See discussions, stats, and author profiles for this publication at: [Link]

net/publication/281635643

CO2 emissions, real output, energy consumption, trade, urbanization and

financial development: testing the EKC hypothesis for the USA

Article in Environmental Science and Pollution Research · January 2016

DOI: 10.1007/s11356-015-5323-8

CITATIONS READS

1,161 3,596

2 authors:

Eyup Dogan Berna Turkekul

Abdullah Gül University Ege University
95 PUBLICATIONS 14,037 CITATIONS 27 PUBLICATIONS 1,450 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Eyup Dogan on 08 March 2018.

The user has requested enhancement of the downloaded file.

Environ Sci Pollut Res (2016) 23:1203–1213
DOI 10.1007/s11356-015-5323-8

RESEARCH ARTICLE

CO2 emissions, real output, energy consumption, trade,

urbanization and financial development: testing the EKC
hypothesis for the USA
Eyup Dogan 1 & Berna Turkekul 2

Received: 4 August 2015 / Accepted: 25 August 2015 / Published online: 9 September 2015
# Springer-Verlag Berlin Heidelberg 2015

Abstract This study aims to investigate the relationship be- energy consumption, from financial development to output,
tween carbon dioxide (CO2) emissions, energy consumption, and from urbanization to financial development. In light of the
real output (GDP), the square of real output (GDP2), trade long-run estimates and the Granger causality analysis, the US
openness, urbanization, and financial development in the government should take into account the importance of trade
USA for the period 1960–2010. The bounds testing for openness, urbanization, and financial development in controlling
cointegration indicates that the analyzed variables are for the levels of GDP and pollution. Moreover, it should be noted
cointegrated. In the long run, energy consumption and urban- that the development of efficient energy policies likely con-
ization increase environmental degradation while financial de- tributes to lower CO2 emissions without harming real output.
velopment has no effect on it, and trade leads to environmental
improvements. In addition, this study does not support the Keywords CO2 emissions . Financial development .
validity of the environmental Kuznets curve (EKC) hypothe- Urbanization . Trade . Energy . Output
sis for the USA because real output leads to environmental
improvements while GDP2 increases the levels of gas emis-
sions. The results from the Granger causality test show that Introduction
there is bidirectional causality between CO2 and GDP, CO2
and energy consumption, CO2 and urbanization, GDP and The environmental Kuznets curve (EKC) hypothesizes that
urbanization, and GDP and trade openness while no causality the relationship between environmental quality and real out-
is determined between CO2 and trade openness, and gas emis- put has an inverted U-shaped feature as environmental degra-
sions and financial development. In addition, we have enough dation first increases until a certain level of GDP and then
evidence to support one-way causality running from GDP to decreases with increases in GDP. Akbostancı et al. (2009),
Lee and Lee (2009), Fodha and Zaghdoud (2010), and
Saboori et al. (2012) among others analyze the possible pres-
Responsible editor: Marcus Schulz
ence of the EKC hypothesis in Turkey, panel of 109 countries,
We hereby declare that the manuscript has not been published before nor Tunisia, and Malaysia, respectively. In addition, a group of
submitted to another journal for the consideration of publication.
study focus on the relationship between GDP per capita (or
We would like to thank the editor and two anonymous referees for their
comments and suggestions. Any remaining errors are solely ours. economic growth) and energy consumption for a variety of
countries and regions (Soytas and Sari 2003; Wolde-Rufael
* Eyup Dogan 2005; Ozturk 2010; Shahbaz and Lean 2012; Smyth and
[Link]@[Link] Narayan 2014; Dogan 2014, 2015a, 2015b; Aslan 2014;
Berna Turkekul Komal and Abbas 2015; Shahbaz et al. 2015). As one of the
[Link]@[Link] leading works in the literature, Ang (2007) combines the
1
Department of Economics, Abdullah Gul University, energy–output nexus and environment–output nexus under a
Sumer Campus, Office #B206, Kayseri, Turkey modified EKC framework in which carbon dioxide (CO2)
2
Department of Agricultural Economics, Ege University, emissions as a proxy for environmental degradation are
Bornova, Turkey regressed on energy consumption, GDP per capita, and the
1204 Environ Sci Pollut Res (2016) 23:1203–1213

