University of Ghana http://ugspace.ug.edu.gh
UNIVERSITY OF GHANA
DEPARTMENT OF ECONOMICS
ENERGY CONSUMPTION AND ECONOMIC GROWTH: EVIDENCE FROM THE
WEST AFRICAN SUB-REGION
BY
KWAME SUMAILA IDDRISU
(10348016)
THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN
PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF MPHIL
ECONOMICS DEGREE
JULY, 2017 University of Ghana http://ugspace.ug.edu.gh
DECLARATION
This is to certify that I, KWAME SUMAILA IDDRISU have undertaken this research on my own and that to the best of my knowledge, this work does not contain without due reference in the text any information previously published or submitted in any other institution for the reward of a degree.
STUDENT ………………………………………… KWAME SUMAILA IDDRISU (10348016) ...………………………………………. DATE
SUPERVISORS
…………………………………………… …………………………………..
DR. DANIEL KWABENA TWEREFOU DR. WILLIAM BEKOE DEPARTMENT OF ECONOMICS DEPARTMENT OF ECONOMICS UNIVERSITY OF GHANA UNIVERSITY OF GHANA
…………………………………….. …………………………………..
DATE DATE
ii
University of Ghana http://ugspace.ug.edu.gh
DEDICATION
This thesis is dedicated to my mum Mrs. Janet Korantengmaa for her immense support throughout the years. I would also like to dedicate to my family and every individual who contributed in diverse ways to making this study a reality.
iii
University of Ghana http://ugspace.ug.edu.gh
ACKNOWLEDGEMENT
I cannot proceed with my acknowledgements without first recognizing the one who gives breath to every human being, God. For this reason, my first acknowledgement goes to God for without
Him and the life He gives, it would not have been possible to undertake this study.
Having done that, I also desire to extend my sincere gratitude to my supervisors Dr. Daniel
Twerefou and Dr. William Bekoe for their guidance, advice and suggestions. Your contribution made this work a reality and I am very much grateful for your time and effort.
I would also like to appreciate the contribution of my family through their encouragement and prayers. I can never forget the efforts of my mother towards the successful completion of my
Master of Philosophy program. I am very grateful for your sacrifices.
Finally, to every individual who contributed in one or the other to the successful completion of this study, I am very much grateful for your effort. I pray the almighty God replenish whatever you may have lost.
iv
University of Ghana http://ugspace.ug.edu.gh
ABSTRACT
The availability of reliable energy supply to meet the exigency of the growing population in
West Africa is important in achieving sustainable development and reducing poverty in the continent. Diverse studies have sought to examine the link between energy consumption and growth. However, the consensus with reference to the causal link if any between energy consumption and growth is not explicit. Conflicting conclusions have been espoused on the energy-growth nexus and this has necessitated this study.
In this study, we employ panel cointegration and granger causality to examine the relationship between energy consumption disaggregated into total energy consumption, petroleum consumption and electricity consumption for seventeen West African countries for the period
1990 to 2013.
The study finds that in the short run, there is no causal relationship running from total energy consumption, electricity consumption and petroleum consumption to growth and from growth to total energy consumption. However, in the short run, the conservation hypothesis is established as there is a unidirectional relationship running from growth to electricity consumption. As well, electricity consumption has a significant effect on petroleum consumption. In the long run, total energy consumption has a negative and significant effect on growth with electricity and petroleum consumption having a positive and significant effect on growth. Also, in the long run, there is no causal relationship among the variables. The study therefore recommends that policies that will enhance access to electricity are implemented.
v
University of Ghana http://ugspace.ug.edu.gh
TABLE OF CONTENTS
CONTENT PAGE
DECLARATION ……………………………………………………………………. ii
DEDICATION……………………………………………………………………….. iii
ACKNOWLEDGEMENT…………………………………………………………… iv
ABSTRACT…………………………………………………………………………… v
TABLE OF CONTENTS……………………………………………………………... vi
LIST OF TABLES……………………………………………………………………. xii
LIST OF FIGURES…………………………………………………………………... xiii
LIST OF ABBREVIATIONS………………………………………………………... xv
CHAPTER ONE ……………………………………………………………………..... 1
INTRODUCTION………………………………………………………………………. 1
1.1 Background to the study……………………………………………………………... 1
1.2 Statement of problem………………………………………………………………... 4
1.3 Research Questions………………………………………………………………….. 7
1.4 Research objectives………………………………………………………………….. 7
vi
University of Ghana http://ugspace.ug.edu.gh
1.5 Justification of the study……………………………………………………………. 8
1.6 Scope of study…………..……………………………………………………………9
1.7 Organization of the study…………………………………………………………….10
CHAPTER TWO……………………………………………………………………… 11
OVERVIEW……………………………………………………………………………. 11
2.0 Introduction…………………………………………………………………………. 11
2.1 Sources of energy in Africa………………………………………………………….11
2.2 Overview of the Energy sector in Africa…………………………………………….12
2.3 Energy consumption mix in selected West African countries……………………….16
2.3.1 Energy consumption mix in Benin…………………………………………………16
2.3.2 Energy consumption mix in Burkina Faso………………………………………….19
2.3.3 Energy consumption mix in Cote d’Ivoire………………………………………….21
2.3.4 Energy consumption mix in Ghana…………………………………………………23
2.3.5. Energy consumption mix in Nigeria………………………………………………..26
2.4 The West African economy……………………………………………………………29
2.5 Analysis of standard of living in West Africa…………………………………………32
vii
University of Ghana http://ugspace.ug.edu.gh
2.6. Relationship between energy consumption and economic
growth in some selected West African countries……………………………………………33
2.7 Conclusion………………………………………………………………………………..36
CHAPTER THREE…………………………………………………………………………38
LITERATURE REVIEW……………………………………………………………………..38
3.0 Introduction………………………………………………………………………………..38
3.1 Theories of growth………………………………………………………………………...38
3.1.1 The basic growth model………………………………………………………………....38
3.1.2 Endogenous technical change……………………………………………………………40
3.1.3 Growth models with natural resources…………………………………………………...41
3.2. Critique and alternative views……………………………………………………………..43
3.2.1 Ecological economists and Mainstream Views on Growth………………………………43
3.2.2 Limits to substitution……………………………………………………………………..44
3.2.3 Limits to technological change……………………………………………………………45
3.3 Factors influencing the relationship between energy and growth…………………………. 45
3.3.1. Relationship between energy and capital: substitution and complementarity……………46
viii
University of Ghana http://ugspace.ug.edu.gh
3.3.2. Innovation and energy efficiency………………………………………………………..46
3.3.3 Energy quality and changes in the constituents of energy input…………………………47
3.3.4 Shifts in the composition of output……………………………………………………….47
3.4 Review of the empirical literature…………………………………………………………..48
3.5 Conclusion…………………………………………………………………………………..54
CHAPTER FOUR……………………………………………………………………………...55
METHODOLOGY………………………………………………………………………………55
4.0 Introduction…………………………………………………………………………………..55
4.1 Theoretical framework……………………………………………………………………….55
4.2 The Empirical Model………………………………………………………………………...61
4.3 Estimation Technique………………………………………………………………………. 63
4.3.1 Panel Granger causality test……………………………………………………………….65
4.3.2 Testing for Unit Roots in a Panel Context…………………………………………………66
4.3.2.1 Levin, Lin and Chu Panel Unit root test…………………………………………………67
4.3.2.2 Im, Pesaran and Shin Panel unit root test………………………………………………..68
4.3.3 Panel Cointegration………………………………………………………………………..69
4.3.3.1 Kao residual based test…………………………………………………………………..70
4.3.3.2 Pedroni (1999) Cointegration test………………………………………………………..71
ix
University of Ghana http://ugspace.ug.edu.gh
4.4 Estimating the long run relationship………………………………………………………..72
4.5 Testing for causality………………………………………………………………………...74
4.6 Description of variables...……………………………………………………………………75
4.7 Source of data……………………………………………………………………………….76
4.8 Conclusion…………………………………………………………………………………..76
CHAPTER FIVE……………………………………………………………………………….78
DISCUSSION OF RESULTS…………………………………………………………………...78
5.0 Introduction…………………………………………………………………………………..78
5.1 Descriptive statistics of variables…………………………………………………………….79
5.2 Panel unit root test…………………………………………………………………………...80
5.3 Panel Cointegration results…………………………………………………………………..83
5.3.1 Pedroni Panel Cointegration test (GDP, Electricity consumption
and petroleum consumption)…………………………………………………………………….85
5.4 Estimating the long run Relationship………………………………………………………...87
5.5 Short run analysis and policy implications…………………………………………………..94
5.6 Summary of Main findings...………………………………………………………………...99
x
University of Ghana http://ugspace.ug.edu.gh
CHAPTER SIX………………………………………………………………………………101
CONCLUSION AND RECOMMENDATIONS……………………………...... 101
6.0 Introduction………………………………………………………………………………..101
6.1 Conclusion………………..……..…………………………………………………………101
6.2 Recommendations of the study…………………………………………………………….103
6.3 Area for further study………………………………………….……………………………105
REFERENCES….……………………………………………………………………………...106
APPENDICES………………………………………………………………………………….120
xi
University of Ghana http://ugspace.ug.edu.gh
LIST OF TABLES
Table Page
Table 5.1: Descriptive statistics of variables employed………………………………….79
Table 5.2: Levin, Lin & Chu Panel Unit root test………………………………………...80
Table 5.3: Im, Pesaran and Shin Unit root test……………………………………………81
Table 5.4: ADF – Fisher Chi Square……………………………………………………...82
Table 5.5: Results of Pedroni Panel test Cointegrating (GDP Per Capita and Total Energy Consumption)………………………………………………………….83
Table 5.6: Kao Cointegration test results…………………………………………………84
Table 5.7: Fisher Cointegration test (GDP and total energy consumption)……………....84
Table 5.8: Pedroni cointegration results (GDP, electricity and petroleum consumption)...85
Table 5.9: Results of Kao residual cointegration test……………………………………...86
Table 5.10: Fisher Cointegration test (GDP, electricity and petroleum consumption)……86
Table 5.11: Fully Modified Ordinary least squares………………………………………..87
Table 5.12: Dynamic Ordinary least squares………………………………………………88
Table 5.13: Results of short run analysis…………………………………………………..95
Table 5.14: Results of Granger causality test………………………………………………96
xii
University of Ghana http://ugspace.ug.edu.gh
LIST OF FIGURES
Figure Page
Figure 2.1: Average Primary Energy Consumption in West Africa (1980-2013)……14
Figure 2.2: Access to Electricity in West Africa, 2012 (% of Population)…………...15
Figure 2.3: Energy Consumption Mix in Benin………………………………………17
Figure 2.4: Energy use per capita in Benin (kg of oil equivalent per capita)…………19
Figure 2.5: Energy Consumption Mix in Burkina Faso……………………………….20
Figure 2.6: Energy Consumption Mix in Cote D’Ivoire……………………………….21
Figure 2.7: Energy use per capita in Cote D’Ivoire…………………………………….23
Figure 2.8: Energy Consumption Mix in Ghana……………………………………….24
Figure 2.9: Energy use per capita in Ghana…………………………………………….25
Figure 2.10: Energy consumption mix in Nigeria………………………………………27
Figure 2.11: Energy consumption per capita in
Nigeria (kg of oil equivalent per capita)………………………………………………..28
Figure 2.12: Real per Capita GDP Growth Rate and Real GDP growth in West Africa (annual %)………………………………………………………………30
Figure 2.13: Contributions to growth by various sectors of the economy (% of GDP), 2013……………………………………………………………. 31
xiii
University of Ghana http://ugspace.ug.edu.gh
Figure 2.14: GDP per capita for West African countries………………………………...32
Figure 2.15: Relationship between energy use and economic growth in Benin………….33
Figure 2.16: Relationship between energy use per capita and growth for Cote d’Ivoire…34
Figure 2.17: Movement of Energy use per capita and growth in Ghana………………….35
Figure 2.18: Relationship between energy consumption and growth in Nigeria………….36
Figure 3.1: The Solow model……………………………………………………………...39
xiv
University of Ghana http://ugspace.ug.edu.gh
LIST OF ABBREVIATIONS
ARDL Autoregressive Distributed Lag
ADF Augmented Dickey Fuller
BTU British Thermal Units
ECOWAS Economic Community of West African States
EASE Energizing Access to Sustainable Energy
DOLS Dynamic Ordinary Least Squares
FMOLS Fully Modified Ordinary Least Squares
GDP Gross Domestic Product
KwH Kilowatt-Hours
TwH Terawatt-Hours
LPG Liquefied Petroleum Gas
REEP Renewable Energy and Energy Efficiency Partnership
REMP Renewable Energy Master Plan
USD United States Dollars
WAGP West African Gas Pipeline
xv
University of Ghana http://ugspace.ug.edu.gh
CHAPTER ONE
INTRODUCTION
1.1 Background to the study
Prior to the emergence of the oil crisis in 1973, the world had failed to underscore the importance of energy in the production process in an economy. However, the crisis accentuated the importance of energy (Erbaykal 2008). Dunkerly et al., (1981) stipulate that energy has been considered an essential input in tandem with capital and labour in production following the oil crisis. This assertion is further reiterated by Yemane (2008) as he posits that the conceivable improvements to economic growth may rely on the extent to which capital, energy and labour function as complements.
Energy is a resource necessary to drive growth in an economy (Pokharel, 2007; Augutis et al.,
2011). The importance of energy has been underscored since the industrial revolution. Over the years energy use has increased globally and in individual countries. This increase has been in tandem with the increase in gross domestic product. This, therefore, makes it evident that energy is inextricably linked to growth.
The key to enhanced health, enhanced education, better economic opportunities and prolonged life is access to energy (Africa Energy Outlook, 2014). Energy use has over the years revamped the agricultural sector of most economies. Hitherto, agriculture involved planting seedlings, expecting a significant amount of rainfall for seedlings to mature and then harvesting takes place.
However, modern agriculture requires the use of energy in each stage of the agricultural process.
The modern agriculture energy requirements can be dichotomized into direct and indirect energy use. The direct energy use comprises of energy for irrigation, cultivation, farm machinery, water
1
University of Ghana http://ugspace.ug.edu.gh
management as well as harvesting. After harvesting, energy is needed for food processing, storage and in the transportation of the produce to markets. The indirect energy needs are in the form of sequestered energy in fertilizers, herbicides, pesticides, and insecticides (Food and
Agriculture Organization, 2000).
Furthermore, energy is an essential tool for the growth of the manufacturing sector. The availability of cheap energy to drive machines in factories was one of the defining features of the first industrial revolution (Green and Xiao 2013). For any industrialized economy, consistent energy supply is required to power the machines used in production. For this reason, inadequate supply of electricity is likely to have dire consequences on the manufacturing output of any economy: It has been suggested that measures must be put in place to guarantee that energy is supplied in enough and dependable quantum to the industrial sector in order for the maintenance and improvement of its role in the Ghanaian economy (Kwakwa, 2011).
Africa’s energy sector remains a significant component of the continent’s economic development agenda. Africa is well endowed in energy resources which include the sun, wind, hydro just to mention a few. Therefore, the energy sector has resources which if properly harnessed and managed will be sufficient to meet domestic needs. In Northern and Western Africa, there is a colossal dependence on fossil fuel in the generation of electricity as a result of the concentration of oil and gas in these regions. In contrast, countries in Central and the Eastern part of Africa largely depend on hydropower in generating electricity In the Southern part of Africa, the power sector is generated using Coal and to some degree hydro power (Energy Sector in Africa, 2011).
More so, in West Africa, about 5 -7 percent of the continent’s hydroelectric potential has been tapped and only 0.6 percent of its geothermal potential (World Bank, 1997).
2
University of Ghana http://ugspace.ug.edu.gh
Notwithstanding the colossal energy potential in Africa, energy consumption as a whole and electricity consumption specifically is very low compared to other continents. (Economic
Commission for Africa, 2004). For instance, according to data from World Development
Indicators’ (2016) the share of the populace who are able to access electricity in a country like
Liberia is estimated at approximately 10% which as against a 100% access to electricity in the
United States. While the energy sector has the potential, it is poorly managed as more than two- thirds of its populace does not have access to modern energy (Africa Energy Outlook, 2014).
The energy sector in West Africa has over the years been crippled with various challenges such as the inconsistency in the supply of energy. In spite of these energy challenges, West African economies have not performed badly. Growth in West African economies had been slow over the 1990s. Growth had reduced from a value of 4.0% recorded in 1993 to 2.1% in 1994 (The
African economy, 1994). This has partly been attributed to the slow rate of growth experienced by large economies in the sub region specifically Nigeria as well as political instabilities in some of these countries. In recent times, however, according to the (African Economic Outlook, 2015), growth in West Africa was estimated to be 6 % in 2014 and growth is expected to rise to 6.1% in the year 2016.
In Ghana for instance, despite the numerous challenges that have engulfed the economy, an annual average growth of 4 percent has been recorded over the years. Cote d’Ivoire post-civil war has recorded significant improvements in the economy. The economy recorded a robust growth of 8.3 percent in 2014 (African Development Bank, 2015). Countries including Benin,
Niger, and Togo are on a path to sustained growth. Following the recent instabilities in Mali and
Guinea Bissau, growth contracted slightly. However, growth in these two countries after the unrest shot to 5.8 percent and 2.6 percent in 2014 in Mali and Guinea-Bissau respectively.
3
University of Ghana http://ugspace.ug.edu.gh
Nigeria has the largest economy in the sub-region and has recorded positive growth rates despite the numerous challenges. Overall, the African Economic Outlook (2015) envisages complex tests ahead for West Africa with regards to the economy. However, it also predicts bright prospects for these West African economies.
1.2 Statement of problem.
Extensively, it is acclaimed that the readiness of contemporary, consistent and productive energy services constitutes an essential and undeniably important tool for economic development.
Again, complete access to energy systems that provide a consistent and sufficient supply of power to homes, firms and service providers is a condition for persistent human development.
However, in Africa, attempts to alleviate poverty are being challenged by the insufficient energy supply and access (Panel, 2015).
In West Africa, the populace has over the years endured hardships as a result of limited access to electricity (Gnansounou, 2008). In recent times, the electricity crisis has aggravated in some
West Africa countries. A classic example is the inconsistent power supply that has bedeviled
Ghana for over five years beginning from the year 2010. Again, with the reduction in the level of water in the Akosombo Dam in 1998, Benin and Togo were subjected to intermittent power outages. This disruption which lasted for several months consequently stifled growth in these economies causing a recession (Gnansounou, 2008).
Furthermore, Nigeria for some years has had difficulties in consistently supplying electricity to its citizenry and industries. This is because electricity demand has consistently been more than the supply of electricity in the country. This coupled with the inability of the country to harness its vast gas resource potential to generate electricity has led to acute electricity problems in the
4
University of Ghana http://ugspace.ug.edu.gh
country (Sambo et al., 2010). Some other West African countries including Senegal, Mali, and
Guinea have also experienced inefficient electricity supply. In Senegal for instance, the IMF
(2010) postulate that over the years 2006 to 2009, the disruptions in electricity supply were frequent and this consequently adversely affected business operations in the country. Indeed, most West African countries have experienced interruptions in electricity due to the excess demand for electricity that occurs at some point.
These inconsistencies and insufficient power supply have impacted negatively on growth in these economies. The interruptions mean that the manufacturing sector that relies hugely on electricity suffer greatly. For those companies that can afford generators and plants to augment power supply, they will operate at a higher production cost. This transcends to higher prices of their output and ultimately they are less competitive compared to their counterparts in other parts of the world who operate at a relatively lower cost due to a consistent power supply. In Ghana, for instance, companies are losing 15 percent of the value of sales as a result of power shortages
(US Agency for International Development, 2014). Also, Africa (2009) reports that with the availability of reliable electricity supply, tropical producers would experience a decline in energy cost by 60 percent. This quagmire has hindered the alleviation of poverty in a region reported to have the highest proportion of the world’s poor people.
Different studies have sought to examine the energy – growth nexus (Ouedraogo, 2013;
Twerefou et al., (2008); Akinlo, 2008). Diverse conclusions have been reached due to the different methodology and data employed in the various studies. Some studies have found a causal link running from energy consumption to economic growth (Cheng and Lai, 1997).
Meanwhile, other studies have discovered that no causal link exist between the two variables,
(Glasure and Lee, 1997). Also, others have concluded that a bidirectional causality links energy
5
University of Ghana http://ugspace.ug.edu.gh
consumption to economic growth (Asafu-Adjaye, 2000; Akinlo, 2008). Some studies have concentrated on Africa; however, in spite of the vast studies that have sought to investigate the energy-growth nexus, with regards to West Africa, there are limited studies focusing on the sub- region.
It is essential to ascertain the influence of disaggregated energy consumption on economic growth as well as the direction of the causality for policy purposes. Investigating the direction helps to know if energy conservation policies should be undertaken or not. For those countries that have causality running from energy consumption to growth, policies that will ensure sustained energy supply will be essential. On the other hand, for those countries with causal link running from economic growth to energy consumption, undertaking energy conservation policies will have a negligible effect on economic growth (Ouedraogo, 2010).
According to Behera (2015), four likely conclusions could be reached on the causality between energy consumption and economic growth. Firstly, there could be a unidirectional relationship running from energy consumption to growth. This implies that energy is an important element in the growth process and for that matter, a fall in energy generation as well as inconsistent and insufficient energy supply, would have dire consequences on economic progress. This causality is also called the growth hypothesis. Secondly, there can exist a unidirectional causality running from growth to consumption of energy. This is also termed the conservation hypothesis. Thirdly, it is possible to have a bidirectional causality running from energy consumption to economic progress. This is where there exist causality from economic progress to energy consumption and concurrently, a causality running from energy consumption to economic progress. This is termed the feedback hypothesis. Fourthly, there could be a neutral relationship between energy consumption and economic progress, where no relationship exists between the two variables.
6
University of Ghana http://ugspace.ug.edu.gh
This will inform policy makers that, the consumption of energy has no influence on the economic progress in the West African sub-region.
1.3 Research Questions.
Examining the relationship between energy consumption and economic growth is important for the purpose of policy making. In West Africa, not many studies have ascertained the effect of disaggregated energy consumption on economic growth. Therefore, this studies seeks to fill the gap in the literature by finding answers to the following research questions;
What is the nature of the causality between Total energy consumption and economic
growth for the West African region?
What is the nature of the causality between petroleum energy consumption and economic
growth in West Africa?
What is the nature of the causality between electricity consumption and economic growth
in West Africa?
1.4 Research Objectives.
The main objective of the study is to explore the causal link between energy consumption and economic growth.
Specific objectives
Examine the causality between Total energy consumption to economic growth for the
West African region.
Empirically ascertain the causal linkage between petroleum energy consumption and
growth for the West African sub–region.
7
University of Ghana http://ugspace.ug.edu.gh
Investigate the causal linkage between electricity consumption and economic growth for
the West African region.
1.5 Justification of the study
Energy is an essential input promoting the plight of people as well as providing economic and social progress (Diskiene et al., 2008). However, with concerns of increasing exigency for energy along with the insufficiency in energy supply make energy supply a concern for all countries (Šliogerienė et al., 2009).
