You are on page 1of 15

Energy 73 (2014) 44e58

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Economic growth, energy consumption and CO2 emissions in Gulf


Cooperation Council countries
Mohammad Salahuddin a, *, Jeff Gow b, c
a

Australian Digital Future Institute and School of Commerce, University of Southern Queensland, Toowoomba, QLD 4350, Australia
School of Commerce, Faculty of Business, Economics, Law and Arts, University of Southern Queensland, Toowoomba, QLD 4350, Australia
c
Department of Agricultural Economics, Stellenbosch University, Stellenbosch, South Africa
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 11 January 2014
Received in revised form
5 May 2014
Accepted 14 May 2014
Available online 4 July 2014

This study examines the empirical relationship between economic growth, energy consumption and
carbon dioxide emissions, calculates the trend of decoupling effects and nally analyzes the evolution of
inequality in CO2 emissions in GCC (Gulf Cooperation Council countries) countries. Results indicate a
positive and signicant association between energy consumption and CO2 emissions and between
economic growth and energy consumption both in the short- and the long-run. No signicant relationship is found between economic growth and CO2 emissions. Energy consumption and CO2 emissions
Granger cause each other while unidirectional causal link running from economic growth to energy
consumption is also found to exist. Both absolute and relative decoupling occurred in all the GCC
countries except Saudi Arabia during the study period. Divergences in the Gini index values contributed
towards different levels of emissions inequality in the region. CO2 emissions inequality signicantly
declined both in energy carriers as well as in the economic sectors over time. Despite some optimistic
ndings, the GCC countries are still signicant contributors to CO2 emissions and as such, the study
recommends pursuing favorable regulatory policies that would promote various initiatives to reduce
emissions. The overall ndings will help GCC countries assess its position better in future climate change
negotiations.
2014 Elsevier Ltd. All rights reserved.

Keywords:
CO2 emissions inequality
Decoupling effects
Economic growth
Energy consumption
GCC countries

1. Introduction
The GCC (Gulf Cooperation Council countries) comprising Saudi
Arabia, The UAE (United Arab Emirates), Qatar, Bahrain, Kuwait and
Oman own approximately 40% of the world's proven oil reserves
and around 25% of natural gas reserves. Three of the GCC countries,
Saudi Arabia, The UAE and Qatar have already been identied as the
largest per capita CO2 emitters in the world [61]. In 2009, GCC
countries contributed to approximately 8% of the global CO2
emissions [8]. Although these countries are expected to continue as
suppliers of oil and gas to the rest of the world for the decades to
come, the region can not afford the luxury of being complacent for
being a source of cheap energy. Instead, they are poised to confront
acute energy shortage in the coming decades and years due to
rapidly growing domestic demand for energy, in particular, for
electricity. Electricity is now a critical factor for the environmental

* Corresponding author.
E-mail addresses: salahuddin.mohammad@usq.edu.au, salahuddin0000@gmail.
com (M. Salahuddin), Jeffrey.Gow@usq.edu.au (J. Gow).
http://dx.doi.org/10.1016/j.energy.2014.05.054
0360-5442/ 2014 Elsevier Ltd. All rights reserved.

sustainability in the GCC countries. As such, the sustainability of the


energy systems in the region is likely to be vulnerable if the
anticipated energy crisis, in particular, electricity crisis and the CO2
emissions issues are not addressed appropriately.
Although economic growth is always accompanied by
increasing domestic demand for energy, this is not the case with
the GCC countries. The GCC countries experienced very high rates
of per capita electricity consumption which has already surpassed
the level of major industrial countries [61]. The rapid growth of
electricity demand has been attributed mainly to three factors; rst
of all, the phenomenal growth of urban population and the increase
in their purchasing power. The massive inux of migrant workers to
the GCC countries has reinforced their national population growth.
Second, the process of economic diversication and industrialization. These two have combinedly led to the growth of industries
that require electricity rather than other forms of energy. Third,
relatively low price of energy to industry as well as to the nal
consumer in the region both in the form of liquid fuels and in the
form of electricity. As electricity is the major source of CO2 emissions, this region has per capita emissions which is among the
highest in the world. However, energy intensity is an undeniable

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

feature and analysis of both the issues of transition to less energy


intensity and CO2 emissions inequality across the region are of
prime importance.
In the light of the above environmental adversity in the GCC
countries, this study investigates three important issues; rst, it
empirically estimates the short- and the long-run relationship between economic growth, energy consumption and CO2 emissions.
Second, it attempts to estimate contemporary trends in the relationship between production and natural resources inputs, in other
words, it involves the estimation of decoupling of energy consumption from economic growth in GCC countries. Third, it further
investigates how the energy mix and the sectoral composition of a
country's energy use determine inequality in CO2 emissions in the
region. This is to the best of our knowledge, the rst attempt ever, to
address these issues concerning GCC countries.
The rest of the paper is structured as follows; Section 2 discusses
the empirical evidence from literature while Section 3 is devoted to
methodology and data. Section 4 discusses the results, whilst
conclusions and discussions are covered in Section 5.
2. Literature review: the relationship between CO2 emissions,
economic growth and energy consumption, decoupling effect
and inequality in CO2 emissions
2.1. CO2 emissions and economic growth
The dominating relationship between economic growth and
CO2 emissions has received great attention in the literature. The
relationship between CO2 emissions and economic growth is
termed the EKC (Environmental Kuznets Curve) and the relationship is described in linear and non-linear terms of GDP per capita.
The EKC states that in the early stages of economic growth, CO2
emissions increase but as the economy keeps on growing and after
a certain threshold level, these emissions decline and the environmental quality improves.
The EKC hypothesis was initially proposed and tested by Ref.
[53]. It was then extensively tested in numerous studies such as
[38,45,59,78,88,122,124,133]: and [34]. Existing studies seem to
present mixed empirical evidences on the validity of EKC
[37,64,116] and [139] supported the existence of EKC in China.
Ref. [4] did not support the existence of EKC in Turkey. They argued
that CO2 emissions are automatically reduced due to the rapid pace
of economic growth. The ndings of Ref. [43] revealed the existence
of EKC between the SO2 emissions and economic growth but not for
the CO2 emissions in Tunisia. In contrast, Ref. [141] tested the EKC
hypothesis for 43 developing countries including Middle Eastern
countries. They nd that for the Middle Eastern panel, the income
elasticity in the long run is smaller than the short run implying that
carbon dioxide has fallen with the rise in income. Ref. [92] tests the
EKC hypothesis for 12 Middle East countries and nd little support
for it. Only for three countries does the study nd support for the
EKC hypothesis, for a further six countries it nds the U-shaped EKC
which is in contrast with the EKC hypothesis and no causal link was
found for the other three countries.
2.2. Energy consumption and economic growth
This strand of the literature suggests that energy consumption
and economic growth are closely related. Higher levels of economic
growth cause energy consumption to rise [5]. The relationship
between energy consumption and economic growth has been
examined extensively in a number of countries in recent years
[14,89,94,99]. To date, there exist three views on the causal relationship between energy consumption and economic growth. The
rst view is that energy consumption Granger-causes economic

45

growth. This means that economic growth is dependent on energy


consumption. According to the proponents of this view, a decrease
in energy consumption would probably have a negative effect on
economic growth [154,90]. The second view, however, maintains
that it is the growth of the real sector that drives energy consumption. This view is known as the growth-led energy consumption view.
According to the proponents of this view, the demand for energy
is supposed to be driven largely by the growth of the real sector.
This implies that a country need not entirely depend on energy for
its economic growth. Hence, energy conservation policies could be
implemented with little or no adverse effects on economic growth.
The third view, which may be regarded as the middle-ground,
however, maintains that both energy consumption and economic
growth Granger-cause one another. In other words, there is a
bidirectional causality between energy consumption and economic
growth. Table 1 gives an overview of some of the empirical ndings
associated with these three views.
The relationship between energy consumption and economic
growth has been investigated extensively as well. For example [68],
for USA [80], for Taiwan and Korea [17], for Pakistan [131], for African countries [84], for Fiji [107], for Soviet Union [29], for Malaysia
[85], for Middle Eastern and [135] for South American countries
concluded that increased energy consumption is positively linked
with economic growth. Opposite causality was also found by
Refs. [9] and [55] for Turkey, [121] for OPEC [137], for China and [89]
for Tanzania. Bivariate causality between energy consumption and
economic growth was found in a study by Ref. [19] for Thailand and
the Philippines.
Recent literature focused on the association among economic
growth, energy consumption and environmental pollution to
investigate the validity of EKC. There have also been plenty of both
time series and panel data studies on this relationship. In the time
series studies [108], investigates the bi-directional long-run relationship between energy consumption, CO2 emissions and economic growth in the road transport sector of OECD (Organization of
Economic Cooperation and Development) countries. The study uses
FMOLS (Fully Modied Ordinary Least Squares) method and conrms that there exist positive signicant bi-directional relationship
between CO2 emissions and economic growth, road sector energy
consumption and economic growth and between CO2 emissions
and road sector energy consumption. The study also uses IAA
(Innovation Accounting Approach) that includes the generalized
impulse response function to examine the effects of a shock in one
variable on other variables. Their ndings indicate that most of the
CO2 emissions occur due to energy consumption. The study emphasizes long-run policies that aim to enhance energy efciency
and to shift to other options for energy such as biofuel, renewable
and nuclear energy.
Ref. [6] undertakes a huge time series study involving seven
different regions, namely East Asia and Pacic, East Europe and
Central Asia, Latin America and the Caribbean, Middle East and
North Africa, South Asia, Sub-Saharan Africa and Western Europe.
This study uses FMOLS (Fully Modied Ordinary Least Squares)
method to capture the relationship between urbanization, energy
consumption and CO2 emissions. The results reveal that there is a
positive long-run relationship between the variables in 84% of the
countries while ndings are mixed for the rest 16%. Some countries
demonstrate negative relationship while others especially the low
income countries do not show any relationship at all.
Ref. [12] nds a stable long run relationship for France while
Ref. [13] also got similar results for Malaysia. Ref. [12] shows that
causality runs from economic growth to energy consumption and
CO2 emissions in the long run but increased energy consumption
causes economic growth in short run. In the case of Malaysia [13],

46

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

Table 1
Selected empirical ndings on the direction of the causality between economic
growth or income to energy consumption.
Author(s)