square of GDP per capita. Ang (2007) supports the validity of opportunity to make more investment and buy new machines
an inverted U-shaped relationship between carbon dioxide and equipment, resulting in more energy consumption and
emissions and output in France as CO2 is positively impacted CO2 emissions. Because financial development likely links
by GDP per capita and negatively impacted by the square of to cheaper personal loan rates, it may trigger consumers to
GDP per capita. The EKC hypothesis begins to be widely purchase houses, cars, and durable goods (i.e., refrigerator
analyzed in the energy literature after energy consumption is and dish washer), which increases output, energy consump-
inserted as an additional explanatory variable into the conven- tion, and gas emissions. On the other hand, financial develop-
tional EKC model (Soytas et al. 2007; Zhang and Cheng ment may detracts energy consumption and gas emissions as it
2009; Soytas and Sari 2009; Ozturk and Acaravci 2010; Wang can potentially stimulate the efficiency of business perfor-
et al. 2011; Nasir and Rehman 2011; Kanjilal and Ghosh mance as well as energy efficiency (Tamazian et al. 2009).
2013; Salahuddin and Gow 2014; Kasman and Duman Sadorsky (2010) and Aslan et al. (2014) find significantly
2015; Bastola and Sapkota 2015; Baek 2015). positive relationship between energy consumption and
The existing literature about the energy–environment–out- financial development. Zhang (2011) claims that financial de-
put nexus is abundant (Al-Mulali et al. 2015a). Therefore, the velopment negatively impacts environmental degradation. Is-
linkage between GDP, energy consumption, and environmen- lam et al. (2013) find financial development and economic
tal degradation should be tested by taking into account partic- growth to have impact on energy consumption. Tang and
ular segments of the economy rather than by testing the valid- Tan (2014) reveal that financial development affects economic
ity of the EKC hypothesis using the simple econometric mod- growth and energy consumption influences financial develop-
el in which CO2 emissions are regressed on real output, the ment. Al-Mulali and Lee (2013) indicate that GDP, financial
square of real output, and energy consumption. By means of development, urbanization, and trade positively impact energy
including additional variables into the simple model, the state- consumption. In addition, Tamazian et al. (2009), Tamazian
of-the-art attempt to eliminate the omitted-variable bias prob- and Rao (2010), Jalil and Feridun (2011), Ozturk and
lem. Trade openness can be considered as a commonly used Acaravci (2013), Shahbaz et al. (2013a, b), and Omri et al.
variable in the literature (Ang. 2009; Halicioglu 2009; Jalil (2015) include trade and financial development into the sim-
a n d M a h m ud 2 0 0 9 ; N a s i r a n d R e h m a n 2 0 11 ; ple EKC model.
Jayanthakumaran et al. 2012). Farhani et al. (2014) decom- Al-Mulali et al. (2015b) and Farhani and Ozturk (2015)
pose the effect of trade on pollution, energy consumption, and include trade, urbanization, and financial development as sep-
output into three components, namely scale, composition, and arate sectors of the economy into the simple EKC model so as
technique. The scale effect basically implies that the increases to attempt to obtain unbiased effect of energy consumption on
in the amount of trade influence output, energy consumption, carbon dioxide emissions and to test the validity of EKC hy-
and thus, CO2 emissions. The composition effect refers to the pothesis. Only these two studies, to the best of our knowledge,
re-allocation in a country’s traded goods basket. In other simultaneously take into account the possible effects of trade,
words, free trade enables the country to specialize on the pro- urbanization, and financial development, and only a few stud-
duction of goods for which it has comparative advantage. ies consider the impacts of one or two variables among finan-
Hence, the use of energy and environmental quality may in- cial development, trade, and urbanization while the influence
crease or decrease depending on whether or not the sectors of energy consumption and GDP on environmental quality is
that the country specializes need more energy. The technique investigated. Furthermore, these studies reach different con-
effect means that trade liberalization leads to environmental clusions in terms of whether trade, urbanization, and financial
improvements since the technology gets better in producing development negatively or positively impact carbon dioxide
goods and using energy more efficiently. emissions and whether or not the EKC hypothesis is present.
Another variable that recently begins to be introduced into In light of the abovementioned arguments, the fundamental
the simple model is urbanization. Martínez-Zarzoso and contribution of this study is that for the first time in the liter-
Maruotti (2011) argue the possible impact of urbanization on ature ,this study aims to analyze the relationship between car-
environmental degradation through several channels. At the bon dioxide emissions, energy consumption, real output,
most basic interpretation, the increase in urban population trade, urbanization, and financial development for the USA
results in higher industrial output, transportation, and energy in an econometric model based on the EKC hypothesis. Given
consumption and gas emissions. Hossain (2011), Sharma that the analyzed variables are connected to each other, this
(2011), and Kasman and Duman (2015) add trade and urban- study also tries to eliminate the omitted-variable bias problem.
ization into the simple EKC model. Because this type of investigation is a significant gap in the
Financial development is the recent additional variable literature and, thus, the outcome has a high value in terms of
used by the recent works in the literature. Financial develop- policy implications, we focus on a single-country study rather
ment may lead to lower financing costs and better and larger than a panel study. According to the World Development In-
financing networks through which enterprises can have higher dicators, the USA is famous for large amount of energy
Environ Sci Pollut Res (2016) 23:1203–1213 1205

