the employment rate is rising. Company Taxes: Company taxes have a direct impact on output of most production firms. Unlike income tax and indirect taxes like sales tax, which most of the time fall on the final consumer of the good, corporate taxes fall on the profits of company’s and affects output. This is because part or all of the profits earned is usually reinvested into the firm’s business. If profits are taxed, the amount of profit is reduced and its serves as a disincentive to reinvest in the firm. Depending on the type of goods being produced in the economy, theory suggest a negative effect on output is expected if the goods produced by the company in question are elastic in nature and would not have a negative effect on output if it is inelastic in nature since the burden of the tax can be pushed to the consumer. Hence, corporate tax variable is expected to be inversely related to manufacturing value-added out and productivity. Most theories also find corporate taxes to be an impediment to increasing employment in any economy or economic sector. As found in Ljungqvist and Smolyansky (2016), increases in corporate tax rates lead to substantial reductions in employment and income since profits that accrue to organizations for reinvestments are reduced there is little income available for employment at competitive wages. Similarly, tax cuts in corporate firms is seen to boost economic activity. Therefore, corporate tax variable is also expected to be inversely related to manufacturing employment rate. Foreign Exchange rate: Foreign exchange rate is important in our study because industries in Ghana import most of their manufacturing equipment, machines and other capital inputs. Trend 62 University of Ghana http://ugspace.ug.edu.gh from data shows a persistent depreciation of Ghana’s domestic currency relative to other national currency measured in cedis per foreign currency unit (usually US dollar). In Ghana, foreign exchange rates are calculated using the price quotation. According to the quotation of the nominal exchange rate, an increase in the rate denotes a depreciation and a decrease denotes an appreciation. Effects of exchange rate can either be expansionary through aggregate demand (imported goods, exports) or negatively on aggregate supply through its effect on the cost of imported intermediate inputs. From theory, a devaluation negatively impacts on aggregate supply by making imported goods much more expensive even if the net effect on aggregate demand is expansionary. This is based on the assumption that the intermediate imported good is a necessary input in the production process. So then, more expensive inputs decrease the demand for the product and the amount to produce. According to Agenor (1991) an average of 65% of raw materials utilized in the manufacturing sector is imported, contraction in aggregate supply is likely to dominate the expansion of aggregate demand. Thus, a weak local currency relative to foreign currencies causes imports to be more expensive, which leads to higher inflation, weakened economic growth and productivity and eventually leads to job losses. However, a weak currency can be better for an economy and for firms that export goods to other countries. As noted above Ghana imports more annually and thus there is the expectation that there will be output and job losses if currency mostly devalues. Therefore, the foreign exchange variable is expected to have a negative relationship with manufacturing value- added output, labour productivity and employment variables. 63 University of Ghana http://ugspace.ug.edu.gh 4.5 Model Specification for Manufacturing Firm Level Analysis The firm level analysis for this study reformulates the work of Abeberese (2013) which examines the responses of firms to electricity costs in India, where the firm’s outcome variables were firm output and firm labour productivity. The model for estimation used in Abeberese (2013) is given as; 𝑦 = 𝛽 + 𝛽 𝑙𝑜𝑔(𝑒𝑙𝑒𝑐𝑡𝑟𝑖𝑐𝑖𝑡𝑦 𝑝𝑟𝑖𝑐𝑒) + 𝛽 𝑋 + 𝛽 𝑆 + 𝛾 + 𝜋 + 𝛿 𝑡 + 𝜖 ………………………………. (4.10) Where 𝑦 is an outcome variable for firm i in region r in year t, electricity priceisrt is the price paid by a firm per kilowatt-hour of electricity, 𝑋 represented price of coal used by a firm (an alternative source of energy), Sst is log of state GDP per capita and state population and 𝛾 is the firm fixed effects. In this model the log of electricity price was mainly determined by the interaction between thermal source of energy and log of coal price. Modifying the model above by Abeberese (2013) for this study, the three manufacturing performance indicators (Value-added output, labour productivity and employment) at the manufacturing firm level are each regressed on the cost of electricity, the cost of raw materials and wages. Price of electricity is mainly determined by electricity generation costs and tariff rates in Ghana. Wages and cost of raw materials are added as explanatory variables because they are expenses made by most manufacturing firms in the country. This is to compare the effect expenditure on electricity affects firm performance in relation to these other expenses. The models are estimated econometrically using the Panel Fixed Effects model. Therefore the models to be estimated are: Model 4: The relationship between electricity expenditure consumption and firm level manufacturing value-added output. 64 University of Ghana http://ugspace.ug.edu.gh 𝑙𝑛𝑚𝑣𝑎 = 𝛽 + 𝛽 𝑙𝑛𝑎𝑒𝑐 + 𝛽 𝑙𝑛𝑤𝑎𝑔𝑒 + 𝛽 𝑙𝑛𝑎𝑐𝑟𝑚 + 𝑓𝑖 + 𝜀 …………….. (4.11) Where, 𝑙𝑛𝑚𝑣𝑎 is the log of manufacturing value-added output, 𝑙𝑛𝑎𝑒𝑐 is the log of electricity costs, lnwages is the log of wages, and 𝑙𝑛𝑎𝑐𝑟𝑚 is the log of raw materials and 𝑓𝑖 is the firm fixed effects. Model 5: The relationship between electricity expenditure consumption and firm level manufacturing labour productivity. 𝑙𝑛𝑟𝑣𝑎𝑝𝑤 = 𝛽 + 𝛽 𝑙𝑛𝑎𝑒𝑐 + 𝛽 𝑙𝑛𝑤𝑎𝑔𝑒 + 𝛽 𝑙𝑛𝑎𝑐𝑟𝑚 + 𝑓𝑖 + 𝜀 …………….. (4.12) Where, 𝑙𝑛𝑟𝑣𝑎𝑝𝑤 is the log of real value-added output per worker which measures labour productivity, 𝑙𝑛𝑎𝑒𝑐 is the log of electricity cost, lnwages is the log of wages, 𝑙𝑛𝑎𝑐𝑟𝑚 is the log of cost of raw materials and 𝑓𝑖 is the firm fixed effects. Model 6: The relationship between electricity expenditure consumption and firm level manufacturing employment. 𝑙𝑛𝑚𝑙 = 𝛿 + 𝛿 𝑙𝑛𝑎𝑒𝑐 + 𝛿 𝑙𝑛𝑤𝑎𝑔𝑒 + 𝛿 𝑙𝑛𝑎𝑐𝑟𝑚 + 𝑓𝑖 + 𝜀 …………….. (4.13) Where, 𝑙𝑛𝑚𝑙 is the log of labour employed in a manufacturing firm, 𝑙𝑛𝑎𝑒𝑐 is the log of electricity cost, lnwages is the log of wages and 𝑙𝑛𝑎𝑐𝑟𝑚 is the log of cost of raw materials, and 𝑓𝑖 is the firm fixed effects. 4.6 Explanatory Variable Description and Interpretations (Firm level Data variables) Electricity cost: Electricity expenditure is the cost a firm pays for consuming electricity. Electricity costs by a firm is measured in prices at which units of electricity is purchased. Research and theory has shown a negative relationship between cost of electricity and output. This is in the sense that as electricity prices increase, output growth decreases since increases in the price of electricity per kilowatt-hour consumed by firms leads to higher costs of electricity and goods 65 University of Ghana http://ugspace.ug.edu.gh become expensive to produce hence lower output. Labour productivity is also decreased since output per labour produced becomes smaller. This further has a negative effect on employment growth as fewer workers are hired with increasing electricity expenditure and cost of production. It is expected that electricity cost negatively affects manufacturing output, productivity and employment. Wages: Wage variable here is defined as annual wage bill for each firm. Higher wages normally tend to increase growth and productivity from literature. Yellen (1984) suggested that higher wages create the conditions for workers to be more productive, pointing to "reduced shirking by employees due to a higher cost of job loss; lower turnover; an improvement in the average quality of job applicants and improved morale." Also according to Holzer (1990) "high-wage firms can sometimes offset more than half of their higher wage costs through improved productivity and lower hiring and turnover cost." Cost of Raw material: Raw materials are accounted for as intermediate goods by manufacturing companies and when production is complete it is accounted for as part of cost of final goods sold. Higher costs of raw materials from several literature tend to have a negative impact on manufacturing performance. This is because increases in cost of materials increases cost of production, making final produce expensive. Demand for such goods declines which reduces the amount of output produced for sale, thereby leading to a decline in productivity and employment growth. 