Region/country

Direction of causality

A: Studies in favor of a unidirectional causality from economic


growth to energy consumption
[68]
USA
GNP / Energy
consumption
[144]
USA
Energy
consumption/
employment
[3]
USA
No causal link
between Energy
consumption and
economic growth
No causal and
[136]
USA, UK, Poland,
unidirectional link
Philippines,
were found for
South Korea
the countries
of study.
[1]
USA
GNP / Energy
consumption
[33]
Taiwan
GDP / Energy
consumption
[145]
India
GDP / Energy
consumption
[120]
G-7 countries and
GDP / Energy
emerging markets
consumption in
Italy and Korea
[57]
Sweden
Economic growth /
Energy consumption
[12]; [13]
France,
Economic growth /
Malaysia
Energy consumption
B: Studies in favor of a unidirectional causality from energy
consumption to economic growth
Energy
[136]
USA, UK,
consumption / GNP
Poland, Korea,
in Korea and
The Philippines
the Philippines
[142]
USA
Energy
consumption/ Income
Energy consumption /
[79]
India, Pakistan,
Real income in India
Malaysia, Singapore,
Indonesia,
The Philippines
[32]
Brazil, Mexico
Energy
and Venezuela
consumption / GDP
in Brazil
[19]
Asian developing
Energy
countries
consumption / Income
in India and Indonesia
[31]
Taiwan
Energy
consumption / Output
[120]
G-7 countries and
Energy
emerging markets
consumption / GDP
in Turkey, France,
Germany and Japan
[140]
Shanghai
Energy
consumption / Real GDP
[73]
Developing
Energy consumption / GDP
countries
[89]
Tanzania
Energy consumption/ GDP
[72]
Energy consumption to
economic growth
[129]
China
Energy consumption/ GDP
C: Studies in favor of a bidirectional causality between energy
consumption and economic growth
[80]
2 NICs: Korea
Energy consumption 4
and Taiwan
Real income
[52]
South Korea
Energy consumption 4 GDP
and Singapore
[19]
Asian developing
Energy consumption 4 Income
countries
in the Philippines and Thailand
[51]
Korea
Energy consumption 4
Economic growth
[120]
G-7 countries and
Energy consumption 4
emerging markets
GDP in Argentina

Table 1 (continued )
Author(s)

Region/country

Direction of causality

[151]

India

[131]

17 African countries

[134]

4 Asean countries

[90]

South Africa

[130]

Liberia

[46]

[91]

Portugal, Italy, Greece,


Spain and Turkey
60% of Latin American
and Caribbean countries
14 MENA countries

[115]

Angola

Energy consumption 4
Economic growth
Electricity consumption 4
Economic growth in Egypt,
Gabon and Morocco
Electricity consumption 4
Economic growth in
Malaysia and Singapore
Electricity consumption 4
Economic growth
Electricity consumption 4
Economic growth
Electricity consumption 4
Economic growth
Energy consumption 4
Economic growth
Energy consumption 4
Economic growth
Electricity consumption4
Economic growth

[7]

Notes: /, 4 denote unidirectional causality, bidirectional causality, respectively.

reported that per capita GDP increases CO2 emissions and energy
consumption. Ref. [93] studies the relationship between income,
energy consumption, emissions .and employment in Turkey. They
found that neither carbon emissions per capita nor energy consumption per capita Granger-cause real GDP growth per capita [49]
nds no long run causality between economic growth and CO2
emissions and bivariate short run causality in India. Ref. [97]
examine the dynamic relationship between pollutant emissions,
energy consumption and the output for Brazil. They nd that in the
long run, emissions appear to be output inelastic and energy use
elastic. The causality results indicate that there is a bi-directional
strong causal link between income, energy consumption and CO2
emissions.
Ref. [96] applies the coingration technique and causality test to
examine the dynamic relationships between emissions, energy use
and real output using time series data for the period 1990e2007 for
Russia. They nd no support for the EKC hypothesis. However, the
study further reveals that there exist strong bidirectional causal
relationship between output, energy use and emissions. The ndings recommend that Russia should pursue such environmental
policy that would increase infrastructure investment to improve
energy efciency and step up energy conservation policies to avoid
unnecessary wastage of energy. Ref. [2] investigates the short- and
the long-run equilibrium relationship between CO2 emissions,
economic growth, technical efciency and industrial structure for
Ghana, Senegal and Morocco. The ndings suggest a multiple longrun relationship between the variables for Ghana and Senegal
while one way long-run relationship for morocco. The Toda and
Yomamoto Granger causality test shows that CO2 emissions were
not found to Granger cause in Senegal but it is found as a limiting
factor for the other two countries. The variance decomposition
analysis used in the study reveal that economic growth contributes
to future CO2 emissions in all these countries.
Ref. [50] conducts a time series analysis for India using data for
the period 1971e2008. They employ Threshold cointegration tests
complemented by ARDL (Autoregressive Distributed Lag) and
Johansen and Jusilius maximum likelihood procedure. Their results
indicate long-run cointegrating relationship among the variables
having endogenous structural breaks or regime shifts. TY (TodaYamamoto) Granger causality results support unidirectional causal
link running from energy consumption to economic activity and
from economic activity to urbanization. The study recommends

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

that rapid urbanization and migration of Indian people to urban


areas should be an integral part of energy policies of India.
For the panel studies [109], in a panel of select Southeast Asian
nations use ARDL (Autoregressive Distributed Lag) model and
Granger causality test based on VECM (Vector error correction
model) to examine the empirical link between energy consumption, CO2 emissions and economic growth. They nd signicant
positive relationship between CO2 emissions and energy consumption both in the short- and the long-run for all the countries.
They also nd that CO2 emissions increase over time with respect to
energy consumption in the region. The study further nds evidence
in support of EKC hypothesis for Singapore and Thailand. There is
also long-run bi-directional causal link between CO2 emissions and
economic growth in all the ASEAN (Association of Southeast Asian)
countries. [126] examine the relationship in a panel of 98 countries
to nd a threshold effect between the two variables. The results do
not support the EKC hypothesis.
Ref. [110] investigates the relationship using a national panel
dataset for Iran and nd that there exist a strong positive correlation between energy consumption and CO2 emission. Ref. [22] in a
panel of 25 OECD countries suggest that nancial development
dominate the long run relationship between energy consumption
and real GDP in the region. [98] in a panel of BRICS (Brazil, Russia,
India, China and South Africa) countries recommend the methods
of managing both energy demand and FDI and increasing investment both in energy supply and energy efciency to reduce CO2
emission without compromising these fast growing countries'
competitiveness.
Ref. [16] investigates this relationship for six Central American
economies using panel VECM (Vector error correction mechanism).
It is evident from that study that energy consumption is positively
linked with CO2 emissions. Ref. [72] reaches the same conclusion
for the case of ASEAN (Association of Southeast Asian Nations)
countries. Empirical evidence from Ref. [83] also supports the EKC
hypothesis for 43 low income countries. In addition [72], note long
run causality running from energy consumption and CO2 emissions
to economic growth in a study of ASEAN countries. On the other
hand [14,15], nd that energy consumption and economic growth
cause CO2 emissions while bivariate causality exists between energy consumption and economic growth and between energy
consumption and CO2 emissions in the Commonwealth of Independent States.
Ref. [35] in a study on BRICS countries nd support for neutrality
hypothesis for Brazil, India and China indicating no link between
electricity consumption and economic growth. However, regarding
GDP-CO2 emissions nexus, a feedback hypothesis for Russia, a oneway Granger causality running from GDP to CO2 emissions and a
reverse relationship were found for Brazil resulting in inconclusive
policy implications. Ref. [112] nds positive relationship between
economic growth and energy consumption. The authors also nd a
negative impact of CO2 emissions on energy demand in a time series study involving the UAE. Ref. [5] nds positive relationship
between CO2 emissions and economic growth in case of Saudi
Arabia. They also conclude that electricity is less polluting than
other sources of energy.
Ref. [115] shows a bi-directional causal link between electricity
consumption and economic growth in Angola. Ref. [25] nds that
economic growth drives not carbon emissions rather its intensities
in the USA. Ref. [10] examines the intensity of CO2 emissions in
Portugal using 'complete decomposition' technique. It shows that
CO2 emissions intensity declined signicantly during the period
1996e2009. The authors attribute this decline to the substitution of
fossil fuels for less polluting fuels. The study further conrms the
bidirectional causality between CO2 emissions intensity and the
share of fossil fuels in total energy consumption.

47

Ref. [95] employs NGBM (non-linear grey Bernoulli model) to


forecast CO2 emissions, energy consumption and economic growth
in China. The cointegration ndings indicate the long-run equilibrium relationship among CO2 emissions, energy consumption and
economic growth in China. CO2 emissions are found to be output
inelastic while it is energy consumption elastic. The study further
advocates that China should adopt the dual strategy of increasing
energy efciency, reducing the loss of power transmission and
distribution and strengthening its conservation policies to avert any
wastage of energy.
Ref. [152] uses scenario analysis approach and projects China's
possible CO2 emissions during the period from 2010 to 2020. [77]
employs TY (Toda-Yamamato) causality analysis to investigate
economic growth, CO2 emissions and fossil fuels consumption in
Iran. Their empirical ndings support the unidirectional causal
link running from GDP and energy consumption to CO2 emissions. Na causal link is found running from fossil fuel consumption to CO2 emissions in the long-run. The results further
indicate that CO2 emissions, energy consumption and fossil fuel
consumption do not cause economic growth but gas consumption does. Based on the ndings, the study recommends that
China should adopt a low carbon industrialization model for the
sustainability of its economy.
The literature studying the relationship between CO2 emissions
and economic growth involving only GCC countries is rare although
there have been several studies on APEC, (Association of Petroleum
Exporting Countries) MENA (Middle East and North African) and
high income countries which included all or some of the GCC
countries. Ref. [119] investigates the relationship between carbon
emissions, income energy and total employment in ve selected
OPEC countries including Saudi Arabia. They nd a cointegrating
relationship between the variables. Ref. [65] tests the EKC hypothesis for 36 high income countries including three MENA
countries; Bahrain, Oman and UAE. The ndings indicate that CO2
emissions have fallen with the rise in income in the long run.
Ref. [18] in a study on 12 MENA countries show that energy
consumption has a positive signicant impact on CO2 emissions
and that real GDP demonstrates a quadratic relationship with CO2
emissions. Ref. [91] in a study on 14 MENA countries nds that
there is a bi-directional causal link between energy consumption
and economic growth in the region. Ref. [75] shows that there is a
positive long-run relationship between real GDP per capita and
energy consumption variables. Ref. [76] in a large panel of 132
countries nd that factors like urbanization, population density,
trade, energy mix, and economic environment impact the absolute
level of carbon dioxide emissions. Ref. [7] in a panel study of Latin
American and the Caribbean countries nd that 60% of those
countries have a positive bi-directional long run relationship between energy consumption, CO2 emissions and economic growth.
Ref. [30] nds a bi-directional causal link between energy consumption and GDP and between GDP and CO2 emissions. It further
suggests that increased energy consumption will stimulate CO2
emissions.
2.3. Decoupling effect
Another strand of literature has focused on the decoupling issue.
Ref. [148] rst proposed the concept of 'decoupling' which occurs
when the growth of environmental pressure is slower than the
economic growth. Ever since, OECD countries have attached great
importance to the research on the decoupling theory and its
application dividing the decoupling concept into relative decoupling and absolute decoupling. Relative decoupling occurs when
emissions grow at a slower rate than economic growth while absolute decoupling happens when emissions decline with the