consumption, output, and gas emissions. The USA is also one pollution function using CO2 emissions (as an endogenous
of the top countries in the world in terms of trade openness, variable) and GDP, energy use, and trade openness (as exog-
urbanization, and financial development. In addition, the USA enous variables) over the annual period of 1953–2006. The
is an important country in terms of its impressive regional and findings indicate that more energy use, GDP, and trade
world affairs and its role in the energy market since it is among openness lead to more CO2 emissions. In the same way,
the NAFTA countries, one of the G7 countries, and one of the Halicioglu (2009) suggests the dynamic causal relationships
five permanent members of the United Nations. The rest of the between CO2 emissions, GDP, energy consumption, and for-
study is as follows: the BLiterature review^ section provides a eign trade in Turkey over the period of 1960–2005. Jalil and
literature review, the BMethods and data^ section explains the Mahmud (2009) extend the methodology of Halicioglu (2009)
methods and data, the BEmpirical results^ section reveals the for China over the period of 1975–2005. The findings also
empirical results, and the BConclusions^ section concludes the indicate that CO2 emissions can be determined by GDP and
aims and findings. energy consumption, while trade has insignificant impact on
CO2 emissions in the long run. Jayanthakumaran et al. (2012)
test the long-run and short-run relationships between CO2
Literature review emissions, growth, energy use, trade, and endogenously de-
termined structural breaks for both China and India over the
A number of existing studies including Ang (2007), Iwata period of 1971–2007. Using the autoregressive distributed lag
et al. (2010), Hamit-Haggar (2012), Saboori et al. (2012), (ARDL) approach to cointegration, the findings indicate that
Tiwari et al. (2013), Lau et al. (2014), Yavuz (2014), and CO2 emissions in China are determined by real GDP, energy
Osabuohien et al. (2014) among others investigate the rela- consumption, and structural changes while no causal relation-
tionship between income and carbon dioxide emissions and ship is detected for India.
validate the existence of the EKC hypothesis in a variety of The financial development factor is recently included in the
countries and regions. In addition, various studies validate the environmental function through the works of Jalil and Feridun
EKC hypothesis using the following panel data: for instance, (2011), Ozturk and Acaravci (2013), and Shahbaz (2013). The
Skaza and Blais (2013) for 190 developing and developed study of Jalil and Feridun (2011) discusses the impact of eco-
countries; Al-Mulali and Sheau-Ting (2014) for 189 countries nomic growth energy consumption, trade openness, and
from six different regions, namely Asia Pacific, Eastern Eu- financial development on carbon emissions in China from
rope, the Americas, Middle East and North Africa (MENA), 1953 to 2006. The findings show that financial development
Sub-Saharan Africa, and Western Europe; Omri et al. (2015) has no significant impact on carbon emissions in the long run,
for MENA countries; Ziaei (2015) for European, East Asian, while economic growth, energy consumption, and trade
and Oceania countries; and Al-Mulali et al. (2015c) for 93 openness present significant impacts on carbon emissions. In
countries. Moreover, Ang (2007), Jalil and Mahmud (2009), addition, Ozturk and Acaravci (2013) investigate the causal
Alam et al. (2012), Ozturk and Acaravci (2013), Shahbaz et al. relationship between carbon emissions, GDP, energy con-
(2013a), Alkhathlan and Javid (2013), and Boutabba (2014), sumption, trade openness, and financial development in Tur-
using time series data, also support the empirical presence of key over the period of 1960–2007. The findings show that an
the EKC hypothesis for France, China, Turkey, Bangladesh, increase in trade openness leads to an increase in carbon emis-
South Africa, Saudi Arabia, and India. On the other hand, Al- sions, while financial development has no significant effect on
Mulali et al. (2015a) and Farhani and Ozturk (2015) find con- carbon emissions in the long run. Finally, Shahbaz (2013)
troversial results not supporting the validity of EKC for Viet- examines the relationship between financial instability and
nam and Tunisia. the environmental degradation within the presence of GDP,
As seen above, different studies reach conflicting results as energy consumption, and trade openness in Pakistan over the
to the effect of income on the environment. These differences period of 1971–2009. The empirical findings indicate that the
may occur from omitted-variable bias problem, the choice of long-run relationship between variables can be detected and
specific functional forms (econometric techniques), and sam- financial instability may increase the environmental
ple selection bias. To overcome the omitted-variable bias degradation.
problem, several studies include different variables ranging The inclusion of urbanization in the environmental func-
from financial development, trade, foreign direct investment, tion presents an intense debate for discussion, especially in
and energy consumption to energy prices, labor, gross fixed terms of environmental and regional development. There
capital formation, and urbanization (Farhani and Ozturk 2015; are, however, limited works (Hossain 2011; Sharma 2011;
Shahbaz et al. 2015; Al-Mulali et al. 2015a; Komal and Abbas Kasman and Duman 2015) that have documented the
2015). In the literature, there are some efforts attempting to importance of the inclusion of urbanization in the rela-
examine environmental pollution including the impact of tionship between CO2 emissions, economic growth, energy
trade. Ang (2009) explores the estimation of the Chinese consumption, and trade. Hossain (2011) investigates the
1206 Environ Sci Pollut Res (2016) 23:1203–1213

