Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

Energy consumption, carbon dioxide emissions and economic development: MARK

Evaluating alternative and plausible environmental hypothesis for
sustainable growth
⁎
Khalid Zamana, , Mitwali Abd-el. Moemenb
a
Department of Economics, University of Wah, Quaid Avenue, Wah Cantt, Pakistan
b
College of Law and Political Science, King Saud University, Riyadh, Saudi Arabia

A R T I C L E I N F O A BS T RAC T

Keywords: Pursuit of excellence in economic development, in the midst of damaging the natural environment, is a
Carbon dioxide emissions shameless growth. The economic impacts on environmental degradation are quite visible in industrialized
GDP per capita economies where human health is compromised by rapid economic growth and energy induced emissions. This
Energy Consumption study examines the interrelationship between energy consumption, economic growth and carbon dioxide (CO2)
Environmental Kuznets curve
emissions under the six alternative and plausible hypothesis including Environmental Kuznets Curve (EKC),
Pollution haven hypothesis
Pollution Haven Hypothesis (PHH), population based emissions (IPAT), energy led emissions, sectoral growth
Panel GMM technique
emissions and Emissions emancipated Human Development Index (eHDI) in the context of low and middle
income countries, high income countries and in aggregated panel, over the period of 1975–2015. The results
supported the EKC hypothesis, IPAT hypothesis, energy induced emissions, and sectoral growth emissions in
different regions of the world, while PHH and eHDI hypothesis does not confirm across regions. This study
exclusively determines the key socio-economic and environmental problems in a large pool of the world
economies to understand the need of development policy agenda for sustainable growth.

1. Introduction
The global growth is projected around 3.4% in 2016 and it would be
added at 0.2% in 2017 that make around 3.6%. This estimate would
larger than 0.3% (2016) and 0.5% (2017) from the year 2015, which is
currently estimated at 3.1% points. Although, this projection as
simulated by World Economic Outlook [64] largely emphasized on
the positive growth trajectory in some BRICS countries and some
countries of the Middle East, however, the slow down & rebalancing
the Chinese economy and distress economic growth in emerging
markets, the IMF report provoked to see some positivity in the global
economic growth.
According to the latest report of International Energy Outlook [30],
the devastating rate of CO2 emissions resulting due to the combustion
of fossil fuels, which is further linked with the climate change debate.
The report projected that global energy- induced CO2 emissions would
increase around 35.6 billion metric tons in 2020 which will further
added up to 7.6% in 2040 to make a figure up to 43.2 billion metric
tons, which would be far greater than the estimates of 2012 i.e., 32.3
billion metric tons. The report further argued that emissions’ growth is
highly sensitive in the developed nations that continue to rely heavily
on fossil fuels to gear the pace of economic growth to utilized energy
demand. OECD emissions will be contribute around 13.8 billion metric
tons in 2040 while non-OECD emissions contributes around 29.4
billion metric tons, which is about 8% and 51% higher than the existing
emissions level of 2012.
The recent trends of economic growth and energy -induced CO2
emissions strike the need of well-balanced economic model that
fundamentally approved by sustainable agencies for better health and
wealth. This study takes an initiative to propose an interactive
environmental model that aligned with the key factors of green growth
and renewable energy resources to support sustainable growth agenda
across the globe. The importance of sustainable growth in a policy
liaison is broad-based that need careful examination to evaluate
energy, environment and economic growth in a heterogeneous panel
of countries by including financial and trade liberalization, social
expenditures, and sectoral value added. This study attempts to explore
these factors in a panel of low, medium, and high income countries.
The relationship between CO2 emissions and per capita income
extensively debated in the academic arena. Kuznets [58] explored an
inverted U-shaped relationship between income inequality and per
capita income. The followers of Kuznets school of thought added CO2

⁎
Corresponding author.
E-mail address: Khalid_zaman786@yahoo.com (K. Zaman).