48

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

economy growing. Ref. [149] proposed the concept of primary


decoupling, secondary decoupling and double decoupling. According to him, primary decoupling refers to delinking natural resources consumption from economic growth while secondary
decoupling is decoupling of environmental pollution from consumption of natural resources. Double decoupling occurs when
primary
decoupling
and
secondary
decoupling
occur
simultaneously.
Empirical research focusing on decoupling issue is scarce. [138]
examines the occurrence of decoupling between growth rates in
economic activity and CO2 emissions from energy consumption
during 1995e2009 in Jiangsu province which is one of the most
developed regions in China. They show that during the study
period, Jiangsu experienced weak decoupling and strong decoupling except during 2003e2005. The decoupling states for the
secondary and tertiary industries are similar to that of the whole
economy. Ref. [128] in another study in the same region show that
economic activity is the critical factor in the growth of energy
related CO2 emissions and the energy intensity effect plays vital
role in reducing CO2 emissions.
Ref. [11] uses decomposition analysis to assess the progress in
decoupling economic growth from CO2 emissions in Italy. They split
data for the periods of 1998e2002 and 2002e2006. The study
considers ve key sectors and four explanatory variables, CO2
emissions, CO2 intensity, energy intensity, structural changes and
economic activity. The ndings indicate that Italian economy did
not perform absolute decoupling during both the periods in terms
of economic growth and CO2 emissions and that economic growth
and energy intensity are mostly responsible for CO2 emissions. The
highest level of decoupling is observed in 2009.
Ref. [150] investigate the trend of decoupling effects in nonferrous metal industry in China. The study observes four decoupling stages in the industry; strong negative decoupling stage
(1996e1998), weak decoupling stage (1999e2000), negative
decoupling stage (2001e2003) and weak decoupling stage
(2004e2008). The study further suggests that the rapid growth of
the industry is the most important factor responsible for the increase of CO2 emissions. The increase in electric energy consumption contributed to increased CO2 emissions. Ref. [44] examines the
state of decoupling of growth and CO2 emissions from energy
consumption during the period 2004e2009 in Brazil. Using
decomposition analysis based on LMDI (Log-mean Divisia Index),
the study nds that carbon intensity and energy mix are the two
factors that need to be addressed to curb CO2 emissions. They also
observe several periods of relative decoupling in the country.
[69,70] and [71] found that use of larger vehicles, increased average
loads and empty running were the key factors that contributed
towards the decoupling of trafc growth from economic growth in
Denmark. Refs. [81] and [117] had similar ndings for the United
Kingdom.
2.4. Inequality in CO2 emissions
Inequality in CO2 emissions emerged as an issue of central
climate debate. It has been extensively studied and investigated
through various methods. Ref. [58] used Atkinson index [21]. His
ndings indicated a declining trend in global CO2 emissions
inequality during 1960 and 1999. Ref. [153] used Theil Index
[125] for inequality and [67] to decompose emission inequality
across countries in terms of carbon intensity of energy, energy
intensity and afuence. They found that income differences were
mainly responsible for emissions inequality between countries.
Ref. [40] uses the method to analyze the contribution of sectoral
changes and energy intensity to the inequality of nal pare capita
energy consumption among OECD countries. They nd that the

diffusion of energy efcient technologies contributed to declining


energy intensities in among these countries leading to a fall in
per capita energy consumption in OECD countries. Ref. [26]
studies how projected emissions inequality from climate
change assessment models is and predicts that GDP per capita
differences would be responsible for future emission
concentration.
Ref. [41] uses ecological footprint and nd that the key
contributor for inequality in per capita ecological footprint is the
difference in GDP per capita. Ref. [143] used Gini Index and show
that emissions inequality is closely related with income
inequality. Ref. [27] uses the Gini Index to analyze emissions
inequality from climate policy scenarios. They nd that tradable
permits matter for future emission inequality. They also nd
strong association between inequality in GDP and emissions.
Ref. [39] compares a set of inequality measures such as Gini Index, Theil index, Atkinson index as well as the coefcient of
variation method. He concludes that different methods provide
different results due to differences in their distributive sensitivity. Ref. [74] decomposes the Gini coefcient in a within and
between group component. In our study, we apply this method
to decompose CO2 emissions inequality by its source of different
energy carriers and as well as sectors.
3. Data and methodology
3.1. Empirical relationship between economic growth, energy
consumption and CO2 emissions
3.1.1. Data
The model based on variables in natural logarithms for estimation took the form of:

Cit bi b1i Eit b2i Yit it

(1)

The coefcients, b1i and b2i represent the long run elasticity
estimates of CO2 emissions with respect to energy consumption
and per capita GDP. According to the discussion above, an increase
in energy consumption and income are expected to cause an increase in CO2 emissions.
To conduct the analysis and investigate the relationship between CO2 emissions, energy consumption and economic growth,
values for the following variables for the GCC countries are
required.
- CO2 emissions (C)
- Energy consumption (E)
- Per capita real GDP (Y)
Annual data for the period 1980e2012 was obtained from the
World Development Indicators database 2013 for all GCC countries: Saudi Arabia, Bahrain, Kuwait, UAE, Oman and Qatar [132].
For income (Y), real GDP per capita which is measured at constant 2000 US$ was used, per capita energy use (kilotons of oil
equivalent) and per capita CO2 emissions were obtained by
dividing CO2 emissions by the mid-year population. All variables
were transformed to natural log form in order to avoid the
problems of non-linear modeling and heteroscedasticity and to
obtain the growth rate of the relevant variables by their differenced logarithms.
The estimation of the model proceeds as follows: (i) a CD (crosssectional dependence) test is performed to verify whether there is
cross-sectional dependence across the panel; (ii) recognizing the
presence of cross-sectional dependence, an appropriate panel unit
root test (i.e. CIPS (cross-sectionally augmented IPS)) is conducted

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

to examine whether the series are stationary; (iii) the Pedroni


cointegration test is implemented to verify the long-run relationship among the variables; (iv) a PMG (pooled mean group estimation) estimation is employed to examine the short-run and longrun relationships among the variables and the speed of adjustment
of the short-run disequilibrium towards the long-run equilibrium
and (v) a SUR is estimated to check the robustness of the PMG
results.
3.1.2. Testing for unit roots
While dealing with long time-series data, long-run parameters
are conventionally expected to exhibit cointegrating relationships
among a set of I(1) variables [20]. Because the time period of the
dataset is fairly long (33 years), it is expected that the macroeconomic variables in the model will be characterized by a unit root
process [86]. Therefore, it is necessary to check the variables for the
order of integration before examining any long-run relationship.
Hence, unit root tests for all variables in the dataset are imperative.
However, it is extremely likely that there will be cross-sectional
dependence among the GCC countries. This may be due to some
shocks such as global nancial crisis or oil price shock, which affects
all countries but to various magnitude. To verify the existence of
contemporaneous correlation across individuals in the panel, a CD
(cross-sectional dependence) test developed by Ref. [102] to
examine the null hypothesis of cross-sectional dependence among
individuals in the panel was applied. Ref. [102] denes the CD
statistic as:

CD



TNN  1 1=2
b
r;
2

where


b
r

2
NN  1

3.1.3. Testing for panel cointegration


Because the results from the CIPS unit root test indicate cointegrating relationships in the dataset and because the primary
focus is to investigate the long-run relationship between CO2
emissions, economic growth and energy consumption, several
panel cointegration tests suggested by Refs. [100,101] were conducted. The key advantage of the Pedroni cointegration test over
similar tests is that it controls for country size and heterogeneity
allowing for multiple regressors (as in this case) for the cointegration vector to vary across various sections of the panel. [100]
provides seven panel cointegration statistics for seven tests. Four
of these are based on the within-dimension tests, and the other
three are based on the between-dimension or group statistics
approach.
The starting point for the residual-based panel cointegration
test statistics of [101] is the computation of the residuals of the
hypothesized cointegrating regression as follows:

yi;t ai b1i ; x1i;t b2i x2i;t / bMi xMi;t i;t

N1 XN
i1

b
r
ji1 ij

Dyit ai ki t bi yit1 gi yt1 fi Dyt it ;


t 1::::::::::::::T and i 1:::::N;

(2)

where,

XN

where T is the number of observations over time, N denotes the


number of individual members in the panel, and M is the number of
independent variables. It is assumed here that the slope coefcients
b1i,bMi and the member-specic intercept ai can vary across each
cross-section. To compute the relevant panel cointegration test
statistics, the panel cointegration regression in (1) should be estimated rst. For the computation of the panel-r and panel-t statistics, the rst difference of the original series is taken and used in
the estimation of the residuals of the following regression:

CIPSN; T N 1

N
X

ti N; T

i1

where ti (N, T) is the t statistic of bi in equation (2). The critical


values of CIPS (N, T) are available in Table 2(c) of Ref. [103].

(4)

Using the residuals from the differenced regression with a [87]


b 2i;t , which is symbolized as
estimator, the long run variance of p
2
b
L 11i was calculated:
2
1
b
L 11i
T

T
X

b 2i;t
p

t1

 X
ki 
T
2X
s
b p
b
1
p
T s1
ki 1 ts1 i;t i;ts

For panel-r and group-r statistics, estimation of the regression


b i;t and using the residuals b
bib
using the b
e i;t g
e i;t from the
e i;t1 u
cointegration regression (2) was made. Then computing the long2
b it
run variance (b
s 2i ) and the contemporaneous variance (bs i ) of u
where,
2
b
si

T
1X
b
u
T t1 i;t

b
s 2i


ki 
T
1X
2X
s
b it
1
u
T t1
T s1
ki 1

y
i1 it

is the cross-sectional mean of yit. The purpose of including the


cross-sectional mean in the above equation is to control for
contemporaneous correlation among yit. This is a modied version
of the IPS [62] test that is referred to as the CIPS (cross-sectionally
augmented IPS) test [103]. The null hypothesis of the test can be
expressed as H0: b1 0 for all i against the alternative hypothesis
H0: b < 0 for some i.
The test statistic provided by Ref. [103] is given by:

(3)

t 1; ; T; i 1; :; N; m 1; ; M

yi;t b1i ; Dx1i;t b2i Dx2i;t / bMi DxMi;t pi;t

X

in which b
r ij is the pair-wise cross-sectional correlation coefcients
of residuals from the conventional ADF regression; T and N are
sample and panel sizes, respectively.
Because it indicates a cross-sectional dependence in the panel,
the following CADF (cross-sectionally augmented Dickey-Fuller)
regression was used:

yt N 1

49

T
X

b i;t u
b i;ts the train platform no
u

ts1

where ki is the lag length. In addition to this, calculation of the


term:

li


1 2
2
b
si  b
si
2

On the other side, for panel-t and group-t statistics using again
the residuals of b
e of cointegration regression (1), estimation of
Pi;t
b
b *i;t was made.
b ik Db
bib
e i;t g
e i;t1 kt1 g
e i;t1 u