relationship between gas emissions, energy consumption, real empirical findings of the Granger causality test indicate that
output, energy consumption, trade, and urbanization for newly long-run causality runs from output, energy consumption, fi-
industrialized countries over the period 1971–2007. The em- nancial development, openness, and urbanization to gas emis-
pirical findings show that unidirectional causality runs from sions as well as from CO2, output, energy consumption, open-
real output and trade openness to CO2, from real output to ness, and urbanization to financial development.
energy consumption, from trade openness to real output, from According to the above survey of the literature, the empir-
urbanization to GDP, and from trade openness to urbanization ical studies fail to achieve unanimous conclusion regarding
in the short run, although there is no long-run causal relation- the effects of urbanization, financial development, and trade
ship between the analyzed variables. Sharma (2011) examines openness as well as the validity of the EKC hypothesis. The
the relationship between environmental quality, energy con- main reason for the discrepancy in results in the previous
sumption, GDP, energy consumption, openness, and urbani- research comes from data characteristics, estimation tech-
zation for a panel of 69 countries for the years 1985–2005. niques (cointegration methods and causality tests), and devel-
The results indicate that trade openness, output per capita, and opment level of the country on which a study is conducted.
energy consumption lead to environmental degradation while
CO2 is negatively impacted by urbanization. Kasman and
Duman (2015) analyze the causal linkage between gas emis-
sions, energy consumption, real output, energy consumption, Methods and data
trade, and urbanization for new European Union members and
candidate countries over the period 1992–2010. The study Following the works of Al-Mulali et al. (2015b) and Farhani
supports the evidence of the EKC hypothesis in the analyzed and Ozturk (2015), the model that we are going to use is
countries; in addition, the fully modified ordinary least
squares (FMOLS) regression presents that openness and ur- ðCO2 Þt ¼ β 0 þ β1 GDPt þ β2 GDPt 2 þ β3 ECt
banization have a positive effect on the level of gas emissions.
þ β4 URBt þ β5 TRt þ β6 FDt þ et ð1Þ
Moreover, one-way causality is detected from energy con-
sumption, trade openness, and urbanization to CO2; from
GDP to energy consumption; from GDP, energy consumption, where CO2 is the carbon dioxide emissions per capita,
and urbanization to trade openness; from urbanization to GDP is the real gross domestic product per capita, GDP2
GDP; and from urbanization to trade openness. is the square of real gross domestic product per capita, EC
Al-Mulali et al. (2015b) and Farhani and Ozturk (2015) are is the energy consumption measured in kilograms of oil
the only published studies considering the effects on environ- equivalent per capita, URB is the urbanization measured
mental quality of urbanization, trade, and financial develop- by urban population to total population, TR is the trade
ment in conjunction with energy consumption and GDP. Al- openness measured by total trade as a share of GDP, and
Mulali et al. (2015b) analyze the long-run relationship be- FD is the financial development measured by domestic
tween CO2, energy consumption, real output, urbanization, credit to private sector. The time series data are from
trade openness, and financial development for a panel of 129 1960 to 2010 and obtained from the World Development
countries from four groups: namely low-income countries, Indicators ([Link] We use the longest
lower middle-income countries, upper middle-income coun- available time series data. All variables are transformed
tries, and high-income countries. The results from dynamic into their natural logarithmic forms.
ordinary least squares (DOLS) show that energy consumption The relationship between per capita carbon emissions,
leads to environmental degradation in all groups while finan- per capita real income, and the square of per capita real
cial development improves the environmental quality in the income, per capita energy consumption, trade openness,
four groups. In addition, urbanization and GDP have negative urbanization, and financial development in the USA is per-
and positive effects on CO2 in the three groups, respectively. formed in four steps. First, we test the integration proper-
Finally, trade openness has no significant effect in one group, ties of CO2, GDP, GDP2, EC, TR, URB, and FD. Second,
negative effect in two groups, and positive effect in one group. in case that they are non-stationary, the long-run relation-
Farhani and Ozturk (2015) investigate the linkage between ship among the analyzed variables is investigated using the
gas emissions, GDP, the square of GDP, energy consumption, ARDL bounds testing approach of cointegration. Third,
urbanization, trade, and financial development for Tunisia assuming that the variables are cointegrated, the short-run
over the period 1971–2012. According to the results obtained and long-run coefficients on real output, the square of real
from ARDL approach, all of the analyzed variables lead to output, energy consumption, openness, urbanization, and
environmental degradation. In addition, the EKC hypothesis financial development are estimated. Last, we test the caus-
is not valid in Tunisia since the coefficients on GDP and the al relationship between the analyzed variables using the
square of GDP are positive and statistically significant. The error correction-based causality models.
Environ Sci Pollut Res (2016) 23:1203–1213 1207