In West Africa, the importance of energy is accentuated by Economic Community of West
African States projects such as the West African Gas Pipeline (WAGP) which is to produce and channel natural gas from Nigeria to other member states for the generation of electricity and the
West Africa Power Pool which is also to ensure inter-connectivity of national electricity grids.
This implies that the importance of energy in the sub-region cannot be overemphasized.
However, the performance of the energy sector in the sub-region has been abysmal.
There are vast literature examining the connection between energy consumption and economic growth in both the developed and the developing world. The consensus as to the direction of the causality is still not explicit. That is, the empirical findings with regards to the causality between the consumption of energy and economic progress have been varied. Most of the studies that explore the energy-growth connection such as (Akinlo, 2008) have sought to ascertain the relationship between total energy consumption and economic growth in addition to electricity consumption and growth. This study posits that different types of energy including, petroleum
8
University of Ghana http://ugspace.ug.edu.gh
energy could impact on growth differently and therefore taking energy consumption in totality without disaggregating it into the various types consumed may not provide the true picture of the causal relationship.
Again, knowledge on the causality between energy consumption and economic growth is important for policy purposes. If a unidirectional causality from growth to energy is arrived at, then the implication is that policies that aim at conserving energy can be implemented without any dire consequence on economic growth. Juxtaposing this with the scenario where a unidirectional causality runs from energy consumption to economic growth, the policy implication will be putting in place measures that will ensure consistent energy supply so as to spur growth. However, if no causality runs in either direction then it could imply policies targeted at conserving energy would not influence economic growth (Akinlo 2008). The study will therefore, be of immense benefit to ECOWAS as a regional organization and also to the individual countries.
1.6 Scope of the study
The study covers the period 1980 to 2013. The choice of the time period is based on the unavailability of data with regards to periods before 1980 as well as periods after 2013. There are seventeen West African countries includes in the study. The seventeen West African countries are Benin, Burkina Faso, Cote d’Ivoire, Cabo Verde, The Gambia, Ghana, Guinea, Guinea
Bissau, Niger, Nigeria, Mali, Mauritania, Senegal, Liberia, Sao Tome and Principe, Sierra Leone and Togo
9
University of Ghana http://ugspace.ug.edu.gh
1.7 Organization of the study
The study is organized into six chapters. The remainder of this study is organized therefore as: chapter two contains an overview of the energy sector in the West African sub-region. It also contains an overview of the economic performance of West African States. Chapter three focuses on the review of both empirical and theoretical literature. Chapter four also focuses on the methodology employed in the study. Chapter five contains the estimation, analysis, and discussion of results and finally, Chapter six includes the summary of the study as well as policy recommendations.
10
University of Ghana http://ugspace.ug.edu.gh
CHAPTER TWO
OVERVIEW OF ENERGY CONSUMPTION AND ECONOMIC GROWTH IN WEST
AFRICA.
2.0 Introduction
In this chapter, we present an extensive analysis of energy consumption and growth in West
Africa. This study evaluates the level of access to electricity in some selected West African countries. Again, the chapter analyses the energy consumption mix in the various countries along with the structure of the economy in the countries.
2.1 Sources of Energy in Africa.
The sources of energy in Africa comprise wind, water, sun, coal just to mention a few. In
Northern Africa, there is the presence of oil and gas in colossal quantities. Libya, for instance, is stipulated to have approximately 50 percent of oil reserves on the continent. Southern Africa dominates in coal and uranium or nuclear reserves. It is estimated that southern Africa has 91 percent of the continent’s coal reserves.1 The use of hydro resources is also quite predominant in
Central and Western Africa. Eastern Africa is known for its huge renewable energy potential.
Furthermore, Africa is endowed with many natural resources including fossil fuel, crude oil, biomass, solar, wind just to mention a few; some of these resources such as petroleum, natural gas, coal, crude oil have been fairly exploited and have consequently been harnessed to provide energy in African countries. Again, some of these resources such as wind, solar have not been exploited to the optimal level and therefore their potential level of use has not been reached.
Coal, for instance, is a major source of energy is the Southern part of Africa. Oil and gas energy
1 Energy in Africa; accessed online from en.m.wikipedia.org
11
University of Ghana http://ugspace.ug.edu.gh
production is also quite predominant in the Western part of Africa. With regards to renewable energy, Africa is tagged as the continent with the best capacity for renewable energy. This is due to the enormous renewable energy available on the continent.
Africa is endowed with a considerable amount of sunlight as a result of its geographical positioning as it is the closest continent to the equator. The sun could be harnessed to provide solar energy for African countries. However, despite the solar energy potential of Africa, the continent lags behind in terms of solar energy. It is worth highlighting that in recent ties, attempts have been made by some countries to exploit their solar energy potential. Examples include the 160-megawatt solar plant that is under construction in Morocco. Countries like
Rwanda and Kenya has also invested in solar energy with the former known for having the fastest built solar energy plant.
The continent also has colossal wind energy potential. Countries like Kenya have the reputation for having built the continent’s biggest wind energy farm. Unfortunately, in spite of this potential, the energy sector still faces significant challenges. Africa also harbors many rivers including the Nile, Congo, and Volta among others which have the potential of providing energy for the continent. These rivers jointly constitute 13% of the hydropower potential in the world
(UNECA, 2007).
2.2 Overview of the Energy sector in Africa.
Energy in Africa consists of fossil fuel and renewable energy; with growing concerns of climate change, investment in renewable energy production had become necessary. Africa is estimated to have 9.5%, 5.6%, as well as 8% oil, coal, and gas resources respectively (BP, 2006). The continent also has significant renewable energy potential. However, despite this potential, the
12
University of Ghana http://ugspace.ug.edu.gh
energy sector in Africa is characterized by extremely low access to energy. It is interesting to find a whopping 64% of the populace with no access to electricity2. For most of the Sub-Saharan
African countries, access to electricity remains a major challenge. These challenges are existent in a geographical area that has been described as heavily endowed in terms of resources for energy production. It then becomes evident that the issue with African countries is not the availability of resources; however, it is how these resources can be harnessed and used to provide energy to feed the growing population.
In West Africa, according to data from the US Energy information administration, energy consumption has varied over the years among the member states. Particularly, over the years
1980 to 2013, the average energy consumption in quadrillion British Thermal Units recorded were significantly different. The average energy consumption in Burkina Faso was recorded as
0.4 quadrillion Btu. Nigeria has the largest economy in the sub-region and therefore energy consumption is relatively higher in the country compared to her counterparts in the sub-region.
Nigeria recorded the highest average energy consumption figure of approximately 20.1 quadrillion Btu over the period 1980 to 2013. Figure 2.1 highlights the average primary energy consumption figures of the countries in West Africa over the period 1980 to 2013. Primary energy consumption includes energy consumed from its direct source. Stated differently, primary energy is energy that has not been subjected to any conversion or transformation process; energy consumed from its direct source.3 It includes consumption of the energy sector, losses in the process of transformation (for instance from oil or gas to electricity) and distribution of energy as well as the final consumption by end users.4
2 Energy sector in Africa 3 Glossary of environmental statistics; retrieved from stats.oecd.org/glossary/detail.asp 4 Eurostats statistics explained; accessed online from ec.europa/Eurostat/statistics-explained/index.php
13
University of Ghana http://ugspace.ug.edu.gh
25
20
15
10
5
0
Figure 2.1: Average Primary Energy Consumption in West Africa (1980-2013). Source: Constructed by Author with data from EIA.
In Figure 2.1, Total energy consumption varies among the countries in the sub-region. We find that Nigeria recorded the highest average energy consumption figure of 20.1 Million tons of oil equivalent. As argued by Adenikinju (2008), the size of the Nigerian economy is relatively large compared to the other countries and it is the most populous of all the countries. This is partly why it has over the years recorded the highest energy consumption figure. Ghana, Cote d’Ivoire,
Senegal, and Benin all recorded significant figures of 3.2, 2.5, 1.5, and 0.7 Million tons of oil equivalent respectively. However, we find that countries like Sao Tome and Principe, Guinea
Bissau, Liberia among others recorded very minimal energy consumption figures. This could be as a result of the inability of these countries to properly harness their energy resources to produce energy for consumption as well as the relatively small population size of these countries.
14
University of Ghana http://ugspace.ug.edu.gh
Again, access to electricity remains a significant challenge to the energy sector in West Africa.
Access to electricity is very low in urban centres and rural areas. The situation is worse, particularly in the rural areas. In Ghana, access to electricity is estimated by the US energy information administration to be 72 %; that is to say 72% of the populace have access to electricity with urban areas access to electricity estimated as 92% as against that of the rural areas of 50%. This clearly shows the incongruence in access to electricity in the sub-region.
Figure 2.2 illustrates the share of the population with access to electricity in West African states.
Togo 31.5 Sierra 14.2 Senegal 56.5 Sao Tome 60.5 Nigeria 55.6 Niger 14.4 Mauritania 21.8 Mali 25.6 Liberia 9.8 Guinea B 60.6 Guinea 26.2 Ghana 64.1 Gambia 34.5 Cote 55.8 Cabo verde 70.6 BF 13.1 benin 38.4
Figure 2.2: Access to Electricity in West Africa, 2012 (% of Population). Source: Constructed by Author with data from World Development Indicators’ (2016).
It is evident from Figure 2.2 that, approximately 71% of the entire populace in Cape Verde have access to electricity as at the year 2012. Countries like Ghana, Guinea Bissau follow with
15
University of Ghana http://ugspace.ug.edu.gh
approximately 64% and 61% of their population with access to electricity. For countries like
Burkina Faso and Liberia, they have quite a minute percentage of their total population with access to electricity.
2.3 Energy consumption mix in selected West African countries.
In this sub-section, the study discusses the various components of energy consumed in some selected countries in the sub-region. The countries are selected based on the availability of data pertaining to the country.
2.3.1 Energy Consumption Mix in Benin
Energy in Benin as in many West African countries is dominated by biomass use energy sources.
The country has a population of a little over 10 million in 2015 with an estimated gross domestic product per capita of 779.07 at current USD in 2015 (World Development Indicators’ 2016). The consumption of fossil fuel as a share of the total energy consumption was estimated to be
42.88% in 2013 (Energypedia). In 2009, Benin recorded a total energy consumption of 3475 ktoe according to the World Bank, (2009). Biofuels and waste constituted 58.7% of the total primary energy supply with oil constituting 41.3%. With regards to electricity consumption, the
International Energy Agency estimated the country’s electricity consumption as 1.03 TWh in
2014. However, it is worth noting that the country is also highly dependent on electricity imports from neighboring countries like Ghana, Cote d’Ivoire, and Nigeria in order to satiate its electricity demands. Precisely, 85% of its electricity is imported from these countries mentioned earlier (Energypedia). That notwithstanding, the country has a significant renewable energy potential which unfortunately has not been exploited much. The country is quite dependent on petroleum products.
16
University of Ghana http://ugspace.ug.edu.gh
In Figure 2.3, an illustration of the primary energy consumption mix in Benin sourced from the
International Energy Statistics is presented. In Benin, petroleum consumption is a key component of the total energy consumption mix. This shows that the Benin hugely depends on imported petroleum. The use of coal, as well as hydro energy, are quite negligible in the country.
There is also no major use of nuclear in the country.
A number of investments have been made in the energy sector in Benin. The Benin government has adopted a long term electricity investment of increasing electricity generation to 4000MW through partnership with the private sector. In 2013, a company named Combustion Associates
Incorporated completed the installation of an 80MW gas turbine in Benin. This is to help augment the already plants in generating electricity. Among the policies put in place to strengthen the energy sector include promoting energy efficiency in all sectors, promoting private investments in the power sector. Benin also wants to ensure that there is equal access to electricity in the Benin. With regards to renewable energy, Benin is part of the countries under the ECOWAS Regional Centre for Renewable Energy and Energy Efficiency which essentially is to ensure that harnessing of renewable energy for sustained growth and energy supply.
Figure 2.3 shows the energy consumption mix in Benin.
17
University of Ghana http://ugspace.ug.edu.gh
Figure 2.3: Energy Consumption Mix in Benin.
Source: International Energy Statistics, 2014
It is evident from Figure 2.3 that petroleum constitutes a substantial component of the energy consumption mix in Benin. This evidence backs the earlier assertion made about the dependence of the country on petroleum products. Again, a significant quantum of renewable energy is consumed to some extent. Though yet to be fully exploited, there is huge capacity for renewable energy exploitation. Overall, total primary energy consumption has been consistently rising since the year 2002 reaching its peak in the year 2011.
The energy use per capita as recorded over the years is presented in Figure 2.4.
18
University of Ghana http://ugspace.ug.edu.gh
450
400
350
300
250
200
150
100
50
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Figure 2.4: Energy use (kg of oil equivalent per capita) in Benin Source: Constructed by Author with data from World Development Indicators’ (2016)
In Figure 2.4, it is evident that energy use in Benin has over the years being increasing. It is evident from Figure 2.4 that energy consumption peaked at a value of 393.38 kg of oil equivalent per capita in 2014. Despite the quite substantial decrease in energy use from a value of 314.34 in
2005 to 305.44 kg of oil equivalent in 2006, any other decrease in energy consumption over the years has been marginal.
2.3.2 Energy Consumption Mix in Burkina Faso
Biomass use is quite prevalent in Burkina Faso; in the rural areas, biomass is the main source of energy. With a surface area of 274,222 km squared and a population of over 18 million as at
2015 (Energypedia), the country according to the World Bank recorded a GDP Per Capita of
683.95 USD in 2013. The country imports electricity from neighboring countries Ghana and
Cote d’Ivoire. The total installed electricity capacity in 2013 was estimated as 247MW. With
19
University of Ghana http://ugspace.ug.edu.gh
regards to renewable energy, the installed solar energy potential in 2014 was estimated to be around 400kWp. In spite of the country’s huge solar potential, solar energy constitutes a minimal proportion of the energy consumed. The country has a restricted potential for wind energy due to the geographical location of the country (REEEP policy database, 2014). A survey of hydropower shows that the country has enormous hydro potential; hydroelectricity utilization covers about 20% of the national electricity grid. Figure 2.5 shows the energy consumption mix in Burkina Faso.
Figure 2.5: Energy Consumption Mix in Burkina Faso Source: International Energy Statistics, 2014
In Burkina Faso, there is a colossal dependence on petroleum products in terms of the energy consumption mix in the country. Despite the huge potential of solar energy, there is very
20
University of Ghana http://ugspace.ug.edu.gh
minimal use or investment in renewable energy. There is also no use of nuclear energy in the country.
With regards to policy, the country initiated energy sector reforms with the inauguration of the
Energy Sector Development Policy paper in December 2000. An important component of this policy is to encourage private participation in the energy sector. The government of Burkina
Faso has planned on increasing electricity production by 1000MW from 2015-2025.
2.3.3 Energy Consumption Mix in Cote d’Ivoire
Cote d’Ivoire is stipulated to have four primary energy sources; these include hydropower, oil, natural gas, and biomass. However, a greater proportion of energy consumption comes from biomass.5 The country is estimated to have a natural gas reserve of 1 trillion cubic feet.6 With regards to the country’s electricity generation, it is estimated that hydropower and thermal energy are the major components of the country’s electricity generating capacity. Thermal is estimated to have accounted for 62 percent of the generating capacity in 2002 whilst hydropower accounted for 38 percent.7 Figure 2.6 presents the energy consumption mix in Cote d’Ivoire.
5 REEEP Policy Database; Retrieved online from www. Reegle.info/policy-and-regulatory-overviews/CI 6 Energy in Ivory Coast Retrieved online from en.wikipedia.org/wiki 7 Energy in Ivory Coast; Retrieved online from www.en.wikipedia.org/wiki
21
University of Ghana http://ugspace.ug.edu.gh
Dry Natural Gas Consumption Total Petroleum Consumption Hydro Electricity, Nuclear and other Renewables .15
.10
.05
QuadrillionBtu 0.00 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 2.6: Energy Consumption Mix in Cote D’Ivoire.
Source: International Energy Statistics, 2014
In Cote d’Ivoire, there is significant use of natural gas. Indeed, the generation of electricity in the country has over the years depended on domestic natural gas. The country imports a substantial quantum of petroleum and for that matter, petroleum constitutes a component of the energy mix.
Again, hydro energy is quite significant in the country as it also used in generating electricity.
Cote d’Ivoire is one of the model countries in terms of investments in energy in the sub-region.
The country emerging from a recent civil war is now able to export electricity to neighbouring countries such as Ghana. A major investment in energy in Cote d’Ivoire is the 6.6 billion dollar investment by the government of Cote d’Ivoire to increase the production of electricity. The government again an adopted an audacious plan of adding 150 megawatts of electricity to its current national grid by the year 2020.
22
University of Ghana http://ugspace.ug.edu.gh
700
600
500
400
300
200
100
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Figure 2.7: Energy use per capita in Cote D’Ivoire
Source: Constructed by Author with data from World Development Indicators’ (2016)
In Cote D’Ivoire, energy use has over the years experienced some fluctuations. At a very low point of 377.23 kg of oil equivalent per capita, consumption shot up to 522.34 kg of oil equivalent per capita. The political instabilities experienced in the country over the same period could have resulted in the substantial reduction in energy use over the period. However, it is important to highlight the fact that since 2010, energy use has continually been increasing.
2.3.4 Energy consumption mix in Ghana.
In Ghana, biomass constitutes a larger proportion of energy consumed; the use of biomass is predominant in most Ghanaian households.8 Another important energy endowment existent in
8 REEEP Policy Database
23
University of Ghana http://ugspace.ug.edu.gh
the country is its renewable energy potential. Solar energy is available in colossal volume but yet to be fully harnessed. The country also makes use of its hydro potential in generating electricity.
Furthermore, the use of coal is minimal. However, petroleum constitutes a significant component of the energy consumption mix over the years. Again, renewable energy including biomass with its huge potential has been explored to some extent as its consumption has been rising over the years. In recent years, however, there have been attempts to explore solar energy as an option in generating electricity. Figure 2.8 displays the energy consumption mix in Ghana.
Figure 2.8: Energy Consumption Mix in Ghana
Source: International Energy Statistics, 2014
From Figure 2.8, it is evident that the consumption of renewable energy since the year 2010 has gradually been increasing in the country. In terms of renewable energy, the country has huge potential. Solar is available in abundance in the country. The monthly solar irradiation is between
4.4 and 5.6 kWh/m2/day with a sunshine duration between 1800 and 3000 hours per annum
(REEEP policy database). Biomass is one resource that is predominantly used in the country.
24
University of Ghana http://ugspace.ug.edu.gh
Biomass fuels in Ghana primarily include charcoal, wood fuel as well as plant residue. Ghana has quite exploited its hydro potential generating electricity from its exploitation. The hydropower potential is estimated to be about 2420MW.
Until recently, the country had failed to largely exploit its solar potential. However, in recent times, a 2 megawatt-peak solar photovoltaic solar plant has been completed in Navrongo. Again, with regards to electricity, the FPSO Nkrumah along with the recent Ameri power plant acquisition has added to the country’s existing national grid. With regards to energy policies, the government of Ghana has a national energy policy which the objective of increasing access to electricity in every part of the country by the year 2020. Another important energy-related policy with the objective of increasing the renewable energy component in the final energy consumption mix is the National Renewable Energy Policy. The policy aims at raising the proportion of renewable energy consumption in the final energy consumption to 10% by the year
2020.9 Again, the national electrification scheme established has been quite successful increasing access to many communities in the country.
3000
2500
2000
1500
1000
500
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Figure 2.9: Energy use per capita in Ghana
Source: Constructed by Author with data from World Development Indicators’ 2016
9 REEEP Policy Database
25
University of Ghana http://ugspace.ug.edu.gh
Ghana has recorded relative higher levels of energy use per capita compared to the other West
African countries such as Benin and Burkina Faso. However, since 2011, energy use has continually been decreasing. The energy crisis experienced over recent times could partly be the reason for this trend. That notwithstanding, the per capita energy use values recorded are still higher than other countries in the sub-region.
2.3.5 Energy Consumption Mix in Nigeria
Nigeria is one of the oil producing countries in the West African sub-region. In 2005, the country recorded the highest figure of crude oil production which was estimated at 2.44 million b/d
(Energy information Administration). However, due to civil and political instabilities such as the
February 2000 conflict between Christians and Nigerians the Biafra war just to mention a few, there have been persistent fluctuations in crude production in the country. In terms of the energy consumption mix, the country consumes quite a significant volume of petroleum. Again, biomass in Nigeria comprises of shrubs, wood and forage grasses, animal waste and other waste from forestry, agriculture as well as aquatic biomass. Again, traditional biomass constitutes a major component of the energy consumption. Traditional biomass and waste account for 83% of total primary consumption.10 With regards to hydropower, it is estimated that the gross hydropower potential for the country is approximately 14750MW with current hydropower generation estimated as constituting 14 percent of the country’s hydropower potential.11 Figure 2.10 shows the energy consumption mix in Nigeria.
10 Energy in Nigeria; retrieved online from Wikipedia 2016. 11 Nigeria: Rapid Assessment and Gap analysis; sustainable energy for all; accessed online from se4all.org/sites/default
26
University of Ghana http://ugspace.ug.edu.gh
Figure 2.10: Energy consumption mix in Nigeria
Source: International Energy Statistics, 2014.
From Figure 2.10, it is evident that in 2014, petroleum consumption constituted a significant component of energy consumption in the country. Again, Nigeria is one of the countries in the sub-region that make good use of its natural gas potential. Also, natural gas is consumed in a significant proportion in the country.
Nigeria has a colossal solar energy potential largely due to its geographical position. It is estimated that if Nigeria were to properly harness its solar potential by covering approximately 1 percent of its natural land surface with solar collectors, the country could generate 1850 x 103
GWh of solar electricity per year and this would have much more sufficient to feed the country’s energy demand that its current generation capacity (REEEP policy database, 2014). With regards to its energy framework, Nigeria launched a program named Renewable Energy Master Plan
(REMP) in 2006 which basically is to ensure that renewable energy potential is exploited to
27
University of Ghana http://ugspace.ug.edu.gh
generate energy for consumption in the country. Among the targets of the program include a
40MW wind energy generated in the country by the year 2025 and the building of small hydro plants which would cumulatively generate 2000MW of electricity by the year 2025. Again, there is the Energizing Access to Sustainable Energy (EASE) which also is aimed at improving at mitigating issues of deforestation in the country and encouraging the efficient use of energy in the country. It is also aimed at improving access to energy and assuaging the impact of climate change by encouraging the decline of greenhouse gas emissions.
Figure 2.11 illustrates the per capita energy consumption in Nigeria.
820 800 780 760 740 720 700 680 660 640 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Figure 2.11: Energy use per capita in Nigeria. Source: Constructed by Author with data from World Development Indicators’ (2016)
From Figure 2.11, it is evident that energy use in Nigeria have been fluctuating over the years. In
2010, the country experienced a substantial reduction in the per capita energy use. In 2010, the crude oil and natural gas sectors were estimated to have recorded a decline in production.12
These resources also constitute a key component of the energy consumption mix in the country.
12 Nigeria: Rapid assessment and gap analysis; sustainable energy for all
28
University of Ghana http://ugspace.ug.edu.gh
For this reason, the reduction in energy use over the period could be attributed to the decline in the production of these resources over the period.