http://dx.doi.org/10.1016/j.rser.2017.02.072
Received 25 November 2015; Received in revised form 2 February 2017; Accepted 21 February 2017
Available online 17 March 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.
K. Zaman, M.A.-e. Moemen
emissions in relation with the per capita income and evaluate the EKC
hypothesis, which is still the icon for the academic arena (See,
[46,53,57,21,51]; etc). Seldon and Song [46] investigated the EKC
hypothesis in the panel of cross sectional countries and found that air
quality pollutants including suspended particulate matter, sulfur
dioxide (SO2), nitrogen oxides, and carbon monoxide increases along
with the increase per capita income, while at the later stages of
development, these air pollutants significantly decline over time, hence
the EKC hypothesis was found in all of the four air pollutants. Stern
et al. [53] discussed the EKC hypothesis in relation with the global air
pollutants and per capita income, and found a significant rise in SO2
emissions and deforestation till the years 2025. Torras and Boyce [57]
investigated the triangular relationship between per capita income,
income inequality and pollution and found that literacy, political rights
and civil liberties significantly impact on environmental quality in low
income countries. Dinda [21] evaluated the global literature on EKC
hypothesis and argued that income per capita does not identify the
income range where environmental pollutants start to decline.
However, EKC hypothesis holds with the local pollutants. Stern [51]
presented the history of EKC hypothesis and discussed the rise and fall
of EKC in different countries/ regions. Stern and Common [52]
evaluated EKC hypothesis in relation with the SO2 emissions and per
capita income and found that EKC hypothesis was not related with the
income of the country/region, however, it was pronounce with the time
factor. In addition, EKC hypothesis holds in high income countries with
the SO2 emissions. The diverse literature is available on EKC hypoth-
esis which needs to be critically examine in a large pool of countries for
robust inferences. The subsequent sections critically discussed the
different alternative and plausible environmental hypothesis for sus-
tainable growth.
1.1. Tradeoff between CO2 emissions and economic growth
(Environmental Kuznets Curve approach)
The relationship between air pollutants and income per capita is
widely discussed in the environmental economics that linked with the
theory of ‘Environmental Kuznets Curve (EKC)’. The recent strikes of
the literature on the EKC hypothesis is the ray of hope that provide sun
shine to the other regions of the world i.e., Tutulmaz [59] investigated
the possible reasons of controversial EKC results for similar model and
selected a case study of Turkish economy, by using the time series data
ranging from 1986 to 2007. The results discussed some major rationale
by using a selection of model in a country's perspectives and in a panel
of developing countries, and discussed the policy implications for
sustainable growth agenda. The study of Yang et al. [66] reinvestigated
the EKC hypothesis in relationship with the different air pollutants in
29 provinces of China, by using extreme bound analysis from the
period of 1995–2010. The results do not confirm the inverted U –
shaped relationship in relation with different pollutants and China's
per capita income. Bernard et al. [15] examined the EKC hypothesis
with the CO2 emissions and SO2 emissions and found the validity of
EKC hypothesis in the context of OECD countries. Robalino-López
et al. [44] followed the EKC hypothesis in Venezuela for the period
1980–2025, and were unable to find any significant sign of EKC
hypothesis during the study time period. Jeblie and Youssef [31]
included renewable energy sources, non-renewable energy sources,
trade, and economic growth in EKC model for Tunisia during the
period of 1980–2009. The results show that non-renewable energy
sources and trade both have a positive and significant impact on CO2
emissions, while renewable energy sources decreases CO2 emissions at
nationwide. In addition, EKC hypothesis does not hold in the country's
context. [6,9] investigated the EKC hypothesis in the panel of 93
countries by using ecological indicator and confirmed the EKC
hypothesis across countries. Mazur et al. (2015) investigated the
relationship between environmental quality and growth per capita in
the panel of European Union countries, by using the data from 1992 to
1120
Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130
2010. The results do not confirm the EKC hypothesis in the 28 current
EU member states. Ozturk and Al-Mulali [40] investigated the EKC
hypothesis in Cambodia by using the two stage least square and
generalized method of moments technique and found that macroeco-
nomic variables including energy demand, trade liberalization, urba-
nization and income per capita significantly increases CO2 emissions in
a country, however, EKC hypothesis was not visible at nationwide.
The above literature confirmed the viability of EKC hypothesis in
different countries and regions of the world. This study followed the
pattern of EKC modeling and builds the theoretical framework in order
to assess the EKC hypothesis in a panel of selected countries for open
policy discussions.
1.2. Tradeoff between CO2 emissions, trade openness and FDI inflows
(pollution haven hypothesis)
The relationship between CO2 emissions and trade (FDI inflows)
are widely discussed topic in the trade and environmental literature.
The link between CO2 emissions and trade-FDI inflows come to the
meaningful conclusion that we called “Pollution Haven Hypothesis
(PHH)”. The PHH is the most conceivable hypothesis that has been
investigated in different countries across the globe. The major motiva-
tion behind all of the studies is that polluting industries damage the
natural environment of the world's precious resources that should have
to be preserved with the effective environmental regulation policies.
Shahbaz et al. [47] investigated the relationship between FDI inflows
and CO2 emissions in the panel of low, middle and high income
countries and confirmed the visibility of PHH in different regions of the
world. Aktas [4] found that globalization promotes polluting industries
in Turkey, while Ali et al. [6] argued that environment is on the stake,
as rapid progression of air pollutants are merely due to the higher
economic growth. Tang and Tan [55] investigated the PHH in the
context of Vietnam, using the time series data set ranging from 1976 to
2009. The results show that energy demand, economic growth, and FDI
inflows have a strong determinant to influence CO2 emissions in a
country. The policies should be taken with care while handling
environmental issues with respect to FDI and economic growth
together. Salahuddin et al. [45] examined the long-run relationship
between energy demand, per capita income, financial development,
and CO2 emissions in the selected Gulf cooperation countries and
found that energy demand and per capita income both increase CO2
emissions while financial development significantly decreases CO2
emissions. Ibrahim and Law [28] investigated the impact of trade
openness, institutional supremacy, and their interaction term on CO2
emissions in a panel of selected Sub-Sahara African countries and
found that trade openness significantly increases environmental quality
of those countries that have a strong institutional supremacy and hurt
environmental quality where institutional quality reforms are poor.
The above discussion confirmed the strong nexus between CO2
emissions, FDI inflows and trade in different countries settings. The
PHH is valid in different countries’ contexts, and we find the traces of
significant association of FDI and trade with the CO2 emissions from
the literature. The present study attempts to explore PHH in the panel
of selected countries.
1.3. Tradeoff between CO2 emissions and population growth (IPAT
hypothesis)
The impact of population growth on CO2 emissions is widely visible
topic and a normative debate in the environmental arena, however, the
blame on population growth to deteriorate environment is not fairly
justice. McGee et al. [36] investigated the impact of economic growth,
technology and population on carbon emissions for the panel of 173
countries and concluded that stochastic impacts by regression on
population, affluence and technology have considerable policy implica-
tions on future technology generation. Zhang and Liu [67] investigate
K. Zaman, M.A.-e. Moemen
the role of ICT industries on CO2 emissions in China by using STIRPAT
model, over the period of 2000–2010. The results show that ICT
industry is less sensitive to CO2 emissions, as it is significantly decline
CO2 emissions during the study time period. Hassan and Salim [26]
investigated the impact of population ageing and economic growth on
CO2 emissions in the panel of OECD countries and supported the EKC
hypothesis. Wang et al. [60,61] modeled population growth in relation-
ship between emissions and growth and found the inverted U-shaped
EKC relationship under the premises of population growth that makes
the EKC steeper and reached up to the higher peak. Alam et al. [5]
found the positive relationship between CO2 emissions and population
growth in India and Brazil, however, this relationship is insignificant in
the context of China and Indonesia. Halkos et al. [24] found that higher
per capita income and population growth tend to worsen the environ-
mental efficiency that required sustainable policy actions to reexamine
environmental actions for healthier region. The above discussion
significantly indicates the tradeoff between population and carbon
emissions across the globe. This study attempts to investigate the
impact of population growth on CO2 emissions in the panel of selected
countries.
1.4. Tradeoff between CO2 emissions and energy consumption
(energy led emissions hypothesis)
The relationship between CO2 emissions and energy consumption is
one of the ongoing debates in the energy economics literature since last
two decades. The search for renewable energy sources is one of the
fundamental ways to reduce/ minimize CO2 emissions from the atmo-
sphere. Mudakkar et al. [37] examined the causal relationship between
energy demand and various macroeconomic factors, and found the
unidirectional causality running from nuclear energy to industrial
GDP, water resources, and CO2 emissions, while energy demand
Granger cause agriculture GDP. The study further finds a feedback
relationship between energy demand and Pakistan's population den-
sity. Akhmat et al. [3] investigated the relationship between nuclear
energy demand and climatic factors in the panel of 35 developed
countries and found that nuclear energy consumption significantly
decline air pollution and greenhouse gas (GHG) emissions. [2] in the
similar lines found that energy consumption significantly increases CO2
emissions in the panel of SAARC countries. Khan et al. [32] investi-
gated the impact of energy demand on global GHG emissions and
found that energy demand Granger cause GHG emissions in the
different regions of the world. Apergis and Payne [13] emphasized
the need of renewable energy sources for increasing economic growth
and mitigating the climate change containing the CO2 emissions. Ajmi
et al. [1] investigated the impact of energy consumption and economic
growth on CO2 emissions and found the bidirectional time varying
causality running between CO2 emissions and energy demand in the
U.S, while unidirectional causality running from energy consumption
to carbon emissions in France. Cheema and Javid [16] examined the
relationship between economic growth, CO2 emissions, and disaggre-
gate energy consumption in Asian perspectives and found that all forms
of energy demand pollute environment that exhibit the EKC hypothesis
in the region.
The above studies emphasize the role of energy demand that played
a vital role to increase economic growth, however, it may compromised
to the environment that may lead to further cause of climatic change
worldwide. The present study seeks to investigate the energy led
emissions in the panel of low, middle and high income countries.
1.5. Tradeoff between CO2 emissions and sectoral growth (sectoral
induced carbon emissions)
There is undeniable fact that industrialization or sectoral produc-
tivity played a vital role in economic development, however, the society
bears the unmanageable cost in terms of environmental damage which
1121
Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130
affect the health and wealth of the countries. Rahman et al. [41]
examined the relationship between energy demand and sectoral
productivity in Malaysia and found that energy consumption signifi-
cantly increases Malaysia's economic growth at aggregate level, how-
ever, at disaggregate level, this result is evaporated. The study further
confirmed the strong association between CO2 emissions and sectoral
growth at country level. Ouyang and Lin [39] examined the energy led
emissions in China's industrial sector and show that industrial activity
significantly increases CO2 emissions while energy intensity declined
CO2 emissions at nationwide. In the similar lines, Cheng et al. [17]
empirically investigated the impact of CO2 emissions on Chinese
exports and found that regulations on CO2 emissions significantly
increases Chinese exports. Robaina-Alves et al. [43] investigated the
impact of energy consumption and CO2 emissions on Portuguese
tourism industry and found that energy and CO2 emissions both
significantly affected tourism industry in a country. Xu and Lin [65]
examined the relationship between urbanization, industrialization and
its resulting impact on CO2 emissions in the provinces of China and
confirmed the EKC hypothesis for urbanization and industrialization
growth across region.
The above stated studies indicate the sectoral productivity and CO2
emissions nexus. There is no doubt about the efficiency of manufactur-
ing sector; however, the policy should be taken care with the handling
of air quality indicators and technological innovations that may
increase the CO2 emissions during the production process. The present
study attempted to investigate the sectoral associated emission in the
panel of selected countries.
1.6. Tradeoff between CO2 emissions and human development
indicators [emissions emancipated human development index (eHDI)]
Now-a-days, the term emissions emancipated human development
index (eHDI) is widely appeared in the economic literature which
associated with the human efforts in terms of attaining better health,
education and wealth, however, human efforts lead to increase CO2
emissions that deteriorate the environmental quality across the globe.
Koroneos and Rokos [33] critically evaluated the term sustainable
development and coined the new word i.e., ‘Worth-living Integrated
Development” for policy making purpose. The worth-living integrated
development, basically the amalgamation of economic prosperity,
social inclusion and environmental protection that desirable for human
development and sustainable development together. Nussbaumer et al.
[38] created the link between energy demand and millennium devel-
opment goals including climate change and income poverty across the
globe. The results emphasized the role of energy services that required
both for human development and mitigating climatic variations. The
statistical link clearly shows the strong positive correlation between
provision of energy services and human development index. Dancy and
Wiebelhaus-Brahm [20] explored the relationship between transitional
justice with three human development indices i.e., economic develop-
ment, income inequality and human development. The results con-
firmed the statistical positive correlation between human development,
income inequality and transitional justice; however, it is uncorrelated
with the economic development. Ranganathan et al. [42] explored
different sustainable development goals in order to improve the well-
being of the common peoples in terms of child mortality and CO2
emissions. Sweidan and Alwaked [54] statistically examined the impact
of economic development on the energy intensity of human well-being
(EIWB). The data set is used from 1995 to 2012 in the panel of GCC
countries for estimation purpose. The results show that there is a
significant and positive association between economic development
and EIWB in the region.
The above cited studies indicate the positive relationship between
human development and sustainable policies in order to mitigate
environmental hazards across the globe. The present study seeks to
examine the interrelationship between CO2 emissions and eHDI in the
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