Unit root tests The null hypothesis of no cointegration in Eq. 2 (H0: γi =

0; ∀ i=1,…,7) is tested against the alternative hypothesis
This study uses the following unit root tests: the augmented of cointegration (H1: γi ≠0; ∀ i=1,…,7). The critical value
Dickey–Fuller test due to Dickey and Fuller (1979) and the bounds are computed by stochastic simulations using 20,000
Zivot–Andrews test with one structural break due to Zivot and replications because the actual critical values for relatively
Andrews (2002). The augmented Dickey–Fuller (ADF) unit small sample sizes can potentially differ from the critical
root test is employed to test the integration level of the vari- values posted in Pesaran et al. (2001).
ables. A well-known weakness of the ADF unit root test is its Having established the existence of a long-run relationship
potential confusion of structural breaks in the series as evi- based on the F- test, the second step of the ARDL analysis is
dence of non-stationarity. In other words, it may fail to reject to estimate the long-run and the associated short-run coeffi-
the unit root hypothesis if the series has a structural break. For cients. The long-run effects of GDP, GDP2, EC, URB, TR,
the series that is found to be I, there may be a possibility that and FD on CO2 are the estimates of − (γ2\γ1), − (γ3\γ1),
they are, in fact, stationary around the structural break(s), I(0), −(γ4\γ1), −(γ5\γ1), −(γ6\γ1), and −(γ7\γ1) in Eq. 2. Further-
but are erroneously classified as I(1). To overcome this, the more, the short-run effects of each explanatory variable on the
Zivot–Andrews (ZA) unit root test is employed. The ZA unit response variable are posed by the coefficient estimates of the
root test allows for one structural break. In this test, the null first-differenced series in Eqs. 3–8. For instance, the short-run
hypothesis is that the series has a unit root with structural effects of EC and TR on gas emissions are posed by the esti-
break against the alternative hypothesis that they are stationary mates of δ4k and δ6k in Eq. 3. The order of the lags in the
with break. We apply the ZA test with one-time changes in the ARDL model is selected using the Akaike information crite-
level and slope of the trend function of the series. rion (AIC) ensuring that there is no evidence of residual serial
correlation, functional form misspecification, non-normality,
ARDL approach to cointegration and heteroscedasticity.