50

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

In this study, the step-down procedure and the Schwarz lag


order selection criterion are used to determine the lag truncation
order of the ADF t-statistics:
*2
b
si

T
1X
b *2 ;
u
T t1 i;t

bs *2
i;t

N
1 X
*2
b
si
N
i1

The next step is the calculation of the relevant panel cointegration statistics using the following expressions.
Paneler statistic
N X
T
p
p X
2 2
b
T NZbr N;T1 T N
e i;t1
L 11i b

!1

i1 t1


2 
b
e i;t  b
li
L 11i b
e i;t1 Db

N X
T
X
t1

Panelet statistic
*
ZtN;T

~s*2
N;T

N X
T
X

2 2
b
e i;t1
L 11i b

!1=2

i1 t1

N X
T
X

2
b
e i;t
e i;t1 Db
L 11i b

t1

Grouper statistic
1=2
~
TN 1=2 Z
br N;T1 TN

N
X

T
X

i1

t1

!1
2
b
e i;t1

T 
X

b
e i;t  b
li
e i;t1 Db

t1

Groupet statistic

~ * TN1=2
N 1=2 Z
t;N;T

N
X

T
X

i1

t1

bs *2
b2
i e i;t1

!1=2

T
X

b
e i;t
e i;t1 Db

that consider different levels of heterogeneity across countries


while estimating both the short run and the long run effects along
with the speed of short-run adjustment towards the long run
equilibrium.
The PMG (pooled mean group estimation) technique is
employed in this study. The justication for employing this technique is that GDP per capita growth in GCC countries is expected to
be affected by the long-run homogeneous conditions while the
short-run adjustment depends on country-specic characteristics
such as vulnerability to domestic and external shocks (for example,
the recent global nancial crisis), monetary and scal adjustment
mechanisms. nancial-market imperfections, change in political
regime, etc.
Refs. [106] and [104] show that PMG can render consistent and
efcient estimates of the parameters in a long-run relationship
between both integrated and stationary variables and that inference on these parameters can be conducted using standard tests.
Furthermore, these methods do not require pre-testing of the
order-of integration conformability given that they are valid
whether the variables of interest are I(0) or I(1). [105] term this as
the ARDL (autoregressive distributed lag) approach to long-run
modeling. In the ARDL, the lags of the dependent and the independent variables are distributed over different periods of time. An
important assumption for the consistency of the PMG estimates is
that the residual are serially uncorrelated. In practice, contemporaneous correlation across residuals arises from omitted common
factors. To eliminate the inuence of these common factors, timespecic effects in the estimated regression are allowed for. In order to comply with the requirements for standard estimation and
inference, equation (1) is embedded into an ARDL (p, q) model. In
error correction form, this can be written as follows:

t1

Lastly, the appropriate mean and variance adjustment terms to


each panel were applied to the cointegration test statistic so that
the test statistics are standard normally distributed:

p
cN;T  m N
p
> N0; 1
v
where cN,T is the appropriately standardized form of the test statistic and the functions of moments of the underlying Brownian
motion functionals. The appropriate mean and variance adjustment
terms for the various number of regressors (m is the number of
regressors without taking the intercept into account) and the
various panel cointegration test statistics are given in Table 2 in Ref.
[101].
The null hypothesis of no cointegration for the panel cointegration test is the same for each of the statistics: H0: gi 1 for all i.
The alternative hypothesis for the between-dimension-based and
within-dimension-based panel cointegration tests differ. The
alternative hypothesis for the between-dimension-based statistics
is H1: gi < 1 for all i, where a common value for gi g is not
required. For the within-dimension-based statistics, the alternative
hypothesis is H1: g gi < 1 for all i, and it assumes a common value
for gi g. Under the alternative hypothesis, all of the panel cointegration test statistics considered in this paper diverge to negative
innity. Thus, the left tail of the standard normal distribution is
used to reject the null hypothesis.
3.1.4. Pooled mean group testing
One shortcoming of the Pedroni tests is that these tests do not
estimate for the short-run relationship and the speed of adjustment
of the short-run disequilibrium towards the long run equilibrium
relationship [82]. A number of alternative methods are available

Dyi t

p1
X

gij Dyi tj

j1

q1
X

h
oi
dij Dxi tj 4i yi t1  bi1 Xi t1

j0

bi0 mt it
(5)
where yi is the dependent variable (CO2 emissions), Xis are independent variables (energy consumption and economic growth), gij
and dij are short run coefcients, 4i is the error correction adjustment speed, bi1 are the long- run coefcients, bi0 , mt and it are
country-specic xed effects, time-specic effects and stochastic
error term respectively.
The use of Pooled Mean Group regression in the panel is quite
justied as it is expected that all the countries in the GCC have some
common long-run parameters that brought them under one umbrella. They have similarities in terms of their economic and
monetary structures. Therefore, it is not unrealistic to assume homogeneity in the long-run parameters across the panel of these
countries. However, there may be different types of short-run
shocks in different GCC countries due to local laws, regulations
and political regimes allowing for short-run country heterogeneity.
3.1.5. SUR (seemingly unrelated regression) testing
Since the number of cross sections (N 6) is less than 10,
another technique known as SUR (seemingly unrelated regression)
is applied to check for the robustness of the results.
3.1.6. Panel Granger causality test
Since all the variables in our study are rst difference stationary
[I(1)], this study proceeds further to determine the causal link between them [146]. The exact direction of causal link helps with
better policy implications of the ndings [155].

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

3.2. Decoupling of energy consumption from economic growth


3.2.1. Data
We employ historical data from the International Energy Agency
[147] on per-capita CO2 emissions over the period from 1971 to
2008 and emissions scenario data over the period from 1980 to
2012. The IEA dataset contains data on CO2 emissions disaggregated
by primary energy carriers in GCC (oil and gas), whereas, the data
on economic sectors (manufacturing, transport, and residential) are
obtained from WDI (World Development Indicators database).
3.2.2. Methodology
Following Ref. [23]; we estimate decoupling index for energy
and GDP per capita ratio. The DI (decoupling index) refers to the
ratio of the change in the rate of consumption of a given resource, to
the change in the rate of economic growth (in terms of GDP), within
a certain time period (typically one year). DI (decoupling index) is
estimated from the following formula;

DI

Et  Et1 =Et1
GDPt  GDPt1 =GDPt1

When DI > 1, no decoupling is taking place.


When DI 1, it is the turning point between absolute coupling
and relative decoupling is represented.
When 0 < DI < 1, relative decoupling is taking place.
When DI 0, it is implied that the economy is growing while
resource consumption remains constant. This is the turning
point between relative and absolute decoupling.
When DI < 0, the relationship can be described as absolute
decoupling.
3.3. Methodology for the analysis of inequality in CO2 emissions
across GCC countries
3.3.1. The Gini decomposition
Following Ref. [74] we estimate decomposed Gini coefcient for
CO2 emissions across the GCC countries.
Let,

y x; k 1; 2; 3; K while mk mean of xk ; and Fk


cumulative distribution of xk

51

of total emissions (i.e. a large Sk) has a more important impact on


overall inequality. For instance, an innitesimal small emission
source can only have an innitesimal small effect on overall
inequality, even if its distribution is highly unequal.
Finally, the inequality and share of any emission source play
more important roles, the higher their correlation with overall
emissions (i.e. the higher Ck) are. This is because, with low correlations countries featuring high emissions from this source do not
necessarily also have high emissions from other sources. For
instance, an emission source could be unequally distributed across
countries in a way that it is higher for countries having comparatively low levels of emissions from other sources and vice versa (i.e.
Ck 0). In this case, the negative correlation would actually work
towards a decrease in overall inequality.
3.4. Laspeyres decomposition
In order to gain a deeper understanding of the factors driving
the evolution of overall inequality, we employ the Laspeyres
decomposition method [123]. The Laspeyres decomposition allows
us to break down changes in overall emissions inequality to
changes in its single components G, S, and R. With D denoting the
difference between the year 2010 and 1980, we can decompose the
change in the Gini index of total per-capita emissions between two
points in time into the joint contribution of three underlying effects
E for every source of emissions.

DG

K
K
X
X
Ck DCk Gk DGk Sk DSk 
Ck Gk Sk
k1

k1

This can be expressed as:

DG

K 
X

EkC EkG EkS

k1

The individual effects can be derived from carrying out the


multiplication in the above equation and dividing the residuals (i.e.
changes of second and third order) evenly across factors. This is
demonstrated exemplarily for EkS , i.e. the change in inequality that
can be attributed to a change of the share of source k in total percapita emissions, below:

EkS DSk Gk Ck

1
1

2DSk DGk Rk Gk DCk 3DSk DGk DRk

my mean of y and Fy cumalative distribution of y


Then, we can show the overall Gini coefcient of CO2 emissions
as follows

4. Results and discussion

i
h
h

u1 i
ucov xk ; 1  Fk u1 m
K cov xk ; 1  Fy
P
k
i
h
Gu
u1
mk
my
k1 cov xk ; 1  Fk

In this section, at the beginning, the ndings on the empirical


relationship between energy consumption, economic growth and

K
P

Table 2
Descriptive statistics.

Ck Gk Sk

k1

where Ck is rank correlation between emission component k and


total emission, Gk is the Gini of component k, and Sk is the share of
component k in total emissions. Therefore, the contribution of each
single source k to overall inequality in per-capita emissions is given
by CkGkSk. This decomposition is rather intuitive. Firstly, the more
unequally emissions in any source k are distributed (i.e. the higher
the Gk), the more it matters for total inequality G, however, if their
distribution was entirely equal (i.e. Gk 0), it would not affect the
overall inequality. Secondly, a source that constitutes a large share

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis
Jarque-Bera
Probability
Sum
Sum Sq. Dev.
Observations

LCO2

GDPPC

LEU

3.041997
3.097144
4.227344
1.492101
0.601584
0.451316
2.996736
6.721719
0.034705
602.3154
71.29498
198

8.914616
9.048599
10.04801
6.688322
0.663755
1.016824
4.124082
44.54410
0.000000
1765.094
86.79255
198

10.01649
10.02496
11.31383
8.711521
0.630448
0.096702
1.626723
15.86718
0.000358
1983.264
78.30043
198

52

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

Table 3
Panel unit-root test results.

Table 5
Pooled mean group results.

Panel unit root test

LGDPC
LCO2
LEU

CO2 emissions (dependent variable)

Pooled mean group

CD

CIPS

CIPS (1st diff.)