2.4 The West African economy
The West African economy has over the years recorded positive growth rates. In 2014, some economies in the sub-region were debilitated due to the outbreak of Ebola. Specifically, countries like Liberia, Sierra Leone, and Guinea were struck with the Ebola virus and this led to the reduction in the growth rates recorded in these economies. In other countries like Cote d’Ivoire, Guinea, the presence of civil war at a point in the history of these countries also had a weakening effect on the economy of these countries. However, in spite of these setbacks, West
African countries have recorded robust growth rates in recent years. In 2015, Cote d’Ivoire recorded growth rate of 8.8% (Sahel and West Africa Club, 2016).
Figure 2.12 shows the average growth of GDP and real GDP Per Capita over the period 2000-
2014 for the West African region.
8.00
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00 GDP Growth Per Capita GDP Growth
29
University of Ghana http://ugspace.ug.edu.gh
Figure 2.12: Real per Capita GDP Growth Rate and Real GDP growth in West Africa (annual %) Source: Constructed by Author with data from African Development Bank, 2016
From Figure 2.12, it is evident that the West African sub-region has since the year 2000 recorded positive GDP per capita growth rates. Over the period 1980 to 2013, the West African sub-region recorded an average GDP growth rate of approximately 6.8% whilst the average per capita GDP growth is 3.9%. Growth has been consistently rising and this is reflected in the overall wellbeing of the populace in the sub-region. However, as this study asserts, the Ebola epidemic disease experienced in some of the countries in the sub-region coupled with the general economic misfortunes experienced by some of the countries have resulted in the slow growth rates recorded in recent times particularly in the year 2014. The African Development Bank annual report stipulates that the presence of the Ebola disease in countries including Guinea, Sierra
Leone, Liberia and parts of Nigeria contributed immensely to the growth value of 6 percent recorded in the sub-region in 2014. The report also argues that the persistent conflicts along with the sharp decline in oil prices meant that some of the countries recorded fiscal deficits and this consequently led to the marginal increase in growth rate recorded in 2014.
In West African economies, whilst some countries are dependent on agriculture, the industrial sector and the services sector contribute significantly to GDP. Again, a few other countries including new actors like Mali, Niger together with Nigeria, Ghana are oil exporting countries
(The Socio-economic context of West Africa, 2006).
In Figure 2.13, there is an illustration of the contributions of the various sectors of the economy to GDP.
30
University of Ghana http://ugspace.ug.edu.gh
80
70
60
50
40
% OF % GDP 30
20
10
0
AGRIC INDUS SER
Figure 2.13: Contributions to growth by various sectors of the economy (% of GDP), 2013
Source: Constructed by Author with data from World Development Indicators’ 2016
As illustrated in Figure 2.13, apart from countries like Guinea Bissau, Mauritania and Sierra
Leone, the services and other sectors of the economy excluding the agriculture and Industrial sector contributed the highest proportion to growth in these economies. However, for countries including Sierra Leone and Guinea Bissau, agriculture contributes more to GDP than other sectors in the economy. For Mauritania, the industrial sector plays a significant role in growth as it contributes more to GDP than the other sectors of the economy.
31
University of Ghana http://ugspace.ug.edu.gh
2.5 Analysis of standard of living in West Africa.
In this sub-section, the study analyses GDP per capita as recorded in the various countries. GDP per capita essentially is employed as a measure of the wellbeing of the populace in the countries.
As such, in analyzing GDP per capita, the study is able to paint a picture of the living conditions of citizens in these countries.
Figure 2.14 shows average GDP per capita recorded for West African countries over the period
1980 to 2013.
1800 1600 1400 1200 1000 800 600 400 200 0
Figure 2.14: Average GDP per capita for West African countries (1980-2013). Source: Constructed by Author with data from World Development Indicators’, 2016
In West Africa, Cape Verde recorded the highest average GDP per capita of 1583.5 dollars over the period 1980 to 2013. Nigeria follow with a per capita GDP of 749.9 dollars. Countries including Niger, Liberia, and Gambia have a relatively lower standard of living as they recorded
32
University of Ghana http://ugspace.ug.edu.gh
lower average per capita GDP’s. Some of the countries record lower GDP per capita values due to challenges in their economies. Countries like Liberia, Sierra Leone, for instance, are still in the process of recovering from the Ebola epidemic.
2.6 Relationship between energy consumption and economic growth in some selected West
African countries.
In this sub-section, the study analyses the trend between energy use and economic growth for some selected countries. The countries for the analysis are selected based on the readiness of data on the variables of interest for that country. Energy use in this analysis is measured as primary energy that has not been subjected to any transformation or conversion process. Economic growth is measured using GDP per capita; which is given by the total GDP of a country in that particular year divided by the total number of its population at that particular time.
Figure 2.15 illustrates the relationship between energy consumption and economic growth in
Benin over the period 2001 to 2014.
1000 900 800 700 600 500 400 300 200 100 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Energy use/per capita GDPper capita
Figure 2.15: Relationship between energy use and GDP Per Capita in Benin.
Source: Constructed by Author with data from World Development Indicators’ (2016)
33
University of Ghana http://ugspace.ug.edu.gh
In Figure 2.15, it is evident that over the years starting from the year 2001, energy use per capita increases steadily initially. This is followed by subsequent increases in GDP per capita; therefore, as energy use increase, GDP per capita also increases. However, along the years particularly in 2006, as energy use per capita reduces, GDP per capita recorded in the same year\ increases from a value of 570.68 kg of oil equivalent per capita in 2005 to 587.08 kg of oil equivalent in 2006.
Figure 2.16 illustrates the Relationship between energy consumption and GDP Per Capita in
Cote d’Ivoire between the period 2001 and 2014.
1600
1400
1200
1000
800
600
400
200
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Energyuse per capita GDP per capita
Figure 2.16: Relationship between energy use and GDP Per Capita for Cote d’Ivoire.
Source: Constructed by Author with data from the World Development Indicators’ (2016)
In Cote d’Ivoire, Figure 2.16 makes it evident that energy use per capita rises initially whilst
GDP per capita experience a marginal fall between 2002 and 2003. Energy use per capita has,
34
University of Ghana http://ugspace.ug.edu.gh
however, been increasing at a rate faster than GDP per capita. An interesting point worth noting in the figure is that, between 2009 and 2010, energy use per capita reduced substantially. Within the same period, GDP per capita also reduced. This trend could be attributed to the civil war which destabilized the country within that period.
2000
1800
1600
1400
1200
1000
800
600
400
200
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Energy use per capita GDP per capita
Figure 2.17: Movement of Energy use per capita and GDP Per Capita in Ghana.
Source: Constructed by Author with data from World Development Indicators’ (2016).
In Ghana, there is an interesting connection between energy consumed and GDP Per Capita. The evidence available shows that for West African countries like Benin among others, energy consumption per capita increases at a faster rate than GDP per capita, in Ghana GDP per capita increases at a rate faster than energy consumed per capita. This could imply that energy consumed has over the years had no relationship with growth as is being argued by Twerefou et al., (2008) in Ghana.
35
University of Ghana http://ugspace.ug.edu.gh
Figure 2.18 illustrates the Relationship between energy consumption and GDP Per Capita between the period 2001 and 2014.
3500
3000
2500
2000
1500
1000
500
0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Energy use per capita GDP per capita
Figure 2.18: Relationship between energy consumption and GDP Per Capita in Nigeria.
Source: Constructed by Author with data from World Development Indicators’ (2016).
It is patent from the Figure 2.18 that in the years between 2001 and 2005, energy consumed per capita was more than GDP per capita recorded in the country. However, we find that growth in the economy is more than energy consumed per capita from the year 2006 and beyond.
2.7 Conclusion
In this chapter, the study has explored the various components of the energy consumption mix in the various countries in the sub-region. The study finds that biomass constitutes a greater
36
University of Ghana http://ugspace.ug.edu.gh
proportion of energy consumed in the sub-region. Again, petroleum constitutes a major component of the energy consumption in countries like Benin. The study also finds that some countries have a relatively higher rate of access to electricity than others. That is to say, in countries like Ghana, the share of the population with access to electricity is estimated to be approximately 64 %. Again, Cape Verde has the highest rate of access to electricity with a proportion of about 70% of its population with access to electricity. Countries like Liberia,
Burkina Faso have the lowest rate of access to electricity with an estimated proportion of 10% and 13% respectively of their population with access to electricity. In terms of the economic structure of the various countries, the study finds that the structure of the economy of the countries vary. Whilst countries like Sierra Leone, Guinea Bissau have the agricultural sector contributing more to GDP, other countries like Cape Verde, Benin have the services and the industrial sector contributing a higher percentage to growth than the other sectors.
37
University of Ghana http://ugspace.ug.edu.gh
CHAPTER THREE
LITERATURE REVIEW
3.0 Introduction
This chapter elaborates the theoretical perspective of energy- growth nexus. The theoretical review is based largely on the work done by Stern (2004) and Stern & Cleveland (2004). Stern
(2004) analyses the role of energy in economic growth by ascertaining the importance of energy in the production process. Similarly, Stern & Cleveland (2004) reviews the pertinent biophysical theory, mainstream and resource economics models of growth as well as some criticisms of the mainstream model of growth. The thesis then reviews some of the empirical work available so far on the connection between energy consumption and economic growth.
3.1 Theories of Growth
3.1.1 The Basic Growth Model
The simplest growth model in economics was propounded by Solow (1956) and this has become known as the Solow model. The model just like other core mainstream growth models failed to recognize resources or energy as important in the production process. According to Solow
(1956), output produced is a function of capital, labour and technology. Labour and technology enter the production function as a single variable known as effective labour. Solow assumes that both labour and technology grow at a constant rate. The model assumes a constant returns to scale to capital and labour; that is to say, if capital and labour are changed by a common factor, output varies by the same factor. Again, as capital employed in the initial levels of production increases, output also increases at an increasing rate. However, as capital increases continually, output starts increasing at a diminishing rate. Figure 3.1 depicts the Solow model.
38
University of Ghana http://ugspace.ug.edu.gh
Figure 3.1: The Solow Model
푌
∗ Y=f(k) 푦
푛 + 푔 + 푑 푘
푆 = 퐷
푆 = 푠푓 푘
∗ 푘 푘
The diagram shows the relationship between output and capital; the uppermost curve represents the output curve. y* and k* represent the equilibrium output and capital stock respectively. sf(k) is the actual investment and this includes a constant proportion of the output. (n+g+d)k is the breakeven investment. According to Solow, at k* actual investment is equal to breakeven investment as depicted above and therefore the economy is in a long run equilibrium. In order words, at k*, capital stock does not change and so it is called steady state capital stock. For any point below k*, actual investment is more than breakeven investment and hence capital rises.
Stating differently, for a point below k*, the addition to capital stock is more than the depreciation of capital and capital must increase. Again, for any point above k*, the depreciation of capital is more than the addition to capital and capital must fall. Any deviation from k* brings the economy back to k* and that is why it is the steady state capital stock. For countries with a smaller capital stock, growth tends to be faster due to the high marginal product of capital
39
University of Ghana http://ugspace.ug.edu.gh
associated with the smaller capital stock. No country can grow perpetually by reason of diminishing returns to capital stock. An increase in the savings rate shifts the output curve upwards and takes the economy to a new equilibrium capital stock. Again, according to the model, capital and output continually increase provided the labour force multiplies at a fixed rate continually. However, in equilibrium, per worker output and per capita output stays constant.
The Solow model is exogenous in that it assumes that technological changes are unexplained.
This gave rise to the emergence of endogenous models which sought to find variables that could explain technological changes and these include the neoclassical growth theory. As argued by the proponents of the neoclassical growth theory, the only basis for persistent economic growth is an improvement in technology. With an improvement in the level of technology, a given amount of input would produce greater or better quantities of output. Improving the wealth of knowledge derived from technology available raise the rate of return to capital and this consequently offsets the diminishing returns to capital accumulation that would have otherwise resulted in stagnation of growth.
3.1.2 Endogenous technological change
New growth models endeavor to explain technological change as dependent on certain factors and decisions. They argue that advancement in technology within the growth model is the consequence of firms and individuals decisions. Endogenous growth models include the “AK” model and the Research and development model. In the endogenous growth models, the link between output and capital is modeled as Y = AK, and A is given as a constant as well as capital
K comprising of manufactured capital and knowledge-based capital. Endogenous growth models
40
University of Ghana http://ugspace.ug.edu.gh
have established that with the constant term, ‘A’ in the model, there will be an unlimited growth as capital is accrued. The endogenous growth models treat technological knowledge as a type of capital and it is amassed through the process of Research and Development (R&D). Through research and development, the economy benefits from the spill overs of such innovations. That is, there is some component of external effects that the economy enjoys through research and development and for that matter the benefit to the economy is more than that enjoyed by the original innovator. These benefits are able to drive growth in the economy. As new capital are installed by firms, this process inclines towards product and process advancement. The motivation to commit more resources to innovation is derived from the short term monopoly profits derived from fruitful innovations. Therefore, the growth of capital (K) implies the growth of a composite stock of capital and intangible technological knowledge. For this reason, output is in this scenario not subjected to diminishing returns but able to increase as a stable share of the composite capital stock. The economy is also influenced by a higher savings rate. The greater the rate of savings, the greater the level of growth in the economy. Therefore, in an endogenous growth model, there is a sustained growth in output and diminishing returns to capital are counterbalanced by the external benefits derived from technological growth.
3.1.3 Growth models with natural resources
All the models that have been discussed so far have assumed that there are no natural resources in the economy. With the exception of sunlight and deuterium that may exist in large quantities, most natural resources exist in finite quantities. Again, some environmental resources are non- renewable (that is, they cannot reproduce through natural processes); those resources that are renewable are also expendable. The exhaustibility and the finiteness of renewable and non- renewable resources respectively make the concept of unlimited economic growth difficult.
41
University of Ghana http://ugspace.ug.edu.gh
The presence of more than one input in the production process implies that economic growth could take many alternative paths; the institutional arrangements assumed determine the path that economic growth takes. Overtime, analysts have assessed growth models that attempt to attain sustainability and growth models that anticipate real world economies with the existence of proper market structures.
The neoclassical literature on growth focus on the circumstances that make a perpetual increase in economic growth possible or a non-reducing consumption or utility level. These conditions according to the literature are the technical and institutional arrangements in the economy. The institutional arrangements include the structure of the market system (that is, competition as against central planning), and the system of property rights. The technical conditions include the blend of renewable and non-renewable resources, the extent to which the inputs are substituted for each other and the initial endowments of natural and capital resources.
According to Solow (1974), achieving a continued increase in economic growth is possible in a model that is fixed and nonrenewable natural resource with no extraction costs and non- depreciating capital. Again, the elasticity of substitution between capital and the natural resources should be unity and other technical conditions must be met. With these, Solow (1974) argues that economic growth can increase perpetually. Dasgupta and Heal (1979) also argue that when a society continually invests in enough capital to replace the exhausted natural resources then sustainability would be possible.
Technical change can, therefore, be effectively used to separate economic growth from energy and other resources. Through technical change, resources that have been exhausted can be replaced with other human capital such as machines. But this is not the case; as argued earlier, the foremost interest of neoclassical economist are the institutional arrangements that will
42
University of Ghana http://ugspace.ug.edu.gh
eventually result in sustainability in the economy and not the technical arrangements. Therefore, neoclassical economists find out ways by which institutional arrangements may lead to sustainability. According to Solow (1974), situations where the elasticity of substitution between nonrenewable resources and capital is greater or less that unity can be discarded. In a case where the elasticity is greater than unity, sustainability is possible; however, in situations where the elasticity is less than unity, sustainability is not possible if the economy relies only on nonrenewable resources. Again, in the presence of renewable resources, sustainability is possible at least when there is no population growth.
3.2 Critique and Alternative Views
3.2.1 Ecological Economics and Mainstream Views on Growth
Ecological economists have a view of the structure of an economy that is quite different from that of the neo classical economists. Whilst ecological economists argue that the environment remains an important and significant component of the economy and could form the foundation of the economy, main stream growth theory focus on the institutional restrictions to growth.
Proponents of the growth theory argue that technological progress and substitution among inputs would cause the economy to grow sustainably. However, critics of the growth theory center their argument on the possible limits to substitution and limits to technological change as means of alleviating the lack of resources. If resources and technological change are replaced with manufactured capital, then more output could be derived from the restricted resource input and this could help avoid the inability of the natural environment to reduce the impact resource and energy use by absorbing the impacts. However, if these two means are restricted, then extreme environmental impact may have a negative restriction on growth.
43
University of Ghana http://ugspace.ug.edu.gh
3.2.2 Limits to substitution
There will be many reasons why substitution will be restricted due to the fact that there are many forms of substitution between inputs. There is the possibility of having substitutions that exist in a class of related production and between diverse sets of inputs. There is also a difference in substitution made at the micro level and at the macro level. Again, some other forms of substitution that are probable in a country may not be imaginable universally.
The within category form of substitution is very essential. It includes the substitution of a renewable resource for a nonrenewable resource. In economies that are heavily industrialized, there is a high tendency for the substitution of certain resources for the other. There is the substitution of coal for wood, oil for coal, natural gas for oil and then electricity is substituted for oil. There is also the possibility of the elasticity of substitution for substitution of inputs in related class or group to be higher than unity. The implication of this would be that some individual inputs are not important.
The substitution of manufactured capital for natural capital is one the ecological economists place much emphasis on as they find it to be very essential. The importance of natural capital is inherent in the fact that it is required for capturing energy and for mitigating the effect of using energy and resource. All production activities will in one way or the other have an impact or interrupt the natural environment even in cases where a minimal volume of energy is used in production. In most cases, a particular type environmental interruption such as pollution is substituted with another form of environmental disruption such as the construction of a hydro dam.
44
University of Ghana http://ugspace.ug.edu.gh
3.2.3 Limits to Technological Change
In situations where changes in technology are boosted by natural capital and are unrestricted in scope, then it is possible to achieve sustainability even if substitution possibilities are restricted.
The case for technological change as the answer to sustainability would be more persuading if technological change was a concept which is quite different from the concept of substitution.
According to the neoclassical approach, an unlimited number of productive techniques exist at any particular time and that substitution takes place among these methods. There is an advancement in technology when modern proficient methods are discovered. Nevertheless, stated differently, these new techniques signify the replacement of factors of production with knowledge. More skilled workers, as well as improved capital goods, represent the knowledge.
More so, skilled workers and capital goods can only produce in when there is an availability of energy, materials and for this reason, there are still thermodynamic limitations on the degree to which energy and material flows can be decreased regardless of how sophisticated the worker and machinery become.
3.3 Factors influencing the relationship between energy and growth
A general production function is represented by the equation (1);
(푄1,…,, 푄푚)' = f (A,푋푖 , 퐸푖 …………………….. 1
Where 푄1 represent the various output that could possibly be produced using 푋푖 inputs available including capital, labour, 퐸푖 includes the various energy inputs that may be used in the production process and A represents the total factor productivity. The connection between energy and output may, therefore, be influenced by the following factors;
Replacement of energy with other inputs
45
University of Ghana http://ugspace.ug.edu.gh
A change in total factor productivity; that is change in technology
Change in the composition of the energy input.
Change in the composition of output
3.3.1 Relationship between Energy and Capital: Substitution and Complementarity
Diverse arguments have been made by various studies as to the nature of the connection between capital and energy. Some studies including Apostolakis (1990) have stipulated that capital and energy behave as substitutes in the long run and then in the short run, they act as complements using time series regression techniques. However, with the emergence of the concept of
Cointegration, one could argue the uncertainties with regards to the conclusion by Apostolakis
(1990). Frondel and Schmidt (2002) having reviewed the work of Apostolakis (1990) also argue that with a smaller cost proportion of energy in production, it is probable to have a complement relationship between energy and capital. Again, Thompson & Taylor (1995) stipulate that using the Morishima elasticity of substitution and not the Allen-Uzawa elasticities concludes in capital and energy having a substitutable relationship. According to Blackorby & Russell (1989), the substitutability or the complementarity of energy and capital can only be ascertained using the indication of the cross price elasticity which essential has an identical sign as the Allen-Uzawa substitution elasticities. It appears therefore that energy and capital are no less than best weak substitutes and probably complements.
3.3.2 Innovation and energy efficiency
Various studies have attempted to establish that a technological innovation could consequently improve energy use and have an impact on economic growth. Stern (2002) argues that innovations in technology provide households with appliances that require energy to function
46
University of Ghana http://ugspace.ug.edu.gh
thereby increasing energy use in households and industries benefit and industries benefit from technological innovations as it provides opportunities to save energy. Autonomous energy efficiency index (AEEI) measures variations in the energy/GDP ratio that are distinct from variations in the relative price of energy. According to the Khazzoom-Brookes Postulate
(Brookes, 1990; Khazzoom, 1980) or “rebound effect”, energy saving innovations lead to more monetary expenses on some other goods which would also require energy to function.
3.3.3 Energy quality and changes in the constituents of energy input.
Energy quality refers to the economic efficacy per heat equal unit of diverse fuels and electricity.
Energy quality can be determined by the marginal product of fuel. The marginal product of fuel measures the change in output produced as a result of an additional heat unit of fuel. Energy quality is not fixed as it changes depending on the marginal product of fuel which varies depending on the activities it is meant for, the quantum and type of capital, labour along with other materials that augment it. Schurr & Netschert (1960) argue that the quality of energy has a significant impact on output produced. This according to the literature is because a higher quality fuel decreases the quantum of energy needed to create goods worth a dollar of GDP. Similarly,
Berndt (1990) stipulates that employing a higher quality of energy input increases total factor productivity.
3.3.4 Shifts in the composition of output
As an economy grows, the structure of that economy changes. For some economies, there is a shift from agricultural dependence to industrialization as it grows. This shift which consequently changes the nature of output produced in the economy could affect energy use in the economy.
47
University of Ghana http://ugspace.ug.edu.gh
In the early stages of progress, there is a change from dependence on agriculture towards investment in heavy industries which as widely known require more energy.
3.4 Review of the empirical literature.
In the previous section, we looked at the theoretical perspective of the relationship between energy and economic growth. In this section, we elaborate on some of the empirical work that have been done in relation to the research area.
Neither correlation nor ordinary linear regression of variables can be used to ascertain a causal relationship among the variables that may be employed in ascertaining the connection between energy consumption and economic growth. In this regard, various studies as stipulated by
Granger (1969) and Engle and Granger (1987) have employed Granger causality tests and
Cointegration analysis to establish the causal linkage between the two variables.
Hamilton (1983) and Burbridge & Harrison (1984) both discovered that variations in prices of oil does granger cause GNP. This results were supported by Stern (1993) who employed a multivariate framework and proxied growth with GDP. Again, capital as well as labor inputs, and a quality-adjusted index of energy input were used as a measure of gross energy use. He tests for granger causality and finds energy to granger cause GDP. His conclusion therefore is that energy spurs economic growth.
Masih & Masih (1996) investigate the possibility of a connection between energy and output in six Asian countries in the long run. They find the existence of cointegration in countries including India, Pakistan, and Indonesia. However, they find no cointegration in countries like
48
University of Ghana http://ugspace.ug.edu.gh
Malaysia, Singapore, and Philippines. The conclusion therefore is that a causal relationship runs from energy consumption to growth in India but the reverse exists in the other countries.