66 University of Ghana http://ugspace.ug.edu.gh 4.7 Model Estimations The estimation of the time series models will be based on the Autoregressive Distributive Lag model developed by Peseran et al (2001). It is an estimation tool that expresses the long run relationship between variable in the presence of more than 1 explanatory variable. This is termed as Co-integration. The existence of co-integration implies that short term disturbances that occur will not distort the long run equilibrium relation that exists between variables. The ARDL model produces reliable and unbiased estimates of the long run coefficients that are asymptotically normal regardless of whether the regressors understudy are I(1) or I(0). This estimation method generally produces valid t-statistics even when some of the regressors are endogenous (Harris, 2003). Another advantage ARDL has is that it has good small sample properties compare to alternative techniques. The general formula for ARDL model is; 𝜙(𝐿, 𝑝)𝑦 = ∑ 𝛽 (𝐿, 𝑞 )𝑥 + 𝛿 𝑤 + 𝑢 ……………………... (4.14) Where 𝜙(𝐿, 𝑝) = 1 − 𝜙 𝐿 − 𝜙 𝐿 − ⋯ − 𝜙 𝐿 𝛽 (𝐿, 𝑞 ) = 𝛽 + 𝛽 𝐿 + ⋯ + 𝛽 𝐿 𝑖 = 1,2, … 𝑘 …….…………… (4.15) Where L is the lag operator such as 𝐿 = 𝐿 and 𝑤 a Vector of deterministic variables (intercept, trends, exogenous variables, etc). The long run parameters can be obtained by solving equation 1 for all possible values of 𝑝 = 1,2 … 𝑘, 𝑞 = 1,2 … 𝑘 𝑎𝑛𝑑 𝑖 = 1,2. . 𝑛 The general formula for estimation in this study is; ∆𝑙𝑛𝑌 = 𝛼 + ∑ 𝛼 ∆𝑙𝑛𝑋 + ∑ 𝛿 ∆𝑙𝑛𝑋 + ∑ 𝜗 ∆𝑙𝑛𝑋 + ∑ 𝛿 ∆𝑋 + 𝜑 𝑙𝑛𝑙𝑛𝑋 + 𝜑 𝑙𝑛𝑙𝑛𝑋 + 𝜑 𝑙𝑛𝑙𝑛𝑋 + 𝜑 𝑙𝑛𝑙𝑛𝑋 ……..……….. (4.16) H = φ = φ = φ = φ H = φ ≠ 0, φ ≠ 0, φ ≠ 0 φ ≠ 0 67 University of Ghana http://ugspace.ug.edu.gh 4.8 Diagnostic Test 4.8.1 Unit Root Test for Stationarity It is important to examine if a time series is stationary or not since it can strongly influence its behaviour and properties. This is because a series can experience shocks that is a change or an unexpected change in a variable or perhaps simply the value of the error term during a particular time period (Brooks, 2008). A stationary series can be defined as one with a constant mean, constant variance and constant auto covariances for each given lag. If a series is stationary, ‘shocks’ to the system will gradually die away. This can be contrasted with the case of non- stationary data, where the persistence of shocks in a series in time t will always affect the series from year to year till time t+infinity. Non-stationary series leads to spurious regressions where t- ratios are not statistically significant and the value of the R2 tends to be very low. This seems obvious, for the variables are not related to one another. There are some regressions that tend to have higher R2 but the two variables are not related. That type of model is termed a ‘spurious regression. The early work on unit root tests in time series was done by Dickey and Fuller (Fuller, 1976; Dickey and Fuller, 1979). The basic objective of the test is to examine the null hypothesis that 𝜑 = 1 in 𝑦𝑡 = 𝜑𝑦𝑡 − 1 + 𝑢𝑡 …………………….. (4.17) against the one-sided alternative 𝜑 < 1. Thus the hypotheses of interest are H0: series contains a unit root versus H1: series is stationary. In practice, the following regression is employed, for ease of computation and interpretation 68 University of Ghana http://ugspace.ug.edu.gh 𝑦𝑡 = 𝜓𝑦𝑡 − 1 + 𝑢𝑡 …………………….. (4.18) However, the presence of autocorrelation in the dependent variable ∆𝑦 would make the test invalid. An augment form of the test called Augmented Dicky Fuller test is used to correct the problem. Augmented Dickey-Fuller test equation takes the form ∆𝑦 = 𝛼 + 𝜑𝑦 + ∑ 𝛽 ∆𝑦 + 𝑒 ……………….….. (4.19) Where 𝑦 is any variable in the model. 𝑘 lags of the dependent variable ∆𝑦 is incorporated into the model, that is (∆𝑦 ) forms an augmentation in the model. The introduction of the 𝑖𝑡ℎ number of lags of ∆𝑦 ensures that the 𝜇 is white noise and autocorrelation is absorbed in the ∆𝑦 (Dickey and Fuller,1981) A regression test is run on each variable against its lagged terms and lagged differenced term of the kind. To include the constant 𝛼 or the constant term and the linear trend t or none in the equation, a line graph of each variable is plotted to observe their characteristics. The appropriate test with respect to the data characteristics is applied as follows. If the calculated t-ratio of the coefficient 𝜑with negative sign is less than its critical value, then is said to be stationary. Philips and Perron (1988) developed an alternative method for testing for the presence of unit root. The difference between the Augmented Dickey-Fuller test and the Phillip-Perron test is how serial correlation is controlled for in each test. The Philip-Perron Test employs a nonparametric autoregression for in DF test equation and modifies the t-ratio of the coefficient so that serial correlation does not affect the asymptotic distribution of the test statistic. 69 University of Ghana http://ugspace.ug.edu.gh 4.8.2 Test for Heteroscedasticity An assumption which is critical in the OLS estimation is the assumption of Homoscedasticity; that is the variance of the error term must be constant over time. When this assumption does no hold and the variances of the error term is non-constant, it is said to be heteroscedastic. Even in the presence of heteroscedasticity, the OLS estimators will still give unbiased coefficient estimates but these estimates would not be the Best Linear Unbiased Estimator (BLUE). Hence any conclusion drawn on these estimates will be erroneous and misleading because the OLS standard errors will be too big. The study would employ a White test to check whether the variances of error term is constant over time. 4.8.3 Test for Autocorrelation Also, it is assumed that the error term are uncorrelated with one another. The term “autocorrelation” is when values of error terms are correlated. When such a situation in occurs, the coefficient estimates of the OLS would be inefficient with the standard errors relatively smaller than the real standard errors. Moreover the R-squared would be inflated. The study would employ Breusch–Godfrey test to check for the presence of autocorrelation. 4.8.4 Stability Test Stability tests are important to determine whether the long run and short relationships found were stable across the study period and therefore the essential to test for stability of the parameters. The cumulative sum (CUSUM) and the cumulative sum of squares (CUSUMSQ) tests proposed by Brown et al. (1975) are the methods used to test for the stability of the parameters. The CUSUM test applies the cumulative sum of recursive residuals based on the first n observations and is updated recursively and plotted against break point. The CUSUMSQ makes use of the squared 70 University of Ghana http://ugspace.ug.edu.gh recursive residuals and follows the same procedure. If the plot of the CUSUM and CUSUMSQ lies within the 5 per cent critical bound the null hypothesis that all coefficients are stable cannot be rejected. If however, either of the parallel lines lie across the critical bounds then the null hypothesis (of parameter stability) is rejected at the 5 per cent significance level. 4.9 Firm Level Analysis The study at the firm level is estimated using the fixed effects model. The fixed effect model is suitable for this panel data analysis due to its ability to take care of individual firm difference where these differences do not change over time but have a correlation with other factors of production. First the panel data is tested to see if there are persistent differences between firms (individual heterogeneity). The firm differences come about by factors that explain output apart from the traditional inputs of production that is labour and capital. Some examples are conditions of working environment, management style differences, and technological differences. These factors are embedded in an additional error term fi introduced to capture the individual heterogeneity present in the estimation equation. Most factors causing differences between firms are not observable that is why they are treated as errors. The index 𝑖 in the error term 𝑓𝑖 reflects the fact that it is individual specific and constant through time. Using the normal Ordinary Least square technique might lead to spurious results and inaccurate conclusions if there is the case that manufacturing firms exhibit different production techniques since differences across units can have influence on the performance of firms. The test for heterogeneity can be done either by using the random effects model or the fixed effects model. We use the Breusch-Pagan test to determine if there is (individual heterogeneity). If there is heterogeneity then the random effect is used. In determining the random effects we estimate the equation using the Generalised Least Squares (GLS) method which has the general formular 71 University of Ghana http://ugspace.ug.edu.gh 𝑌 = 𝛼 + 𝛽𝑋 + 𝑉 + 𝜇 …..……..