Variable

Coefcient

Standard error

0.488
0.298
0.352

2.09*
0.52
4.05***

1.501
1.747
0.876

2.637***
2.664***
3.156***

Long-run coefcients
GDPPC
Per capita energy use
Error correction Coefcient
D GDPG
D Energy Use
Intercept

.2606841
.7011095*
.3528844*
.0348034
.5963364*
2.060458*

0.1218678
0.1918241
.0587911
0.2997785
0.1340185
0.3560433

Notes: *, **, *** demonstrate level of signicance at 1%, 5% and 10% respectively.

CO2 emissions are presented. Table 2 presents the descriptive statistics which shows that the data are fairly dispersed. The standard
deviations show that the data are considerably homogeneous.
Table 3 reports the results of CD test and CIPS unit root test. CD
test conrms that there is cross sectional dependence in two of the
three series (GDPC and energy consumption). CIPS unit root test
suggests that all the variables are stationary at rst difference, i.e.
I(1).
Results from the Pedroni panel cointegration test are presented
in Table 4. It is evident that the statistical values of six out of seven
tests indicate the rejection of null hypothesis of no cointegration.
Nevertheless, among the seven test statistics, group rho statistic
has the best power [54] which also reject the null of no cointegration. Thus, it can be concluded that there is long run cointegrating relationship among the variables. Therefore, it was not
required to conduct a panel cointegration test with structural
breaks.
Table 5 presents results from the PMG estimates. Overall results
suggest that the long run coefcient of energy consumption to CO2
emissions is 0.15 and signicant at 1% level. In other words, energy
consumption enhances CO2 emissions by 15% in the long run. There
is also signicant positive long run relationship between energy
consumption and CO2 emissions. There is no signicant relationship between per capita GDP and CO2 emissions both in the shortand the long-run. Energy consumption has a positive and signicant (at the 5% level) short run effect on CO2 emissions. The error
correction term in the short run is 0.154 at the 1% level of signicance suggesting that approximately 15% of the deviations from
the long run are corrected each year.

Table 4
Pedroni residual cointegration test results.
Pedroni residual cointegration test
Series: LCO2 LEU LGDPC
Date: 12/14/13, Time: 17:45
Sample: 1980 2012
Included observations: 198
Cross-sections included: 6
Null hypothesis: no cointegration
Trend assumption: no deterministic trend
Lag selection: Automatic SIC with a max lag of 7
NeweyeWest bandwidth selection with Bartlett kernel
Alternative hypothesis: common AR coefs. (within-dimension)

Panel
Panel
Panel
Panel

v-Statistic
rho-Statistic
PP-Statistic
ADF-Statistic

Statistic

Prob.

0.266888
1.867851
2.697601
2.703817

0.6052
0.0309
0.0035
0.0034

Table 6 reports results from the SUR estimates. The ndings are
supportive of the PMG results indicating signicant positive relationship between energy consumption and CO2 emissions and
between economic growth and energy consumption. Therefore, the
overall ndings are robust from PMG estimates are robust.
Panel Granger causality ndings are reported in Table 7. The
results suggest bidirectional causal link between energy consumption and CO2 emissions and unidirectional causal link running
from economic growth to energy consumption. No causal link is
found between economic growth and CO2 emissions in the GCC
region.
This follows the presentation of the ndings on decoupling index. Fig. 1 shows estimates of the DI (decoupling index) for E/GDP
per Capita ratio for the years from 1980 to 2012. The DI estimated
from the E/GDP ratio shows all the signs of absolute decoupling,
relative decoupling and no decoupling in GCC countries during the
period of study. In the case of Bahrain, there was absolute decoupling from 1980 to 1992. No decoupling is observed during
1993e2003. It performed relative decoupling during 2004e2005.
Again there was no decoupling during 2006e2008 followed by
relative decoupling in 2009. The country experienced absolute
decoupling during 2010e2012. Qatar experienced relative decoupling for the longest period from 1981 to 2000. During 2001e2004,
it was in the state of absolute decoupling before performing relative
decoupling again during 2005e2006. However, during the period
2007e2011, Qatar was in the state of absolute decoupling.
Oman experienced absolute decoupling during the period from
1981 to 2009 while it performed relative decoupling during
2010e2012. Surprisingly, no decoupling was performed in Qatar for
the entire period of the study although Qatar has been one of the
GCC countries that was pursuing various initiatives to reduce CO2
emissions. Saudi Arabia experienced absolute decoupling for the
whole period of the study. The UAE (United Arab Emirates) performed absolute decoupling during 1981e1982 followed by relative
decoupling in 1983. The country again performed absolute decoupling during 1984e2009. During the period 2010e2012 also, the
UAE experienced relative decoupling. Absolute decoupling was
observed in Kuwait in 1981. This followed relative decoupling for a
long period from 1982 to 1990. Again during 1991e1994, absolute
decoupling was performed in the country. Kuwait experienced the

Weighted
Statistic

Prob.

1.165996
2.164126
3.089946
3.334316

0.8782
0.0152
0.0010
0.0004

Alternative hypothesis: individual AR coefs. (between-dimension)

Group rho-Statistic
Group PP-Statistic
Group ADF-Statistic

Notes: *, **, *** demonstrate level of signicance at 1%, 5% and 10% respectively.

Statistic

Prob.

0.571742
2.197197
2.258127

0.2837
0.0140
0.0120

Table 6
Panel Granger causality tests.

D (LCO2)
D (LCO2)
D (LEU)
D (LGDPC)
ALL

10.736***
1.781
13.350***

Prob.
0.004
0.410
0.009

D (LEU)

Prob.

D (LGDPC)

Prob.

37.258***

0.000

1.548
0.612

0.461
0.736

6.325**
41.396***

0.042
0.000

1.574

0.813

Notes: The asterisks ***, ** and * denote the signicance at the 1, 5 and 10 per cent
level, respectively.

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

inequality in CO2 emissions rose while it signicantly fell to almost


equality level in 2010. It can be observed from the gure that CO2
emissions inequality was more prominent in 1980 in the transport
sector. However, it experienced a declining trend after 1980. In

Table 7
Seemingly unrelated regression results.
Dependent variable: LCO2
Method: panel least squares
Date: 12/14/13, Time: 17:46
Sample (adjusted): 1981 2012
Periods included: 32
Cross-sections included: 6
Total panel (balanced) observations: 192
Period SUR (PCSE) standard errors & covariance (d.f. corrected)
Coefcient

Std. Error

t-Statistic

LGDPC
LGDPC(-1)
LEU
LEU(-1)
C
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob (F-statistic)

0.195043
0.293426
0.648409
0.131958
4.906600
0.884002
0.881521
0.205275
7.879778
34.11058
356.2755
0.000000

0.360297
0.541338
0.366202
0.801267
0.082936
7.818147
0.079221
1.665694
0.376480
13.03284
Mean dependent var
S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
DurbineWatson stat

Bahrain: Decoupling Index CO2 Emission/ GDP per Capita


1.2
1
Prob.

0.8

0.5889
0.4240
0.0000
0.0974
0.0000
3.042047
0.596370
0.303235
0.218405
0.268878
0.672063

0.6

Axis Title

Variable

53

0.4
0.2
0
-0.2

-0.4

Axis Title

Kuwait:Decoupling Index CO2 Emission/ GDP per Capita


2.5
2

Axis Title

1.5
1
0.5
0
-0.5

Axis Title

Oman: Decoupling Index CO2 Emission/ GDP per Capita


1.6
1.4

Axis Title

1.2
1
0.8
0.6
0.4
0.2
0
-0.2
Axis Title

Qatar: Decoupling Index CO2 Emission/ GDP per Capita


1.2
1
0.8

0.6
Axis Title

longest relative decoupling for about a decade and a half during


1995e2011. The overall decoupling effects reveal that almost all
GCC countries experienced some shifts towards higher quality energy sources and increasing energy intensity and the dependence
of economy on natural resources is declining with the passage of
time.
This section gives an overview of changes in per-capita emissions and the associated Gini index between 1980 and 2011 for GCC
countries. Further, in order to gain a better understanding of the
underlying drivers of the developments, it decomposes overall
inequality in per-capita emissions in contributions attributable to
emissions from specic primary energy carriers or economic sectors. Since, GCC countries use very negligible amount of coal,
therefore we only consider oil and natural gas. Likewise, the
emissions from service sector is signicantly low and as such, we
consider three major economic sectors, manufacture, residential
and transport sectors.
Panel A of Fig. 2 reports decomposed Gini index of CO2
emissions from natural gas and oil use. The gure shows that the
concerned countries demonstrate more equality in terms of CO2
emissions from oil rather than from natural gas. The coefcient
of Gini of CO2 emissions from oil was 0.00001 in 1980 followed
by a rise to 0.00005 in 1985. Then, it increased further to
0.00008 and remained stable during the period from 1985 to
1990. It again jumped to 0.00013 in 2000 followed by a decline
to 0.00008 in 2005. It signicantly dropped further to 0.00001 in
2010.
The inequality of CO2 emissions from gas use is higher than that
in oil uses. In 1980, the Gini value was 0.00016, then it increased to
0.00034 in 1985. In 1990, it fell to 0.00019 but again increased to
0.00037 in 1995. It fell to 0.00001 in 2000 but rose to 0.00130 in
2005. Again, it dropped to 0.00052 in 2010. The striking nding is
that the emissions from gas are unevenly distributed than emissions from oil.
Panel B of Fig. 2 reports the emissions distribution in economic
sectors. It is evident that in the manufacturing sector, the level of
inequality in CO2 emissions was quite high in 1980. It then started
declining and achieved a signicantly low level of inequality in
1995. However, during 2000 and 2005, the inequality level rose
drastically. In 2009, the inequality reached a signicantly low level.
In the residence sector, it is evident that inequality signicantly
declined in 1995 from that in 1980. During the period 2000e2005,

0.4
0.2
0
-0.2
-0.4
-0.6

Axis Title

Fig. 1. Decoupling index (DI) of each of the GCC countries.

54

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

Saudi Arabia: Decoupling Index CO2 Emission/ GDP per Capita


1.2
1

Axis Title

0.8
0.6
0.4
0.2

0
-0.2
Axis Title
UAE: Decoupling Index CO2 Emission/ GDP per Capita
2.5

Axis Title

2
1.5
1
0.5
0

Axis Title

Fig. 1. (continued).