Wolde-Rufael (2006) examines the causal relationship between electricity consumption and economic growth for seventeen African countries. He employs the ARDL bounds test approach and finds that for nine of the countries, there exist in the long run a link between electricity consumption per capita and real GDP per capita and Granger causality for only 12 out of the seventeen countries. He again finds that for six of the countries there exist a uni-directional causality running from real GDP per capita to electricity consumption per capita. However, for three countries, there exists a unidirectional relationship running from electricity consumption to real GDP and a bi-directional causality for the three other countries. His results are similar to that of Bildirici (2013) who also finds a bidirectional causality for three countries namely Ghana,
Gabon and Guatemala. Also, the growth hypothesis exists in Cameron, Congo Rep., Ethiopia,
Kenya and Mozambique and the conservation hypothesis in Senegal and Zambia.
Contributing to the argument on the causal direction between energy and growth, Akinlo (2008) investigates the relationship between energy consumption and economic growth in eleven countries in the sub-Saharan African region. His study like Wolde – Rufael (2006) and
Odhiambo (2008) employs the ARDL bounds test. He finds a long run relationship between energy consumption and economic growth in Cameroon, Cote d’Ivoire, Gambia, Ghana,
Senegal, Sudan and Zimbabwe. He then employs the Granger causality test and the results show that in countries like Gambia, Ghana and Senegal, economic growth causes energy consumption whilst energy consumption also causes economic growth. There is therefore a bidirectional causality in these countries. However, employing the Granger causality test, he finds that economic growth Granger causes energy consumption in countries like Sudan and Zimbabwe. In
49
University of Ghana http://ugspace.ug.edu.gh
Cameroon, Cote d’Ivoire as well as in Nigeria, Kenya and Togo, the neutrality hypothesis is confirmed with regards to the relationship between energy consumption and economic growth.
Twerefou et al., (2008) employ the Vector Error correction model using data spanning from
1975 to 2006 to explore the causal link between energy consumption and economic growth for
Ghana. They found that for Ghana, there existed the growth led hypothesis. The authors explained that the Ghanaian economy if driven by the agricultural sector which does not require much energy for its growth as compared to the industrial sector.
Lee and Chang (2008) contribute to the literature on the relationship between the consumption of energy and economic growth. They explore the link between energy and economic progress in
16 Asian countries. Employing data spanning from 1971 to 2002 as well as panel cointegration and panel error correction model, they find that in the short run, there is causal relationship running from GDP to energy consumption. However, in the long run, there is causality running from energy consumption to growth.
Furthermore, a paper by Odhiambo (2008) explores the connection between energy consumption and growth in Tanzania. The author employs the ARDL bounds test approach propounded by
Pesaran et al. (2001) in investigating causal relationships. He employs two proxies of energy consumption; total energy consumption per capita and electricity consumption per capita and time series data spanning from 1971 to 2006. The results of the bounds test show that there is a stable long-run relationship between total energy consumption and growth and electricity consumption and growth. With regards to the causality test, the study finds a causal relationship running from total energy consumption to growth and again a unidirectional relationship running from electricity consumption to economic growth. His study concludes therefore that energy consumption spurs economic growth.
50
University of Ghana http://ugspace.ug.edu.gh
Huang et al., (2008) espouse their views on the relationship between energy consumption and economic progress with focus on 82 countries. These countries include low income countries, middle income countries and high income countries. The data employed spans from the year
1971 to 2002. The study then uses the panel techniques and concludes that for middle income countries, the conservation hypothesis is confirmed. However, for low income countries, the results indicate the existence of the neutrality hypothesis and for high income countries, progress in the economy was found to have a negative connection with the consumption of energy.
Apergis & Payne (2009) sought to examine the causal link between energy consumption and economic growth in six Central American countries. The study employs panel data spanning from 1980 to 2004. The study test for a long run relationship among the variables using panel cointegration techniques. The error correction model is used to ascertain the causal relationship between the variables. The results from the granger causality test indicate the existence of the growth causal relationship running from energy consumption to growth and thereby confirming the growth hypothesis in the countries.
Similarly, Ozturk et al., (2010) explore the relationship between energy consumption and economic growth. Their study focuses on 51 developing countries and data spanning from 1971-
2005. The countries are segregated on the basis of the economic standing of the countries. The 3 groups are lower income countries, lower middle income countries and upper middle income countries. They find that energy consumption does not lead to economic growth in any of the three income groups and therefore energy conservation policies could be pursued in all the countries.
Esso (2010) in ascertaining the relationship between energy consumption and economic growth in seven sub-Saharan Africa employs Cointegration methodology. Sourcing data on energy and
51
University of Ghana http://ugspace.ug.edu.gh
growth within the period 1970–2007, the author finds a long run relationship between energy and economic growth. With regards to the causality test, the author found a unidirectional causality running from real GDP to energy consumption in Ghana.
Kwakwa (2012) studies the causality between energy consumption and economic growth in
Ghana. He disaggregates total energy consumption into electricity consumption and fossil consumption. He further divided economic growth into overall growth, agricultural and manufacturing growth over the period 1971-2007. Employing the Johansen Cointegration test, he found the presence of Cointegration between the variables employed. With regards to the causality test, the paper found a unidirectional causality between overall growth to electricity and fossil consumption. Furthermore, a unidirectional causality from agriculture growth to electricity consumption was evident in Ghana over the time period chosen and a feedback relationship is found between manufacturing and electricity consumption.
Onakoya et al., (2013) in espousing their views on the relationship between energy consumption and growth in Nigeria employ data spanning from 1975 to 2010. They employ the cointegration approach and the ordinary least squares estimation technique. Their findings show that a long run movement between energy consumption and economic growth in the long run. Again, they find that electricity and the aggregate energy consumption have significant and positive relationship with economic growth in Nigeria. They also find that the consumption of coal has a significant but negative relationship with economic growth in Nigeria.
Ouedraogo (2013) focuses on the West African sub-region as she examines the direction of the relationship between energy consumption and economic growth. The author employs panel
Cointegration techniques including panel unit root, panel Cointegration and Granger causality tests. The conclusion is that there is long run movement between GDP and energy consumption
52
University of Ghana http://ugspace.ug.edu.gh
and electricity. The causality results show that there is a unidirectional relationship running from
GDP to energy consumption in the short-run, and from energy consumption to GDP in the long- run. They conclude by recommending that the government should boost the consumption of the various forms of energy used in generating power.
Lu (2016) examines the link between electricity consumption and economic growth for 17
Taiwanese industries. Data within the period 1998 and 2014 is employed. The techniques used in analyzing the data includes Panel unit root test and panel cointegration approaches. The study finds a bidirectional causality in the short run as well as the long run.
Esso & Keho (2016) take the discussion on energy consumption and economic growth nexus further by investigating the linkage between energy consumption, economic growth and carbon emissions focusing in twelve selected sub-Saharan African countries. They employ the bounds test to cointegration and granger causality and use data spanning from 1971 to 2010. The results show that economic growth causes emissions in countries like Benin, Democratic republic of
Congo, Ghana, Nigeria and Senegal. Again, they find a reverse causality running from C02 emissions to economic growth in Gabon, Nigeria and Togo and a bidirectional causality running from economic growth to C02 emissions in the short run for countries including Nigeria and in the long run for Congo and Gabon. Also, in the long run, energy consumption and economic growth cause C02 emissions in Benin, Cote D’Ivoire, Nigeria, Senegal, South Africa and Togo.
Yasar (2017) explores the relationship between the consumption of energy and economic growth for 119 countries. The countries are segregated into four different groups. The division is made on the basis of income ranking of the countries by the World Bank. Data that spans from 1975 to
2015 is employed as well as the panel autoregressive distributed lag and granger causality technique is employed in analyzing the data. The results indicate that for low and lower middle
53
University of Ghana http://ugspace.ug.edu.gh
income groups, the neutrality hypothesis is confirmed in the short run. For upper middle income and lower middle income groups, the results indicate the presence of the conservation hypothesis in the long run. Again, the outcome reinforces the feedback postulation for upper middle income and high income groups in the long run.
3.5 Conclusion
In this chapter, we have analyzed the theoretical underpins of the connection between energy and growth. The study further looked at some of the studies that have sought to explore the link between energy consumption and economic progress and the findings that have been reached on the dynamic causal link between energy and growth. This study finds that there are existing studies that have sought to analyze the linkage between energy and growth. However, the consensus as to the direction of the relationship is still not explicit. Whilst other studies find a unidirectional causality, others find a bidirectional causality and some have also found no relationship between the two variables. This study contributes to the existing literature by examining the relationship between total energy consumption and economic growth. The study further disaggregates energy into electricity consumption and petroleum consumption and ascertains the effect of these components of total energy on economic growth.
54
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FOUR
METHODOLOGY
4.0 Introduction
This chapter is segregated into three sections. The first section concentrates on the theoretical framework underpinning the methodology employed. The second section discusses the empirical model employed in the analysis. The third and final portion contains an analysis of the estimation techniques used in the study, the source of data employed and variables used in the empirical model as well as a summary of the chapter.
4.1 Theoretical framework
Solow (1956) and Swan (1956) propounded the growth model that explains how factors of production drive growth. The model is represented by the equation;
푌 푡 = 퐹 퐾 푡 , 퐴 푡 퐿 푡 (1) where Y represents output produced, K is capital, L is labour and A represents technology or knowledge. According to the model, technology is captured as labour augmenting and hence
A(t)L(t) represents effective labour. Equation (1) can be represented in a Cobb Douglas form as;
훼 1−훼 푌푡 = 퐾푡 퐴푡퐿푡 (2)
Similarly, in this scenario, K represents capital stock with A and L representing the level of efficiency or technology and labour respectively. The Solow model assumes the level of technology to be exogenous. However, subsequent growth models that emerged after the Solow model have sought to find factors that may influence the degree of technology in an economy.
Accordingly, the level of efficiency A is explain by the equation;
55
University of Ghana http://ugspace.ug.edu.gh
푔푡+ 𝜌푡휃 퐴 = 퐴0푒 (3)
In equation 3, the extent of efficiency is explained by factors including the growth rate ‘g’ which is the rate of technological progress and it is supposed to be constant, 𝜌 is the vector representing all the other factors such as energy consumption that may possibly influence the level of technology and productivity in the economy, 휃 is the vector of coefficients associated with these variables whilst the subscript t denotes time.
In deriving how capital ‘K’ evolves overtime, Solow (1956) makes three assumptions; these assumptions are represented by the following equations;
푆 = 푠푌, 0 < 푠 < 1 (4)
퐼 = ∆퐾 + 훿퐾 (5)
Equation (4) can be rewritten as; 퐼 = 퐾̇ + 훿퐾 (6)
퐼 = 푆 (7)
Where I is investment, S denotes savings, s is a fraction of output saved, K is capital, 훿 denotes the depreciation rate and it is a fraction between 0 and 1, 퐾̇ is the change in capital. The first assumption represented by equation (4) implies that a fixed fraction of output is saved in an economy. Equation (5) implies that investment is given by the addition of the change in capital and the depreciation of capital. Equation (7) implies that investment is equal to savings in an economy. From equations (5), (6) and (7) we can write the equation that shows how capital ‘K’ evolves overtime as;
퐾푡̇ = 푠푘푌푡 − 훿퐾푡 (8)
We define y and k which represent output per effective labour and capital per effective labour as;
56
University of Ghana http://ugspace.ug.edu.gh
푌푡 퐾푡 푦푡 = and 푘푡 = (9) 퐴푡퐿푡 퐴푡퐿푡
We can represent the evolution of the physical capital stock by a unit of labor as;
̇ d Kt kt = [ ] (10) dt AtLt
Differentiating equation (9) with respect to time (t) gives;
̇ ′ ̇ 퐾푡 퐴푡퐿푡 −퐾푡 퐴푡퐿푡 푘푡 = 2 (11) 퐴푡퐿푡
̇ ̇ ̇ ̇ 퐾푡 퐴푡퐿푡 −퐾푡 퐴푡퐿푡+퐴푡퐿푡 ⇒ 푘푡 = 2 (12) 퐴푡퐿푡
퐾̇푡 퐴푡퐿푡 퐾푡퐴푡퐿̇ 푡 퐾푡퐴̇푡퐿푡 = 2 − 2 − 2 (13) 퐴푡퐿푡 퐴푡퐿푡 퐴푡퐿푡
̇ ̇ ̇ ̇ 퐾푡 퐾푡 퐴푡퐿푡 퐾푡 퐴푡퐿푡 푘푡 = − ( )( ) − ( ) . ( ) (14) 퐴푡퐿푡 퐴푡퐿푡 퐴푡퐿푡 퐴푡퐿푡 퐴푡퐿푡
̇ ̇ ̇ ̇ 퐾푡 퐿푡 퐴푡 푘푡 = − 푘푡 − 푘푡 (15) 퐴푡퐿푡 퐿푡 퐴푡
But 퐾̇푡 = 푠푘푌푡 − 훿퐾푡
̇ ̇ ̇ 푠푘푌푡−훿퐾푡 퐿푡 퐴푡 ⇒ 푘푡 = − 푘푡 − 푘푡 (16) 퐴푡퐿푡 퐿푡 퐴푡
퐴̇ The growth in technology is assumed exogenous and therefore = 푔 and the growth in labour 퐴
퐿̇ = 푛. 퐿
̇ 푌푡 퐾푡 Therefore; 푘푡 = 푠푘 − 훿 − 푛푘푡 − 푔푘푡 (17) 퐴푡퐿푡 퐴푡퐿푡
57
University of Ghana http://ugspace.ug.edu.gh
푌푡 퐾푡 But 푦푡 = and 푘푡 = 퐴푡퐿푡 퐴푡퐿푡
Therefore;
̇ 푘푡 = 푠푘푦푡 − 푛 + 푔 + 훿 푘푡 (18)
This equation represents the equation of motion of k.
From the Cobb Douglas production function in equation (2), deriving output per effective labour
gives;
1 YKALt t() t t YKtt 1 ()ALtt ALALt t t t
K t yt ( A t L t ).( A t L t ) ALtt
K y t t ()ALtt K y t t AL tt yk tt (19)
Substituting equation (18) into equation (17) gives;
. k s k () n g k k t t (20)
̇ However, at the steady state 푘푡 = 0 and therefore;
58
University of Ghana http://ugspace.ug.edu.gh s k () n g k kt (21)
In deriving the equilibrium steady state capital from equation (20); we have;
skt k() n g k 1 sk k() n g s k1 k ()ng 11 * sk k ()ng (22)
The economy converges toward a steady state denoted by equation (22).
Again, from equation (19);
푌푡 훼 = 푘푡 (23) 퐴푡퐿푡
This implies that equilibrium steady state output can be represented by the equation;
* Yi * ki ALii
*** yi A i k i (24)
푌푖 ∗ Note that = 푦푖 is the output per labour. 퐿푖
The relationship in equation (8) symbolizes the output by the worker at the equilibrium and for every country.
59
University of Ghana http://ugspace.ug.edu.gh
At the equilibrium, technological progress is represented by equation (25);
A* A eii i 0 (25) where 𝜌 represents the factors or variables that may possibly influence the level of technology. 𝜌 consists of the variables reflecting energy consumption in this study.
Substituting equation (25) and (22) into (24) gives;
* s 1 y A e k 0 ng (26)
Including the indications for time and individual countries and taking the natural logarithm of both sides, equation (26) can be written as;
* ln yi, t ln A 0, i i , t ln s k ln n i , t g 11it, (27)
where (퐴0,1 represents the constant of every country, 푠푘푖,푡 denotes the physical capital reserves 푛푖,푡 denotes the growth rate of the labour, g represents the growth rate of technological progress whilst 훿 is the rate of investment depreciation. The rates g and 훿 are supposed to be stable through time and across countries. Replacing 𝜌 with energy consumption in equation (27) gives;
* ln yi, t ln A 0, i i ENER i , t ln s k ln n i , t g 11it, (28)
60
University of Ghana http://ugspace.ug.edu.gh
4.2 The Empirical Model
From equation (28), we can simplify further and write our empirical model that establishes the linkage between total energy consumption and growth as;
LGDPC LENER it i it it (29)
Where LGDPC represents the natural logarithm of GDP per capita, and LENER represents total energy consumption, 𝑖 is the individual country identifier and 푡 denotes time.
Again, as argued by Olofin et al., (2014), the various types of energy consumed could have different impacts on growth. As such, the consumption of electricity measured by electricity consumption in billion kilowatt per hour and petroleum consumption measured in million tons of oil equivalent are introduced. Therefore, the second empirical model estimated is given as;
LGDPC ELEC PE it i it it (30)
Where LGDPC represents Gross Domestic Product Per Capita, ELEC represents electricity consumption and PE is petroleum consumption. Following the work of Odhiambo (2008), equation (29) and (30) are estimated separately to reduce the incidence of multicollinearity.
In the empirical analysis, following the work of Stern (1993), economic growth is measured by
GDP and the consumption of energy is measured by primary energy consumption Following the work of Ouedraogo (2013), the natural logarithm of capital, labour, GDP Per Capita and energy consumption is used in the study to reduce the issue of heteroscedasticity.
Furthermore, in ascertaining the short run and long run dynamics, an error correction model is employed. Econometrically, the generic form of the error correction model is written as follows;
61
University of Ghana http://ugspace.ug.edu.gh
푛 푛
푦푡 = 훼0 + ∑ 훼1푖푦푡−푖 + ∑ 훼2푖푥푡−푖 + 훼3퐸퐶푡−1 + 푢푡 31 푖=1 푖=1
푛 푛
푥푡 = 훽0 + ∑ 훽1푖푦푡−푖 + ∑ 훽2푖푥푡−푖 + 훽3퐸퐶푡−1 + 휀푡 32 푖=1 푖=1
푦푡 and 푥푡 are the variables of interest in this scenario; 푢푡 and 휀푡 represent the serially uncorrelated error terms, 퐸퐶푡−1 is the lagged error correction term obtained from estimating the long run Cointegration equation. From the equations above, 푥푡 is said to granger cause 푦푡 in the long run if 훼2푖 ≠ 0 and 훼3 ≠ 0. The short run causal effects are obtained by performing the F test on the regressors. The significance of the error correction term gives the long run causal relationship.
As argued by Alam (2006) neo classical economist fail to recognize energy as an essential input in the process of production and this is captured in the their concept of the production function as capital, labour and technology are treated as the essential input in production. Neo classical economist take out energy from the economy and as such alienate the economy from the sources of energy or the ecology. The use of energy is only treated as an intermediate good and not as an input in production. This argument is however countered by ecological economists such as Ayres and Nair (1984) who have argued that energy is the sole most important input in the production process as determined by the laws of thermodynamics. Ecological economists have argued from a thermodynamic perspective the presence of an a priori relationship between energy use and output production and treat capital and labour as intermediate inputs that need energy for their production and maintenance (Stern, 1993). In examining the connection between energy and growth, some studies such as Burgess (1984); Siddiqui (2004) have employed the neo classical approach where capital and labour are treated as the essential inputs in the production function
62
University of Ghana http://ugspace.ug.edu.gh
and energy is treated as having a relatively minor role in production to analyze the link between energy and growth. However, as argued by Stern (1993), a major flaw of all studies that have sought to employ the production function approach is that it is possible to estimate any form of the production function. Therefore, following the work of Akinlo (2008), Ozturk et al., (2010),
Odhiambo (2008), Wolde-Rufael (2006), Twerefou et al., (2008), Razzaqi et al., (2011) and for the purpose of this study, the relationship between energy consumption and growth is analyzed by employing the variables of interest including Gross Domestic product per capita, total energy consumption, electricity consumption and petroleum consumption.
4.3 Estimation Technique
Following the work of (Ozturk et al., 2010; Wolde-Rufael 2006; Esso, 2010), this study employs a panel data estimation approach in investigating the link between energy consumption and economic growth. Panel data includes data obtained by pooling observations on a cross section of a particular subject which may include firms, households, countries etc. over several time periods. Panel data includes a cross section dimension and a times series component. Panel data has several merits or qualities compared to pooled data some of which include the ability to control for individual differences among the individual units. Panel data is able to account for the fact that individual units may have their own unique characteristics which may make the different from the other units. The inability of time series and cross section data to account for individual differences may possibly lead to biased results (Baltagi, 2005). Furthermore, panel data is able to detect and control for issues of multicollinearity. In time series, issues of multi collinearity is quite rampant. In panel data, the cross sectional dimensional together with the time series dimension makes the data more variable and flexible. As such, issues of high collinearity between any of the variables are reduced to the minimum. Again, the combination of time series
63
University of Ghana http://ugspace.ug.edu.gh
component along with the time series dimension makes the data more informative and reliable.
In light of these benefits derived from the use of panel data together with the fact that the objective of the study is to explore the causal link between energy consumption and economic growth for the seventeen West African countries over a period of years, panel data is used in this study as it suits the investigation of the various objectives of the study.
The generic form of a panel regression is given as follows:
푌푖푡 = 훼푖 + 훽푋푖푡 + 휀푖푡 33
Where 푋푖푡 𝑖푠 푎 퐾 × 1 vector of independent variables, 훽푖푡 𝑖푠 푎 퐾 × 1 vector of parameters to be estimated,𝑖 = 1,2 … . . 푁 푎푛푑 푡 = 1,2 … … . 푇, 훼푖 is the time invariant constant term, 휀푖푡 is the error term and 푌푖푡 represents the dependent variable.
Furthermore, in investigating the dynamic causal relationship between the variables, this study adopts three steps: firstly, the panel unit root test will be employed to test for the stationarity of the series.
Secondly, the panel Cointegration test will be carried out to explore the existence of a long run relationship among the variables. The study employs Pedroni (1999), Kao's (1999) residual panel cointegration tests, and Fisher's test.
Once Cointegration is established, the next step is to estimate the long run relationship between energy consumption and growth. The study employs the Fully Modified Ordinary Least Squares
(FMOLS) proposed by Pedroni (2001) and the Dynamic Ordinary Least Squares to establish the long run relationship.
64
University of Ghana http://ugspace.ug.edu.gh
4.3.1 Panel Granger causality test.
Having established Cointegration, the dynamic error correction model will be used to investigate the long run and short run Granger causality between energy and economic growth.
4.3.2 Testing for Unit Roots in a Panel Context.
Panel unit root testing is a recent phenomenon (Baltagi, 2005). Most researchers have prior to the emergence of panel unit root testing techniques ignored the issue of non-stationarity in panel data models. This have often led to spurious empirical findings in models that are employed as this cannot be avoided if issues of non-stationarity of panel data are not properly checked. There are various types of panel unit root testing techniques. Each technique has its own unique characteristics. The equation below represents the general panel unit root testing equation;
푌푖푡 = 훿푖푌푖푡−1 + 휆푋푖푡 + 푈푖푡 34
Given that |훿푖| < 0 then 푌푖푡 is stationary. However, if |훿푖| = 1, 푌푖푡 is non stationary and therefore has unit root. Subtracting 푌푖푡−1 from both sides of equation (34) gives the Augmented Dickey
Fuller model of unit root testing; this gives;
∆푌푖푡−1 = 훿푖 − 1 푌푖푡−1 + 휆푋푖푡 + 푈푖푡 35
Assuming that 휙푖 = 훿푖 − 1 , equation (35) becomes;
YYXU it i it1 it it (36)
There are two assumptions that are made for the purposes of testing; the first is that we assume that the persistence parameters are common across the cross sections so that 휙푖 = 휙 and secondly, we assume that 휙푖 varies with cross sections.