panel of selected countries. The overall results of the literature survey
confirmed all of the six plausible hypotheses which have a close
association with the sustainable development across the globe.
1.7. Contribution of the study
Although, the study topic is widely assessed area in academic and
research arena, however, there is still a significant gap in the literature.
The previous literature majorly focused on EKC hypothesis with
different air pollutants (see, [66,15], etc), while some studies added
renewable energy, non-renewable energy, energy demand, trade open-
ness, and urbanization in EKC framework (see, [31,40], etc). The
present study follows the sequential empirical illustrations from the
existing literature and included few more variables in the EKC frame-
work i.e., FDI inflows, population growth, social expenditures (i.e.,
education and health indicators) and sectoral value added indicators
(i.e., agriculture, industry and services value added) for robust policy
inferences. Secondly, the previous studies added FDI inflows (see,
[47,55]), financial development indicators (see, [45]), trade openness
and institutional factors (see, [28]) to determine the Pollution Haven
Hypothesis (PHH) across diversified economic settings, while as per
authors knowledge, none of the study added FDI inflows, trade
openness, social expenditures, and sectoral value added in EKC
framework to investigated the PHH in a large panel settings. Thirdly,
the previous studies added different energy sources in relation with
carbon emissions including nuclear energy (see, [37,3]), renewable
energy (see, [13]), energy demand (see, [1,2]), and disaggregated
energy demand (see, [16]), while the present study used energy
demand along with social expenditures and sectoral value added under
the EKC framework to support the energy induced emissions across
nations. Fourth, the relationship between carbon emissions and
sectoral value added is less explored area, and mainly exercised on
Malaysia (see, [41]) and Chinese economy (see, [39], Cheng et al. [17],
[65] etc.) while this study added the larger set of variables and larger
pool of the countries to estimate the sectoral induced carbon emissions
across the globe. Finally, the emissions emancipated human develop-
ment index (eHDI) mainly explored in relationship between carbon
emissions and income poverty (see, [38]), income inequality (see, [20])
and child mortality (see, [42]), while this study included more policy
intervening variables in the regression apparatus i.e., education and
health expenditures to analyze the eHDI hypothesis in a larger set of
countries panel. These differences clearly indicate the need of inter-
active environmental model where the entire hypothesis included in a
single regression equation to the diversified panel of countries for
robust inferences. The present study focused on these alternative and
plausible environmental hypotheses and proposed an interactive
sustainable model across the globe.
2. Data source and methodological framework
The study used number of promising variables to examine the
impact of growth factors, sectoral growth and human development
factors on CO2 emissions in the panel of selected low income countries
(income ranges from $1035 or less), lower middle and upper middle
income countries (income ranges from $1036 to $ 12,615), and high
income countries (income ranges from $12,616 or more). The study
used stratified random sampling technique, as World Bank classified
world income groups in three strata i.e., low income countries, middle
income countries, and high income countries. The study randomly
selected 90 countries from these three strata including 25 countries
from low income countries, 42 from lower middle and upper middle
income countries, and 23 high income countries for the general
representation of the respective world income groups for the period
of 1975–2015. The classification of countries has been taken from
World Bank [62] on the basis of Gross National Income (GNI) per
capita. The following variables are used in order to assess the growth
factors including energy consumption, foreign direct investment (FDI)
inflows, GDP per capita, trade openness, and annual percentage of
population growth; while, sectoral growth comprises agricultural value
added, industry value added and services value added. Human devel-
opment factors include health expenditures per capita, and government
expenditures on education. CO2 emissions served as a dependent
variable in order to assess the environmental quality in the panel of
low, middle and high income countries. The annual time series data of
90 countries from 1975 to 2015 taken from World Development
Indicators (WDI) published by World Bank [63] and International
Financial Statistics (IFS) published by IMF (2015). The panel data set
is unbalanced in many countries across the region. Table 1 shows the
list of variables and their expected signs.
Table 1 shows the list of variables and their expected signs. The
relationship between GDP per capita and CO2 emissions is expected to
be positive while it tends to become negative after squaring the per
capita GDP, which supports the EKC hypothesis across countries. The
impact of FDI inflows and trade openness on CO2 emissions is expected
to be positive which support the ‘pollution haven hypothesis’ that
describes the intensity of carbon emissions due to industrial growth in
the developed countries. However, as time surpass, developed econo-
mies strictly regulate their environmental policies that restrict the
pollution intensive products and imposed high taxes for the polluting
industry. This regulation shifts the burden from developed countries
towards developing countries, where low wages and less environmental
regulations provide pollution haven to the industries (see, Akbostnci
et al., 2007, [7] etc.). The study further evaluated the IPAT hypothesis
which indicates that massive population hindrance to achieve environ-
mental sustainability. The relationship between energy demand and

Table 1
List of Variables.

Variables Symbol Measurement Expected Sign Theoretical hypothesis

Carbon dioxide emissions CO2 Metric tons per capita

GDP per capita GDPPC Constant 2010 US $ + Environmental Kuznets curve
GDP per capita square GDPPC2 Constant 2010 US $ –
Foreign direct investment, FDI % of GDP + Pollution Haven Hypothesis
inflows
Trade Openness TOP % of GDP +
Population POP Annual % + Environmental Impacts (I) are the product of Population (P), Affluence (A)
and Technology (T) i.e., IPAT hypothesis
Energy use ENRG kg of oil equivalent per capita + Energy induced emissions (EIE)
Agriculture, value added AGRV Annual % of growth + Primary source of air pollution that not recycled waste.
Industry, value added INDV Annual % of growth +
Services, value added SERV Annual % of growth +
Health expenditures per capita HEXPPC Current US $ + Emissions emancipated Human Development Index (eHDI)
Government expenditures on GEE % of GDP +
education