1995 and 2000, there was almost no inequality in CO2 emissions. In


2010, no inequality in CO2 emissions was found in the transport
sector.
In this section, we discuss results from the Laspeyres decomposition. Fig. 3 demonstrates the contribution of each of the three
factors affecting the value of the Gini index to total inequality, i.e.
(i) changes in the Gini index within this source, (ii) changes in the
share of this source in total per-capita emissions, and (iii) changes
in its rank correlation with total emissions for every individual
source of emissions (i.e. primary energy carrier or economic
sector). The results reveal that the reduced shares of emissions
from gas and oil in total CO2 emissions, declining Gini coefcients
in both these sources of energy and the subsequent lower rank
correlations explain the declining contribution of the Gini index
per capita CO2 emissions. For natural gas, the higher values of Gini
index are mainly attributed to its higher contribution towards
total CO2 emissions. It is also evident that the Gini index is
working in the opposite direction and the change in the rank
correlation is found insignicant. It is observed that both the
energy carriers demonstrate declining inequality in CO2 emissions
over time.
For emissions from manufacturing and residential sectors, the
lower Gini index in 2010 as compared to 1980 signicantly
impacted the fall in these sectors contributions towards lowering
overall emissions inequality. In the case of transport sector,
increased shares in total emissions increase inequality in total per
capita CO2 emissions and the decreased Gini coefcient work in the
opposite direction. Overall, it is observed that emissions inequality
has declined in the GCC countries during the sample period of the
study and that the diverging changes in Gini coefcient for different
sectors contributed towards differences in emissions inequality in
the GCC region.

5. Conclusions and discussions


This study addresses three important issues in GCC countries.
First it investigates the relationship among per capita GDP, energy
consumption and CO2 emissions in GCC countries using panel data
for the period of 1980e2012. CIPS panel unit root tests are conducted that account for cross sectional dependence and nd that all
are rst difference stationary. Pedroni cointegration test conrms a
cointegrating relationship among the variables. In the long-run, per
capita GDP has a positive but insignicant impact on CO2 emissions
and in the short run the effect is negative and insignicant. Energy
consumption has a positive and signicant effect on CO2 emissions
both in the short- and the long-run. The error correction term in the
short-run is 0.352 suggesting that approximately 35% of the deviations from the long run are corrected each year.
The robustness of the ndings is checked by employing SUR. The
SUR estimates support the ndings of PMG. Energy consumption
and CO2 emissions both are found to Granger cause each other. This
nding is consistent with most of the earlier empirical panel
studies such as [30,91,108,109] and [18]. There is unidirectional
causal link running from economic growth to energy consumption.
This nding is also supported by many previous empirical studies.
However, no causal link was found between economic growth and
CO2 emissions in the region which is justied given that the GDP of
GCC countries are largely driven by oil and natural gas exports
which do not cause internal GHG (Greenhouse Gas) emissions.
Second, the study estimates the decoupling effects in the region.
The ndings of the decoupling of energy consumption from economic growth reveal that all GCC countries but Qatar experience
some kind of absolute or relative or both kinds of decoupling during
the period of study. Bahrain experiences no decoupling for some
period (1993e2003, 2006e2008). However, in the rest of the
period of the study, it experienced relative decoupling and absolute
decoupling. Oman and the UAE also experienced absolute and
relative decoupling during the sample period. Saudi Arabia performed only absolute decoupling throughout the entire study
period. Bahrain, Saudi Arabia, Oman and the UAE all these GCC
countries achieved some energy efciency reected through their
decoupling effects. Qatar experienced relative decoupling for the
longest period of the sample. It was in the state of absolute
decoupling. The absolute decoupling scenarios of Qatar and Saudi
Arabia may partly be attributed to the fact that they are at the
forefront of the GCC countries in terms of emissions reduction
initiatives. Kuwait observed both absolute and relative decoupling
during the period of study. It experienced the longest relative
decoupling for about a decade and a half during 1995e2011.
Third, the study analyzes inequality in CO2 emissions in the
region both in terms of energy carriers and the vital economic
sectors of the GCC countries. Gini index of CO2 emissions from
natural gas and oil use shows that the concerned countries
demonstrate more equality in terms of CO2 emissions from oil
rather than from natural gas. The coefcient of Gini of CO2 emissions from oil was 0.00001 in 1980 and signicantly dropped to
0.00001 in 2010. The inequality of CO2 emissions from gas use is
higher than that in oil uses. In 1980, the Gini value was 0.00016,
Again, it dropped to 0.00052 in 2010. The striking nding is that the
emissions from gas are unevenly distributed than emissions from
oil.
From the ndings of the emissions distribution in economic
sectors, it is evident that in the manufacturing sector, the level of
inequality in CO2 emissions was quite high in 1980. In 2009, the
inequality reached a signicantly low level. In the residence sector,
it is evident that inequality signicantly declined in 1995 from that
in 1980. It signicantly fell to almost equality level in 2010. It was
also found that CO2 emissions inequality was more prominent in

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

GAS

Transport

OIL

0.00160

55

Residence

Manufacture

0.014
0.00130

0.00140

0.012

0.00120
0.01

Gini

0.00100

0.0080.010256583

0.00080
0.00052

0.00060
0.00040

0.00034

0.006

0.00037
0.00019

0.004

0.00001
0.00020 0.00016
0.000130.00008
0.00000 0.000010.000050.000080.00008
0.00001
1980

1985

1990

1995

2000

2005

0.002
0

2010

Year

0.000677782
9.47535E-07
0.000723994
5.75751E-06
6.82884E-07
2.57126E-06
1980 1985 1990 1995 2000 2005 2010

Year

Fig. 2. Contribution of individual primary energy carriers and economic sectors to total inequality in per-capita CO2 emissions measured by the Gini index.

1980 in the transport sector. However, it experienced a declining


trend after 1980. In 2010, no inequality in CO2 emissions was
observed in the transport sector.
Also, from the ndings of Laspeyers decomposition, it is evident
that inequality in CO2 emissions has declined over the study period
especially in the case of oil consumption. Inequality also dropped in
all the economic sectors considered in the study i.e. in transport,
manufacture and the residence sector. Also the diverging changes
in Gini coefcient within each of the three sectors and changes in
the share of each sector in total per capita emissions contribute
towards the differences in emissions inequality in the region.
Despite some very optimistic ndings such as expected decoupling effects and the signicant fall in the inequality CO2 emissions
over time both in energy carriers and the economic sectors, the GCC
countries are still very signicant contributors of CO2 emissions and
as such this study recommends three potential measures for GCC
countries to reduce emissions. First of all, they should adopt and
expand CCUS facilities as CCUS (Carbon Capture, Utilization and
Storage) appears to be an effective method to combat CO2 emissions in the region [8]. Since these countries have a unique opportunity to reduce CO2 emissions through its storage while
simultaneously reducing oil eld depletion and have enormous
advantage of freeing natural gas (ng) from the oil elds through
replacing it with CO2, they can increase oil production through EORCO2 ('Enhanced Oil Recovery') scheme in the short run. Also the
freed natural gas could either be sold in the international market or
be used as relatively clean fuel for rapidly increasing domestic
energy demand.

Share

Rank

For long-run storage of CO2, CCUS involves capturing CO2,


transporting it via pipelines or ships nally injecting into suitable
rock formations. Although limited scale operation of EOR-CO2 are
already in place in the UAE, Saudi Arabia and Qatar, more efforts are
required to promote CCUS in the rest of the GCC countries. From
policy perspectives, these countries need to take further initiatives
to ensure a favorable regulatory framework to guide the CCUSbased activities that will promote the growth of CCUS plants.
Also, the governments might invite the private entrepreneurs and
build PPP (publiceprivate partnership) that might play a signicant
role in boosting investment funds for CCUS plants in the region.
Second, GCC countries should promote the use of renewable
sources for power generation such as solar and wind in which they
have clear advantage over many other countries. The average
annual solar radiation falling in the region is 2200 kWh (th)/m2
[60]. The vast uninhabited land in the region offers immense potential for harnessing the energy of sun. Although Saudi Arabia and
the UAE are the most active countries in the region in terms of
investing in solar energy research [61], other GCC countries also
must tap this opportunity for alternative sources of producing
electricity.
Third, apart from adopting and expanding CCUS and solar energy technologies, building nuclear energy is another valid option
for GCC countries for power generation. Usually nuclear energy
plants involve huge investment and the benets are likely to be due
only in the very long-run. Since GCC economies are generally
characterized by large liquid balances, large scale investment in
nuclear energy is not very challenging for them. Nevertheless, the

Gini

Rank

Share

Gini

% of Absolute Change 1980-2010

3
4

2.5

1.5

-2
-4
-6
-8

Gas

Oil

2.262

2.430

-0.960
Residence

-0.846
Manufecture

-0.988

-0.955

1.635

0.5
0
-0.5

-0.999
Transport

-1

-10

-1.5

-12

-2

-1.000

-2.5
Fig. 3. Laspeyres decomposition to determine the source of absolute changes in the Gini index of total per-capita emissions.