65
University of Ghana http://ugspace.ug.edu.gh
In testing for stationarity of the variables employed in this study, three different techniques are used. These are; Levin, Lin and Chu (2002), Im, Pesaran, and Shin (2003). These three tests are employed so as to arrive at robust results in terms of the stationarity of the panel data.
For each of the three panel unit root tests, the unit root test are carried out for all variables and are tested both in levels and in first differences.
4.3.2.1 Levin, Lin and Chu Panel Unit root test.
Levin, Lin and Chu (2002) propose the use of a more robust panel unit root testing mechanism instead of running individual unit root test for each cross section in a panel data. This mechanism is able to eliminate the limitations associated with performing individual unit root test for each cross section. The null and alternate hypothesis for this panel unit root test technique is given as;
퐻0: 푒푎푐ℎ 𝑖푛푑𝑖푣𝑖푑푢푎푙 푡𝑖푚푒 푠푒푟𝑖푒푠푎푠 ℎ 푎 푢푛𝑖푡 푟표표푡 … … … . . 푁푢푙푙 ℎ푦푝표푡ℎ푒푠𝑖푠
퐻푎: 푒푎푐ℎ 푡𝑖푚푒 푠푒푟𝑖푒푠 𝑖푠 푠푡푎푡𝑖표푛푎푟푦 … … … … … … 푎푙푡푒푟푛푎푡푒 ℎ푦푝표푡ℎ푒푠𝑖푠
The hypothesis is represented by the equation below;
pi yit y i,1 t iL y it L mi d mt it L1 (37)
푚 = 1, 2, 3
Levin, Lin and Chu (2002) specifies performing distinct augmented Dickey Fuller (ADF) regressions for each cross section; the specification is given below;
pi yit y i,1 t iL y it L mi d mt it L1 (38)
66
University of Ghana http://ugspace.ug.edu.gh
The lag order is allowed to vary across individual and once it is obtained, Levin Lin & Chu
(2002) suggest running two auxiliary regressions orthogonalized residuals;
Run ∆푦푖푡 on ∆푦푖,푡−퐿 퐿 = 1, … , 푝푖 and 푑푚푡 to get residuals 푒̂푖푡
Run ∆푦푖푡 on ∆푦푖,푡−퐿 퐿 = 1, … , 푝푖 and 푑푚푡 to get residuals 푣푖,̂ t-1
Standardize these residuals to control for different variances across i
푒̂푖푡 = 푒푖푡/𝜎̂휀푖 and 푣푖,̂ t-1 = 푣̂/푖푡 푒̂푖푡 where 𝜎̂휀푖 represents the standard error from each ADF regression, for 𝑖 = 1, … , 푁.
The second step involves estimating the ratio of the long run to short run standard deviations.
Given the null hypothesis of unit root, the long run variance of equation 4 is computed by;
푇 퐾̅ 푇 1 1 𝜎̂2 = ∑ ∆푦2 + 2 ∑ 푤 [ ∑ ∆푦 ∆푦 ] 39 푦푖 푇 − 1 푖푡 퐾퐿 푇 − 1 푖푡 푖,푡−퐿 푡=2 퐿=1 푡=2+퐿
Where 퐾̅ is a truncation lag that can be data dependent and should be obtained in a way that
̂2 ensures the consistency of 𝜎푦푖. The average standard deviation is computed by;
1 푁 푆̂푁 = ∑ 푆̂푖 40 푁 푖=1
The last stage as stipulated by Levin et al., (2002) involves computing the panel test statistics.
This is done by running the pooled regression; 푒̃푖푡 = 𝜌푣̃푖,t-1 + 휀̃푖푡 based on 푁푇̃ observations where 푇̃ represents the average number of observations per individual in the panel with 푝̅ =
푁 ∑푖=1 푝푖 /푁. 푝̅ is the mean number of individual ADF regressions. The conventional t statistic is
𝜌̂ given as 푡 = , 𝜌 𝜎̂ 𝜌̂
67
University of Ghana http://ugspace.ug.edu.gh
where
푁 푇 푁 푇 ̂2 𝜌̂= ∑ ∑푖 푣 ,̂ t − 1 푒̃푖푡⁄ ∑ ∑ 푣푖 , t − 1 41
푖=1 푡=2+푃푖 푖=1 푡=2+푃푖
푁 푇푖 ̂2 𝜎̂ 𝜌̂ = 𝜎̂휀̃ ⁄ [∑ ∑ 푣푖 , t − 1] 42
푖=1 푡=2+푃푖
Levin, Lin and Chu (2002) then computes the adjusted t statistics as given below;
푡 − 푁푇̃푆̂𝜎̂−2𝜎̂ 𝜌̂ 휇∗ ∗ 푝 푁 휀̃ 푚푇̃ 푡푝 = ∗ 43 𝜎푚푇̃
Levin, Lin and Chu (2002) argue that in using their panel unit root test, it is important that one specifies the independent variables in the test equation. Again, according to Levin, Lin & Chu
(2002), for panels of very large sizes, it is advisable to run individual unit root time series test for each cross section. The Levin, Lin & Chu (2002) is limited in that it assumes that all cross sections have or do not have a unit root. Also, the Levin, Lin & Chu (2002) unit root test breaks down in the presence of cross sectional correlation.
4.3.2.2 Im, Pesaran and Shin Panel unit root test
The limitations of the Levin, Lin and Chu (2002) consequently led to the emergence of the Im,
Pesaran and Shin panel unit root test. Im et al., (2003) proposed an alternative testing procedure based on averaging individual unit root test statistics.
The null hypothesis is that each series in the panel has unit roots and the alternative hypothesis allows for some (but not all) of the individual series to have unit roots. This is represented by the equation below;
68
University of Ghana http://ugspace.ug.edu.gh
퐻0: 𝜌푖 < 0 푓표푟 𝑖 = 1,2, … . . ,1 푁
퐻푎: 𝜌푖 = 0 푓표푟 𝑖 =1 푁 + 1, … , 푁
Im et al., (2003) define the 푡 ̅ statistics as the average of the individual ADF statistics and computes it as below; the first stage includes stipulating the 푡 ̅ statistics as;
푁 1 푡̅ = ∑ 푡푝 44 푁 푖 푖=1
Where 푡푝푖 is the individual test statistics for testing 퐻0: 𝜌푖 = 0. In situations where the lag order is always zero, Im et al., (2003) provide simulated critical values for 푡 ̅ for different number of cross-sections N, series length T and Dickey–Fuller regressions containing intercepts only or intercepts and linear trends. However, in cases where the lag order may be non-zero for some cross sections, Im et al., (2003) shows that the t bar statistic has an asymptotic N (0,1) distribution.
The second stage involves estimating the standardized 푡 ̅ statistic which is given by;
1 [푡̅ ∑푛 ̅ √푛 푛푇 − 푛 푖=1 퐸(푡푖푇 푝푖 )] 푍푡̅푛푇 = ~푁 0,1 45 1 √ ∑푛 푣푎푟(푡̅ 푝 ) 푛 푖=1 푖푇 푖
The expected mean and variance expressions given by 퐸(푡푖푇̅ 푝푖 ) and 푣푎푟(푡푖푇̅ 푝푖 ) respectively
′ are provided by Im et al., (2003) for a range of values of the T and 푝푠.
4.3.3 Panel Cointegration.
Having established the non-stationarity of the variables, the panel Cointegration techniques will be employed to establish the long run relationship among the variables. Specifically, this study
69
University of Ghana http://ugspace.ug.edu.gh
employs the Kao residual based test, the Pedroni Cointegration test and the Fisher test to investigate the nature of the long run relationship among the variables.
4.3.3.1 Kao residual based test
Kao (1999) argue that the null hypothesis of no Cointegration can be tested using the DF and
ADF type unit root test for the residual. In the general panel regression equation below;
′ ′ 푦푖푡 = 푥푖푡훽 + 푧푖푡훾 + 푒푖푡 , the DF type test can be computed from the fixed effects residual as;
̂ 푒̂푖푡 = 𝜌푒̂푖푡−1 + 푣푖푡 where 푒̂푖푡 = 푦̃푖푡 − 푥̃훽푖푡 and 푦̃푖푡 = 푦푖푡 − 푦̅푖.
According to Kao (1999), the first step in testing the null hypothesis of no Cointegration (i.e.
퐻0: 𝜌 = 1 involves estimating 𝜌 using OLS and this gives;
푁 푇 ∑푖=1 ∑푡=2 푒̂푒̂푖푡 푖푡−1 𝜌̂= 푁 2푇 46 ∑푖=1 ∑푡=2 푒̂푖푡
And the t statistic is given by the equation below;
푁 푇 2 𝜌̂−1 √∑푖=1 ∑푡=2 푒푖푡̂ −1 푡𝜌 = (47) 푆푒
1 where 푆2 = ∑푁 ∑푇 푒̂ − 𝜌̂푒̂ 2 푒 푁푇 푖=1 푡=2 푖푡 푖푡−1
Kao stipulated four DF type test given as;
√푁푇 𝜌̂−1 +3√푁 퐷퐹 = (48) 𝜌 √10.2
퐷퐹푡 = √1.25푡𝜌 + √1.875푁 (49)
70
University of Ghana http://ugspace.ug.edu.gh
2 3√푁𝜎̂푣 √푁푇 𝜌̂−1 + 2 ∗ 𝜎̂0푣 퐷퐹𝜌 = (50) 4 36𝜎̂푣 √3+ 4 5𝜎̂0푣
√6푁𝜎̂푣 푡𝜌+ ∗ 2𝜎0푣 And 퐷퐹푡 = (51) 2 𝜎̂0푣 3𝜎̂푣 √ 2+ 2 2𝜎̂푣 10𝜎̂0푣
2 ̂ ̂ ̂−1 2 −1 where 𝜎̂푣 = ∑푦푦 − ∑푦푥∑푥푥 and 𝜎̂0푣 = Ω̂푦푦 − Ω̂푦푥Ω̂푥푥
With regards to the ADF test, Kao (1999) runs a regression of the equation below to arrive at the t statistics.
푝 푒̂푖푡 = 𝜌푒̂푖푡−1 + ∑푗=1 휗푗 ∆푒̂푖푡−푗 + 푣푖푡푝 (52)
The ADF t statistics is computed as;
6푁𝜎̂ 푡 + √ 푣 퐴퐷퐹 2𝜎̂ 퐴퐷퐹 = 0푣 (53) 2 2 𝜎̂0푣 3𝜎̂푣 √ 2+ 2 2𝜎̂푣 10𝜎̂0푣
Where 푡퐴퐷퐹 is the t statistic of 𝜌.
4.3.3.2 Pedroni (1999) Cointegration test
The uniqueness of Pedroni’s Cointegration test is inherent in its ability to allow for extensive heterogeneity in the panel data. The null hypothesis is that there is no Cointegration and it is tested against the alternate hypothesis of Cointegration among the panel variables. The first step in Pedroni’s test is to estimate an auxiliary regression of the form as given below;
71
University of Ghana http://ugspace.ug.edu.gh
𝜌푖 ∆푒푖푡 = 𝜌푖푒푖푡−1 + ∑푗=1 휑푖푗∆푒푖푡−푗 + 푣푖푡 (54)
Having estimated the regression equation (54), Pedroni (1999) then estimates the within group test statistics as follows;
Panel v statistic:
3 2 2 3 푇 푁2 푇 푁 푍푉푁푇 = 푇 푁 −2 2 (55) ∑푡=1 ∑푛=1 퐿1,1푒̂푖푡
Panel 𝜌 statistics:
푇 푁 −2 2 ̂ 푇√푁[∑푡=1 ∑푛=1 퐿1,1(푒̂푖푡)− 휆푖] 푇√푁푍𝜌푁푇 = 푇 푁 −2 2 (56) ∑푡=1 ∑푛=1 퐿1,1푒̂푖푡
Panel t statistic:
2 푇 푁 −2 2 푇 푁 −2 2 2 ̂ 푍푡푁푇 = √𝜎푁푇 ∑푡=1 ∑푛=1 퐿1,1푒̂푖푡−1 [∑푡=1 ∑푛=1 퐿1,1(푒̂푖푡)∆푒̂푖푡 − 휆푖] (57)
Pedroni (1999) argues that these standardized statistics are asymptotically normally distributed.
4.4 Estimating the long run relationship
Once Cointegration is established, the next step is to estimate the long run relationship between energy consumption and growth. The study employs the Fully Modified Ordinary Least Squares
(FMOLS) and the Dynamic Ordinary least squares (DOLS) proposed by Pedroni (2001) to
72
University of Ghana http://ugspace.ug.edu.gh
establish the long run relationship. In a panel framework, provided the regressors are not strictly exogenous, using the OLS to estimates the long run relationship may lead to biased estimates of the parameters. According to Ouedraogo (2013), the FMOLS is preferred over the DOLS as the former requires less assumptions.
Considering equation (61), the FMOLS and DOLS can be computed subsequently as follows;
ki Yit i i ENER it ik EC it k it kk i (58)
Where 𝑖 = 1,2 … … 푇, 푌푖푡 represents the log of GDP per capita over time and across the countries,
ENER is energy consumption, EC is the error correction term and 휇푖푡 is the serially uncorrelated error term.
From equation (61), two assumptions are made to arrive at the equation for the FMOLS and
DOLS. First, let 휉푖푡 = 휇̂푖푡 and ∆퐸퐶푖푡 be a stationary vector which is made up of the estimated residuals from the cointegration regression and differences in energy consumption.
−1 푇 푇 Again let, Ω푖푡 = 푙𝑖푚푇 → ∞퐸[푇 ∑푡=1 휉𝑖푇 ∑푡=1 휉𝑖푇 ] represent the long run covariance.
Subsequently, the FMOLS estimators is given as;
∗ −1 푁 푇 ̅̅̅̅̅̅̅̅ 2 −1 푇 ̅̅̅̅̅̅̅̅ ∗ 훽퐹푀푂퐿푆 = 푁 ∑1 ∑푡=1 퐸푁퐸푅푖푡 − 퐸푁퐸푅푖 ∑푡=1 퐸푁퐸푅푖푡 − 퐸푁퐸푅푖 푦푖푡 − 푇 ⌢ 훾푖 (59)
And the DOLS estimators is computed as;
∗ −1 푁 푇 푖 −1 푇 ∗ 훽퐷푂퐿푆 = 푁 ∑푖=1(∑푡=1 푍푖푡푍푖푡) ∑푡=1 푍푖푡푌푖푡 (60)
Where 푍푖푡 is a 2 퐾 + 1 1 vector of regressors.
73
University of Ghana http://ugspace.ug.edu.gh
4.5 Testing for causality
Having established the presence of Cointegration among the variables, the next objective of the study is to ascertain the direction of the causal relationship among the variables. Cointegration helps to establish whether there is a long run relationship among the variables, however, it does not tell the direction of the causality among the variables. It therefore becomes imperative to carry out the granger causality test so as to establish the causal relationship among the variables.
The granger causality test is employed as it provides more robust results irrespective of the sample size. Also, the granger causality based error correction model helps to estimate and distinguish between both the short run and the long run granger causality. The expression for the granger causality test which is based on an error correction model in this study is given as;
qq LGDPCit 1i 11ik LGDPC it k 12 ik LENER it k kk11
EC it1 it (61)
qq LENERit 1i 11 ik LENER it k 12 ik LGDPC it k kk11 EC it1 it (62)
qq LGDPCit 1i 11 ik LGDPC it k 12 ik ELEC it k kk11 q 13ik PE it k EC it 1 it k1 (63)
qq ELECit 1i 11 ik ELEC it k 12 ik LGDPC it k kk11 q 13ik PE it k EC it 1 it k1 (64)
74
University of Ghana http://ugspace.ug.edu.gh
qq PEit 1i 11 ik PE it k 12 ik LGDPC it k kk11 q 13ikELEC it k EC it 1 it k 1 (65)
Where LGDPC is gross domestic product per capita, LENER represents total energy consumption, ELEC is electricity consumption, PE represents petroleum consumption, Δ denotes the difference operator; ECT is the lagged error correction term derived from the long-run
Cointegration relationship; 훼푖, 훽푖, and λ are adjustment coefficients; k is the number of lags determined by the Alkaike Information Criterion (AIC) and Schwarz information criterion (SIC) and 휇 is the serially uncorrelated error term.
4.6 Description of variables
Economic growth: Economic growth measures the rise in the total production of goods and services in an economy over a period of time. GDP is also defined as the total value of final goods and services produced in an economy within a period of time usually a year. GDP is used as a measure of economic activity in an economy. GDP per capita is useful in analysis involving several countries due to its ability to measure the standard of living present in each country. GDP per capita is given by total GDP divided by the total population in the country. Following the work of Ouedraogo (2013) GDP per capita is used as a proxy for economic growth. The study uses GDP measured in constant (2005 dollars). It is represented hereafter by LGDPC.
Total Energy consumption: Total energy consumption measures the consumption of primary energy. Primary energy consumption includes energy consumed from its direct source. That is, energy that has not been subjected to any conversion process. In this study, following the work
75
University of Ghana http://ugspace.ug.edu.gh
of Odhiambo (2008), total energy consumption is proxied by energy consumption per capita. It is denoted hereafter by LENER.
Electricity consumption: Electricity consumption measures the total amount of electricity consumed in a country for various purposes. It includes electricity used by industries, households, institutions and all other segments of the country. Electricity consumption is measured in billion kilowatt per hour. Following the work of Sama & Tah (2016), electricity consumption is represented by electricity consumption in billion kilowatt per hour. It is denoted hereafter by ELEC.
Petroleum consumption: petroleum consumption includes the consumption of petroleum energy in the domestic country. It excludes the export of petroleum. In this study, following the work of Sama & Tah (2016), petroleum consumption is measured by petroleum consumption in million tons of oil equivalent. It is represented hereafter by PE.
4.7 Source of Data
Data on key macroeconomic electricity consumption and petroleum energy consumption was sought from the United States Energy Information Administration. Data on GDP per capita and total energy consumption were also sought from the World Development Indicators’ (2016).
4.8 Conclusion
In this chapter, the methodology employed in finding answers to the various objectives posed in the study has been clearly outlined. Basically, the study chose to employ the Im et al., (2003) and the Levin, Lin & Chu (2002) unit root test in testing for the stationarity of the panel data
76
University of Ghana http://ugspace.ug.edu.gh
employed. This choice was basically based on the robust nature of the results provided by these techniques. Again, the study finds that once the variables are non-stationary, then it is possible that some variables may have a long run relationship. For this reason, it became imperative to employ the Pedroni (1999) and the Kao residual Cointegration test in order to investigate the possibility of a long run relationship among the variables. Also, once Cointegration is established, then the next objective which is to establish the causal relationship among the variables has to be investigated. The study employs the granger causality based error correction model in this regard.
77
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FIVE
DISCUSSION OF RESULTS
5.0 Introduction
Having presented an extensive analysis of the methodology in chapter four, the results obtained are presented and discussed in this chapter. The results of our findings include that of the unit root test, the panel cointegration test along with the short run and long run panel granger causality test.
5.1 Descriptive statistics of variables
This section of the study presents an analysis of the descriptive statistics of the variables employed in the study. The descriptive statistics is essential so as to know the characteristics and other qualities of the variables. The mean, median and other essential characteristics of the variables employed are reported in this section. The descriptive statistics are presented for Gross
Domestic Per Capita, Electricity consumption, petroleum consumption and total energy consumption. Table 5.1 presents the descriptive statistics of the variables.
78
University of Ghana http://ugspace.ug.edu.gh
Table 5.1: Descriptive statistics of variables employed
Variables Mean Median Maximum Minimum Standard Obs. Skew Kurtos deviation ness is Gross 6.38 6.33 7.83 5.52 0.43 237 0.42 3.24 Domestic Product Per Capita Electricity 3.11 1.09 24.78 0.03 4.42 237 2.53 10.27 consumption Total 5.79 5.86 6.68 4.09 0.54 237 -0.99 4.53 Energy Consumptio n Petroleum 2.72 1.11 15.91 0.03 4.37 237 2.01 5.38 consumption Source: Computed by Author using Eviews 9 and data from EIA.
In the descriptive statistics reported in Table 5.1, the number of observations for all the variables is 237. Gross Domestic Product Per Capita has a mean value of 6.38 along with a median of
6.33. The maximum and minimum values of GDP are 7.83 and 5.52 respectively. It also has a standard deviation of 0.43. Again, electricity consumption recorded a mean value of 3.11, maximum value and minimum value of 24.78 and 0.03 respectively. It also recorded a standard deviation of 4.42. Total energy consumption recorded a mean, median and a standard deviation of 5.79, 5.86 and 0.54 respectively. Also, petroleum consumption recorded a mean value of 2.72.
It has a maximum value of 15.91 and a standard deviation of 4.37. It is evident from the descriptive statistics that electricity consumption is the most dispersed with a standard deviation of 4.42 whilst GDP Per Capita is the least dispersed with a standard deviation of 0.43.
Essentially, the descriptive statistics recorded in Table 5.1 implies that the average Gross
Domestic Product Per Capita recorded over the years 1980 to 2013 for all the West African countries is 6.38 constant (2005 dollars). Again, the average electricity consumed in the seventeen countries over the period specified is 3.11 billion kilowatt per hour over. Similarly, the
79
University of Ghana http://ugspace.ug.edu.gh
average total energy and petroleum consumed over the years 1980 to 2013 is 5.79 and 2.72 million tons of oil equivalent respectively. Furthermore, the highest quantum of electricity consumed is 24.78 billion kilowatt per hour with 0.03 billion kilowatt being the least. The descriptive statistics therefore summarizes and describes the data set. Beyond that, the descriptive statistics cannot be used in reaching conclusions regarding any hypotheses that has been made or providing answers to the research questions posed in this study.
5.2 Panel unit root test.
As established in the previous chapter, the study tests for the stationarity or otherwise of the variables employed. Two panel unit root techniques were used in testing for unit roots; these are the Im, Pesaran and Shin & Levin, Lin & Chu unit root test. The results of the Augmented
Dickey Fuller Fisher Chi square and the PP Fisher Chi square are also reported to solidify the findings of the Im et al., (2003) test and the Levin, Lin &Chu (2002) techniques. The variables were tested for unit root at levels and then at first difference. The results of the panel unit root test are reported in Table 5.2.
Table 5.2: Levin, Lin & Chu Panel Unit root test
At levels At first difference Order of Variables T - statistics Probability t statistics Prob. integration value value GDP Per Capita 0.77054 0.7795 -14.9561*** 0.0000 I(1)
Total Energy 1.73905 0.9590 -74.2759*** 0.0000 I(1) Consumption Electricity 7.94968 1.0000 -21.9985*** 0.0000 I(1) Consumption Petroleum 6.67136 1.0000 -16.1627*** 0.0000 I(1) Consumption Source: Computed by Author using Eviews 9 and data from EIA. Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively.
80
University of Ghana http://ugspace.ug.edu.gh
From Table 5.2, in testing for unit roots at levels, the results reveal that the null hypothesis of unit root cannot be rejected at both 1% and 5% significance level because the probability values are greater than 0.05. Therefore, at levels, the Levin, Lin & Chu (2002) test conclude that the variables have unit roots. However, when the first difference of the variables are taken, the probability values become very small and highly significant. As such, we reject the null hypothesis of unit roots and conclude that the variables become stationary at first difference and are therefore integrated of order (1).