…………….. (4.20) We only use the random effects if the heterogeneity term fi are not correlated with the regressors, if they are we use the fixed model. Fixed effects explores the relationship between predictor and outcome variables within an entity (country, person, company, etc.). Under FE it is assumed that there is something unique within an individual which may affect or bias the predictor or outcome variables and there is the need to control for this. FE eliminates the effect of those time-invariant characteristics so as the net effect of the predictors on the outcome variable are assessed. Under FE, TI variables are absorbed by the intercept. An advantage of the RE over the FE is that you can include time invariant (TI) variables (i.e. gender). To differentiate between the two models the hausman Taylor panel data estimation technique is to test if the fi is correlated to the regressors or not. 4.10 Data Sources This study makes use of annual time series data spanning from 1990 to 2015. The time period was chosen for this study largely because of unavailability of data on electricity consumed by industries for earlier years. Data on industrial electricity consumption was obtained from the Electricity Company of Ghana. Yearly statistics of the Ghana Statistical Service (GSS). Corporate taxes as a percentage of direct taxes was also taken from the State of the Ghanaian Economy (ISSER) annual reports. Information on manufacturing output (value added), manufacturing industrial labour and GDP per capita were taken from the World Bank’s World Development Indicators. Labour productivity was derived by author by dividing the manufacturing value-added by labour employed in manufacturing. Lastly, Yearly Foreign exchange data was obtained from the Bank of Ghana. In 72 University of Ghana http://ugspace.ug.edu.gh terms of the firm level analysis, a comprehensive data from a panel survey of firms operating within the Ghanaian manufacturing sector collected by the Centre for the Study of African Economies (CSAE), the University of Oxford, the University of Ghana, Legon and the Ghana Statistical Office. It covers 12 years of data, collected in seven rounds over the period 1992 to 2004. 73 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE ESTIMATION AND DISCUSSION OF RESULTS 5.1 Introduction In this chapter, the study looks at the summarized descriptive statistics and stationarity properties of the data set variables. An empirical estimation and analysis of the models that deal with the manufacturing aggregate level time series data and firm level panel data are undertaken using Autoregressive Distributed Lag (ARDL) and fixed effects models of estimation respectively. The models to be estimated are the relationship between electricity consumption by manufacturing industries on; value-added output, labour productivity growth and employment growth. A Panel analysis on the effect of electricity expenditure on firm level manufacturing value-added output, productivity and employment are also estimated. Diagnostics tests is done for all models to ensure the authenticity and credibility of the results that accrue from the estimations. 5.2 Descriptive Statistics of Data 5.2.1 Time Series Dataset Data was collected from World Development Indicators, Electricity Company of Ghana, and Ghana Statistical Service. The period of observation is from 1990 to 2015. This makes the total number of observations for all variables in the time series data set is 26. From Table 5.1 below, the yearly manufacturing electricity consumption rate was 1,236 Gwh. The maximum and minimum manufacturing electricity consumption rate in a year were 2,134 Gwh and 580 Gwh respectively. In terms of manufacturing value-added output, the highest recorded output was 2.54 billion cedis and the lowest output realized is 449 million cedis. The highest total employment recorded within the manufacturing sector was 1.783 million and the lowest recorded was 531 thousand. 74 University of Ghana http://ugspace.ug.edu.gh The average, maximum and minimum values observed for GDP per capita is 758, 1,783 and 265 respectively. Averagely, the foreign exchange rate in terms of the US dollar was 0.96 cedis per dollar1. The highest foreign exchange rate with the US dollar recorded within the period of the study was 3.79 cedis per dollar and the lowest of 0.03 cedis per dollar. Company tax rate averaged 47% of direct taxes over the study period and observed maximum and minimum tax rates were 64.7% and 36% of direct taxes over the same period. Table 5.1: Summarized Data Description (Time Series) Variable Obs. Mean Maximum Minimum Std Dev ECI (Gwh) 26 1236.007 2133.791 580.1300 516.8918 MVA (billion) 26 1.19 2.45 0.449 0.759 ML 26 1,149,953 1,783,662 531,810 381,903.3 GDPPC 26 757.6237 1827.101 264.7026 512.7502 CT(%) 26 47.03346 64.70000 36.31000 7.976330 FX(Ghc/US$) 26 0.957115 3.794800 0.034400 0.958468 PR 26 979.3468 1620.085 395.3113 383.0989 In terms of graphical representation of variables (Appendix A, Fig 1-7), electricity consumed by manufacturing industries (Fig 1), nominal foreign exchange (Fig 5), manufacturing labour (Fig 2) all show a persistent upward trend. With Company taxes (Fig 6), the variable shows an inconsistent pattern of increasing and decreasing movements along a negative slope and does not follow a particular trend. Similarly manufacturing value-added output (Fig 3) and manufacturing labour productivity output (Fig 4), do not move along a trend or follow any particular trend. Gross Domestic Product per capita (Fig 7) also shows an inconsistent upward trend 1 Foreign exchange rates variables from 2007 till date are based on the redenomination of the cedi. 75 University of Ghana http://ugspace.ug.edu.gh 5.3 Stationarity Tests The Augmented Dickey-Fuller tests and Phillips-Peron tests is used to test variables to see if they are integrated at levels I(0) or of order one I(1) since an Augmented Distributed Lag model would produce consistent long run estimates if variables are either I(0) or I(1). In the time series models where we look at how electricity affects manufacturing value-added, labour productivity and employment, variables that were not significant at levels were labour productivity output (LPR), electricity consumption by industries (LECI), and foreign exchange variables (LFX). However, company tax variable was significant at all significant levels. Since some of the variables were not stationary at levels, we tested the stationarity of those variables at first difference. Variables that were tested at first difference (LMVA, LPR, LML, LGDPPC, LECI) were all statistically significant at 1%, 5% and 10% levels. Table 5.2: Augmented Dickey-Fuller (at levels) Variables T-statistics 1% level 5% level 10% level Prob LMVA -0.826635 -3.724070 -2.986225 -2.632604 0.7939 LPR -1.275793 -3.724070 -2.986225 -2.632604 0.6243 LML -2.073268 -4.374307 -3.603202 -3.238054 0.5349 LECI -1.741520 -4.374307 -3.603202 -3.238054 0.7020 LGDPPC -1.654577 -4.374307 -3.603202 -3.238054 0.7410 LFX -1.636208 -4.374307 -3.603202 -3.238054 0.7489 LCT -3.605530 4.374307 --3.603202 -3.238054 0.0498** *, **, *** mean p-value <10%, 5% and 1% respectively Table 5.3: Augmented Dickey-Fuller (at First Difference) Variables T-statistics 1% level 5% level 10% level Prob LMVA -3.891172 -4.394309 -3.612199 -3.243079 0.0287** LPR -3.882910 -3.737853 -2.991878 -2.635542 0.0072*** LML -4.023818 -4.394309 -3.612199 -3.243079 0.0218** LECI -3.767519 -4.394309 -3.612199 -3.243079 0.0368** LGDPPC -3.680855 -4.394309 -3.612199 -3.243079 0.0437** LFX -4.008170 --4.394309 -3.612199 -3.243079 0.0226** *, **, *** mean p-value <10%, 5% and 1% respectively 76 University of Ghana http://ugspace.ug.edu.gh Table 5.4: Phillips-Peron (at Levels) Variables T-statistics 1% level 5% level 10% level Prob LMVA -3.678571 -3.724070 -2.986225 -2.632604 0.7239 LPR -1.625991 -3.724070 -2.986225 -2.632604 0.4550 LML -2.073268 -4.374307 -3.603202 -3.238054 0.5349 LECI -1.951361 -4.374307 -3.603202 -3.238054 0.5983 LFX -1.636208 -4.374307 -3.603202 -3.238054 0.7489 LCT -1.676154 -3.752946 -2.998064 -2.638752 0.0429** Table 5.5: Phillips-Peron (at First Difference) Variables T-statistics 1% level 5% level 10% level Prob LMVA -4.087032 -3.737853 -2.991878 -2.635542 0.0254** LPR -5.089002 -3.737853 -2.991878 -2.635542 0.0004*** LML -4.023818 -4.394309 -3.612199 -3.243079 0.0218** LECI -3.775098 -4.394309 -3.612199 -3.243079 0.0362** LFX -4.018252 -4.394309 -3.612199 -3.243079 0.0221** LCT -5.566728 -3.788030 -3.012363 -2.646119 0.0002*** *, **, *** mean p-value <10%, 5% and 1% respectively 5.4 Estimated Results for Model 1: The relationship between electricity consumed by manufacturing industries and manufacturing sector value-added output. To find the model to use for the estimation of Model 1, the optimal lag selection criteria based on the Akaike Information Criterion was used. Results shown in appendix A indicated that ARDL model (1,0,1,0,1) is best suited for the model. This model is preferred over other models since the model with the lowest lag values is deemed the best lag structure for estimation of a model. To confirm the existence of the long run relationship between dependent variable and the explanatory variables, we use the co-integrating bounds test of Pesaran, Shin, and smith (2001). The bounds test employed for this model showed the F-statistics (5.852249) is greater than the upper bound critical value at all significant levels. The null hypothesis of no co-integration is therefore rejected giving an indication of a long run relationship existing between log of manufacturing value-added and the log of causal variables (ECI, FX, CT and GDPPC). 