1122
K. Zaman, M.A.-e. Moemen
CO2 emissions is expected to be positive to support energy induced
emissions, while the sectoral growth including agriculture value added,
industry and service value added also exposed to the environment due
to not recycled wastes of manufacturing industries, power generation,
road, rail & air transport and agriculture & timber production (see,
[25,56,23] etc.). Costa et al. [19] agreed the notion ‘Emissions
emancipated Human Development Index (eHDI)’ which is more prone
to social expenditures induced emissions that need to be addressed
caution with the environmental degradation. The lists of sample
countries are shown in appendix A-Table A1.
The study evaluated the following hypothesis in order to assess the
relationship between economic development and CO2 emissions in the
panel of selected countries i.e.,
H1. There will be an inverted U-shaped relationship between CO2
emissions per capita and per capita income in the context of low,
middle, high income countries.
H2. Trade openness and FDI inflows both will increase CO2 emissions.
H3. The rapid increase in population growth will significantly associate
with the increase CO2 emissions.
H4. Higher energy demand deteriorates the environmental quality.
H5. The rapid increase in the sectoral growth will damages the
environmental quality, and
H6. Human development index will be correlated with the increase
CO2 emissions in the panel of selected countries.
The following non-linear regression is used to examine the impact
of economic growth on CO2 emissions in the panel estimation mode
i.e.,
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (λ )i, t
+ εi, t
Where CO2 is carbon dioxide emissions, GDPPC is GDP per capita,
GDPPC 2 is the square of GDP per capita, λ is the set of other
explanatory variables, ‘i′ is cross-section identifiers, ‘t′ is time period.
Eq. (1) shows the EKC hypothesis, as expected β1 > 0, β2 < 0 that
show the inverted U-shaped relationship between CO2 emissions and
GDP per capita. λ is the set of explanatory variables including FDI and
trade openness, both would facilitate to specify the industry intensive
pollution growth, called ‘pollution haven hypothesis’. Eq. (1) now
becomes a more elaborative form i.e.,
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (FDI )i, t
+ β4Log(TOP )i, t +
β5Log(λ )i, t + εi, t
Where FDI is foreign direct investment inflows and TOP is trade
openness. As expected both .. which implies the pollution haven effect
across the countries. Further, we added population growth in Eq. (2)
that provides the existence of IPAT, i.e., environmental impact depends
upon the population growth, followed by the affluence and technology.
We have already included ‘wealth’ (i.e., GDPPC) in Eq. (2) that help to
capture the impact of ‘affluence’ on carbon emissions, while for
technology indicator; both the trade openness and FDI has been added
in Eq. (2) that corresponds to the technological stocks. The more
advanced form of Eq. (2) is as follows:
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (FDI )i, t
+ β4Log(TOP )i, t +
β5Log(POP )i, t + β6Log(λ )i, t + εi, t
Where, POP is annual percentage of population growth.
Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130
As IPAT hypothesis based on technology acceptance model, there-
fore, for utilizing the technological factors in the regression equation,
we added energy demand factor to evaluate energy induced emissions,
i.e.,
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (FDI )i, t
+ β4Log(TOP )i, t +
β5Log(POP )i, t + β6Log(ENRG )i, t + β7Log(λ )i, t + εi, t
(4)
Where, ENRG is energy demand.
Sectoral growth variables considered the primary sources of pollu-
tion, therefore, these variables further included in Eq. (4) i.e.,
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (FDI )i, t
+ β4Log(TOP )i, t +
β5Log(POP )i, t + β6Log(ENRG )i, t + β7Log(AGRV )i, t + β8Log(INDV )i, t
+ β9Log(SERV )i, t
β10Log(λ )i, t + εi, t
(5)
Where, AGRV is agriculture value added, INDV is industry value
added, and SERV is service value added.
Finally, human development indicators added in Eq. (5) to assess
the Emissions emancipated Human Development Index (eHDI) in-
cluding health expenditures per capita (HEXPPC) and government
expenditures on education (GEE) i.e.,
Log(CO2i, t ) = β0 + β1Log(GDPPC )i, t + β2Log(GDPPC )i2, t + β3 log (FDI )i, t
+ β4Log(TOP )i, t +
β5Log(POP )i, t + β6Log(ENRG )i, t + β7Log(AGRV )i, t + β8Log(INDV )i, t
(1) + β9Log(SERV )i, t
β10Log(HEXPPC )i, t + β11Log(GEE )i, t + εi, t
(6)
The present study employed different panel unit root tests i.e., Im
et al. [29] and ADF & PP Fisher chi-square unit root test [35] to check
the stationarity series of the candidate variables, while Johansen
system cointegration test is used for long-run relationships between
the variables. The study further employed difference Generalized
Method of Moments (Panel GMM) estimator and dynamic fixed effect
regression model that avoid the spurious regression by using appro-
priate instrumental list of the variables (in difference GMM estimator),
i.e.,
ΔLog(CO2i, t ) = β0 + β1ΔLog(CO2)i, t −1 + β2ΔLog(GDPPC )i, t
(2)
+ β3ΔLog(GDPPC )i2, t + β4Δ log (FDI )i, t
+β5ΔLog(TOP )i, t + β6ΔLog(POP )i, t + β7ΔLog(ENRG )i, t + β8ΔLog(AGRV )i, t
+ β9ΔLog(INDV )i, t
+β10ΔLog(SERV )i, t + β11ΔLog(HEXPPC )i, t + β12ΔLog(GEE )i, t + zi, t + Δεi, t
(7)
Where, Δ shows first difference and ‘z′ is the list of instrumental
variables which are taken the first lag of the explanatory variables. Eq.
(7) is used to access the impact of growth factors, sectoral growth and
human development indicators on CO2 emissions in the different
regions of the world i.e., low income countries (i=25), middle income
countries (i=42), high income countries (i=23) and total countries
(i=90).
The major motivation is to employed difference GMM estimator to
produce correct and robust inferences without affected serial correla-
(3) tion in the stochastic terms. Arellano and Bond [14] provide a sound
debate on the GMM estimator to fix endogeneity problems and
1123
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

heteroskedasticity, which produces efficient parameter estimates. By
first differencing GMM estimator not only eliminates the country-
specific effects while it further account for possible endogeneity among
regressors. The first difference ‘response variable i.e., CO2 emissions in
this study used first lagged period that provide a valid instrument in the
idiosyncratic disturbance terms. The first differenced regressors with
lagged period served as instrumental variables for unbiased parameter
estimates. The difference GMM estimator further gives an analytical
estimation tool to find an evidence of possible autocorrelation in the
given model, for this purpose, we applied AR(1) and AR(2) serial
correlation tests to investigate whether the stochastic term exhibits an
evidence of serial correlation. The positive and significant coefficient
value of lagged dependent variable would indicate the inertia in the
level of carbon emissions that provide the sound rationale to form
dynamic panel model. The insignificant statistics value of AR(2) would
implies the absence of serial correlation at second order in the
differenced errors thus to support the validity of instruments in the
given model. Thus, the log-log model is used to estimate the given
models in order to interpret the results in elasticity form.
3. Results
Table 2 shows the descriptive statistics of the variables, which
includes maximum value, minimum value, mean, standard deviation,
skewness and kurtosis. In addition, time period and cross-section
identifiers also presented for the ready reference. These statistics is
shown as a rough sketch of the candidate variables in the panel of
selected countries.
Table 3 shows the correlation matrix between the candidate
variables. The results indicate that there is a positive correlation
between CO2 emissions and energy demand that confirmed the energy
induced CO2 emissions, while the positive relationship between CO2
emissions, FDI inflows, and trade openness support the pollution
haven hypothesis in a region. The relationship between CO2 emissions
and per capita income shows the monotonic increasing relationship
between them, whereas, the negative relationship between population
growth and CO2 emissions rejected the IPAT hypothesis across nations.
There is a negative correlation between sectoral value added and CO2
emissions, which does not support the sectoral induced CO2 emissions,
however, the correlation results validate the eHDI emissions, as public
spending on education and health significantly correlated with the CO2
emissions in a panel of countries.
After analyzing the descriptive statistics and correlation matrix,
there is a substantial need to check the stationarity properties of the
given variables. Table 4 shows the different panel unit root tests to
check the order of integration of the given variables.
The results of IPS panel unit root test show that FDI inflows (FDI),
agriculture value added (AGRV), industry value added (INDV), and
service value added (SERV) stationary at their level, while remaining
variables are differenced stationary. Both the ADF Fisher chi-square
and PP Fisher chi-square tests indicate that per capita GDP (GDPPC),
health expenditures (HEXPC) and education expenditures (GEE) are
differenced stationary variables while the remaining variables are level
stationary. The high volatility in the data series of the variables show
insignificant probability values that tend to be significant at their first
difference, which confirm the order of integration is one, i.e., I(1).
Table 5 shows the cross-section dependency estimates and confirmed
that all the selected variables have a strong dependency in the cross-
sectional identifier that gives a good justification to used dynamic panel
estimators that reduced the problem of simultaneity from the given
models.
The study employed difference panel GMM estimator and panel
fixed effect regression on aggregate data, low, middle & upper middle
income countries, high income countries, and aggregated panel of
countries and presented the results in Table 6.
The results confirmed the existence of EKC hypothesis by using
both of the panel estimators (i.e., panel GMM and panel fixed effect) in
the panel of aggregate countries. This result confirmed that at the
initial level of development, CO2 emissions increases along with the
increase in per capita income (GDPPC), however, at the later stages of
development, CO2 emissions significantly decreases. This result pro-
vides good insight about the existence of carbon - EKC hypothesis
across countries. Policy makers should have to pursue sustainable
policies in order to reap the maximum benefit from economic growth,
caution with care the handling issues of environmental pollution. The
results are in line with the previous studies of Anand and Kanbur [10],
Stern et al. [53], Andreoni and Levinson [11], Dinda [21], [6,8],
Apergis and Ozturk [12] etc., as these studies described the different
shapes of EKC in diversified panel of countries. The results do not
validate the pollution haven hypothesis in the aggregated data, as FDI
inflows and trade openness (TOP) both does not signify their positive
impact on CO2 emissions, however, we may not ignore the significant
impact of sectoral value added on CO2 emissions, as industry value
added (INDV) and services value added (SERV) both significantly
correlated with the CO2 emissions in the panel fixed effect regression
model, however, agriculture value added (AGRV) significantly de-
creases CO2 emissions by using panel GMM estimator. The previous
literature confirmed that FDI inflows and trade openness (TOP) have a
positive impact on CO2 emissions i.e., Akbostnci et al. (2007), Shahbaz
et al. [47], Cole et al. [18], Al-Mulali and Ozturk [7] etc., that
substantiate the viability of polluting industries all across the world.
In the fixed effect regression model, it is evident that energy demand

Table 2
Descriptive Statistics.
Source: [63].