Table 5.3: Im, Pesaran and Shin Unit root test. At levels At first difference Order of Variables W - statistics Probability t statistics Prob. integration value value GDP Per 2.24502 0.9876 -14.5679*** 0.0000 I(1) Capita Total Energy 0.87863 0.8102 -4.87547*** 0.0000 I(1) Consumption Electricity Consumption 9.86622 1.0000 -21.4229*** 0.0000 I(1) Petroleum Consumption 6.17016 1.0000 -17.7604*** 0.0000 I(1) Source: Computed by Author using Eviews 9.
Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
From Table 5.3, the Im, Pesaran and Shin test conclude that the null hypothesis of unit root cannot be rejected at 1% and 5% significance level for all the variables. However, the probability values become significant in testing for unit root at first difference. Therefore, the variables are non-stationary at levels but become stationary at first difference.
81
University of Ghana http://ugspace.ug.edu.gh
Table 5.4: ADF – Fisher Chi Square At levels At first difference Order of Variables T - statistics Probability t statistics Prob. integration value value GDP Per 34.9950 0.4206 252.263*** 0.0000 I(1) Capita Total Energy 6.01106 0.9997 210.771*** 0.0000 I(1) Consumption Electricity 7.95675 1.0000 377.190*** 0.0000 I(1) Consumption Petroleum 25.7306 0.8450 305.152*** 0.0000 I(1) Consumption Source: Computed by Author using Eviews 9.
Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
From Table 5.4, the ADF Fisher test confirms the results of the other tests employed. The null hypothesis of unit root cannot be rejected at levels for all the variables given that the probability values are not significant. The probability values are significant at first difference and therefore the test concludes that the variables are integrated of order one. Results from the Phillips Perron unit root as displayed in Appendix X also indicate that all the variables are integrated of order one.
In summary, results from the various panel unit root test employed in the study reveal the null hypothesis of unit root cannot be rejected given that the probability values are all not significant at 5% level. That is to say, the variables (GDP Per Capita, Total Energy, Electricity consumption and Petroleum consumption) are non-stationary at levels. This implies that the variables have unit roots. The study therefore tests for unit root by differencing each of the variables. The results of panel unit root testing at first difference reveal that all the variables become stationary at first difference.
82
University of Ghana http://ugspace.ug.edu.gh
5.3 Panel Cointegration results.
Having concluded that all the variables are integrated of order one, a long run movement among the variables may be possible. As argued by Kyle & Miguel (2015), in cases where a linear combination of two non-stationary series results in a stationary process, it is possible that the series are cointegrated and therefore they may move together in the long run. For this reason, the study tests for cointegration among the variables. The results of cointegration are presented for
Total energy consumption and GDP as well as for electricity consumption, petroleum consumption and GDP Per Capita.
Table 5.5: Results of Pedroni Panel test Cointegrating (GDP Per Capita and Total Energy Consumption)
Alternative hypothesis: common AR coefs. (within-dimension) Weighted Statistic Prob. Statistic Prob. Panel v-Statistic 0.429698 0.3337 -0.370485 0.6445 Panel rho-Statistic -0.059573 0.4762 0.162618 0.5646 Panel PP-Statistic -0.155580 0.4382 0.095177 0.5379 Panel ADF-Statistic 0.270005 0.6064 0.381884 0.6487
Alternative hypothesis: individual AR coefs. (between- dimension)
Statistic Prob. Group rho-Statistic 1.035942 0.8499 Group PP-Statistic -0.556300 0.2890 Group ADF-Statistic -0.358336 0.3600
Source: Computed by Author using Eviews 9.
Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
Table 5.5 reports results of the Pedroni Cointegration test. From the results, the null hypothesis of no cointegration cannot be rejected at 1% and 5% level of significance. The probability values
83
University of Ghana http://ugspace.ug.edu.gh
generated by the within and between dimensions of the Pedroni test lead to the conclusion of no long run relationship among the variables. The Kao residual cointegration test and the Fisher test are performed to solidify the findings of the Pedroni test.
Table 5.6: Kao Cointegration test results
t-Statistic Prob. ADF 3.677032 0.0001
Source: Computed by Author using Eviews 9.
Table 5.6 reports results of the Kao cointegration test. The probability value of 0.0001 implies that the variables (GDP Per capita and total energy consumption) are cointegrated at 1% significance level. The Fisher cointegration test is performed to confirm the results of a possible cointegration among the variables.
In employing the Fisher cointegration test, it is important that an optimal lag length is chosen using the various lag length criteria. The Akaike Information Criterion (AIC) and the Schwarz
Information Criterion (SIC) are performed so as to determine the optimal lag length. The AIC and SIC indicate that the optimal lag length is one. A lag length of one is chosen in testing for cointegration using the Fisher test. The results are reported in table 5.7.
Table 5.7: Fisher Cointegration test (GDP and total energy consumption) Hypothesized Fisher stats Prob. value Fisher test Prob. value No of CE. (Trace test) (Max-eigen value) None 37.15*** 0.0000 38.86*** 0.0004 At most 1 12.58*** 0.5595 12.58*** 0.5595 Source: Computed by Author using Eviews 9.
Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
84
University of Ghana http://ugspace.ug.edu.gh
The results from Table 5.7 indicate that the null hypothesis of no cointegration is rejected at a
1% level of significance given that the probability value is significant. The conclusion therefore is that the variables have a long run relationship with at most one cointegration equation. This finding is corroborated by the work of Akinlo (2008), Ouedraogo (2013, Nondo et al., (2010).
5.3.1 Results of Pedroni Panel Cointegration test (GDP, Electricity consumption and petroleum consumption).
The Pedroni Cointegration test is performed to investigate the existence of a long run relationship between GDP, electricity consumption and petroleum consumption. The results are reported in Table 5.8.
Table 5.8: Pedroni cointegration results (GDP, electricity and petroleum consumption)
Alternative hypothesis: common AR coefs. (within-dimension) Weighted
Statistic Prob. Statistic Prob. Panel v-Statistic 2.826710 0.0024*** 0.618473 0.2681 Panel rho-Statistic -0.719868 0.2358 -0.510445 0.3049 Panel PP-Statistic -1.445245 0.0742* -1.571804 0.0580* Panel ADF-Statistic -2.156041 0.0155** -1.617479 0.0529*
Alternative hypothesis: individual AR coefs. (between- dimension)
Statistic Prob. Group rho-Statistic 0.821504 0.7943 Group PP-Statistic -1.361984 0.0866* Group ADF-Statistic -1.502827 0.0664*
Source: Computed by Author using Eviews 9
Results of the Pedroni test as reported in Table 5.8 show the presence of a long run relationship between GDP, electricity consumption and petroleum consumption. The results indicate that
85
University of Ghana http://ugspace.ug.edu.gh
seven out of the eleven test statistics have probability values less than 0.1. The seven significant probability values are 0.0024, 0.0742, 0.0155, 0.0580, 0.0529, 0.0866 and 0.0664. The null hypothesis of no cointegration is therefore rejected at 10% level of significance. This means that the variables (GDP, electricity and petroleum) have a long run relationship. The Kao test is performed and the results are reported in Table 5.9
Table 5.9: Results of Kao residual cointegration test
t-Statistic Prob. ADF -2.900750 0.0019
Source: Computed by Author using Eviews 9.
From the results of the Kao cointegration test displayed in Table 5.9, the probability value of
0.0019 implies that the null hypothesis of no cointegration cannot be rejected at 1% significance level. There is therefore a long run relationship between GDP, electricity consumption and petroleum consumption. The Fisher cointegration test is then performed to solidify the other approaches to cointegration. The results are reported in Table 5.10.
Table 5.10: Fisher Cointegration test (GDP, electricity and petroleum consumption)
Hypothesized Fisher stats Prob. value Fisher test Prob. value No of CE. (Trace test) (Max-eigen value) None 98.11*** 0.0000 82.64*** 0.0004 At most 1 43.16 0.1348 39.42 0.2403 At most 2 40.07 0.2190 40.07 0.2190 Source: Computed by Author using Eviews 9. Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
86
University of Ghana http://ugspace.ug.edu.gh
The results of the Fisher cointegration test also confirm the presence of a long run relationship among the variables. Both the Trace test and the Max-eigen value test reject the null hypothesis of no cointegration at 1% significance level. The variables (GDP, Electricity consumption, petroleum consumption) move together in the long run.
5.4 Estimating the long run Relationship
Having established the existence of a movement among the variables in the long run, the
FMOLS and the DOLS are employed to estimate the nature of this long run relationship. As discussed in Chapter four, the FMOLS produces more robust results as compared to the DOLS and therefore the FMOLS forms the focus of discussion in this study. The results of the FMOLS and DOLS are reported in the Table 5.11 and 5.12 respectively.
Table 5.11: Fully Modified Ordinary least squares
Dependent variable Independent variables GDP Per Capita Total Energy Electricity Petroleum Consumption -0.139887*** 0.107273*** 0.058452*** (0.0000) (0.0010) (0.0023) Source: Computed by Author using Eviews 9 Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively. P values are given in brackets. The results of the FMOLS as reported in Table 5.11 show that in the long run, total energy consumption has a negative and significant effect on GDP. Similarly, electricity consumption has a positive and statistically significant impact on GDP in the long run. Also, petroleum consumption has a positive and statistically significant relationship with GDP.
87
University of Ghana http://ugspace.ug.edu.gh
Table 5.12: Dynamic Ordinary least squares
Dependent variable Independent variables GDP Per Capita Total Energy Electricity Petroleum consumption consumption -0.159404*** 0.049647 0.026938 (0.1121) (0.0000) (0.2168) Source: Computed by Author using Eviews 9 Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively. P values are given in brackets.
The results from the FMOLS indicate that in the long run, Total energy consumption has a negative and significant impact on growth. The probability value of 0.0000 is significant and therefore a 1% increase in Total energy consumption in the long run reduces growth by 0.14%. It is very difficult to get works that are able to segregate Total energy consumption into its various components. However, this result contradicts the findings of Ouedraogo (2013); Ozturk et al.,
(2010). Ouedraogo (2013) finds that in the long run, total energy consumption has a positive relationship with GDP Per Capita. Ozturk et al., (2010) find that in countries including Ghana,
Congo, Kenya, Pakistan, India, Togo, Zambia and Zimbabwe, there is a positive relationship between total energy consumption and economic growth in the long run.
With regards to the long run negative impact of Total energy consumption on GDP, a plausible explanation for this finding could be that in West Africa, a look at the components of Total energy consumed reveals that traditional biomass constitutes a greater proportion of energy consumed. Specifically, about 80% of energy consumed is traditional biomass (Adenikinju,
2008). This trend is unlikely to change substantially in the long run notwithstanding the attempt to introduce the consumption of modern fuels in the form of LPG gas, kerosene among others and to increase access to electricity (Karekezi et al., 2008). In the rural areas, the consumption of
88
University of Ghana http://ugspace.ug.edu.gh
traditional biomass is quite predominant and for this reason traditional biomass is expected to constitute a greater proportion of total energy consumed in the long run. The consumption of biomass has several negative effects ranging from health to environmental problems
(Ravindranath & Rao, 2005). The incomplete and inefficient burning of wood fuels indoors by households is the cause of diseases such as pneumonia which is estimated to be the major cause of premature deaths (Bruce et al., 2000; Ruta, 2010). The WHO (2006) argues that enclosed air pollution is the cause of approximately 1.5 million implusive deaths and a greater proportion of these deaths (more than 85%) are due to the use of biomass. Arbex et al., (2004) have argued that the burning of biomass indoor for cooking and the smoke generated through this activity has led to diseases such pulmonary infections and this primarily affects developing countries. The
United Nations Environment programme stipulates that the continued use of biomass in Africa exposes an estimated 90 percent of the populace to the undesirable health effects of biomass use.
For this reason, the continued consumption of biomass is likely to endanger the health of the population in the long run. As stipulated by Cole & Neumayer (2006), poor health in a country has negative consequences on productivity. The use of biomass is likely to transcend to a reduction in productivity in the sub-region.
More so, the use of biomass has undesirable effects on the environment. Ravindranath & Rao,
(2005) argue that the use of biomass leads to deforestation. Deforestation refers to the felling of trees or the clearing of forest for alternative purposes other than forest use. The forest provides huge environmental benefits such as preservation of biodiversity, prevention of soil erosion, prevention of climate change among others. Smith (2001) stipulates that the tropical forest is rapidly diminishing at a rate of 5 percent per annum as they are cleared to serve as agricultural land, provide biofuels among others. A ramification of deforestation is its undesirable impact on
89
University of Ghana http://ugspace.ug.edu.gh
the global atmosphere (Chakravarty et al., 2012). The destruction of the forest leads to an increase in carbon dioxide concentration in the atmosphere as trees that serve as the primary terrestrial sink of carbon are destroyed. Deforestation therefore contributes greatly to climate change which occurs as result of the increase concentration of Greenhouse Gases in the atmosphere (Chakravarty et al., 2012). More so, Houghton (2005) argues that an estimated two billion tonnes of carbon are released into the atmosphere due to the activities of deforestation.
Therefore, the continued felling of trees for use is likely to worsen the impact of climate change in the sub-region.
Again as argued by Fay et al., (2010), West African countries are unable to enforce measures put in place to reduce pollution and the rate of emissions. Consequently, large power plants and factories that are fueled by petroleum products as well as small scale diesel generators are likely to emit undesirable gases into the atmosphere thereby affecting the local air quality. This may possibly lead to undesirable effects on human health as well as damage to the natural environment thereby affecting growth negatively. For these reasons, the negative effects of traditional biomass use coupled with the undesirable effect from pollution and emissions is likely to reduce GDP in the long run.
Again, electricity consumption has a positive and significant impact on GDP Per Capita in the long run. Specifically, in the long run, an increase in electricity consumption by one unit increases GDP Per Capita by approximately 11% as estimated by the FMOLS. This finding is corroborated by Bildirici (2013); Enu & Dodzi (2014); Lu (2016). Bildirici (2013) finds a long run significant relationship between electricity and growth for Gabon, Guatemala and Senegal.
Enu & Dodzi (2014) conclude that in the long run, an increase in electricity consumption increases Real GDP per capita. Similarly, Lu (2016) employs the DOLS and finds for a panel of
90
University of Ghana http://ugspace.ug.edu.gh
seventeen Taiwanese industries a long run positive impact of electricity consumption on economic growth.
A plausible explanation for the long run positive impact of electricity consumption on growth could be that in Figure 2.13 in chapter two of this study, there is ample evidence that suggest that growth is driven by the services sector in most West African countries and to a large extent the industrial sector. According to Hollinger (2015), the services sector dominates the economy, contributing 42% to GDP on average for the past decade. The agricultural sector accounts for
(35%) and then the industrial sector (23%). However, the agricultural sector employs a greater proportion of the total labour force in West Africa. It is estimated that 65 percent of the labour force are employed by the sector. In countries like Liberia and Sierra Leone, agriculture contributed more than 72% and 65% to GDP respectively in 2011 (Kanu et al., 2014). Electricity consumption in agriculture is significantly low in the sub-region due to the fact that agriculture is mainly subsistence (Breman, 2003). Conversely, electricity is used predominantly in the services and the industrial sector as electricity is needed to drive processes in the industrial and services sector (Bergasse et al., 2013; Cali, 2008). For this reason, the services and industrial activities which drive growth will increase electricity consumption.
Furthermore, petroleum consumption has a positive and significant impact on GDP per capita in the long run as shown in Table 5.11. An increase in petroleum consumption by one unit increases
GDP Per Capita by 6%. This finding is similar to that of Behmiri & Manso (2013); Olusanya
(2012); Gbadebo & Okonkwo (2009). Behmiri & Manso (2013) seek to investigate the relationship between crude oil consumption and economic growth in a panel of twenty three sub-
Saharan African countries. They find a long run significant and positive impact of crude oil consumption on economic growth for the whole panel of countries. They estimate that a 1%
91
University of Ghana http://ugspace.ug.edu.gh
increase in oil consumption increases economic growth by 0.21%. Olusanya (2012) and Gbadebo
& Okonkwo (2009) both find a positive relationship between petroleum consumption and economic growth in Nigeria.
A plausible explanation for this finding could be that petroleum is used mainly in the transport sector and in the generation of electricity through thermal plants.
In West Africa, road transport is the most dominant mode of transportation as it facilitates movement of persons and goods in Africa and accounts for an estimated 80% of goods and 90% of passenger traffic (UNECA, 2009). Transport serves as an input used to compliment the production process. It is required in the production process as a factor input to make it complete and efficient. It is an important requirement for economic activities (Alam et al., 2013). For this reason, an increase in GDP will consequently result in an increase in transport services (Muktar,
2011). More so, an increase in transport services subsequently results in an increase in fuel consumption (Alam et al., 2013; Ribeiro et al., 2007).
Furthermore, in West Africa, public transport remains poorly organized and under-developed.
Most of the countries lack a proper and formal public transport system (Trans-Africa consortium,
2008). A ramification of the lack of a mass transport system coupled with the lack of comfortability, accessibility with regards to the existing public transport system is the proliferation of private cars which could result in an increase in fuel consumption. This is because private cars are less fuel efficient compared to mass transportation using commercial vehicles (Akoena &Twerefour, 2000). The increase in the use of private cars which are fuel inefficient at the expense of mass transfer is bound to increase petroleum consumption in an environment where the services sector is growing.
92
University of Ghana http://ugspace.ug.edu.gh
Again, Porter (2012) argues that many roads in West Africa are in poor condition notwithstanding the colossal investments in road network that have been made over the years.
This poor nature of roads has led to the excessive use of Sport Utility Vehicles (SUV) which are known to consume more fuel relative to saloon cars. The increase in use of SUV’s therefore increases petroleum consumption which all other things being equal will increase GDP.
Also, Coffin et al., (2016) find that in low income countries, there is a taste or preference for used vehicles. They allude this assertion to three main reasons; the first is that only a restricted number of new vehicles is imported and offered for sale in most developing countries and therefore it is less difficult to acquire a used vehicle relative to a new car. Secondly, the cost of used vehicles are lower than that new vehicles and therefore due to lower income levels in most developing countries, people are able to afford used vehicles than new ones. Thirdly, the cost of repairing used vehicles in low income countries is relatively less due to the lower cost of labour services. Therefore, a consumer in a low income country will be willing to pay more for a used vehicle in spite of the vehicle’s tendency to require repairs frequently compared to a consumer in a higher income country. Therefore, there is a proliferation of used vehicles in most developing countries and West Africa for that matter. Busse et al., (2010) argue that with new vehicles, an increase in fuel prices leads to a higher market share for fuel efficient vehicles. This is because new vehicles are fuel efficient and for that matter with an increase in fuel prices, consumers would save more by using fuel efficient vehicles. However, with used vehicles, an increase in fuel prices results in a reduction in the purchase of low mileage per gallon used vehicles as they are fuel inefficient and therefore with an increase in fuel prices, it is cost inefficient to purchase a used vehicle. Again, fuel efficiency worsens with the age of vehicles (Sustainable Energy
Consumption in Africa, 2004). As used vehicles are relatively older than new vehicles, it is
93
University of Ghana http://ugspace.ug.edu.gh
expected that used vehicles are more fuel inefficient. Therefore, the prevalent use of used vehicles tend to increase fuel consumption which all other things being equal increases GDP.
Again, electricity consumption is expected to increase in the long run if growth is to be propelled by the services and the industrial sector as has been argued earlier. In recent times, there is gradual shift from the generation of electricity using hydro power to the generation of electricity using thermal plants (Acheampong, 2016). More so, Harto & Yan (2011) stipulate that issues of climate change that has consequently led to unexpected drought at certain times has affected hydro generation of electricity than thermoelectric generation. Again, the high initial cost of renewable energy investments such as solar, wind has significant resulted in relatively lower investments in renewables (Vandaele & Porter, 2015). Therefore, increase in growth will lead to an increase in electricity consumption which will consequently increase petroleum consumption.
5.5 Short run analysis and policy implications
Having established the nature of the long run link among the variables, the short run analysis and granger causality as well as the long run causality are presented in this section. The optimal lag length used in the analysis as chosen by the AIC and SIC is one. The results are presented in
Table 5.13 and 5.14.
94
University of Ghana http://ugspace.ug.edu.gh
Table 5.13: Results of short run analysis
Dependent Independent variables variable
퐷퐿퐺퐷푃퐶푡 퐷퐿퐺퐷푃퐶푡−1 퐷퐿퐸푁퐸푅푡−1 퐷푃퐸푡−1 퐸퐶푇푡−1, 퐸퐶푇푡−1, 퐷퐸퐿퐸퐶푡−1 LGDPC, ELEC, PE ---- 0.044435 -0.004800 0.006214 -0.000254 2.35 (0.2650) (0.3334) (0.4816) (0.8034) 0.1077
퐷퐿퐸푁퐸푅푡 퐷퐿퐸푁퐸푡−1 퐷퐿퐺퐷푃퐶푡−1 퐸퐶푇푡−1,
---- 0.010498 0.002762 (0.8171) (0.0406)
퐷퐸퐿퐸퐶푡 퐷퐸퐿퐸퐶푡−1 퐷퐿퐺퐷푃퐶푡−1 퐷푃퐸푡−1 퐸퐶푇푡−1,
----- 0.080375 0.001437 0.001021 (0.0564)* (0.9688) 0.0000
퐷푃퐸푡 퐷푃퐸푡−1 퐷퐿퐺퐷푃퐶푡−1 퐷퐸퐿퐸퐶푡−1 퐸퐶푇푡−1,
----- 0.031051 0.056492 0.000393 (0.2965) (0.0687)* (0.0001) Source: Computed by Author using Eviews 9. Note: ***, ** and * indicates the statistical significance of the estimated parameters at 1%, 5% and 10% respectively; p-values are given in brackets and ECT shows the estimated error-correction term.
From the short run analysis displayed in Table 5.13, Total energy consumption has an insignificant impact on growth in the short. More so, both electricity consumption and petroleum consumption have insignificant relationship with growth in the short run. Again, it is evident from Table 5.13 that electricity consumption has a positive and statistically significant relationship with lagged GDP Per Capita in the short run. A one percent increase in GDP Per
Capita increases electricity consumption by 0.08 billion kilowatt per hour. Also, from Table
95
University of Ghana http://ugspace.ug.edu.gh
5.13, electricity consumption has a statistically significant and positive relationship with petroleum consumption in the short run. Further, a one unit increase in electricity consumption increases petroleum consumption by approximately 0.06 units.
The long run analysis is centered on the statistical significance of the error correction term. From
Table 5.13, none of the error terms is negative and statistically significant. Therefore, there is no long run causal relationship running from Total energy consumption to growth or from growth to total energy consumption as well as from electricity and petroleum consumption to GDP.
Table 5.14: Results of Granger causality test.