77 University of Ghana http://ugspace.ug.edu.gh Table 5.6: ARDL Bounds Test Model 1 ARDL Model (1,0,1,0,1) Test Statistic Value K F-Statistic 5.852249 4 Critical Value Bounds Significance I(0) I(1) 10% 2.45 3.52 5% 2.86 4.01 2.50% 3.25 4.49 1% 3.74 5.06 Results from the estimations in table 5.7 and table 5.8 showed the existence of a positive relationship between electricity consumed by manufacturing industries and manufacturing value added growth in Ghana in both the short and the long run, however statistically insignificant. The positive short and long run estimates are in line with the expected sign. This means that when there is a percentage decrease in gigawatt-hour of electricity consumed by manufacturing industries as a result of inadequate power supply, there would be a decrease in value added output of manufacturing industries in Ghana by 0.06% in the short run and an increased fall in value added output by 0.29% in the long run. Similarly, when there is an increase in power supply, value added output will increase by 0.06% and 0.29% in the short and long run respectively. This is supported by empirical study of Nwankwo (2013) in the case of Nigeria where amount devoted to electricity infrastructure had a positive relationship with manufacturing industry production expenditure. Likewise, Soytas and Sari (2006) found out a positively significant effect of positive electricity consumption on manufacturing output. Similarly, GDP per capita had highly significant and positive relationship with manufacturing value-added output both in the short and long run. The results indicated that as GDP per capita 78 University of Ghana http://ugspace.ug.edu.gh increased by 1% value-added in manufacturing increased by 1.24% in the short run. This means increases in real incomes improves people’s ability to purchase goods which creates the demand for goods and therefore leads to increases in the production of manufacturing goods. However, the effect increasing incomes have on manufacturing production reduced to 0.53% in the long run. This is supported by empirical studies of Singariya & Sinha (2005) Results also revealed a negative relationship between company taxes and value-added output growth in both the short and long run. As company taxes increases by 1%, value-added output fell by 0.09% in the short run whereas the rate of fall in output is bigger in the longer term at a rate of 0.48%. Foreign exchange rate also had a negative relationship with value-added output in the long run where a 1% increase in exchange rate, that is, a fall in the Ghanaian cedi by 1% leads to a fall in value-added output by 0.04% in the long run. Similar results are found in Lawal (2016) where she found that exchange rate has a positive short and long run relationship with manufacturing sector output in Nigeria The result shows that Error Correct Term, ECM(-1) of the short run error correction model is correctly signed and also significant at 5% level. The significance of the ECM(-1) means that if all other influences are held constant, changes in D(LECI), D(LFX), D(LCT) and D(LGDPPC) will help restore deviations in D(LMVA) to its long run equilibrium value should the actual value of D(LMVA) deviate from the value consistent with its long term equilibrium. The Value of the ECM(-1) is -0.2001. This means the speed D(LMVA) takes to adjust from its short run deviation to the long run equilibrium by changes in D(LECI), D(LFX), D(LCT) and D(LGDPPC) is about 20%. 79 University of Ghana http://ugspace.ug.edu.gh Table 5.7: Model 1Short-run Coefficients with Error Correction term Variable Coefficient Prob. D(LECI) 0.0582 0.6020 D(LGDPPC) 1.2420*** 0.0000 D(LCT) -0.0955 0.4535 D(FX) 0.2026** 0.0021 CointEq(-1) -0.2001*** 0.0227 *, **, *** mean p-value <10%, 5% and 1% respectively Table 5.8: Model 1 Estimates of Long Run Coefficients Variable Coefficient Prob LECI 0.2909 0.6204 LGDPPC 0.5333** 0.0169 D(LCT) -0.4770 0.4756 D(FX) -0.0410 0.8187 C 16.7289 0.0007 *, **, *** mean p-value <10%, 5% and 1% respectively 5.5 Estimated Results for Model 2: The relationship between electricity consumed by manufacturing industries and manufacturing sector labour productivity. The second objective is to estimate the effect of electricity consumption on labour productivity growth in manufacturing industries in Ghana. The optimal model based on the Akaike Information Criterion selected for this objective is (2,1,2,2,1) as seen in appendix. The ARDL bounds test undertaken for this model showed that there is a long run relationship between electricity consumption and labour productivity since the F value (7.352983) was higher than the upper bound critical values at all critical values showed in table 5.9 below. 80 University of Ghana http://ugspace.ug.edu.gh Table 5.9: Bounds test for Model 2 ARDL Model (2,1,2,2,1) Test Statistic Value K F-Statistic 7.352983 4 Critical Value Bounds Significance I(0) I(1) 10% 2.45 3.52 5% 2.86 4.01 2.5% 3.25 4.49 1% 3.74 5.06 The ARDL co-integration test in this model revealed a positive relationship between electricity and labour productivity growth. It can be seen in tables 5.10 and 5.11 that an increase in electricity consumption by one percent leads to an increase in productivity in the manufacturing sector by 1.56%. This implies that when there is a reduction in power supply by a percent labour productivity falls by the same 1.56% in the longer run. However, electricity consumption was negatively related to labour productivity in the short run. This can be explained by the fact that inadequate power supply does not negatively affect the labour productivity in the immediate time period (between 5 and 10 years) since most of the manufacturing firms are labour intensive and hence are able to mitigate the effect by engaging more man hours in production. The findings also showed that GDP per capita has a positive relationship with manufacturing labour productivity. An increase in GDP per capita increases labour productivity by 1.9% in the short run and 0.22% in the long run. Similarly to that of output growth, increases in real incomes increases the demand for goods and which increases the production of manufacturing goods. Given labour and wage increases as a result of increases in income, productivity is bound to increase as predicted by theory. This finding is backed by findings of Kitov & Kitov (2008) where they found 81 University of Ghana http://ugspace.ug.edu.gh that the growth rate of labour productivity is explained by the influence of real GDP per capita in USA, Japan and some countries in Europe (France, UK, Italy, Canada and Germany) With respect to foreign exchange variable, there was a positive and insignificant relationship between nominal exchange rate and manufacturing value-added output in the short run but a negative and economically significant relationship in the long run. Findings indicated that a depreciation in the local currency (cedi) by one percent leads to a fall in labour productivity by 0.65% in the long run. This is because of the increased cost of production for manufacturing firms who import goods at a higher cost which brings about reduced output and therefore productivity. This is supported by empirical works in Ghana by Alagidede & Ibrahim (2016) and that in Nigerian manufacturing by Ehinomen & Oladipo (2012). Company taxes was seen to have a negative but insignificant relationship with labour productivity in the long run even though it did not have a negative with labour productivity in the short run. Output grew by 0.03% when taxes increased by 1% in the short run. Increases in company taxes in Ghana by a percentage point affected manufacturing labour productivity negatively in the long run by 1.4%. This is in line with work done by Gemmel et al (2010) where firms with higher corporate profits from increased productivity but situated in higher corporate tax regimes rates are expected to have lower productivity in the longer term than similar firms in lower corporate tax regimes. This is because higher taxes acts as a disincentive towards undertaking innovations that raise factor productivity. The result shows that Error Correct Term, ECM(-1) of the short run error correction model is correctly signed and significant at 10% level. The Value of the ECM(-1) is -0.2991. This means the speed D(LPR) takes to adjust from its short run deviation to the long run equilibrium by changes in D(LECI), D(LFX), D(LCT) and D(LGDPPC) is about 30%. 