Statistics AGRV CO2 ENRG FDI GDPPC Square of GEE HEXPC INDV POP SERV TOP
GDPPC
(% of GDP) (metric (kg of oil (% of GDP) (constant (constant ( %of (current (annual (annual %) (annual (% of
ton per equivalent 2010 US$) 2010 US$) GDP) US$) % of % of GDP)
capita) per capita) growth) growth)

Mean 1.964 5.536 2449.197 4.525 18539.100 3.87E+08 4.621 1358.803 3.147 0.942 4.024 82.863
Maximum 39.551 19.748 7974.711 142.416 75424.280 5.74E+09 10.679 9386.514 49.255 5.322 42.671 439.657
Minimum −41.008 0.065 135.210 −79.737 244.137 13325.39 0.010 3.010 −73.661 −2.258 −15.898 16.439
Std. Dev. 7.855 4.273 1806.447 8.207 18805.120 7.73E+08 1.450 1782.222 6.888 1.071 4.105 53.188
Skewness −0.291 0.714 0.706 6.199 0.877 2.874706 0.111 1.616 −1.182 0.012 0.675 3.087
Kurtosis 7.383 3.273 2.917 115.620 2.510 13.23798 3.199 5.134 24.804 3.263 15.031 17.796
Countries 90 90 90 90 90 90 90 90 90 90 90 90
Observations 2695 3098 2779 2981 3147 3147 2572 2585 2593 3489 2600 3071
Time Period 1975–2015 1975– 1975–2015 1975–2015 1975–2015 1975–2015 1975– 1975– 1975– 1975–2015 1975– 1975–
2015 2015 2015 2015 2015 2015

1124
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

Table 3
Correlation Matrix.

Correlation

Probability AGRV CO2 ENRG FDI GDPPC GEE HEXPC INDV POP SERV TOP

AGRV 1
–

CO2 −0.091* 1
(0.005) –

ENRG −0.095* 0.898* 1

(0.004) (0.000) –

FDI −0.038 0.144* 0.162* 1

(0.248) (0.000) (0.000) –

GDPPC −0.112* 0.676* 0.802* 0.168* 1

(0.000) (0.000) (0.000) (0.000) –

GEE −0.032 0.312* 0.377* 0.080** 0.389* 1

(0.331) (0.000) (0.000) (0.015) (0.000) –

HEXPC −0.087* 0.607* 0.727* 0.111* 0.909* 0.381* 1

(0.008) (0.000) (0.000) (0.000) (0.000) (0.000) –

INDV 0.150* −0.114* −0.123* 0.058*** −0.165* −0.162* −0.201* 1

(0.000) (0.000) (0.000) (0.076) (0.000) (0.000) (0.000) –

POP 0.092* −0.387* −0.383* −0.027 −0.222* −0.189* −0.194* 0.136* 1

(0.005) (0.000) (0.000) (0.415) (0.000) (0.000) (0.000) (0.000) –

SERV 0.068** −0.221* −0.250* 0.048 −0.285* −0.209* −0.310* 0.367* 0.249* 1
(0.040) (0.000) (0.000) (0.147) (0.000) (0.000) (0.000) (0.000) (0.000) –

TOP −0.069** 0.167* 0.228* 0.347* 0.173* 0.120* 0.031 0.052 −0.002 0.071** 1
(0.0365) (0.000) (0.000) (0.000) (0.000) (0.000) (0.348) (0.1130) (0.9468) (0.0305) –

Note: Small bracket shows probability values. *, **, and *** shows 1%, 5%, and 10% level of significance respectively.

Table 4
Results of Panel Unit Root.

Variables Im, Pesaran, Shin (IPS) ADF Fisher Chi-square PP Fisher chi-square

Level 1st Difference Level 1st Difference Level 1st Difference

CO2 1.430 −49.234* 266.436* 2030.35* 522.375* 2112.56*

(0.923) (0.000) (0.000) (0.000) (0.000) (0.000)
GDPPC 25.115 −28.260* 61.766 1136.19* 57.975 1169.71*
(1.000) (0.000) (1.000) (0.000) (1.000) (0.000)
FDI −14.315* −64.137* 597.123* 2466.17* 589.143* 2480.78*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
POP −0.402 −24.396* 296.403* 920.627* 298.636* 978.898*
90.343) (0.000) (0.000) (0.000) (0.000) (0.000)
TOP −0.571 −47.540* 206.465*** 1948.99* 202.198 2149.85*
(0.283) (0.000) (0.085) (0.000) (0.123) (0.000)
ENRG 2.260 −39.524* 226.866* 1524.26* 233.531* 1531.12*
(0.998) (0.000) (0.000) (0.000) (0.000) (0.000)
AGRV −58.391* −88.200* 2167.31* 2814.91* 2665.60* 4711.33*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
INDV −30.207* −64.332* 1236.25* 2517.25* 1292.38* 2898.94*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
SERV −27.200* −62.650* 1121.75* 2424.43* 1120.23* 2455.83*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
HEXPPC 18.751 −18.463* 24.164 659.883* 25.496 682.240*
(1.000) (0.000) (1.000) (0.000) (1.000) (0.000)
GEE 2.047 −18.634* 146.990 707.630* 146.999 1048.55*
(0.979) (0.000) (0.852) (0.000) (0.852) (0.000)

Note: AGRV is agriculture value added, CO2 is carbon dioxide emissions, ENRG is energy use, FDI is foreign direct investment inflows, GDPPC is GDP per capita, GEE is government
expenditures on education, HEXPPC is health expenditures per capita, INDV is industry value added, POP is population growth, SERV is service value added and TOP is trade openness.
* and *** indicates 1% and 10% significant level. Small bracket shows probability values.

1125
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

Table 5
Estimates of Cross-Section Dependency Test.

Methods AGRV CO2 ENRG FDI GDPPC HEXPC INDV POP SERV TOP

Breusch-Pagan LM 4551.789* 44128.70* 42207.68* 18549.08* 88105.23* 63453.38* 8245.584* 42459.1* 8340.808* 5010.903*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Pesaran scaled LM 9.483* 447.310* 469.362* 164.342* 938.676* 663.232* 52.6860* 428.656* 53.799* 210.492*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Bias-corrected scaled LM 8.408* 446.126* 468.298* 163.230* 937.551* 660.864* 51.611* 427.531* 52.724* 210.204*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Pesaran CD 4.721* 28.037* 49.644* 89.484* 234.177* 248.792* 41.553* 65.145* 38.633* 63.243*
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

Note: * shows 1% level of significance. Small bracket shows probability value.

Table 6
Estimates of Panel GMM Estimator and Fixed Effect Regression.

Variables Aggregate Data High Income Countries Low income, lower Middle and Upper Middle
Income Countries

Difference GMM Fixed Effect Difference GMM Fixed Effect Difference GMM Estimates Fixed Effect
Estimates Estimates

C −0.910** −6.264* −14.748 7.052 −9.088* −6.370*

(0.045) (0.000) (0.143) (0.674) (0.000) (0.000)
ln(CO2)i,t−1 0.960* 0.593* 0.973* 0.308* 0.587* 0.555*
(0.000) (0.000) (0.000) (0.003) (0.000) (0.000)
ln(GDPPC)i,t 0.181** 0.737* 2.849 −2.500 1.445* 0.857*
(0.025) (0.000) (0.137) (0.439) (0.000) (0.000)
ln(GDPPC^2) i,t −0.010** −0.042* −0.132 0.126 −0.083* −0.050*
(0.039) (0.000) (0.143) (0.406) (0.003) (0.000)
ln(FDI) i,t −0.013 −0.00059 0.0013 −0.025* −0.0086 0.0060
(0.138) (0.849) (0.858) (0.001) (0.358) (0.344)
ln(TOP) i,t 0.010 0.013 −0.326*** −0.143*** 0.013 0.0023
(0.354) (0.462) (0.055) (0.062) (0.796) (0.933)
ln(POP) i,t 0.005 0.0082 0.021** −0.011 0.018 −0.013
(0.390) (0.185) (0.029) (0.541) (0.589) (0.310)
ln(ENRG) i,t 0.004 0.478* −0.030 0.932* 0.462* 0.414*
(0.778) (0.000) (0.233) (0.000) (0.007) (0.000)
ln(HEXPC) i,t 0.014 −0.0045 −0.024 −0.048 −0.013 0.026
(0.390) (0.727) (0.167) (0.091) (0.777) (0.281)
ln(GEE) i,t 0.019 0.0074 0.016 0.021 0.015 0.019
(0.212) (0.697) (0.585) (0.821) (0.774) (0.442)
ln(AGRV) i,t −0.031** −0.00035 −0.00031 −0.0052 −0.0019 8.02E−05
(0.013) (0.266) (0.663) (0.219) (0.526) (0.984)
ln(INDV) i,t 0.041 0.0011* 0.0059* −0.0055 0.0057* 0.0058
(0.112) (0.006) (0.000) (0.421) (0.009) (0.305)
ln(SERV) i,t −0.011 0.0020* −0.0087** −0.0060 −0.010 0.019**
(0.515) (0.004) (0.017) (0.565) (0.294) (0.035)