56

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

local conditions (social and environmental) for nuclear energy


deployment in the GCC countries are more ideal than many
industrialized countries [61]. However, the success and sustainability of nuclear energy plants also depend on the political
chemistry of the region.
Finally, the results from the econometric investigation, estimation of decoupling effects and the application of Gini index to assess
the contribution of vital energy drivers and the key economic
sectors towards emissions provide a more ne-grained understanding about GCC countries' position in participating in future
climate change negotiations. This was never done in any previous
study involving these countries.
References
[1] Abosedra S, Baghestani H. New evidence on the causal relationship between
United States energy consumption and gross national product. J Energy Dev
1989;14:285e92.
[2] Adom PK, Bekoe W, Amuakwa-Menash F, Mensah JT, Botchway E. Carbon
dioxide emissions, economic growth, industrial structure and technical efciency: empirical evidence from Ghana, Senegal and Morocco on the causal
dynamics. Energy 2012;47:314e25.
[3] Akarca AT, Long TV. On the relationship between energy and GNP: a re-examination. J Energy Dev 1980;5:326e31.
[4] Akbostanci E, Trt-AsIk S, Ipek TG. The relationship between income and
environment in Turkey: is there an environmental Kuznets curve? Energy
Policy 2009;37:861e7.
[5] Alkhathlan K, Javid M. Energy consumption, carbon emissions and economic
growth in Saudi Arabia: an aggregate and disaggregate analysis. Energy
Policy 2013;62:1525e32.
[6] Al-mulali U. Exploring the bi-directional long run relationship between urbanization, energy consumption and carbon dioxide emission. Energy
2012;46:156e67.
[7] Al-Mulali U, Lee YJ, Mohammed AH, Sheau-Ting L. Examining the link; 2013.
[8] Al-Saleh YM, Vidican G, Natarajan L, Theeyattuparampil VV. Carbon capture,
utilization and storage scenarios for the Gulf Cooperation Council region: a
Delphi-based foresight study. Futures 2012;44:105e15.
[9] Altinay G, Karagol G. Structural break, unit root and causality between energy consumption and GDP in Turkey. Energy Econ 2004;26:985e94.
[10] Alves MR, Moutinho V. Decomposition analysis and Innovative Accounting
Approach for energy-related CO2 (carbon dioxide) emissions intensity over
1996e2009 in Portugal. Energy 2013;57:775e87.
[11] Andreoni V, Galmarini S. Decoupling economic growth from carbon dioxide
emissions: a decomposition analysis of Italian energy consumption. Energy
2012;44:682e91.
[12] Ang JB. CO2 emissions, energy consumption, and output in France. Energy
Policy 2007;35:4772e8.
[13] Ang JB. Economic development, pollutant emissions and energy consumption in Malaysia. J Policy Model 2008;30:271e8.
[14] Apergis N, Payne JE. Energy consumption and economic growth: evidence
from the Commonwealth of Independent States. Energy Econ 2010;31:
641e7.
[15] Apergis N, Payne JE. The emissions, energy consumption, and growth nexus:
evidence from the Commonwealth of Independent States. Energy Policy
2010;38:650e5.
[16] Apergis N, Payne JE. A dynamic panel study of economic development and
the electricity consumption-growth nexus. Energy Econ 2011;33:770e81.
[17] Aqeel A, Butt MS. The relationship between energy consumption and economic growth in Pakistan. Asia Pac Dev J 2001;8:101e10.
[18] Arouri MEH, Youssef AB, Mhenni H, Rault C. Energy consumption, economic
growth and CO2 emissions in Middle East and North African countries. Energy Policy 2012;45:342e9.
[19] Asafu-Adjaye J. The relationship between energy consumption, energy prices
and economic growth: time series evidence from Asian Developing Countries. Energy Econ 2000;22:615e25.
[20] Asteriou D. Foreign aid and economic growth: new evidence from a panel
data approach for ve South Asian countries. J Policy Model 2009;31:
155e61.
[21] Atkinson AB. On the measurement of inequality. J Econ Theory 1970;2:
244e63.
[22] Belke A, Dreger C, Haan F. Energy consumption and economic growth: new
insights into the cointegrating relationship. Energy Econ 2011;33(5):782e9.
between energy consumption, carbon dioxide emission and economic
growth in Latin America and the Caribbean, Renewable and Sustainable
Energy Reviews, 26, 42e48.
[23] Bithas K, Kalimeris P. Re-estimating the decoupling effect: Is there an actual
transition towards a less energy-intensive economy? Energy 2013;51:
78e84.
[25] Burnett WJ, Bergstrom CJ, Wetzstein EM. Carbon dioxide emissions and
economic growth in the U.S. J Policy Model 2013;35:1014e28.

[26] Cantore N. Distributional aspects of emissions in climate change integrated


assessment models. Energy Policy 2011;39:2919e24.
[27] Cantore N, Padilla E. Equality and CO2 emissions distribution in climate
change integrated assessment modelling. Energy 2010;35:298e313.
[29] Chandran VGR, Sharma S, Madhavan K. Electricity consumptionegrowth
nexus: the case of Malaysia. Energy Policy 2009;38:606e12.
[30] Chang C. A multivariate causality test of carbon dioxide emissions, energy
consumption and economic growth in China. Appl Energy 2013;87:
3553e7.
[31] Chang T, Fang W, Wen L. Energy consumption, employment, output, and
temporal causality: evidence from Taiwan based on cointegration and errorcorrection modeling techniques. Appl Econ 2001;33:1045e56.
[32] Cheng BS. Energy consumption and economic growth in Brazil, Mexico and
Venezuela: a time series analysis. Appl Econ Lett 1997;4:671e4.
[33] Cheng BS, Lai TW. An investigation of cointegration and causality between.
1997.
[34] Coondoo D, Dinda S. The carbon dioxide emission and income: a temporal
analysis of cross-country distributional patterns. Ecol Econ 2008;65:375e85.
[35] Cowan WN, Chang T, Inglesi-Lotz R, Gupta R. The nexus of electricity consumption, economic growth and CO2 emissions in the BRICS countries. Energy Policy 2013. http://dx.doi.org/10.1016/j.tenpol.2013.10.081.
[37] Dhakal S. Urban energy use and carbon emissions from cities in China and
policy implications. Energy Policy 2009;37:4208e19.
[38] Dinda S, Coondoo D. Income and emission: a panel data-based cointegration
analysis. Ecol Econ 2006;57:167e81.
[39] Duro JA. On the automatic application of inequality indexes in the analysis of
the international distribution of environmental indicators. Ecol Indic
2012;76:1e7.
[40] Duro JA, Alcantara V, Padilla E. International inequality in energy intensity
levels and the role of production composition and energy efciency: an
analysis of OECD countries. Ecol Econ 2010;69:2468e74.
[41] Duro JA, Teixido-Figueras J. Ecological footprint inequality across countries:
the role of environment intensity, income and interaction effects. Ecol Econ
2013;93:34e41.
[43] Fodha M, Zaghdoud O. Economic growth and pollutant emissions in Tunisia:
an empirical analysis of the environmental Kuznets curve. Energy Policy
2010;38:1150e6.
[44] Freitas L, Kaneko S. Decomposing the decoupling of CO2 emissions and
economic growth in Brazil. Ecol Econ 2011;70:1459e69.
[45] Friedl B, Getzner M. Determinants of CO2 emissions in a small open economy. Ecol Econ 2003;45:133e48.
[46] Fuinhas JA, Marques AC. Energy consumption and economic growth nexus in
Portugal, Italy, Greece, Spain and Turkey: an ARDL bounds test approach
(1965e2009). Energy Econ 2012;34:511e7.
[49] Ghosh S. Examining carbon emissions-economic growth nexus for India: a
multivariate cointegration approach. Energy Policy 2010;38:2613e3130.
[50] Ghosh S, Kanjilal K. Long-term equilibrium relationship between urbanization, energy consumption and economic activity: empirical evidence from
India. Energy 2014;66:324e31.
[51] Glasure YU. Energy and national income in Korea: further evidence on the
role of omitted variables. Energy Econ 2002;24:355e65.
[52] Glasure YU, Lee A-R. Cointegration, error-correction, and the relationship
between GDP and energy: the case of South Korea and Singapore. Resour
Energy Econ 1998;20:17e25.
[53] Grossman G, Krueger A. Environmental impacts of a North American Free
Trade Agreement; 1991. NBER Working Paper, No. 3914, Washington.
[54] Gutierrez L. On the power of panel cointegration test: a Monte Carlo comparison. Econ Lett 2003;80:105e11.
[55] Halicioglu F. An econometric study of CO2 emissions, energy consumption,
income and foreign trade in Turkey. Energy Policy 2009;37:1156e64.
[57] Hatemi A, Irandoust M. Energy consumption and economic growth in
Sweden: a leveraged bootstrap approach (1965e2000). Int J Appl Econ Quant
Stud 2005;2e4:87e98.
[58] Hedenus F, Azar C. Estimates of trends in global income and resource inequalities. Ecol Econ 2005;55:351e64.
[59] Heil MT, Selden TM. Panel stationarity with structural breaks: carbon
emissions and GDP. Appl Econ Lett 1999;6:223e5.
[60] Hertog S. Princes, brokers, bureaucrats: oil and state in Saudi Arabia. Ithaca,
NY: Cornell University Press; 2010.
[61] Hertog S, Luciani G. Energy and sustainability policies in the GCC. The Center
for the Study of Global Governance; November, 2009. Kuwait Program on
Development, Governance and Globalization in the Gulf States.
[62] Im KS, Pesaran MH, Shin Y. Testing for unit roots in heterogeneous panels.
J Econ 2003;115:53e74.
[64] Jalil A, Mahmud S. Environment Kuznets Curve for CO2 emissions: a cointegration analysis for China. Energy Policy 2009;37:5167e72.
[65] Jaunky VC. The CO2 emissions-income nexus: evidence from rich countries.
Energy Policy 2011;39:1228e40.
[67] Kaya Y. Impact of carbon dioxide emissions control on GNP growth; 1989.
[68] Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev
1978;3:401e3.
[69] Kveilborg O. Explaining the decoupling of freight trafc growth and economic growth. In: European Transport Conference; 2004.
[70] Kveilborg O. Decomposing the decoupling of Danish road freight trafc
growth and economic growth. Transp Policy 2007;14:39e48.