Dependent variable Causality direction D(LGDPPC) D(LENER) 1.248978 (0.2650) 퐸푁퐸푅 ↮ 퐺퐷푃 D(ELEC) 0.937307 (0.3334) 퐸퐿퐸퐶 ↮ 퐺퐷푃 D(PE) 0.495921 (0.4816) 푃퐸 ↮ 퐺퐷푃
Dependent variable D(LENER) D(LGDPPC) 0.053622 (0.8171) 퐺퐷푃 ↮ 퐸푁퐸푅
Dependent variable D(ELEC) D(LGDPPC) 3.655941 (0.0564) 퐺퐷푃 ⟶ 퐸퐿퐸퐶 D(PE) 0.001533 (0.9688) 푃퐸 ↮ 퐸퐿퐸퐶
Dependent variable D(PE) D(LGDPPC) 1.092166 (0.2965) 퐺퐷푃 ↮ 푃퐸 D(ELEC) 3.328293 (0.0687) 퐸퐿퐸퐶 ⟶ 푃퐸 Source: Computed by Author using Eviews 9
Notes: Figures represent F-statistic values; p-values are given in brackets, ECT represents the estimated error- correction term, “X”↮ “Y” indicates no causal relationship between “X” and “Y”, “X” → “Y” indicates causality running from “X” to “Y”, “X” ⇔ “Y” indicates a bidirectional causality between “X” and “Y”, and ***, ** and * indicated the statistical significance of the estimated parameters at 1%, 5% and 10% respectively
96
University of Ghana http://ugspace.ug.edu.gh
The vector error correction model based granger causality is employed in investigating the short run dynamics of the variables. Results from the short run granger causality test are consistent with the short run vector error correction model estimates. From the causality analysis as reported in Table 5.14, two significant outcomes are revealed. The first is the existence of a causal relationship running from growth to electricity consumption. This finding is confirmed by
Akinlo (2008), Wolde-Rufael (2004), Wolde-Rufael (2008). Akinlo (2008) finds a unidirectional relationship running from GDP to energy consumption in the short run for countries like Sudan and Zimbabwe. Similarly, Wolde-Rufael, (2004) concludes that for countries including
Cameroon, Ghana, Nigeria, Senegal, Zambia and Zimbabwe, there is a unidirectional causality running from growth to electricity consumption. Again, Wolde Rufael (2008) finds a unidirectional causality running from growth to energy consumption for Zambia. It however contradicts the findings of Chontanawat et al., (2008) who find a bidirectional causality running from energy consumption to growth in countries including Ghana.
A plausible explanation why GDP has a statistically significant and positive relationship with electricity as reported in Table 5.13 and therefore a causal relationship running from GDP to electricity consumption as shown in Table 5.14 is that as the economy grows, income level improves and individuals are able to afford services that require additional energy (Ouedraogo,
2013). Stated differently, individuals are able to increase their consumption of electricity through the use of technology as income increases. Further, as the economy grows, the economic environment becomes generally good for business, investors are attracted into the country to establish industries which tend to increase electricity requirements.
The second is the existence of a short run causal relationship running from electricity consumption to petroleum consumption. A plausible explanation for this relationship is that as
97
University of Ghana http://ugspace.ug.edu.gh
electricity consumption increases, it becomes important to increase the generation capacity so as to meet the increasing demand. With this, petroleum consumption also increases as it is required to fuel thermal plants to generate electricity.
From Table 5.14, there is no short run causal relationship running from Total energy consumption to growth. More so, there is no causal relationship running from GDP to total energy consumption. The results show that the coefficient for Total energy consumption is highly insignificant statistically and for that matter total energy consumption has no influence on growth in the short run. This finding is also confirmed by Nondo et al., (2010); Ciarreta and
Zarraga (2008); Fatai, (2014). Nondo et al., (2010) who sought to explore the link between energy consumption and growth using a panel of nineteen African countries found the coefficients for energy consumption and GDP to be insignificant in the short run. Therefore, they concluded that in the short run, there is no causality running from energy consumption to GDP or from GDP to energy consumption. A plausible explanation for the existence of the neutrality hypothesis running from energy consumption to GDP is due to the energy consumption mix in the sub-region. Biomass constitutes a considerable proportion of the total energy consumed in
Africa and West Africa for that matter ((Sustainable Energy Consumption in Africa, 2004).
Biomass is used mainly by households for cooking and constitutes about 80% of total energy consumed in West Africa (Adenikinju, 2008). Household cooking could have both an indirect impact on growth as cooking provides food which serves as a source of energy for labour and enables labour to do more work. However, there is no established direct link between household cooking and growth and therefore Total energy consumption has no relationship with growth.
In Table 5.13, electricity consumption has an insignificant impact on output in the short run. The coefficient for electricity consumption on GDP is highly insignificant. There is therefore no short
98
University of Ghana http://ugspace.ug.edu.gh
run causality running from electricity consumption to GDP and from growth to electricity consumption as reported in Table 5.14. This result is confirmed by Karanfil & Li (2014); Wolde
Rufael, (2006). A plausible explanation for this finding is that in West Africa, as argued by the
African Development Bank, more than an estimated 57% of the populace in West Africa do not have access to electricity.13 It is also evident from Figure 2.2 that access to electricity in some
West African countries such as Sierra Leone, Liberia and Burkina Faso is very low. The low access to electricity in most West African states implies that limited electricity is fed into productive activities. In other words, production processes that would require electricity are unable to access electricity. The low access to electricity also restricts the opportunities available for people to increase their productivity and their incomes (Scott, 2015). Again, it is also probable that household electricity consumption constitutes a higher proportion of total electricity consumption relative to industrial electricity consumption in West Africa. Households derive utility from consuming electricity for domestic purposes and this domestic consumption of electricity by households could have no bearing on productive activities directly.
5.6 Summary of Main Findings.
In this chapter, it has established that the variables employed are all integrated of order one. As such, it became imperative to test for the existence of a long run relationship among the variables. Employing the Pedroni Cointegration test, the Kao test and the Fisher cointegration test, the study finds that in the long run, there is significant movement among the variables. The
FMOLS and the DOLS are used to estimate the long run relationship. The study finds that in the long run, petroleum consumption and electricity consumption have statistically significant impact on GDP. However, total energy consumption has a statistically significant and negative
13 Africa Power Journal; retrieved online from www.esi-africa.com/news/more-than-57-of-west-africa-population- without-electricity-access-says-afdb
99
University of Ghana http://ugspace.ug.edu.gh
relationship with GDP in the short run. The study finds a short run causality running from GDP to electricity consumption thereby confirming the conservation hypothesis for West Africa.
100
University of Ghana http://ugspace.ug.edu.gh
CHAPTER SIX
CONCLUSION AND RECOMMENDATION
6.0 Introduction
This chapter presents a summary of the findings in the previous chapter. Again, the study elaborates some recommendations that may prove helpful to policy makers in the sub-region.
6.1 Conclusion
The overarching objective of the study is to ascertain the causal relationship between energy consumption and economic growth focusing on the West African sub-region. The study focused on West Africa due to the dearth of studies available on energy consumption and growth in the region. In this regard, data was sought from the United States Energy Information
Administration and the World Development Indicators’ (2015). Specifically, data electricity consumption and petroleum consumption were sought from the Energy Information
Administration. Again, data on Total energy consumption, GDP per capita were also retrieved from the World Development Indicators’ (2015). Having done that, the study then employed a variety of approaches in finding results to the research questions of the study. First, the stationarity or otherwise of the variables were tested using two panel unit root testing approaches. The Levin, Lin & Chu (2002) and the Im et al., (2003) panel unit root testing methods were employed in testing for unit root. The two approaches unanimously reject the null hypothesis of non-stationarity of the variables at levels at 1% level of significance. However, at first difference, all the variables become stationary. Therefore, all the variables are found to be integrated of order one.
101
University of Ghana http://ugspace.ug.edu.gh
With this outcome, it is then necessary to test for cointegration in order to investigate the presence of a possible long run relationship among the variables. Three approaches to cointegration are employed in this study; namely, the Pedroni Cointegration test, the Kao residual cointegration test and the Fisher test. Results of the panel cointegration test as reported indicate the variables are cointegrated and therefore have a long run relationship. The results show that there is a long run relationship between GDP and Total energy consumption as well as between GDP and electricity consumption and petroleum consumption. The FMOLS and the
DOLS are then employed in investigating the long run relationship between energy consumption and growth. The study finds that in the long run, petroleum consumption and electricity consumption have a statistically significant and positive impact on growth. However, Total energy consumption has a statistically significant and negative relationship with economic growth in the long run. With regards to the short run analysis and granger causality, the study finds that in the short run, Total energy consumption, electricity consumption and petroleum consumption have statistically insignificant relationship with growth. Stated differently, there is no causal relationship running from Total energy consumption, electricity consumption and petroleum consumption to growth. However, the study finds a unidirectional relationship running from GDP to electricity consumption. That is to say, in the short run, an increase in GDP increases Total electricity consumption and this is statistically significant at 5 %. Again, electricity consumption has a statistically significant relationship with petroleum consumption in the short run. The study also finds that in the long run, there is no causal relationship running between any of the variables.
102
University of Ghana http://ugspace.ug.edu.gh
6.2 Recommendations of the study
The study makes various recommendations based on its major findings. The study finds that in the long run, electricity consumption has a positive impact on economic growth. It is important that measures that will ensure increased access to electricity are effectively pursued in the West
African sub-region. It is however important to accentuate the fact that ECOWAS currently has policies or projects aimed at increasing the generation capacity of electricity in the sub-region as well as policies to increase access to electricity. In this regard, projects like the West African Gas
Pipeline and the West African Power Pool should be encouraged as these projects will contribute to achieving increased access to electricity in the long run. These regional approaches to increasing access to energy in the sub-region are essential as they allow for the pooling of resources by member countries and thereby resulting in a relatively lower cost incurred compared to the cost the individual countries may have incurred. Again, some countries that are less endowed in some natural resources are able to benefit from the resources of other countries through the regional approach to satisfying the energy demands of the sub-region. The study therefore recommends a regional approach to solving the energy problems in West Africa. Also, it is important to accentuate the fact that in some instances, it would be more economical to adopt a country-level approach to providing solutions in the energy sector due to peculiar scenarios that may be existent in particular countries.
Furthermore, the World Bank stipulates that increasing access to electricity may come at a cost
14 as it may lead to an increase in the level of CO2 emissions in Africa. For some West African countries such as Ghana, the increase in demand for electricity which subsequently caused a shortfall in generation has led to a gradual shift from hydro power to the generation of electricity
14 Addressing the Electricity Access Gap (2010); Background Paper for the World Bank Group Energy Sector Strategy; retrieved online from
103
University of Ghana http://ugspace.ug.edu.gh
with thermal plants which serves as a source of quick and easy electric power generation. This has consequently led to an increase in demand for fossil fuel including petroleum. As argued earlier, if this trend continues, energy consumption will likely have a negative impact on growth in the long run. For this reason, this study recommends that clean sources of energy are harnessed to provide electricity. By clean sources of energy, this study recommends the exploitation of solar energy, wind energy as alternative sources of generating electricity in West
Africa.
Again, the unidirectional relationship running from GDP to electricity consumption in the short run implies that measures could be put in place to conserve energy and this would not have any dire consequence on growth in West Africa. Energy usage particularly electricity consumption in households, both public and private companies could be rationalized and this would save cost incurred in paying utility bills without affecting output. However, it would not be prudent to restrict access to electricity to areas with no access to electricity particularly rural areas. More so, measures that would ensure increase and sustained growth in GDP are important as this would encourage energy consumption in the short run so that in the long run energy consumption has a positive impact on GDP.
Also, the study finds that in the long run, Total energy consumption is expected to have a negative and statistically significant relationship with growth. The study argues that this would be so because of the traditional biomass content in the Total energy consumption. As such, it is important that measures that would reduce the use of traditional biomass are executed in West
Africa. Traditional biomass constitutes a significant proportion of energy consumed in the sub- region. This has encouraged deforestation, land degradation which is making it difficult for the sub-region to mitigate the impact of climate change. Again, the burning of wood, charcoal has
104
University of Ghana http://ugspace.ug.edu.gh
undesirable effects on the health of individuals in the sub-region. For this reason, it is important that measures that would reduce the share of biomass in the Total energy consumption are encouraged. The use of LPG gas should be encouraged to reduce the use of firewood in cooking.
The use of modern biomass should also be encouraged in the sub-region.
6.3 Area for further studies
A major limitation of the study is that the original idea of the researcher was to investigate the relationship between the various components of Total energy consumption and growth in the specific sectors of the West African economy. That is, the impact of electricity consumption, fossil fuel consumption, renewable energy consumption and Total energy consumption in the agricultural sector, the industrial and the services sector. However, due to data constraints, this research objective could not be realized.
105
University of Ghana http://ugspace.ug.edu.gh
REFERENCES
Acheampong T. (2016). The implications of changing power generation mix on energy pricing and security in Ghana. MPRA Paper No. 76703. Available online at https://mpra.ub.uni- muenchen.de/76703/
Adenikinju, A. (2008). West Africa energy security report. University of Ibadan Center for Energy Economics at the University of Texas at Austin Kumasi Institute of Energy, Technology and Environment.
Adenuga, A. O., & Emeka, R. (2013). Electricity consumption, exports and economic growth: Evidence from Nigeria: Open Research Journal of Energy. Vol. 1, No 1, July, 2013, pp. 01-17.
African Development Bank Report. (2014). Retrieved online from www.afdb.org/fileadmin/ on 25th February, 2017.
Africa Energy Outlook. (2014). A focus on Energy prospects in sub-Saharan Africa, world energy outlook special report. International Energy Agency.
African Economic Outlook (2015). Measuring the pulse of Economic Transformation in West Africa. Available on www.afdb.org/en/blogs/measuring-the-pulse-of-economic-transformation- in-west-africa. Accessed on July 25, 2016.
Africa, S. S. (2009). Effects on Infrastructure Conditions on Export Competitiveness. Third Annual Report by United States International Trade Commission–April.
Akinlo, A.E. (2008). Energy consumption and economic growth: evidence from 11 Sub- Sahara African countries. Journal of Energy Economics 30 (2), 2391–2400.
Akoena S. K. K. and Twerefour D. K. (2000). Improving road transport fuel efficiency and consumption in Ghana. Journal of GIMPA Development forum.
Alam M.S. (2006). Economic growth with energy; MPRA Paper No. 1260.
106
University of Ghana http://ugspace.ug.edu.gh
Alam, J. B., Wadud, Z., & Polak, J. W. (2013). Energy demand and economic consequences of transport policy. International Journal of Environmental Science and Technology, 10(5), 1075- 1082. Apergis N., & Payne J.E. (2009). “Energy consumption and economic growth in Central America: evidence from a panel cointegration and error correction model”. Energy Economics 31, 211–216.
Apostolakis, B. E. (1990). “Energy-capital substitutability/complementarity: the dichotomy.” Energy Economics 12: 48-58.
Arbex M. A., Cançado J. E. D., Pereira, L. A. A., Braga, A. L. F., & Saldiva, P. H. D. N. (2004). Biomass burning and its effects on health. Journal Brasileiro de Pneumologia, 30(2), 158-175.
Asafu-Adjaye, J. (2000). ‘The Relationship between Energy Consumption, Energy Prices and Economic Growth: Time Series Evidence from Asian Developing Countries.’ Energy Economics 22 (6): 615–625.
Augutis J., Krikštolaitis R., Pečiulytė S., & Konstantinavičiūtė I. (2011). Sustainable Development And Energy Security Level After Ignalina Npp Shutdown, Technological And Economic Development of Economy 17(1): 5–21.
Ayres R., & Nair I. (1984). Thermodynamics and economics. Physics today, Vol 35, 1984, pp 62-71.
Baltagi B. H. (2005). Econometric Analysis of panel data (3rd Etd); John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England.
Behera, J. (2015). Examined the energy-led growth hypothesis in India: Evidence from time series analysis. Journal of Energy Technology and Policy 5 (6), 15-27.
Behmiri B. N., & Manso J. (2013). How crude oil consumption impacts on economic growth of Sub-Sharan Africa? Energy, vol. 54, issue C, 74-83.
Benin Energy Situation. Retrieved online from energypedia.info/wiki/Benin Energy situation on 12th November, 2016.
107
University of Ghana http://ugspace.ug.edu.gh
Benin Primary Energy Consumption; retrieved online from http://benin.opendataforafrica.org/ on 24th November, 2016.
Bergasse E., Paczynski W., Dabrowski M., & Dewulf L. (2013). “The relationship between energy and socio-economic development in the Southern and Eastern Mediterranean”, paper produced for the MEDPRO project, 2013.
Berndt, E. R. (1990). “Energy use, technical progress and productivity growth: A survey of economic issues.” The Journal of Productivity Analysis 2: 67-83.
Bildirici M. E. (2013). The Analysis of Relationship between Economic Growth and Electricity Consumption in Africa by ARDL Method. Energy Economics Letters, 1(1), 1-14.
Blackorby, C., & Russell R. R. (1989). Will the real elasticity of substitution please stand up? (A comparison of the Allen/Uzawa and Morishima elasticities). The American Economic Review, 79(4), 882-888.
BP (2006). BP Statistical Review of World Energy. British Petroleum, London.
Breman H. (2003). IFA-FAO Agriculture Conference.
Brookes, L. (1990). The greenhouse effect: the fallacies in the energy efficiency solution. Energy Policy 18: 199-201.
Bruce N., Perez-Padilla R., & Albalak R. (2000). Indoor air pollution in developing countries: A major environmental and public health challenge. Bulletin of the world health organization, 78(9), 1078-1092.
Burbridge, J., & Harrison A. (1984). Testing for the effects of oil prices rises using vector autoregressions. International Economic Review 25: 459-484.
108
University of Ghana http://ugspace.ug.edu.gh
Burkina Faso Energy Situation. Retrieved online from http://energypedia.info/wiki on 24th September, 2016
Burkina Faso Primary Energy Consumption; Retrieved from burkinafaso.opendataforafrica.org on 24th September, 2016.
Burgess D. (1984). Energy prices, capita1 formation and potential GNP. The Energy Journal, 5(2), 1-27.
Busse, M. R., Knittel, C., & Zettelmeyer, F. (2010). Pain at the pump: The effect of gasoline prices on new and used automobile markets (No. 0110). CSIO Working Paper.
Cali, M. (2008). The contribution of services to development and the role of trade liberalisation and regulation. Overseas Dev't Institute.
Chakravarty, S., Ghosh, S. K., Suresh, C. P., Dey, A. N., & Shukla, G. (2012). Deforestation: causes, effects and control strategies. In Global perspectives on sustainable forest management. InTech.
Chontanawat, J., Hunt, L.C., Pierce, R., (2008). Does energy consumption cause economic growth? Evidence from systematic study of over 100 countries. The Journal of Policy Modelling 30, 209–220.
Ciarreta Antuñano, A., & Zárraga Alonso, A. (2008). Economic growth and electricity consumption in 12 European Countries: a causality analysis using panel data.
Coffin, D., Horowitz, J., Nesmith, D., & Semanik, M. (2016). Examining barriers to trade in used vehicles.
Cole M. A., & Neumayer E. (2006). The impact of poor health on Total Factor productivity. The Journal of development studies, 42(6), 918-938.
109
University of Ghana http://ugspace.ug.edu.gh
Cote d’Ivoire primary energy consumption: Retrieved from cotedivoire.opendataforafrica.org on 24th November, 2016
Dasgupta, P., & Heal, G. (1974). The optimal depletion of exhaustible resources. The review of economic studies, 41, 3-28.
Diskiene, D., Galiniene, B., Marčinskas, A. (2008). A strategic management model for economic development, Technological and Economic Development of Economy, 14(3), 375-387.
Dunkerley, J., Ramsay, W., Gordon, L., Cecelski, L., (1981). Energy Strategies for Developing Countries. Washington, DC: Resources for the future.
ECOWAS Centre for Renewable Energy and Energy Efficiency. Retrieved online from http://www.ecreee.org/page/ecowas-renewable-energy-policy-erep
Energy in Nigeria. Retrieved online from https://en.wikipedia.org/wiki/Energy_in_Nigeria
Engle, R. F., & Granger, C. W. (1987). Co-integration and error correction: representation, estimation, and testing. Econometrica: journal of the Econometric Society, 251-276.
Enu, P., & Havi, E. D. K. (2014). Influence of electricity consumption on economic growth in Ghana: an econometric approach; International Journal of Economics, Commerce and Management, 2(9), 1-20.
Erbaykal, E. (2008). Disaggregate energy consumption and economic growth: evidence from Turkey. International Research Journal of Finance and Economics, 20(20), 172-179.
Esso, L. J. (2010). Threshold cointegration and causality relationship between energy use and growth in seven African countries. Energy Economics, 32(6), 1383-1391.
110
University of Ghana http://ugspace.ug.edu.gh
Esso, L. J., & Keho, Y. (2016). Energy consumption, economic growth and carbon emissions: Cointegration and causality evidence from selected African countries. Energy, 114, 492-497.
Fatai, B. O. (2014). Energy consumption and economic growth nexus: Panel co-integration and causality tests for Sub-Saharan Africa. Journal of Energy in Southern Africa, 25(4), 93-100.
Fay, M., Iimi, A., & Perrissin-Fabert, B. (2010). Financing greener and climate-resilient infrastructure in developing countries-challenges and opportunities. EIB Papers, 15(2), 34-58.
Ferguson, R., Wilkinson, W., & Hill, R. (2000). Electricity use and economic development. Energy policy, 28(13), 923-934.
Food and Agriculture Organization. Future energy requirements for Africa’s agriculture; Retrieved online from fao.org/ on 24th November, 2016.
Food and Agriculture Organization. Environment and Natural Resources working paper. No. 4. Rome.
Frondel, M., & Schmidt, C. M. (2002). The capital-energy controversy: an artifact of cost shares?. The Energy Journal, 53-79.
Gbadebo, O. O., & Okonkwo, C. (2009). Does energy consumption contribute to economic performance? Empirical evidence from Nigeria. Journal of Economics and International Finance, 1(2), 44.
Ghana Primary Energy Consumption; Retrieved online from http://ghana.opendataforafrica.org on 24th November, 2016.
Gnansounou E. (2008). Boosting the electricity sector in West Africa: An integrative vision. International Association for Energy Economics.
111
University of Ghana http://ugspace.ug.edu.gh
Granger, C. W. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica: Journal of the Econometric Society, 424-438.
Green, R. & Xiao–Ping Z., (2013). Future Role of energy in manufacturing: Future manufacturing project. Evidence paper 11.
Hamilton, J. D. (1983). Oil and the macroeconomy since World War II. Journal of political economy, 91(2), 228-248.
Harto, C. B., Yan, Y. E., Demissie, Y. K., Elcock, D., Tidwell, V. C., Hallett, K., ... & Tesfa, T. K. (2012). Analysis of drought impacts on electricity production in the Western and Texas interconnections of the United States (No. ANL/EVS/R-11/14). Argonne National Laboratory (ANL).
Hollinger, F. (2015). Agricultural Growth in West Africa, Market and policy drivers.
Houghton, R. A. (2005). Tropical deforestation as a source of greenhouse gas emissions. Tropical deforestation and climate change, 13.
Huang, B. N., Hwang, M. J., & Yang, C. W. (2008). Causal relationship between energy consumption and GDP growth revisited: a dynamic panel data approach. Ecological economics, 67(1), 41-54.
IMF (2010) Senegal: Poverty reduction strategy paper annual progress report; IMF country report no. 10/368.
Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of econometrics, 115(1), 53-74.
Kao, C. (1999). Spurious regression and residual-based tests for cointegration in panel data. Journal of econometrics, 90(1), 1-44.
112
University of Ghana http://ugspace.ug.edu.gh
Kanu, B. S., Salami, A. O., & Numasawa, K. (2014). Inclusive growth: an imperative for African agriculture. African Journal of Food, Agriculture, Nutrition and Development, 14(3), A33-A33. Karanfil, F., & Li, Y. (2015). Electricity consumption and economic growth: exploring panel- specific differences. Energy Policy, 82, 264-277.
Karekezi, S., & Kimani, J. (2002). Status of power sector reform in Africa: impact on the poor. Energy Policy, 30(11), 923-945.
Karekezi, S., Kithyoma, W., Muzee, K., & Oruta, A. (2008, April). Scaling-up bioenergy in Africa. In Background paper for international conference on scaling-up renewables in Africa, organized by UNIDO from (pp. 16-18).
Kebede, E., Kagochi, J., & Jolly, C. M. (2010). Energy consumption and economic development in Sub-Sahara Africa. Energy economics, 32(3), 532-537.
Khazzoom, J. D. (1980). Economic implications of mandated efficiency in standards for household appliances. The energy journal, 1(4), 21-40.
Kwakwa, P. A. (2012). Disaggregated energy consumption and economic growth in Ghana.
Johnston, K. A., & Ramirez, M. D. (2015). Foreign Direct Investment and Economic Growth in Cote D’Ivoire: A Time Series Analysis. Business and Economic Research, 5(2), 35-47.
Levin, A., & Lin, C.F., & Chu, C.S. (2002). Unit root tests in panel data: asymptotic and finite sample Operúes. J. Econ. 108 (1), 1–24.
Lee, C.C., & Chang, C.P. (2008). Energy consumption and economic growth in Asian economies: a more comprehensive analysis using panel data. Resource and Energy Economics 30 (1), 50–65.
Lütkepohl, H. (1982). Non-causality due to omitted variables. Journal of Econometrics, 19(2-3), 367-378.
113
University of Ghana http://ugspace.ug.edu.gh
Lu, W. C. (2016). Electricity consumption and economic growth: Evidence from 17 Taiwanese industries. Sustainability, 9(1), 50.
Mahmoudinia, D., Amroabadi, B. S., Pourshahabi, F., & Jafari, S. (2013). Oil products Consumption, Electricity Consumption-Economic growth Nexus in the Economy of Iran: A Bounds Testing Co-integration Approach. International Journal of Academic Research in Business and Social Sciences, 3(1), 353.
Masih, A. M., & Masih, R. (1996). Energy consumption, real income and temporal causality: results from a multi-country study based on cointegration and error-correction modelling techniques. Energy economics, 18(3), 165-183.
Murry, D. A., & Nan, G. D. (1994). A definition of the gross domestic product-electrification interrelationship. The Journal of energy and development, 19(2), 275-283.
Muktar, M. (2011). Impact of Transportation on Economic Growth: An Assessment of Road and Rail Transport Systems. web log post] Retrieved from http://mustaphamuktar. blogspot. com/2011/01/impact-of-transportation-oneconomic. html.
Nigeria Primary Energy Consumption; Retrieved from http://nso.nigeria.opendataforafrica.org on 24th November, 2016.
Nindi, A. G., & Odhiambo, N. M. (2014). Energy consumption and economic growth in Mozambique: an empirical investigation. Environmental Economics, 5(4), 83-92.
Nondo, C., Kahsai, M., & Schaeffer, P. V. (2010). Energy consumption and economic growth: evidence from COMESA countries. Research paper, 1.
Odhiambo, N. M. (2009). Energy consumption and economic growth nexus in Tanzania: An ARDL bounds testing approach. Energy Policy, 37(2), 617-622.
Olofin, O. P., Olayeni, O. R., & Abogan, O. P. (2014). Economic Consequences of Disaggregate Energy Consumption in West African Countries. Journal of Sustainable Development, 7(3), 71.
114
University of Ghana http://ugspace.ug.edu.gh
Olusanya, S. O. (2012). Long-run relationship between energy consumption and economic growth: evidence from Nigeria. Journal of Humanities and Social Sciences, 3(3), 40-51.
Ouédraogo, I. M. (2010). Electricity consumption and economic growth in Burkina Faso: A cointegration analysis. Energy Economics, 32(3), 524-531.
Ouedraogo N. S. (2013). Energy consumption and economic growth: Evidence from the economic community of West African States (ECOWAS). Energy Economics, 36, 637–647.
Ozturk, I., Aslan, A., & Kalyoncu, H. (2010). Energy consumption and economic growth relationship: Evidence from panel data for low and middle income countries. Energy Policy, 38(8), 4422-4428.
Panel, A. P. (2015). Power people planet: seizing Africa's energy and climate opportunities: Africa progress report 2015.
Para a África, C. E. (2004). Economic Report on Africa: Unlocking Africa’s Trade Potential’. Addis Abeba.
Pedroni, P. (1999). Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and statistics, 61(S1), 653-670.
Pedroni, P. (2001). Purchasing power parity tests in cointegrated panels. The review of Economics and Statistics, 83(4), 727-731.
Pesaran, M., Shin, Y., & Smith, R., (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326.
Pokharel, S. (2007). An econometric analysis of energy consumption in Nepal. Energy policy, 35(1), 350-361.
Porter, G. (2012). Reflections on a century of road transport developments in West Africa and their (gendered) impacts on the rural poor. EchoGéo, (20).
115
University of Ghana http://ugspace.ug.edu.gh
Ravindranath, N. H., & Rao U. (2005). Environmental effects of energy from biomass and municipal solid waste In: Jose Goldemberg (Ed.), Interactions: Energy / Environment, Encyclopedia of life support system, UNESCO-ELOSS.
Razzaqi, S., Bilquees, F., & Sherbaz, S. (2011). Dynamic relationship between energy and economic growth: evidence from D8 countries. The Pakistan Development Review, 437-458.
Ribeiro, S. K., Kobayashi, S., Beuthe, M., Gasca, J., Greene, D., Lee, D. S., & Wit, R. (2007). Transportation and its infrastructure. Institute of Transportation Studies.
Ruta, G. (2010). Monitoring Environmental Sustainability.
Sahel and West Africa Club. (2016). Retrieved from www.westafricagateway.org/content/en/ on 28th December, 2016.
Sama, M. C., & Tah, N. R. (2016). The Effect of Energy Consumption on Economic Growth in Cameroon. Asian Economic and Financial Review, 6(9), 510.
Schurr, S. H., & Netschert, B. C. (1960). Energy in the American economy, 1850-1975 (Vol. 960). Baltimore.
Scott, A. (2015). Building electricity supplies in Africa for growth and universal access. Background paper for Power. People, Planet: Seizing Africa’s energy and climate opportunities.
Siddiqui, R. (2004). Energy and economic growth in Pakistan. The Pakistan Development Review, 175-200.
Smith, R. (2001). Global Forest Resources Assessment 2000 Main Report. Food and Agriculture Organization: Rome, Italy.
Stern, D. I. (1993). Energy and economic growth in the USA: a multivariate approach. Energy economics, 15(2), 137-150.
116
University of Ghana http://ugspace.ug.edu.gh
Stern, D. I. (2002). Explaining changes in global sulfur emissions: an econometric decomposition approach. Ecological Economics, 42(1), 201-220.
Stern, D. I. (2004). Economic growth and energy. Encyclopedia of Energy, 2(00147), 35-51.
Stern, D. I., & Cleveland, C. J. (2004). Energy and Economic Growth. Rensselaer Polytechnic Institute (No. 0410). Rensselaer Working Papers in Economics.
Šliogerienė, J. Kaklauskas, A. Zavadskas, E. K. Bivainis, J. & Seniut, M. (2009). Environment factors of energy companies and their effect on value: analysis model and applied method, Technological and Economic Development of Economy 15(3): 490– s521.
Solow, R. M. (1956). A contribution to the theory of economic growth. The quarterly journal of economics, 70(1), 65-94.
Solow, R. M. (1974). Intergenerational equity and exhaustible resources. The review of economic studies, 41, 29-45.
Solow, R. (1993). An almost practical step toward sustainability. Resources policy, 19(3), 162- 172.
Solow, R. M. (1997). Georgescu-Roegen versus Solow-Stiglitz. Ecological Economics, 22(3), 267-268.
Sustainable Energy Consumption in Africa (2004). UN-DESA final report. Retrieved online from http://www.un.org/esa/.../Energy Consumption.
Swan, T. W. (1956). Economic growth and capital accumulation. Economic record, 32(2), 334- 361.
117
University of Ghana http://ugspace.ug.edu.gh
The African economy in 1994. Retrieved from www. africa.upenn.edu/ECA/AfEcl.html on 24th November, 2016.
The energy sector in Africa: Sustainable Energy Regulation and policy making for Africa. Accessed from https://www.yumpu.com/en/document/view/42842389/the-energy-sector-in- africa-reeep-unido-training-package/1
The energy sector in Africa: Sustainable Energy Regulation and policy making for Africa. Module 2.
The socio-economic context of West Africa. (2006). Retrieved from www.oecd.org/migration on 27th November, 2016
Thompson, P., & Taylor, T. G. (1995). The capital-energy substitutability debate: a new look. The Review of Economics and Statistics, 565-569.
Trans-Africa Consortium. (2008). Retrieved online from www.uitp.org/sites/default on 4th March, 2017.
Twerefou D.K., Akoena S.K.K., Egyir-Tettey F.K. and Mawutor G. (2008), Energy consumption and economic growth: evidence from Ghana. Department of Economics, University of Ghana.
Ucan, O., Aricioglu, E., & Yucel, F. (2014). Energy consumption and economic growth nexus: evidence from developed countries in Europe. International Journal of Energy Economics and Policy, 4(3), 411.
United Nations Economic and Social Council for Africa. (2009). The Transport situation in Africa. Sixth session of the Committee on Trade, Regional Cooperation and Integration; Addis Ababa. Available online at http://uneca.org/eca_resources/Publications/UNEA-Publication- toCSD15.pdf.
Vandaele, N., & Porter, W. (2015). Renewable Energy in Developing and Developed Nations: Outlooks to 2040. University of Florida, Summer.
118
University of Ghana http://ugspace.ug.edu.gh
WHO. (2006). Household energy, indoor air pollution and health, in: fuel for life: household energy and health. http://www.who.int/indoorair/publications/fuelforlife/en/index.html (accessed online on 23th June, 2017).
Wolde-Rufael, Y. (2004). Disaggregated industrial energy consumption and GDP: the case of Shanghai, 1952–1999. Energy economics, 26(1), 69-75.
Wolde-Rufael, Y. (2009). Energy consumption and economic growth: the experience of African countries revisited. Energy Economics, 31(2), 217-224.
Wolde-Rufael, Y. (2006). Electricity consumption and economic growth: a time series experience for 17 African countries. Energy policy, 34(10), 1106-1114.
World Development Indicators. (1997). Washington DC. World Bank.
World Bank Development Indicators’. (2016). World Bank.
Yasar, N. (2017). The Relationship between Energy Consumption and Economic Growth: Evidence from Different Income Country Groups. International Journal of Energy Economics and Policy, 7(2).
Yoo, S. H. (2005). Electricity Consumption and Economic Growth: Evidence from Korea. Energy Policy; 33 (12): 1627–1632.
Yoo, S. H., & Kim, Y. (2006). Electricity generation and economic growth in Indonesia. Energy, 31(14), 2890-2899.
Zhao, H., Zhao, H., Han, X., He, Z., & Guo, S. (2016). Economic growth, electricity consumption, labor force and capital input: A more comprehensive analysis on North China using panel data. Energies, 9(11), 891.
119
University of Ghana http://ugspace.ug.edu.gh
APPENDICES
APPENDIX I: Estimation of FMOLS (GDP Per Capita and Total Energy consumption)
Dependent Variable: LGDPC Method: Panel Fully Modified Least Squares (FMOLS) Date: 06/27/17 Time: 10:36 Sample (adjusted): 1981 2013 Periods included: 33 Cross-sections included: 11 Total panel (unbalanced) observations: 223 Panel method: Weighted estimation Cointegrating equation deterministics: C Long-run covariance estimates (Bartlett kernel, Newey-West fixed bandwidth)
Variable Coefficient Std. Error t -Statistic Prob.
LENER - 0.139887 0.031804 - 4.398435 0.0000
R -squared 0.885487 Mean dependent var 6.373403 Adjusted R-squared 0.879517 S.D. dependent var 0.427853 S.E. of regression 0.148511 Sum squared resid 4.653701 Long-run variance 0.012620
APPENDIX II: Estimation of FMOLS (GDP Per Capita, Electricity consumption and Petroleum consumption).
Dependent Variable: LGDPC Method: Panel Fully Modified Least Squares (FMOLS) Date: 06/27/17 Time: 10:47 Sample (adjusted): 1981 2013 Periods included: 33 Cross-sections included: 17 Total panel (unbalanced) observations: 537 Panel method: Weighted estimation Cointegrating equation deterministics: C Long-run covariance estimates (Bartlett kernel, Newey-West fixed bandwidth)
Variable Coeffi cient Std. Error t -Statistic Prob.
ELEC 0.107273 0.013321 8.052804 0.0000
120
University of Ghana http://ugspace.ug.edu.gh
PE 0.058452 0.019071 3.064980 0.0023
R -squared 0.740725 Mean dependent var 6.187026 Adjusted R-squared 0.731715 S.D. dependent var 0.528847 S.E. of regression 0.273923 Sum squared resid 38.86748 Long-run variance 0.051034
APPENDIX III: Estimation of DOLS (GDP Per Capita and Total energy consumption).
Dependent Variable: LGDPC Method: Panel Dynamic Least Squares (DOLS) Date: 06/27/17 Time: 10:53 Sample (adjusted): 1982 2012 Periods included: 31 Cross-sections included: 7 Total panel (unbalanced) observations: 197 Panel method: Weighted estimation Cointegrating equation deterministics: C Fixed leads and lags specification (lead=1, lag=1) Long-run variance weights (Bartlett kernel, Newey-West fixed bandwidth) Warning: one more more cross-sections have been dropped due to estimation errors
Variable Coefficient Std. Error t -Statistic Prob.
LENER - 0.159404 0.099811 - 1.597053 0.1121
R -squared 0.885520 Mean dependent var 6.354463 Adjusted R-squared 0.866440 S.D. dependent var 0.395310 S.E. of regression 0.144469 Sum squared resid 3.506394 Long-run variance 0.052908
APPENDIX IV: Estimation of DOLS (GDP Per capita, electricity and petroleum consumption).
Dependent Variable: LGDPC Method: Panel Dynamic Least Squares (DOLS) Date: 06/27/17 Time: 10:53 Sample (adjusted): 1982 2012 Periods included: 31 Cross-sections included: 7 Total panel (unbalanced) observations: 197
121
University of Ghana http://ugspace.ug.edu.gh
Panel method: Weighted estimation Cointegrating equation deterministics: C Fixed leads and lags specification (lead=1, lag=1) Long-run variance weights (Bartlett kernel, Newey-West fixed bandwidth) Warning: one more more cross-sections have been dropped due to estimation errors
Variable Coefficient Std. Error t -Statistic Prob.
LENER - 0.159404 0.099811 - 1.597053 0.1121
R -squared 0.885520 Mean dependent var 6.354463 Adjusted R-squared 0.866440 S.D. dependent var 0.395310 S.E. of regression 0.144469 Sum squared resid 3.506394 Long-run variance 0.052908
APPENDIX V: Estimation of Equation (61) using Panel EGLS (Cross Section Weights).
Dependent Variable: D(LGDPC) Method: Panel EGLS (Cross-section weights) Date: 06/27/17 Time: 11:10 Sample (adjusted): 1982 2013 Periods included: 32 Cross-sections included: 11 Total panel (unbalanced) observations: 216 Linear estimation after one-step weighting matrix D(LGDPC) = C(1)*( LGDPC(-1) - 2.4781608189*LENER(-1) + 8.12114253921 ) + C(2)*D(LGDPC(-1)) + C(3)*D(LENER(- 1)) + C(4)
Coe fficient Std. Error t -Statistic Prob.
C(1) - 0.000254 0.001020 - 0.249251 0.8034 C(2) 0.254351 0.062396 4.076430 0.0001 C(3) 0.044435 0.039760 1.117577 0.2650 C(4) 0.004685 0.002618 1.789277 0.0750
Weighted Statistics
R -squared 0.084936 Mean dependent var 0.007258 Adjusted R-squared 0.071987 S.D. dependent var 0.045354 S.E. of regression 0.043401 Sum squared resid 0.399328 F-statistic 6.559238 Durbin-Watson stat 2.134715 Prob(F-statistic) 0.000292
122
University of Ghana http://ugspace.ug.edu.gh
Unweighted Statistics
R -squared 0.085359 Mean dependent var 0.006070 Sum squared resid 0.400518 Durbin-Watson stat 2.123671
APPENDIX VI: Estimation of Equation (62)
Dependent Variable: D(LENER) Method: Panel EGLS (Cross-section weights) Date: 06/27/17 Time: 11:38 Sample (adjusted): 1982 2013 Periods included: 32 Cross-sections included: 11 Total panel (unbalanced) observations: 212 Linear estimation after one-step weighting matrix D(LENER) = C(5)*( LGDPC(-1) - 2.4781608189*LENER(-1) + 8.12114253921 ) + C(6)*D(LGDPC(-1)) + C(7)*D(LENER(- 1)) + C(8)
Coefficient Std. Error t -Statistic Prob.
C(5) 0.002762 0.001341 2.060279 0.0406 C(6) 0.010498 0.045334 0.231564 0.8171 C(7) 0.000737 0.063113 0.011673 0.9907 C(8) 0.006729 0.002640 2.548562 0.0115
Weighted Statistics
R -squared 0.021492 Mean dependent var 0.007334 Adjusted R-squared 0.007379 S.D. dependent var 0.056803 S.E. of regression 0.056118 Sum squared resid 0.655029 F-statistic 1.522836 Durbin-Watson stat 2.049729 Prob(F-statistic) 0.209640
Unweighted Statistics
R -squared 0.006561 Mean dependent var 0.0056 99 Sum squared resid 0.666341 Durbin-Watson stat 2.102458
APPENDIX VII: Estimation of Equation (63)
Dependent Variable: D(LGDPC)
123
University of Ghana http://ugspace.ug.edu.gh
Method: Panel EGLS (Cross-section weights) Date: 06/27/17 Time: 11:41 Sample (adjusted): 1982 2013 Periods included: 32 Cross-sections included: 17 Total panel (unbalanced) observations: 518 Linear estimation after one-step weighting matrix D(LGDPC) = C(1)*( LGDPC(-1) + 27.2399848159*ELEC(-1) + 33.6952874177*PE(-1) - 91.0420223014 ) + C(2)*D(LGDPC(-1)) + C(3) *D(ELEC(-1)) + C(4)*D(PE(-1)) + C(5)
Coefficient Std. Error t -Statistic Prob.
C(1) 2.35E -05 1.46E -05 1.611237 0.1077 C(2) 0.250882 0.042565 5.894134 0.0000 C(3) -0.004800 0.004958 -0.968146 0.3334 C(4) 0.006214 0.008824 0.704217 0.4816 C(5) 0.006251 0.001701 3.675125 0.0003
Weighted Statistics
R -squared 0.070735 Mean dependent var 0.012285 Adjusted R-squared 0.063489 S.D. dependent var 0.071063 S.E. of regression 0.068451 Sum squared resid 2.403647 F-statistic 9.762302 Durbin-Watson stat 2.087868 Prob(F-statistic) 0.000000
APPENDIX VIII: Estimation of Equation (64)
Dependent Variable: D(ELEC) Method: Panel EGLS (Cross-section weights) Date: 06/27/17 Time: 11:43 Sample (adjusted): 1982 2013 Periods included: 32 Cross-sections included: 17 Total panel (unbalanced) observations: 518 Linear estimation after one-step weighting matrix D(ELEC) = C(6)*( LGDPC(-1) + 27.2399848159*ELEC(-1) + 33.6952874177*PE(-1) - 91.0420223014 ) + C(7)*D(LGDPC(-1)) + C(8) *D(ELEC(-1)) + C(9)*D(PE(-1)) + C(10)
Coefficient Std. Error t -Statistic Prob.
124
University of Ghana http://ugspace.ug.edu.gh
C(6) 0.001021 0.000118 8.649467 0.0000 C(7) 0.080375 0.042036 1.912051 0.0564 C(8) -0.191158 0.044918 -4.255741 0.0000 C(9) 0.001437 0.036683 0.039160 0.9688 C(10) 0.087133 0.008893 9.797368 0.0000
Weighted Statistics
R -squared 0.137170 Mean depen dent var 0.127229 Adjusted R-squared 0.130443 S.D. dependent var 0.372801 S.E. of regression 0.354757 Sum squared resid 64.56232 F-statistic 20.38883 Durbin-Watson stat 2.024016 Prob(F-statistic) 0.000000
Unweighted Statistics
R -squared 0.133267 Mean dependent var 0.065132 Sum squared resid 98.10614 Durbin-Watson stat 2.151486
APPENDIX IX: Estimation of Equation (65)
Dependent Variable: D(PE) Method: Panel EGLS (Cross-section weights) Date: 06/27/17 Time: 11:45 Sample (adjusted): 1982 2013 Periods included: 32 Cross-sections included: 17 Total panel (unbalanced) observations: 518 Linear estimation after one-step weighting matrix D(PE) = C(11)*( LGDPC(-1) + 27.2399848159*ELEC(-1) + 33.6952874177 *PE(-1) - 91.0420223014 ) + C(12)*D(LGDPC(-1)) + C(13)*D(ELEC(-1)) + C(14)*D(PE(-1)) + C(15)
Coefficient Std. Error t -Statistic Prob.
C(11) 0.000393 9.81E -05 4.003298 0.0001 C(12) 0.031051 0.029712 1.045068 0.2965 C(13) 0.056492 0.030965 1.824361 0.0687 C(14) -0.116022 0.052465 -2.211412 0.0274 C(15) 0.034719 0.007604 4.565792 0.0000
Weighted Statistics
125
University of Ghana http://ugspace.ug.edu.gh
R-squared 0.053884 Mean dependent var 0.052740 Adjusted R-squared 0.046506 S.D. dependent var 0.257881 S.E. of regression 0.254760 Sum squared resid 33.29518 F-statistic 7.304133 Durbin-Watson stat 1.768619 Prob(F-statistic) 0.000010
Unweighted Statistics
R -squared 0.008927 Mean dependent var 0.027118 Sum squared resid 41.85094 Durbin-Watson stat 2.079263
APPENDIX X: RESULTS OF PHILLIPS-PERRON FISHER UNIT ROOT TEST
At levels At first difference Order of Variables T - statistics Probability t statistics Prob. integration value value GDP Per 36.6662 0.3462 288.374*** 0.0000 I(1) Capita Total Energy 4.54629 1.0000 199.839*** 0.0000 I(1) Consumption Electricity 7.83732 1.0000 378.838*** 0.0000 I(1) Consumption Petroleum 25.4975 0.8530 328.837*** 0.0000 I(1) Consumption
126