Estimation of Turning Point for inverted U-shaped EKC

Extremum point = −βˆ /2γˆ 9.05 8.773 8.704 8.570
90% confidence interval, [7.44, 9.58]
Fieller method
90% confidence interval, Delta [7.33, 9.47]
method

Statistical Tests
R-squared 0.989 0.997 0.951 0.985 0.9949 0.9979
Adjusted R-Squared 0.988 0.996 0.948 0.980 0.9941 0.9974
F-statistics – 3263.694* – 179.531* – 2184.444*
J-statistics 1.79E−15 – 7.48E−07 – 5.56E−11 –
Hausman Test (Chi-square – 277.214* – 84.808* – 130.663*
statistics)
J-statistic p > 0.090 – p > 0.090 – p > 0.090 –
AR(1)-prob.value 0.994 – 0.901 0.000
AR(2) – prob-value 0.955 – 0.856 0.737
Number of Observations 1621 2414 328 541 857 1524
Counties 90 90 23 23 67 67

Note: Dependent variable: log (CO2)i, t . *, ** and *** indicates 1%, 5% and 10% significant level. Instrumental variables used as the first lag of the explanatory variables. Small bracket
shows “probability values”.

1126
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

Fig. 1. With or without ‘U′ EKC. a) Robustness Check for Aggregate EKC b) Robustness Check of EKC for low and middle income countries.

(ENRG) has a positive and significant association with the CO2 The study applied robustness check for with or without ‘U′ as
emissions that substantiate the energy induced emissions in an proposed by Lind and Mehlum [34] (in some other studies i.e., [22,27]
aggregated panel data. The results are in line with the previous studies etc) and found that the extremum point of inverted U-shaped Kuznets
of Soytas et al. [50], Zhang and Cheng [68], Soytas and Sari [49] etc., as curve for aggregated data and low income & middle income countries
these studies confirmed the existence of energy induced emissions that fall in the critical values of Fieller [7.44, 9.58] and Delta method
across countries. [7.33, 9.47] at 90% confidence interval i.e., the value of extremium
The results of high income countries show that there is no/flat point for aggregated data by panel GMM estimator is about 9.05 and by
relationship between CO2 emissions and economic growth (GDPPC), by panel fixed effect method is about 8.773, while the extremium point
using both of the panel estimators. The results of both the panel value for low income & middle income countries for panel GMM
estimator describe that trade openness (TOP) have a negative and estimator is about 8.704 and by panel fixed effect regression is about
statistical significant relationship with the CO2 emissions, as if there is 8.570. The point of attention is that the extremium values are quite
1% increase in trade openness (TOP), CO2 emissions decreases by higher in panel GMM estimator as compared to panel fixed effect
−0.326% in panel GMM estimator and −0.143% in panel fixed effect, regression. The other statistical tests confirmed the good explanatory
while by using the panel fixed effect estimates, it is evident that FDI power to explain response variable, while model stability is proven by
inflows also have a negative correlation with the CO2 emissions, as if F-statistics in panel fixed effect regression model. The Sargan-Hansen
there is 1% increase in FDI inflows, per capita income decreases by J-statistics confirmed the instrumental validity that enforces the
−0.025% points. The results do not support the pollution haven reliability of the coefficient estimates by using panel GMM estimator,
hypothesis, as both the FDI inflows and trade openness (TOP) have while the absence of serial correlation at second order confirmed the
an insignificant probability values. In the fixed effect regression model, validity of instruments in the given model. Fig. 1 shows the inverted U-
the result confirmed the energy induced carbon emissions, as 1% shaped EKC relationship in different world income group as per
increase in energy demand (ENRG) increases CO2 emissions by estimated results.
0.932%. The sectoral value added on CO2 emissions have a differential The summary of the hypothesis are presented in Table 7 for the
impact, as industry value added (INDV) contribute to escalate CO2 ready reference.
emissions, while services value added (SERV) reduces the CO2 emis-
sions that partially supported the sectoral induced carbon emissions in Table 7
a panel of countries. The results supported the findings of Hansen et al. Summary of the Hypotheses.
[25], Tegtmeier and Duffy [56], Frankel and Rose [23] etc., as these
S.No. Hypothesis Low income, Lower High income Total
studies confined the findings of sectoral induced CO2 emissions in a upper and middle countries Countries
panel of countries. upper income
Finally, in a panel of low income and middle income countries, the countries
result supports the EKC hypothesis. The impact of energy consumption
1 Existence of EKC √ ≠ √
(ENRG) on CO2 emissions is positive in both of the panel estimator 2 Existence of PHH ≠ ≠ ≠
that confirmed the energy induced carbon emissions in a region. The 3 Existence of IPAT ≠ √ ≠
positive impact of sectoral value added on CO2 emissions is visible, as 4 Existence of Energy √ √ √
panel GMM estimator confirmed that industry value added (INDV) led emissions
5 Existence of eHDI ≠ ≠ ≠
increases CO2 emissions, while panel fixed effect regression model
6 Existence of √ Partial √
confirmed the positive relationship between service value added Sectoral growth accepted
(SERV) and CO2 emissions that support the sectoral induced carbon associated
emissions in a panel of countries. The eHDI hypothesis does not emissions
establish, as the impact of social expenditures on CO2 emissions was
Note: √ indicate the acceptance of hypothesis, while ≠ indicate rejection of the
insignificant. The previous literature, however, confirmed the eHDI
hypothesis.
hypothesis in the studies of Shanthini [48] and Costa et al. [19].