82 University of Ghana http://ugspace.ug.edu.gh Table 5.10: Model 2 Estimates of Short-run Coefficients with Error Correction Term Variable Coefficient Prob D(LPR(-1)) -0.2799 0.1311 D(LECI) -0.5048 0.1007 D(LGDPPC) 1.9102 0.0000*** D(LCT) 0.0369 0.8703 D(LFX) 0.1722 0.1574 CointEq(-1) -0.2991 0.0613* Table 5.11: Model 2 Estimates of long-run Coefficients Variable Coefficient Prob LECI 1.5576 0.2123 LGDPPC 0.2239 0.5482 LCT -1.4027 0.3295 LFX -0.6450 0.0928* C 0.9093 0.8892 *, **, *** mean p-value <10%, 5% and 1% respectively 5.6 Estimated Results for Model 3: The relationship between electricity consumed by manufacturing industries and manufacturing sector employment. This model is also first tested using the ARDL bounds test to determine if there is a long-run relationship between the variables. From Table 5.12 below, the null hypothesis of no co-integration is rejected since the F-statistics (4.228159) from the bounds test is greater than the upper bound critical value I(1) at 5% and 10% significant levels. This reveals that there is a long run relationship existing between employment in the manufacturing sector and the causal variables (ECI, GDPPC, FX, CT). The optimal model chosen for this objective based on the Akaike Information Criterion is (1,2,0,1,2). 83 University of Ghana http://ugspace.ug.edu.gh Table 5.12: Bounds test for Model 3 ARDL Model (1,2,0,1,2) Test Statistic Value K F-Statistic 4.228159 4 Critical Value Bounds Significance I(0) I(1) 10% 2.45 3.52 5% 2.86 4.01 2.5% 3.25 4.49 1% 3.74 5.06 From tables 5.13 and 5.14 the results show a positive relationship between electricity consumption and manufacturing employment, as an increase in electricity causes employment to increase by 0.41% in the short run but is statistically insignificant. In the long run, the electricity coefficient had an unexpected negative value of -0.4372 which indicates that even though there has been power shortages the manufacturing sector has seen an increase in the employment rate by 0.44%. This indicates that the immediate impact of electricity access and supply in Ghana led to the fast expansion in growth as well as manufacturing employment levels. However, power supply has not been expanding for the past two or more decades at the same rate at which manufacturing firms are expanding with its corresponding increase in employment, and since most firms secured private power generators during most of the duration of power outages, the negative effect on employment was not felt much. This is different from literature such as Akarca & Long (1979) where adequate power supply in the case of US economy means more machines operate and less human contact and thus reduction in employment Results also shows a positive and significant relationship between employment growth and GDP per capita in both the short and the long run. It showed that as GDP per capita increases by 1%, 84 University of Ghana http://ugspace.ug.edu.gh increased employment in manufacturing sector by 0.13% in the short run and 0.22% in the long run. This is supported by the findings in Mensah, et al (2013) that showed that when GDP increases, employment growth in the manufacturing sector also increases and vice versa. With respect to the tax variable outcome, in the short run, results showed that increase in taxes a percentage point leads to a reduction in the employment rate by 0.33%. However, an unexpected increase in employment in manufacturing by 0.98% is realized in the long run whenever corporate taxes increases by a percent but however is statistically and economically insignificant. Findings also saw a positive relationship between foreign exchange and manufacturing employment in both short and long run but insignificant in the long run. The result shows that Error Correct Term, ECM(-1) of the short run error correction model is correctly signed and significant at 1% level. The Value of the ECM(-1) is -0.5836. This means the speed D(LML) takes to adjust from its short run deviation to the long run equilibrium by changes in D(LECI), D(LFX), D(LCT) and D(LGDPPC) is about 58.4%. Table 5.13: Model 3 Estimates of Short-run Coefficients with Error Correction Term Variable Coefficient Prob D(LECI) 0.4115 0.1695 D(LECI(-1)) 0.5521 0.1037 D(LGDPPC) 0.1312 0.0041*** D(LCT) 0.1769 0.4397 D(LCT(-1)) -0.3251 0.1112 D(LFX) 0.0027 0.9740 CointEq(-1) -0.5836 0.0092 *, **, *** mean p-value <10%, 5% and 1% respectively 85 University of Ghana http://ugspace.ug.edu.gh Table 5.14: Model 3 Estimates of long-run Coefficients Variable Coefficient Prob LECI -0.4372 0.2229 LGDPPC 0.2248 0.0166** LCT 0.9816 0.1564 LFX 0.4010 0.0027*** C 12.0225 0.0021 5.7 Diagnosis Test 5.7.1 Test for Heteroskedasticity A Breusch-Pagan- Godfrey test for Heteroskedasticity was employed on Models 1, 2 and 3. The null hypothesis of Homoskedasticity is not rejected in all three models. Therefore, it was concluded that the variables of Models 1, 2 and 3 do not suffer for the problem of Heteroskedasticity. The results for the heteroscedasticity test can be seen in tables 4, 5 and 6 in appendix A. 5.7.2 Test for Serial Correlation A Breusch-Godfrey Serial Correlation LM Test is also employed to check the presence of serial correlation in both models. From tables 1, 2 and 3 in Appendix A, the null hypothesis of no serial correlation based on the LM test statistics is not rejected. Models 1, 2 and 3 are therefore concluded to be free of serial correlation. 5.7.3 Stability Test In order to test whether the ARDL model is stable, a Cusum and Cusum of square test is used. Diagrams 1 to 6 in Appendix A, show the residuals and sum of squares of recursive residual for long run equilibrium for all three models are relatively stable since they all lie within the critical bounds. The residuals are indicated by the blue dotted lines and the critical bounds are 86 University of Ghana http://ugspace.ug.edu.gh indicated by the red dotted lines. The results imply that the null hypothesis of no stability for Models 1, 2 and 3 is rejected at 5% significance level. Therefore all three models are stable. 5.8 Manufacturing Firm Level Analysis. In estimating the relationship between electricity expenditure consumption and manufacturing value-added output, productivity and employment at the firm level, the study made use of panel data estimation techniques suitable for all three models. Time effects was first tested for within the models. This is because it is important to check if data is stable over time. In other words, appropriate outcomes are derived if manufacturing firm indicators among all firms are stable over the time period stipulated. After testing for time effects it was realized that time effects is present in all the models as the prob F values equals 0 in each model which means that the null hypothesis of no time effects is rejected. Secondly, heterogeneity within firms was tested for using the Breusch-pagan random effect tests since there is a very high possibility that firms do not have the same production function or firms might differ in terms of their individual mode of operations. It was realized from the tests that there is individual heterogeneity among manufacturing firms that distinguishes between them. These diagnostic tests can be seen in Appendix B Finally, the study applied the fixed effects model because the goal here is to analyze all variables that change within a firm over time and ignore the time invariant variables. A firm’s error term and constant which captures individual characteristics is controlled for so it does not correlate with others. Thus the fixed effect was used to examine the relationship between the explanatory and outcome variables within a firm. 87 University of Ghana http://ugspace.ug.edu.gh 5.8.1 Data Description The number of observations for the selected variables differ from each other. Total observations for log of manufacturing value added output (lmva) is 1903, for log of labour productivity (lpr) is 2012, that of costs of electricity (lace) is 1590, observations for log of wages (lwages) is 1805 and for cost of raw materials (lcrm) is 1998. Below is a table that shows the mean, minimum, maximum, and standard deviations of the data variables. Table 5.15: Summarized Data Description (Panel data) 5.9 Estimated Results for Model 4: The relationship between electricity expenditure consumption and firm level manufacturing value-added output. Findings from estimation of this model reveal a negative relationship between expenditure on electricity and value-added output within a firm. It was found that, a one percent increase in the cost of electricity as electricity prices increases, decreases value-added output in a firm by 0.16% and was statistically significant as seen in table 5.16. This means that when prices of electricity increases as a result of increasing cost of electricity supplied to consumers, cost of consuming electricity by manufacturing firms increases. This either increases firms’ cost of production or firms decrease the amount of electricity they consume thereby affecting output production negatively. This results backs existing empirical results by (Abeberese 2013). 