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58


[71] Kveilborg O, Fosgerau M. Decomposing the decoupling of Danish road freight
trafc growth and economic growth. Danish Transport Institute; 2005.
[72] Lean HH, Smyth R. CO2 emissions, electricity consumption and output in
ASEAN countries. Appl Energy 2010;87:1858e64.
[73] Lee CC. Energy consumption and GDP in developing countries: a cointegrated panel analysis. Energy Econ 2005;27(3):415e27.
[74] Lerman, Yitzhaki. A note on the calculation and interpretation of the Gini
index. Econ Lett 1984;15:363e8.
[75] Li F, Dong S, Li X, Li Q, Yang W. Energy consumption-economic growth relationship and carbon dioxide emissions in China. Energy Policy 2011;39:568e74.
[76] Liao H, Cao H. How does carbon dioxide emission change with economic
development? Statistical experiences from 132 countries. Glob Environ
Change 2013;23:1073e82.
[77] Lotfalipour MR, Falahi MA, Ashena M. Economic growth, CO2 emissions and
fossil fuels consumption in Iran. Energy 2010;35:5115e20.
[78] Lucas REB, Wheeler D, Hettige H. Economic development, environmental
regulation and the international migration of toxic industrial pollution:
1960-1988. Policy Research Working Paper Series, 1062. Washington DC:
World Bank; 1992.
[79] Masih AMM, Masih R. Electricity consumption, real income and temporal causality: results from a multi-country study based on cointegration and errorecorrection modeling techniques. Energy Econ
1996;18:165e83.
[80] Masih AMM, Masih R. On temporal causal relationship between energy
consumption, real income and prices: some new evidence from Asian energy
dependent NICs based on a multivariate cointegration vector error correction approach. J Policy Model 1997;19:417e40.
[81] Mckinon AC. Decoupling of road freight transport and economic growth
economic growth in the UK: an exploratory analysis. Transplant Rev
2007;27:37e64. modeling techniques, Applied Economics 33, 1045-1056.
[82] Murthy NRV. Panel cointegration analysis: an empirical example in cointegration for the applied economist. In: Rao B, editor. Cointegration for the
applied economics. 2nd ed. 2007. pp. 222e56.
[83] Narayan PK, Narayan S. Carbon dioxide emissions and economic growth: panel
data evidence from developing countries. Energy Policy 2010;38:661e6.
[84] Narayan PK, Singh B. The electricity consumption and GDP nexus for Fiji
Islands. Energy Econ 2007;29:1141e50.
[85] Narayan PK, Smyth R. Multivariate granger causality between electricity
consumption, exports and GDP: evidence from a panel of Middle Eastern
countries. Energy Policy 2009;37:229e36.
[86] Nelson C, Plosser C. Trends and random walks in macroeconomic time series:
some evidence and implications. J Monetary Econ 1982;10:130e62.
[87] Newey WK, West KD. A simple positive semi-denite, heteroscedasticity and
auto consistent covariance matrix. Econometrica 1987;55:703e8.
[88] Nohman A, Antrobus G. Trade and the environmental Kuznets curve: is
southern Africa a pollution heaven? South Afr J Econ 2005;73:803e14.
[89] Odhiambo NM. Energy consumption and economic growth nexus in
Tanzania: an ARDL bounds testing approach. Energy Policy 2009;37:617e22.
[90] Odhiambo NM. Electricity consumption and economic growth in South Africa: of omitted variables. Energy Econ 2009;24:355e65.
[91] Omri A. CO2 emissions, energy consumption and economic growth nexus in
MENA countries: evidence from simultaneous equations models. Energy
Econ 2013;40:657e64.
[92] Ozcan B. The nexus between carbon emissions, energy consumption and
economic growth in middle east countries: a panel data analysis. Energy
Policy 2013;62:1138e47.
[93] Ozturk I, Acaravci A. CO2 emissions, energy consumption and economic
growth, in Turkey. Renew Sustain Energy Rev 2010;14:3220e5.
[94] Ozturk I. A literature survey on energy-growth nexus. Energy Policy
2010;38:340e9.
[95] Pao H, Fu H, Tseng C. Forecasting of CO2 emissions, energy consumption and
economic growth in China using an improved grey model. Energy 2012;40:
400e9.
[96] Pao H, Yu H, Yang Y. Modeling the CO2 emissions, energy use and economic
growth in Russia. Energy 2011;36:5094e100.
[97] Pao HT, Tsai CM. Modeling and forecasting the CO2 emissions, energy consumption and economic growth in Brazil. Energy 2011;36:2450e8.
[98] Pao HT, Tsai CM. Multivariate Granger causality between CO2 emissions,
energy consumption, FDI (foreign direct investment) and GDP (gross domestic product) evidence from a panel of BRIC (Brazil, Russian Federation,
India and China) countries. Energy 2011;36:685e93.
[99] Payne JE. A survey of the electricity consumption-growth literature. Appl
Energy 2010;87:723e31.
[100] Pedroni P. Panel cointegration; asymptotic and nite sample properties of
pooled time series tests, with an application to the ppp hypothesis: new
results. Working paper. April: Indiana University; 1997.
[101] Pedroni P. Critical values for cointegration tests in heterogeneous panels
with multiple regressors. Oxf Bull Econ Stat 1999;61:653e70.
[102] Pesaran MH. General diagnostic tests for cross section dependence in
panels; 2004. Cambridge Working Papers in Economics No. 435. University
of Cambridge and CESifo Working Paper Series No. 1229.
[103] Pesaran MH. A simple panel unit root test in the presence of cross-section
dependence. J Appl Econ 2007;22:265e312.
[104] Pesaran MH, Shin Y. An autoregressive distributed lag modeling approach
to cointegration analysis. Chapter 11. In: Strom S, editor. Econometrics and

57

economic theory in the 20th Century: the Ragnar Frisch Centennial Symposium. Cambridge: Cambridge University Press; 1999.
[105] Pesaran MH, Shin Y, Smith R. Pooled mean group estimation of dynamic
heterogeneous panels. J Am Stat Assoc 1999;94:621e34.
[106] Pesaran MH, Smith RP. Estimating long-run relationships from dynamic
heterogeneous panels. J Econ 1995;68:79e113.
[107] Reynolds DB, Kolodzieji M. Former Soviet Union oil production and GDP
decline: Granger causality and the multi-cycle Hubbert curve. Energy Econ
2008;30:271e89.
[108] Saboori B, Sapri M, Baba M. Economic growth, energy consumption and
CO2 emissions in OECD (Organization of Economic Co-operation and
Development)'s transport sector: a fully modied bi-directional relationship approach. Energy 2014;66:150e61.
[109] Saboori B, Sulaiman J. CO2 emissions, energy consumption and economic
growth in Association of Southeast Asian Nations (ASEAN) countries: a
cointegration approach. Energy 2013;55:813e22.
[110] Sahzabi A, Saaki K, Youse H, Sugai Y. CO2 emission and economic growth of
Iran. Mitig Adapt Strat Global Change 2011;16(1):63e82.
[112] Sbia R, Shahbaz M, Hamdi H. A contribution of foreign direct investment,
clean energy, trade openness, carbon emissions and economic growth to
energy demand in UAE. Econ Model 2013;36:191e7.
[115] Solarin SA, Shahbaz M. Trivariate causality between economic growth, urbanisation and electricity consumption in Angola: cointegration and causality analysis. Energy Policy 2013;60:876e84.
[116] Song T, Zheng T, Tong L. An empirical test of the environmental Kuznets
curve in China: a panel cointegration approach. China Econ Rev 2008;19:
381e92.
[117] Sorell S, et al. Decoupling of road freight energy use from economic growth
in the United Kingdom. Energy Policy 2010. http://dx.doi.org/10.1016/
j.enpol.2010.07.007.
[119] Soytas U, Sari R. Energy consumption, economic growth and carbon emissions: challenges faced by an EU candidate member. Ecol Econ 2009;68:
1667e75.
[120] Soytas U, Sari R. Energy consumption and GDP: causality relationship in G-7
and emerging markets. Energy Econ 2003;25(1):33e7.
[121] Sqaulli J. Electricity consumption and economic growth: bounds and causality analyses of OPEC members. Energy Econ 2007;29:1192e205.
[122] Stern DI. The rise and fall of the environmental Kuznets curve. World Dev
2004;32:1419e39.
[123] Sun J, Ang B. Some properties of an exact energy decomposition model.
Energy 2000;25:1177e88.
[124] Suri V, Chapman D. Economic growth, trade and the energy: implications for
the environmental Kuznets curve. Ecol Econ 1998;25:195e208.
[125] Theil H. Statistical decomposition analysis with application in the social and
administrative sciences; 1972 [North Holland, Amsterdam].
[126] Wang M, Han J, Dunn JB, Cai H, Elgowaing A. Well-to-wheels energy use and
greenhouse gas emissions of ethanol from corn sugarcane and cellulosic
biomass for US use. Environ Res Lett 2012;7:4.
[128] Wang W, Liu R, Zhang M, Li H. Decomposing the decoupling of energyrelated CO2 emissions and economic growth in Jiangsu Province. Energy
Sustain Dev 2013;17:62e71.
[129] Wang Y, Wang Y, Zhou J, Zhu X, Lu G. Energy consumption and economic
growth in China: a multivariate causality test. Energy Policy 2011;39:
4399e406.
[130] Wesseh PK, Zoumara B. Causal independence between energy consumption
and economic growth in Liberia: evidence from a non-parametric bootstrapped causality test. Energy Policy 2012;50:518e27.
[131] Wolde-Rufael Y. Electricity consumption and economic growth: a time series
experience for 17 African countries. Energy Policy 2006;34:1106e14.
[132] World Bank. World development indicators. Washington DC, USA: The
World Bank; 2013. available from: http://data.worldbank.org/data-catalog/
world-development-indicators/World Bank-2012 [accessed 03.08.13].
[133] Wyckoff AW, Roop JM. The embodiment of carbon in imports of manufactured products: implications for international agreements on greenhouse
gas emissions. Energy Policy 1994;22:187e94.
[134] Yoo S. The causal relationship between electricity consumption and economic growth in ASEAN countries. Energy Policy 2006;34:3573e82.
[135] Yoo S-H, Kwak S-Y. Electricity consumption and economic growth in seven
South American countries. Energy Policy 2010;38:181e8.
[136] Yu ESH, Choi JY. The causal relationship between electricity and GNP: an
international comparison. J Energy Dev 1985;10:249e72.
[137] Yuan J, Zhao C, Yu S, Hu Z. Electricity consumption and economic growth in
China: cointegration and co-feature collection. Energy Econ 2007;29:
1179e91.
[138] Zhang M, Wang W. Decouple indicators on the CO2 emissions growth linkage: the Jiangsu Province case. Ecol Indic 2013;32:239e44.
[139] Zhang XP, Cheng X-M. Energy consumption, carbon emissions and economic
growth in China. Ecol Econ 2009;68:2706e12.
[140] Wolde-Rufael Y. Disaggregated energy consumption and GDP, the experience of Shanghai, 1952e1999. Energy Economics 2004;22:309e17.
[141] Narayan PK, Popp S. The energy consumption-real GDP nexus revisited:
Empirical evidence from 93 countries. Economic Modeling 2012;29(2):
303e8.
[142] Stern DI. Energy use and economic growth in the USA: a multivariate
approach. Energy Economics 1993;15:137e50.

58

M. Salahuddin, J. Gow / Energy 73 (2014) 44e58

[143] Padilla E, Serrano A. Inequality in CO2 emissions across countries and its
relationship with income inequality: A distributive approach. Energy Policy
2006;34(14):1762e72.
[144] Akarca AT, Long TV. Energy and Employment: A Time Series Analysis of the
Causal Relationship. Resources and Energy 1979;2:151e62.
[145] Cheng BS. Causality between Energy consumption and economic growth in
India: an application of cointegration and error correction modeling. Indian
Economic Review 1999;34(1):39e49.
[146] Granger CWJ. Investigating causal relations by econometric models and cross
spectral methods. Econometrica 1969;36:424e38.
[147] IEA. World energy outlook special report 2013: Rewarding the Energy
Climate Map; 2013.
[148] OECD. Indicators to measure development of environmental pressure from
economic growth [R]. Paris: OECD; 2002.
[149] Juknys R. Transition period in Lithuania e do we move to sustainability?
Journal of Environmental Research Engineering and Management 2003.

[150] Ren S, Hu Z. Effects of decoupling of carbon dioxide emissions by Chinese


non-ferrous metals industry. Energy Policy 2012;43:407e14.
[151] Paul S, Bhattacharya RN. CO2 emissions from energy use in India: a
decomposition analysis. Energy Policy 2004;32:585e93.
[152] He J, Deng J, Su M. CO2 emissions from China's energy sector and strategy for
its control. Energy 2010;35(11):4494e8.
[153] Duro JA, Padilla E. International inequalities in per capita CO2 emissions: a
decomposition methodology by Kaya factors. Energy Economics 2006;28(2):
170e87.
[154] Narayan P, Smyth R. A panel cointegration analysis of the demand for oil in
the middle east. Energy Policy 2007;35(12):6258e65.
[155] Shahbaz M, Zeshan M, Afza T. Is energy consumption effective to spur economic growth in Pakistan? New evidence for bounds test to level relationships and Granger causality. Economic modeling 2012;29(6):2310e9.

You might also like