1127
K. Zaman, M.A.-e. Moemen
The results of Table 7 indicate the existence of different hypothesis
across the world regions. Out of six hypotheses, the visibility of four
hypothesis including EKC hypothesis, energy induced emissions, IPAT
hypothesis and sectoral induced emissions across selected countries,
while the remaining two hypothesis including PHH and eHDI, both
does not find any existence during the study time period. The existence
of EKC hypothesis disappeared in high income countries while in the
aggregated panel data and low income and middle income countries,
the visibility of inverted U-shaped EKC relationship been hold. The
positive impact of FDI inflows and trade openness is statistical
insignificant; however, energy associated emissions and sectoral in-
duced carbon emissions clearly observable which create the doubt of
presence of polluting industries that may affiliate in different regions of
the world. The impact of social expenditures on CO2 emissions is
insignificant, which does not validate the eHDI hypothesis in a panel of
selected countries. The results confirmed the strong association
between CO2 emissions and income per capita, which should need a
strong policy vista for the environmental reforms that need to be
caution with the matter of sustainability across the globe.
4. Conclusions
Environmental Kuznets curve is one of the oldest debates in the
environmental literature. The rise and fall of EKC have different policy
implications that set for policy makers to device sustainable policy for
different regions of the world. Low income countries have a blame to
destroy the natural environment; however, it is evident that polluting
industries have shifted from developed countries to developing coun-
tries because of their low environmental reforms. The developed
countries should have to limit the polluting industries relatively to
provide an opportunity to shift their resources towards developing
countries.
This study provides an opportunity to the policy makers in order to
redesign their environmental strategies for global sustainability. The
objective of the study is to examine the six different hypothesis from
across the countries i.e., EKC, PHH, IPAT hypothesis, energy led
emissions, sectoral growth associated with the CO2 emissions, and
Emissions emancipated Human Development Index (eHDI). The result
does not show any visible sign of eHDI and PHH in the panel of
selected countries. The sectoral growth associated emissions including
industry value added tend to increase CO2 emissions in low & middle
income countries, and high income countries, while agricultural value
added decreases the CO2 emissions in the panel of total 90 selected
countries. In evaluating IPAT hypothesis, population growth increases
CO2 emissions in the panel of high income countries. The impact of
energy demand on CO2 emissions is positive and significant in the
panel of selected countries. The existence of inverted U-shaped EKC
hypothesis is visible across countries, while there is flat/no relationship
between income and emissions in the panel of high income countries.
Appendix A
See the Table A1 here.
1128
Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130
Financial liberalization and trade liberalization policies both are
good for economic growth; however, there is considerable need to
device environmental friendly policies that would not be compromised
with the progressive growth. Developing countries provided the vacant
place for the pollution haven industries; therefore, there is a strong
need to fill this vacant post with the carbon free policies. Energy
demand is considered one of the important vehicle for economic
growth, this vehicle need too much energy for gearing their momentum
in an efficient manner, however, this momentum produce greater CO2
emissions, so policies should be formulated to handle with the care
while using energy intensity to promote economic growth. The inverted
U-shaped relationship between CO2 emissions and per capita income,
have a significant policy implications i.e., at the early stages of
development, CO2 emissions increases, while at later stage of develop-
ment, CO2 emissions significantly go down. The policies should be
device in a way to reduce CO2 emissions at the later stages of economic
development. Pursuit of excellence in world development, in the midst
of damaging the natural bonanza, is a shameless growth. The ‘pollution
haven hypothesis’ in developing countries surpasses the income per
capita that substantially increases CO2 emissions in the region. Apart
from developed countries, the existence of EKC gives food-for-thought
for the environmentalist to device sustainable policies. Energy demand
associated with the high carbon emissions in developed countries need
policy caution for sustainability matters. The results of the study
emphasized the strong policy vista for environmental sustainability
across the globe.
The following policy implication emphasized the need of environ-
mental friendly policies that need to sustain long term policy vista
across the globe. World is like a stadium where every player wants to
win-win situation, whatsoever the cost that it would be bear. This
competition damage the natural flora of the world's resources that
considered is the brazen growth for the countries. Therefore, there is a
strong need to set an optimistic target for growth that would easily be
achieved without the cost of environmental degradation across the
globe. The contribution of the study is to evaluate with or without ‘U′
shaped EKC across countries. In addition, this study confirmed the
energy induced emissions followed by industrial emissions. The
population pressure further escalates the CO2 emissions that support
the IPAT hypothesis in high income countries. This study designed to
evaluate environmental resource policy to attract the policy makers to
device long-term sustainability policies for better health and wealth.
Funding
The authors are thankful to the College of Law and Political Science,
Deanship of Scientific Research, King Saud University, Riyadh Saudi
Arabia for funding research through research group project RGP-VPP-
280.
K. Zaman, M.A.-e. Moemen Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130

Table A1
List of Countries.
Source: [68].

Low Income Lower Middle and Upper High Income

Countries Middle Income Countries Countries

Bangladesh Algeria Austria

Benin Angola Australia
Burkina Faso Argentina Belgium
Central African Bhutan Barbados
Republic Bulgaria Canada
Chad Brazil Cyprus
Ethiopia Botswana Denmark
Ghana Cuba Finland
Haiti China France
Kenya Colombia Germany
Liberia Chile Greece
Malawi Czech Rep. Ireland
Mozambique Costa Rica Italy
Nepal Estonia Japan
Pakistan Fiji Newzealand
Rwanda Georgia Netherland
Senegal Indonesia Portugal
Sudan India Spain
Tajikistan Jordan Singapore
Tanzania Kazakhstan Sweden
Uganda Korea Rep. Switzerland
Uzbekistan Jamaica United Kingdom
Vietnam Lativa United States
Yemen, Rep. Lithuania
Zambia Libya
Zimbabwe Malaysia
Maldives
Morocco
Mexico
Namibia
Philippine
Poland
Romania
Serbia
Slovenia
South Africa
Sri lanka
Tunisia
Thailand
Turkey
Ukraine
Uruguay
Total: 25 42 23

hypothesis in Vietnam. Energy Policy 2015;76:123–31.

References
[1] Ajmi AN, Hammoudeh S, Nguyen DK, Sato JR. On the relationships between CO 2
emissions, energy consumption and income: the importance of time variation.
Energy Econ 2015;49:629–38.
[2] Akhmat G, Zaman K, Shukui T, Irfan D, Khan MM. Does energy consumption
contribute to environmental pollutants? Evidence from SAARC countries. Environ
Sci Pollut Res 2014;21(9):5940–51.
[3] Akhmat G, Zaman K, Shukui T, Sajjad F, Khan MA, Khan MZ. The challenges of
reducing greenhouse gas emissions and air pollution through energy sources:
evidence from a panel of developed countries. Environ Sci Pollut Res
2014;21(12):7425–35.
[4] Aktas MT. Did Globalization Promote Dirty Industries in Turkey?. Int J Bus Soc Sci
2015;6(5):205–12.
[5] Alam MM, Murad MW, Noman AHM, Ozturk I. Relationships among carbon
emissions, economic growth, energy consumption and population growth: testing
Environmental Kuznets Curve hypothesis for Brazil, China, India and Indonesia.
Ecol Indic 2016;70:466–79.
[6] Ali S, Waqas H, Ahmad N. Analyzing the dynamics of energy consumption,
liberalization, financial development, poverty and carbon emissions in Pakistan. J
Appl Environ Biol Sci 2015;5(4):166–83.
[7] Al-Mulali U, Ozturk I. The effect of energy consumption, urbanization, trade
openness, industrial output, and the political stability on the environmental
degradation in the MENA (Middle East and North African) region. Energy
2015;84:382–9.
[8] Al-Mulali U, Saboori B, Ozturk I. Investigating the environmental Kuznets curve
[9] Al-Mulali U, Weng-Wai C, Sheau-Ting L, Mohammed AH. Investigating the
environmental Kuznets curve (EKC) hypothesis by utilizing the ecological footprint
as an indicator of environmental degradation. Ecol Indic 2015;48:315–23.
[10] Anand S, Kanbur SR. The Kuznets process and the inequality—development
relationship. J Dev Econ 1993;40(1):25–52.
[11] Andreoni J, Levinson A. The simple analytics of the environmental Kuznets curve. J
Public Econ 2001;80(2):269–86.
[12] Apergis N, Ozturk I. Testing Environmental Kuznets Curve hypothesis in Asian
countries. Ecol Indic 2015;52:16–22.
[13] Apergis N, Payne JE. Renewable energy, output, carbon dioxide emissions, and oil
prices: evidence from South America. Energy Sources, Part B: Econ, Plan, Policy
2015;10(3):281–7.
[14] Arellano M, Bond S. Some tests of specification for panel data: Monte Carlo
evidence and an application to employment equations. Rev Econ Stud
1991;58(2):277–97.
[15] Bernard JT, Gavin M, Khalaf L, Voia M. Environmental Kuznets Curve: tipping
points, uncertainty and weak identification. Environ Resour Econ
2015;60(2):285–315.
[16] Cheema AR, Javid AY. The Relationship between Disaggregate Energy
Consumption, Economic Growth and Environment for Asian Developing
Economies (No. 2015). Pakistan: Pakistan Institute of Development Economics,
Islamabad; 2015. p. 115.
[17] Cheng H, Li F, Mao Q. The empirical analysis on the influence of CO2 emission
regulation on the export transformation of Chinese manufacturing industries. J
Coast Res 2015;73(sp1):209–15.
[18] Cole MA. Trade, the pollution haven hypothesis and the environmental Kuznets
curve: examining the linkages. Ecol Econ 2004;48(1):71–81.
[19] Costa L, Rybski D, Kropp JP. A human development framework for CO2 reductions.
PLoS One 2011;6(12):e29262. http://dx.doi.org/10.1371/journal.pone.0029262.