88 University of Ghana http://ugspace.ug.edu.gh Results however showed that wages had a positive relationship with value-added output as was expected. Results indicate that as wages increases by a percentage point, manufacturing firm value- added output increases by 0.09% and is statistically significant. This is in line with theory were increase in wages motivates workers to contribute effectively to increasing output. This is backed by empirical work by (Sharpe et al, 2008) Cost of raw materials likewise was found to have an unexpected positive relationship with firm value-added output. The results showed that as cost of raw materials increased by 1%, value-added output of a firm also increases by 0.33%. This might be because, the positive effect raw materials as an intermediate input in the production processes have on value-added output outweighs the effect increase in the cost of raw materials has on value-added output. That is rate of increase in the cost of raw materials did not outweigh the rate of increase in value addition within firms over the period. Table 5.16: Model 4 Results from fixed effects estimations Lmva Coefficient P>|t| Lace -0.1601 0.000 Lwages 0.0982 0.004 Lcrm 0.3265 0.000 Cons 11.2912 0.000 5.10 Estimated Results for Model 5: The relationship between electricity expenditure consumption and firm level manufacturing labour productivity growth. The study found out that electricity consumed by firms and productivity measured by output per worker moved in opposite directions. Results revealed from table 5.17 that, a percentage increase in electricity expenditure led to a strongly significant decrease in manufacturing firm labour productivity by 0.14%. This results is also supported by empirical results in (Abeberese, 2013). 89 University of Ghana http://ugspace.ug.edu.gh Similarly, to that of manufacturing value-added output, it was found that wages had a positive relationship with manufacturing firm labour productivity. According to the findings, as wages increases by a percentage point there is an increase in firm labour productivity by 0.04%, however it was statistically insignificant. This supports the theory that increase in wages serves as an incentive for workers to be more productive and increases in labour productivity in turn increases wages. This is in line with theory and is backed by empirical work by (Sharpe et al, 2008) which shows a positive relationship between labour productivity and real wages in OECD countries. Cost of materials was however found to have a positive relationship with labour productivity as was the case with value-added output. It was realized that anytime cost of raw materials increased by 1% labour productivity in a firm also increases by 0.32%. Table 5.17: Model 5 Results from fixed effects estimations Lpr Coefficient P>|t| Lace -0.1404 0.000 Lwages 0.0454 0.180 Lcrm 0.3161 0.000 Cons 8.5791 0.000 5.11 Estimated Results for Model 6: The relationship between electricity expenditure consumption and firm level manufacturing employment growth. Similarly to Models 4 and 5, the study found out that there was a strongly significant negative relationship between electricity expenditure and labour employed within a firm. Results revealed from table 5.18 that, a percentage increase in electricity expenditure consumption as a result of increases in prices of electricity led to a decrease in labour employed within a firm by 0.02% and was statistically significant. This is supported by evidence from Cox et al (2013) where they found 90 University of Ghana http://ugspace.ug.edu.gh that increases in the electricity prices in Germany leads to an overall decrease in employment in the manufacturing sector It was also realized that wages had an unexpected positive relationship with labour employed in a manufacturing firm. Results showed that as wages increased by a percentage point, employment in a manufacturing firm grew by 0.05%. Findings further revealed that costs of raw materials had a highly positive relationship with labour employment in a manufacturing firm. That is labour employed in a manufacturing firm increases by 0.01 as cost of raw materials increased by 1% however result is statistically insignificant. Table 5.18: Model 6 Results from fixed effects estimations lwkr Coefficient P>|t| lace -0.0196403 0.028 lwages 0.0543124 0.000 lcrm 0.0116819 0.371 cons 2.647968 0.000 5.12 Conclusion It was seen that in the manufacturing sector at the aggregate national level, electricity had a long run relationship with all the manufacturing performance indicators and positively related to manufacturing value-added output growth and labour productivity in the long run. However electricity was negatively related to employment growth in the long run. GDP per capita affected manufacturing value added output and productivity growth positively both in the short and long run. Company taxes and foreign exchange rate both affected value-added output and labour productivity negatively but was not significant in the long run. Looking at firm level manufacturing, the expenditure on electricity consumption had a negative impact on their value- added output, productivity and the numbers of workers the firm employs. Wages and cost of materials were positively related to manufacturing firm level value-added output, labour 91 University of Ghana http://ugspace.ug.edu.gh productivity and employment growth respectively. It can be seen that lower electricity consumption and increased consumption expenditure, as a result of inadequate power supply and increasing tariffs, negatively affects manufacturing performance both at the aggregate and firm levels. 92 University of Ghana http://ugspace.ug.edu.gh CHAPTER SIX SUMMARY, CONCLUSION AND RECOMMENDATION 6.1 Introduction Recent electricity outages experienced throughout Ghana coupled with high tariffs placed on electricity consumers has become a major threat to the growth of the Ghanaian economy. Many producers in the manufacturing sector complained about the intensity of the outages in the country and how expensive power has become which they believe has restricted manufacturing growth in the country. This is the main reason for this study which is to examine the effect the power sector in Ghana plays on the performance of the manufacturing sector. 6.1 Summary The study began by discussing manufacturing development as vital for industrialization and how electricity serves as the foremost factor in manufacturing industrial development in Ghana. The study further discussed how electric energy compares to other sources of energy and the amount of electricity used in Ghanaian manufacturing sector relative to the other sources of energy. An overview of the electricity sector in Ghana was discussed taking a look at the generation and supply trends including electricity access and consumption trends. It further looked at the growth and development of the manufacturing sector over time and how it has evolved over time. The study reviewed theoretical measurements of output and productivity. It looked at the differences between the single factor productivity measurement (labour productivity) and multifactor productivity. It went further to review recent studies on the importance of electricity in production and how electricity is treated as a primary input of production propounded in newer growth theories. Some empirical literature were reviewed and most of the literature have 93 University of Ghana http://ugspace.ug.edu.gh established the contribution of electricity in industrial development and how unavailability of power negatively affects output growth, productivity and employment. It brought to fore various estimation techniques used in these studies. In analyzing annual manufacturing sector performance, the study used a production function with a constrained optimization problem based on Stern’s model of production which uses energy as vital input in industrial development. It employed an Autoregressive distributed lag (ARDL) estimation approach on yearly time series data over the study period from 1990 to 2015. This time period was primarily considered data on electricity use by industries dates back to only 1990. In analyzing