In case where the analyzed variables are found to be either Error correction-based Granger causality analysis
integrated to one order or mixed order, the ARDL bounds
testing procedure introduced by Pesaran et al. (2001) should The ARDL method tests the existence or absence of
be used to expose whether or not gas emissions, GDP, GDP2, cointegration relationship between variables, but not the di-
energy consumption, URB, TR, and FD are cointegrated. rection of causality. If we do not find any evidence for
Therefore, this study uses the ARDL approach to cointegration among the variables, then the specification of
cointegration which estimates the conditional ARDL model the Granger causality test will be a vector autoregression
for CO2 emission, GDP, GDP2, energy consumption, urbani- (VAR) in first difference form. However, if we find evidence
zation, trade, and financial development given in Eq. 1. The for cointegration, then we need to augment the Granger-type
ARDL method can perform well in small samples and irre- causality test model with a one-period lagged error correction
spective of whether the variables are I(0), I(1), or mutually term (ECTt−1). Having found that there is a long-run relation-
cointegrated, and it is unbiased and efficient. The ARDL ap- ship between the analyzed variables, the next step is to esti-
proach for the model given in Eq. 1 takes the following as in mate the vector error correction model (VECM) given in
Eq. 2: Eqs. 3–8 by following Engle and Granger (1987):
X n1 X n2 X n1 X n2
ΔCO2 t ¼ δ0 þ δ ΔCO2 t−k þ
k¼1 1k
δ ΔGDPt−k
k¼0 2k ΔCO2 t ¼ δ0 þ δ ΔCO2 t−k þ δ ΔGDP t−k
X n3 X n4 k¼1 1k k¼0 2k
X n3 X n4
þ δ ΔGDP2t−k þ
k¼0 3k k¼0 4k
δ ΔECt−k þ δ ΔGDP2t−k þ δ ΔECt−k
X n5 X n6 k¼0 3k k¼0 4k
X n5 X n6
þ δ5k ΔURBt−k þ δ ΔUTRt−k
k¼0 6k þ δ ΔURBt−k þ δ ΔTRt−k
Xk¼0
n7 k¼0 5k
X n7
k¼0 6k
þ δ Δ FDt−k þ γ 1 CO2 t−1 þ γ 2 GDPt−1
k¼0 7k þ δ ΔFDt−k þ τ ECTt−1 þ μt
k¼0 7k
þ γ 3 GDP2t−1 þ γ 4 ECt−1 þ γ 5 URBt−1 ð3Þ
þ γ 6 TRt−1 þ γ 7 FDt−1 þ μt X n1 X n2
ð2Þ ΔGDPt ¼ δ0 þ δ ΔCO2 t−k þ
k¼0 1k
δ ΔGDP t−k
k¼1 2k
X n3 X n4
þ δ ΔGDP2t−k þ
k¼0 3k
δ ΔECt−k
k¼0 4k
where Δ denotes the first difference term and μt is the distur- X n5 X n6
bance term assumed to have a mean value of zero and to be þ δ ΔURB t−k þ
k¼0 5k
δ ΔTRt−k
k¼0 6k
uncorrelated with the independent variables. The ARDL ap- X n7
þ δ Δ FD t−k þ τ ECTt−1 þ μt
proach based on the F-statistics is employed to examine the k¼0 7k
existence of cointegration between the analyzed variables. ð4Þ
1208 Environ Sci Pollut Res (2016) 23:1203–1213