1129
K. Zaman, M.A.-e. Moemen
[20] Dancy G, Wiebelhaus-Brahm E. Bridge to human development or vehicle of
inequality? Transitional justice and economic structures. Int J Transit Justice
2015;9(1):51–69.
[21] Dinda S. Environmental Kuznets curve hypothesis: a survey. Ecol Econ
2004;49(4):431–55.
[22] Disli M, Ng A, Askari H. Culture, income, and CO 2 emission. Renew Sustain
Energy Rev 2016;62:418–28.
[23] Frankel JA, Rose AK. Is trade good or bad for the environment? Sorting out the
causality. Rev Econ Stat 2005;87(1):85–91.
[24] Halkos GE, Stern DI, Tzeremes NG. Population, economic growth and regional
environmental inefficiency: evidence from US states. J Clean Prod
2016;112:4288–95.
[25] Hansen L, Peters M, Breneman V, Weinberg M, Cattaneo A, Feather P, Smith M.
Agri-environmental policy at the crossroads: guideposts on a changing landscape.
Economic Research Service, US Department of Agriculture; 2001.
[26] Hassan K, Salim R. Population ageing, income growth and CO2 emission: empirical
evidence from high income OECD countries. J Econ Stud 2015;42(1):54–67.
[27] Ibrahim MH, Law SH. Social capital and CO 2 emission—output relations: a panel
analysis. Renew Sustain Energy Rev 2014;29:528–34.
[28] Ibrahim MH, Law SH. Institutional quality and CO2 emission–trade relations:
evidence from Sub‐Saharan Africa. South Afr J Econ 2016;84(2):323–40.
[29] Im KS, Pesaran MH, Shin Y. Testing for unit roots in heterogeneous panels. J Econ
2003;115(1):53–74.
[30] International Energy Outlook. Chapter 9. Energy-related CO2 emissions. U.S.
Department of energy, Washington D.C. Online available at: 〈https://www.eia.gov/
forecasts/ieo/emissions.cfm〉 [accessed on 31.10.16]; 2016.
[31] Jebli MB, Youssef SB. The environmental Kuznets curve, economic growth,
renewable and non-renewable energy, and trade in Tunisia. Renew Sustain Energy
Rev 2015;47:173–85.
[32] Khan MA, Khan MZ, Zaman K, Naz L. Global estimates of energy consumption and
greenhouse gas emissions. Renew Sustain Energy Rev 2014;29:336–44.
[33] Koroneos CJ, Rokos D. Sustainable and integrated development—A critical
analysis. Sustainability 2012;4(1):141–53.
[34] Lind JT, Mehlum H. With or without U? The appropriate test for a U-shaped
relationship. Oxf Bull Econ Stat 2010;72(1):109–18.
[35] Maddala GS, Wu S. A comparative study of unit root tests with panel data and a
new simple test. Oxf Bull Econ Stat 1999;61(S1):631–52.
[36] McGee JA, Clement MT, Besek JF. The impacts of technology: a re-evaluation of
the STIRPAT model. Environ Sociol 2015;1(2):81–91.
[37] Mudakkar SR, Zaman K, Khan MM, Ahmad M. Energy for economic growth,
industrialization, environment and natural resources: living with just enough.
Renew Sustain Energy Rev 2013;25:580–95.
[38] Nussbaumer P, Bazilian M, Patt A. A statistical analysis of the link between energy
and the Millennium Development Goals. Clim Dev 2013;5(2):101–12.
[39] Ouyang X, Lin B. An analysis of the driving forces of energy-related carbon dioxide
emissions in China's industrial sector. Renew Sustain Energy Rev 2015;45:838–49.
[40] Ozturk I, Al-Mulali U. Investigating the validity of the environmental Kuznets curve
hypothesis in Cambodia. Ecol Indic 2015;57:324–30.
[41] Rahman MS, Junsheng H, Shahari F, Aslam M, Masud MM, Banna H, Liya M.
Long-run relationship between sectoral productivity and energy consumption in
Malaysia: an aggregated and disaggregated viewpoint. Energy 2015;86:436–45.
[42] Ranganathan S, Spaiser V, Nicolis SC, Mann RP, Swain RB, Sumpter DJT. (2015).
Insights into Complex Dynamics of Sustainable Development. Online available at:
http://www.collective-behavior.com/wp-content/uploads/2015/04/SDGbooklet_
final.pdf [accessed on 16.07.15].
[43] Robaina-Alves M, Moutinho V, Costa R. Change in energy-related CO2 (carbon
dioxide) emissions in Portuguese tourism: a decomposition analysis from 2000 to
2008. J Clean Prod 2016;111:520–8.
[44] Robalino-López A, Mena-Nieto Á, García-Ramos JE, Golpe AA. Studying the
relationship between economic growth, CO 2 emissions, and the environmental
Kuznets curve in Venezuela (1980–2025). Renew Sustain Energy Rev
1130
Renewable and Sustainable Energy Reviews 74 (2017) 1119–1130
2015;41:602–14.
[45] Salahuddin M, Gow J, Ozturk I. Is the long-run relationship between economic
growth, electricity consumption, carbon dioxide emissions and financial develop-
ment in Gulf Cooperation Council Countries robust?. Renew Sustain Energy Rev
2015;51:317–26.
[46] Selden TM, Song D. Environmental quality and development: is there a Kuznets
curve for air pollution emissions?. J Environ Econ Manag 1994;27(2):147–62.
[47] Shahbaz M, Nasreen S, Abbas F, Anis O. Does foreign direct investment impede
environmental quality in high-, middle-, and low-income countries?. Energy Econ
2015;51:275–87.
[48] Shanthini R. Emancipating the human development index from its intrinsic ties to
CO2 emissions and the resulting climate change. Online available at: 〈http://www.
rshanthini.com/tmp/EmanciHDI.pdf〉 [accessed on 8.07.15]; 2009.
[49] Soytas U, Sari R. Energy consumption, economic growth, and carbon emissions:
challenges faced by an EU candidate member. Ecol Econ 2009;68(6):1667–75.
[50] Soytas U, Sari R, Ewing BT. Energy consumption, income, and carbon emissions in
the United States. Ecol Econ 2007;62(3):482–9.
[51] Stern DI. The rise and fall of the environmental Kuznets curve. World Dev
2004;32(8):1419–39.
[52] Stern DI, Common MS. Is there an environmental Kuznets curve for sulfur?. J
Environ Econ Manag 2001;41(2):162–78.
[53] Stern DI, Common MS, Barbier EB. Economic growth and environmental
degradation: the environmental Kuznets curve and sustainable development. World
Dev 1996;24(7):1151–60.
[54] Sweidan OD, Alwaked AA. Economic development and the energy intensity of
human well-being: evidence from the GCC countries. Renew Sustain Energy Rev
2016;55:1363–9.
[55] Tang CF, Tan BW. The impact of energy consumption, income and foreign direct
investment on carbon dioxide emissions in Vietnam. Energy 2015;79:447–54.
[56] Tegtmeier EM, Duffy MD. External costs of agricultural production in the United
States. Int J Agric Sustain 2004;2(1):1–20.
[57] Torras M, Boyce JK. Income, inequality, and pollution: a reassessment of the
environmental Kuznets curve. Ecol Econ 1998;25(2):147–60.
[58] Kuznets S. Economic growth and income inequality. Am Econ Rev
1955;45(1):1–28.
[59] Tutulmaz O. Environmental Kuznets Curve time series application for Turkey: why
controversial results exist for similar models?. Renew Sustain Energy Rev
2015;50:73–81.
[60] Wang SX, Fu YB, Zhang ZG. Population growth and the environmental Kuznets
curve. China Econ Rev 2015;36:146–65.
[61] Wang Y, Zhang X, Kubota J, Zhu X, Lu G. A semi-parametric panel data analysis on
the urbanization-carbon emissions nexus for OECD countries. Renew Sustain
Energy Rev 2015;48:704–9.
[62] World Bank. Country’s classification - World Economic Situation and Prospects,
World Bank, Washington DC. 2013.
[63] World Bank . World Development Indicators. Washington, D.C.: World Bank;
2015.
[64] World Economic Outlook. Subdued Demand, Diminished Prospects. World eco-
nomic Outlook update. International Monetary Fund, Washington D.C. Online
available at: 〈http://www.imf.org/external/pubs/ft/weo/2016/update/01/pdf/
0116.pdf〉 [accessed on 31.10.16]; 2016.
[65] Xu B, Lin B. How industrialization and urbanization process impacts on CO 2
emissions in China: evidence from nonparametric additive regression models.
Energy Econ 2015;48:188–202.
[66] Yang H, He J, Chen S. The fragility of the Environmental Kuznets Curve: revisiting
the hypothesis with Chinese data via an “Extreme Bound Analysis”. Ecol Econ
2015;109:41–58.
[67] Zhang C, Liu C. The impact of ICT industry on CO 2 emissions: a regional analysis
in China. Renew Sustain Energy Rev 2015;44:12–9.
[68] Zhang XP, Cheng XM. Energy consumption, carbon emissions, and economic
growth in China. Ecol Econ 2009;68(10):2706–12.