performance of individual manufacturing firms, the study employed panel data estimations on data gathered over the period from 1992 to 2003 using the fixed effects model. There were six models analyzed in this study. The first three models focused on manufacturing sector performance at the aggregate level. The models estimated were to assess the role electricity plays on Ghana’s manufacturing sector output growth in terms of value addition, labour productivity and employment respectively. The other three models were to analyze how expenditure on electricity affects manufacturing value-added output growth, labour productivity and level of employment at the firm level. Empirical analysis and results found for this study showed that electricity use impacted directly on both value-added output and labour productivity at the aggregate national level. This means that growth in value-added output and productivity in the manufacturing sector reduced over the years studied due to the consistent power insecurity and instability. Results showed that value-added output and productivity falls by 0.29% and 1.56% respectively in the long run whenever electricity consumed within the sector falls by a percent. The findings however showed that employment at the aggregate level was inversely related to electricity shortages. That is, even though electricity 94 University of Ghana http://ugspace.ug.edu.gh consumption is lower, manufacturing sector employment rate seems to improve by 0.44% in the long run. At the manufacturing firm level, electricity expenditure had negative impacts on value-added output, labour productivity and workers employed in a firm. Value-added output, labour productivity and employment rates of a firm reduced by 0.16% and 0.14% and 0.02% respectively when firms’ expenditure on electricity consumed annually increases by 1%. 6.2 Conclusion Interpreting the results found for this study, it is realized that electricity provision generally has not encouraged the growth of the manufacturing sector in Ghana. The electricity industry in Ghana has not been able to improve upon the infrastructure which will enable the power industry achieve sustainable levels of power supply. Therefore, majority of Ghanaian manufacturing firms which are small in nature tend to spend more on generators to augment power from national grid because of the power outages experienced. Firms therefore experience a reduction in output production and productivity leading to revenue losses, and also a reduction in the employment within the manufacturing sector. This explains the negative relationship electricity expenditure consumption has with firm output, productivity and employment. Similarly, the aggregate performance of these firms as expected showed a downward turn in output, productivity and employment due to low electricity consumption as electricity supply from the national grid is inadequate. 95 University of Ghana http://ugspace.ug.edu.gh 6.3 Recommendations. The empirical results derived in this study have significant implications for Ghana’s manufacturing sector. Based on the conclusions of the analysis, the study recommends that it is important to find lasting solutions to the recurring energy problems facing the country and should be treated as a principal developmental goal. This is because, improved provision of electricity has positive growth potentials through boosting output, productivity, and employment. This can be done by strengthening of the main institutions in charge of supplying electricity in the country (VRA, GRIDCO and ECG) either financially or improving on quality of labour in the power industry by training of additional technical personnel. Promoting better quality electricity service by investment in infrastructure within the power industry is also important. The study further recommends that electricity tariffs should be reduced and the power that is consumed by manufacturing firms be well subsidized. This is to encourage small and medium firms to invest money that otherwise is expended in electricity to be used to expand their businesses and employ more people. The study recommends further that to mitigate the effect of power crises in the short term, there is the need to reduce uncertainties regarding outage periods by limiting the number of unplanned power outages and disclosing full information of outage schedules. This goes to help firms to effectively plan and organize their production activities. 6.4 Limitations of the Study. Data on industrial manufacturing consumption of electricity was limited because data was not available before the year 1990. Therefore obtaining data on electricity supplied to manufacturing in Ghana for earlier years was a challenge given the fact Time series analysis require longer years of observations. There were a lot of missing data within the Firm level data on electricity 96 University of Ghana http://ugspace.ug.edu.gh expenditure because there were several firms who could not provide information for all the years during the survey. Also the number of years in the both the time series and panel data did not match since data obtained for the panel analysis did not cover recent years. This makes it difficult making an informed conclusion on the extent of electricity consumption on manufacturing sector and firm level productivity and employment generation. 97 University of Ghana http://ugspace.ug.edu.gh REFERENCES Abeberese, A. B. (2013). Electricity Cost and Firm Performance: Evidence from India. 2-27. Ackah, C., Adjasi, C., & Turkson, F. (2014). Scoping Study on the Evolution of Industry in Ghana. University of Ghana. Journal of Economic Literature. Working Paper No.18. Agenor, P.-R. (1991). Output, devaluation and the real exchange rate in developing countries. Review of World Economics. Akarca, A. T., & Long, T. V. (1979). Energy and Employment: A Time Series Analysis of the Causal Relationship. Resources Energy, 2. Alagidede, P., & Ibrahim, M. (2016). On the causes and effects of exchange rate volatility on economic growth: Evidence from Ghana. Journal of Economic Literature . Amoako-Tuffour, J., & Eshun, M. E. (2016). A review of the trends in Ghana's power sector. Energy, Sustainability and Society. Springer Berlin Heidelberg. Association of Ghana Industries (AGI). (2013 - 2015). Business Barometer Annual Reports, Accra. Baah-Nuakoh, A. (1997). Studies on the Ghanaian Economy (Vol. I). Accra: Ghana University Press. Baah-Nuakoh, A. (2003). Studies on the Ghanaian Economy: The Industrial Sector (Vol. III). Accra: Woeli Publishing Services. Beaudreau, B. C. (1995). The impact of electric power on productivity: A study of US manufacturing. Energy Economics, 17(3), 231-236. 98 University of Ghana http://ugspace.ug.edu.gh Bossworth, D. L., & Dawkins, P. J. (1983). Optimal Capital Utilisation in British Manufacturing Industry. (W. Eichorn , R. Henn, K. Neumann, & R. W. Shephard, Eds.) Quantitative Studies on Production and Prices. Physica, Heidelberg. Brown , R. L., Durbin, J., & Evans, J. M. (1975). Techniques for Testing the constancy of Regression Relations Over Time. Journal of the Royal Statistical Society, Series B, 37, 149-163. Bruno, M. (1978). "Duality, Intermediate Inputs and Value-Added" In Production Economics: A Dual Approach to Theory and Applications. (M. Fuss, & D. McFadden, Eds.) Chissokho, L., & Seck, A. (2013). Electric Power Outages and the Productivity of Small and Medium Enterprises in Senegal,. Investment Climate and Business Environment Research Fund (ICBE-RF). Research Report No. 77/13 Dakar. Cornwall, J. (1977). Modern Capitalism. Its Growth and Transportation. New York: St. Martin's Press. Cox , M., Peichl, A., Pestel , N., & Siegloch, S. (2013). Labor Demand Effects of Rising Electricity Prices: Evidence for Germany. Journal of Economic Literature. Dickey, D. A., & Fuller , W. A. (1981). Likelihood Ratio Statistics For Autoregressive Time Series With A Unit Root. Econometrica, 49(4). Dike, C. K., Elekwa, P. O., & Ugwoke, T. I. (2016). Electricity consumption and industrial production in Nigeria. 99 University of Ghana http://ugspace.ug.edu.gh Ehinomen, C., & Oladipo, T. I. (2012). Exchange Rate Management and the Manufacturing Sector Performance in the Nigerian Economy. Journal of Humanities and Social Sciences, 5(5), 01-12. Enerdata. (2016). Electricity Production and consumption. Retrieved from Global Energy Statistical Yearbook.: http://www.enerdata.net Energy Ministry, Ghana. (2010, February). National Energy Policy. Ewusi, K. (1981). The process of Industrialization in Ghana, 1950-1975. Legon: University of Ghana. Fagerberg, J., & Verspagen, B. (1999). 'Modern Capitalism in the 1970s and 1980's'. Macmillan, Setterfield M. Fisher-Vanden, K., Mansur, E., & Wang, Q. (2012). Costly Blackouts? Measuring Productivity and Environmental Effects of Electricity Shortages. NBER Working papers 17741, National Bureau of Economic Research, Inc., Cambridge, MA. Foss, M. F. (1964). The Utilization of Capital Equipment: Postwar Compared to Prewar. Survey of Current Business, 43. Gemmel, N., Kneller, R., Sanz, I., & Sanz-Sanz, J. F. (2010). Corporate Taxation and the Productivity and Investment of Heterogenous Firms: Evidence from OECD Firm-level Data. Ghana Energy Commission . (2017). National Energy Statistics 2007-2016. Ghana Energy Commission. (2006, July). Retrieved from Strategic National Energy Plan 2006- 2020. 100 University of Ghana http://ugspace.ug.edu.gh Ghana Energy Commission. (2016, April). Energy (Demand and Supply) Outlook for Ghana. Ghana Grid Company (GRIDCO). (2010-2015). GRIDCO Annual Reports. Ghana Grid Company Limited. (2010). Ghana Wholesale Reliability Assessment Report. Power Systems Energy Consulting. Ghana Statistical Service (GSS). (1998). Ghana Living Standards Survey (IV), Accra. Ghana Statistical Service (GSS). (2006). "2003 National Industrial Census" Accra. Ghana Statistical Service (GSS). (Various Issues). Data on National Accounts. Accra. Ghana Statistical Service, (GSS). (2008). Ghana Living Standards Survey (V), Accra. Griliches, Z. (1987). "Productivity: Measurement Problems". (J. Eatwell, M. Milgate, & P. Newman, Eds.) The New Palgrave: A Dictionary of Economics. Griliches, Z. (1998). R & D and Productivity: The Econometric Evidence. University of Chicago Press. Heathfield, D. F. (1972). The measurement of capital usage. Using electricity consumption Data for the U.K. Journal of the Royal Statistical Society, 135(2), 208-220. Hicks, J. R. (1963[1932]). The theory of wages. London: Macmillan Palgrave. Holzer, H. J. (1990). The Determinants of Employee Productivity and Earnings. A Journal of Economy and Society, 29(3), 403-422. Institute of Statistical, Social and Economic Research (ISSER). (2005, July). Guide to Electric Power in Ghana. Resource Centre for Energy Economics and Regulation. First Edition. 101 University of Ghana http://ugspace.ug.edu.gh Institute of Statistical, Social and Economic Research (ISSER). (2016). Electricity Insecurity and its Impact on the Economy of Ghana. Retrieved from Legon, University of Ghana. Accra Jorgenson, D. W., & Griliches, Z. (1967). The Explanation of Productivity Change. Review of Economic Studies, 34(3), 249-283. Jorgenson, D. W., Ho, M. S., & Samuels, J. D. (2014). Long-term Estimates of U.S Productivity and Growth. Bureau of Economic Analysis. Ketelhodt, A. F., & Wocke, A. (2008). The impact of electricity crises on the consumption behavi our of small and medium enterprises. Journal of Energy in Southern Africa, 19, 4-12. Kim, Y. C., & Kwon, J. K. (1977). The Utilization of Capital Equipment in a Developing Economy: Case of Korean Manufacturing. Journal of Development Economics, 4. Kitov, I. O., & Kitov, O. I. (2008). The driving force of labor productivity. Journal of Economic Literature. Kuznets, S. (1973). Modern Economic Growth: Findings and Reflections. The American Economic Review, 63(3), 247-258. Kwon, J. k. (1986). Capital Utilization, economies of scale and technical change in the growth of total factor productivity: An Explanation of South Korean manufacturing growth. Journal of Development Economics, 24, 75-89. Lawal , O. E. (2016). Effect of Exchange Rate Fluctuations on Manufacturing Sector Output in Nigeria. Journal of Research in Business and Management, 4(10), 32-39. 102 University of Ghana http://ugspace.ug.edu.gh Ljungvist, A., & Smolyansky, M. (2016). To Cut or Not to Cut? On the Impact of Corporate Taxes on Employment and Income. Finance and Economics Discussion Series. Mensah , M., Awunyo-Vitor, D., & Asare-Menako, H. (2013). Exchange Rate Volatility and Employment Growth in Ghana's Manufacturing Sector. International Journal of Business and Social Science, 4(4). Mensah, J. T. (2016). Bring Back our Light: Power outages and Industrial Performance in Sub- Saharan Africa. Journal of Economic Literature. Moyo, B. (2013). Power Infrastructural quality and manufacturing productivity in Africa: A firm level analysis. Vol 61. Nwankwo, O. C. (2013). The Effect of Electricity Supply on Industrial Production Within the Nigerian Economy (1970-2010). Journal of Energy Technologies and Policy, 3(4). Perron, P., & Phillips, P. C. (1988). Testing for a Unit Root in Time Series Regression. Biometrika, 75(2), 335-346. Peseran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds Testing Approaches To The Analysis of Level Relationships. Journal of Applied Econometrics, 16, 289-326. Robinson, J. (1954). The Production Function and the Theory of Capital. Review of Economic Studies, 21(2), 81-106. Rosenberg, N. (1998). The Role of Electricity in Industrial Development. The Energy Journal, Vol 19. Schreyer, P. (2001). Measuring Productivity: Measurement of Aggregate and industry-level productivity growth. OECD Manual, 11-31. 103 University of Ghana http://ugspace.ug.edu.gh Scott, A., Darko, E., Lemma, A., & Rud, J.-P. (2014). How does electricity insecurity affect business in low and middle income countries? . Overseas Development Institute (ODI), 2- 5. Sharpe, A., Arsenault, J.-F., & Harrison, P. (2008). The relationship between labour productivity and real wage growth in Canada and OECD Countries. CSLS Research Report No. 2008- 8. Sing'andu, N. (2009). An assessment of the Impact of ZESCO's Power Rationing on Firm Productivity and Profitability: A case study of selected Lusaka based firms. Kitwe: the Copperbelt University. Singariya, M. R., & Sinha, N. (2015). Relationships among Per Capita GDP, Agriculture and Manufacturing Sectors in India. Journal of Finance and Economics, 3(2), 36-43. Soytas, U., & Sari, R. (2006). The relationship between energy and production: Evidence from Turkish manufacturing industry. Journal of Economic Literature, 29(6), 1151-1165. Timbergen, J. (1942). "Zur Theorie der langfristigen Wirtschaftsentwicklung". Weltwirtschaftliches Archiv, Band 55:1. Volta River Authority (VRA). (2015). VRA Annual Report. Volta River Authority (VRA). (2016). Power Generation. Generation facts and Figures. Wijayatunga, P. D., & Jayalath, M. S. (2004). Assessment of economic impact of electricity supply interruptions in the Sri Lanka industrial sector. Energy Conversion and Management, 45, 235-247. 104 University of Ghana http://ugspace.ug.edu.gh Yellen, J. L. (1984). Efficiency Wage Models of Unemployment. The American Economic Review, 74(2), 200-205. 105 University of Ghana http://ugspace.ug.edu.gh APPENDIX A: Descriptive Statistics and Diagnostic Tests Line Graphs Fig 1 Fig 2 Fig 3 Fig 4 Fig 5 Fig 6 106 University of Ghana http://ugspace.ug.edu.gh Fig 7 Akaike criteria lag selection structure Model 1 107 University of Ghana http://ugspace.ug.edu.gh Model 2 Model 3 108 University of Ghana http://ugspace.ug.edu.gh Serial Correlation Test Table 1: Model 1 Table 2: Model 2 Table 3: Model 3 Heteroskesdasticity Test Table 4: Model 1 Table 5: Model 2 109 University of Ghana http://ugspace.ug.edu.gh Table 6: Model 3 Stability Tests Plot of Cummulative Sum and Cummulative Sum of Squares of Residuals (CUSUM & CUSUMSQ) Model 1 CUSUM CUSUM OF SQUARES 110 University of Ghana http://ugspace.ug.edu.gh Plot of Cummulative Sum and Cummulative Sum of Squares of Residuals (CUSUM &CUSUMSQ) Model 2 CUSUM CUSUM OF SQUARES Plot of Cummulative Sum and Cummulative Sum of Squares of Residuals(CUSUM & CUSUMSQ) Model 3 CUSUM CUSUM OF SQUARES 111 University of Ghana http://ugspace.ug.edu.gh B: Model Estimations and Results Long Run and Short Run Error Correction Model, Model 1 The relationship between electricity consumed by manufacturing industries and manufacturing sector value-added output. 112 University of Ghana http://ugspace.ug.edu.gh Long Run and Short Run Error Correction Model, Model 2 The relationship between electricity consumed by manufacturing industries and manufacturing sector labour productivity. 113 University of Ghana http://ugspace.ug.edu.gh Long Run and Short Run Error Correction Model, Model 3 The relationship between electricity consumed by manufacturing industries and manufacturing sector employment. 114 University of Ghana http://ugspace.ug.edu.gh The Breush-Pagan Tests Model 4 Model 5 Model 6 115 University of Ghana http://ugspace.ug.edu.gh Hausman-Taylor Test Model 4 Model 5 116 University of Ghana http://ugspace.ug.edu.gh Model 6 Variable Estimation with Fixed Effects model Model 4: The relationship between electricity expenditure consumption on firm level manufacturing value-added output. 117 University of Ghana http://ugspace.ug.edu.gh Model 5: The relationship between electricity expenditure consumption on firm level manufacturing labour productivity. 118 University of Ghana http://ugspace.ug.edu.gh Model 6: The relationship between electricity expenditure consumption on firm level manufacturing employment. 119