X n1 X n2 (GDP), the square of real output (GDP2), energy consumption

ΔECt ¼ δ0 þ δ ΔCO2 t−k þ
k¼0 1k
δ ΔGDPt−k
k¼0 2k
X n3 X n4 (EC), urbanization (URB), trade openness (TR), and financial
þ δ ΔGDP2t−k þ
k¼0 3k
δ ΔECt−k
k¼1 4k
development (FD) in order to exploit the integration properties
X n5 X n6 of the analyzed variables. As mentioned earlier, the ARDL
þ δ ΔURBt−k þ
k¼0 5k
δ ΔTRt−k
k¼0 6k
X n7 approach to cointegration is a reliable method only if the time
þ δ Δ FDt−k þ τECTt−1 þ μt
k¼0 7k
series is either I(0) or I(1). The results are given in Table 1.
The conventional augmented Dickey–Fuller (ADF) unit
ð5Þ
root test indicates that urbanization is stationary at level; how-
ever, CO2, GDP, energy consumption, trade openness, and
X n1 X n2 financial development have unit root at levels but become sta-
ΔURBt ¼ δ0 þ k¼0
δ 1k ΔCO2 t−k
þ δ ΔGDPt−k
k¼0 2k
X n3 X n4 tionary in their first differences. Thereafter, the Zivot–Andrews
þ δ ΔGDP2t−k þ
k¼0 3k
δ ΔECt−k
k¼0 4k
(ZA) unit root test with one structural break is used to confirm
X n5 X n6 the findings of the ADF unit root test. According to the results
þ δ5k ΔURBt−k þ δ ΔTRt−k
k¼0 6k obtained from the ZA unit root test, only urbanization does not
X k¼1
n7
þ δ Δ FDt−k þ τECTt−1 þ μt
k¼0 7k
have a unit root at level; on the other hand, gas emissions,
GDP, EC, TR, and FD are not stationary at levels but become
ð6Þ
stationary in their first differences. Both the ADF unit root test
and the ZA unit root test reach the same conclusion. In short,
X n1 X n2 urbanization is determined to be I(0) while CO2, GDP, energy
ΔTRt ¼ δ0 þ k¼0
δ 1k ΔCO 2 t−k
þ δ ΔGDPt−k
k¼0 2k
X n3 X n4 consumption, trade, and financial development are determined
þ δ ΔGDP2t−k þ
k¼0 3k
δ ΔECt−k
k¼0 4k
to be I(1). Because neither of the analyzed variables I(k) where
X n5 X n6 k>1, we can proceed to the bounds testing for cointegration.
þ δ5k ΔURBt−k þ δ ΔTRt−k
k¼1 6k
X k¼0
n7
þ δ Δ FDt−k þ τECTt−1 þ μt
k¼0 7k
ARDL approach to cointegration
ð7Þ
Given that carbon dioxide emissions, real output, the square of
real output, energy consumption, urbanization, trade, and fi-
X n1 X n2 nancial development are either I(0) or I(1), Eq. 2 is estimated
Δ FDt ¼ δ0 þ k¼0
δ 1k ΔCO2 t−k
þ δ ΔGDPt−k
k¼0 2k
X n3 X n4 based on the ARDL approach to cointegration. Table 2 repre-
þ δ ΔGDP2t−k þ
k¼0 3k
δ ΔECt−k
k¼0 4k
sents the calculated F-statistic and lower and upper critical
X n5 X n6 bounds for 5 % level. The critical value bounds in Table 2
þ δ5k ΔURBt−k þ δ ΔTRt−k
k¼0 6k
X k¼0
n7
are computed by stochastic simulations using 20,000 replica-
þ δ Δ FDt−k þ τECTt−1 þ μt
k¼1 7k
tions because the actual critical values for relatively small
sample sizes can potentially differ from the critical values
ð8Þ
posted in Pesaran et al. (2001). Regarding the estimated
model, f(CO2/GDP, GDP2, EC, URB, TR, FD), in which gas
where τ measures the speed of adjustment to obtain an equi- emissions are the response variable while GDP, the square of
librium in the event of shock(s) to the system and ECTt−1 is GDP, energy consumption, urbanization, trade, and financial
the lagged error correction mechanism attained from the long- development are the explanatory variables, the null hypothesis
run equilibrium relationship. The VECM allows us to capture of no cointegration can be rejected in favor of the alternative
both the short-run and long-run Granger causality. The short- hypothesis of cointegration at 5 % level of significance
run causal effects can be obtained by the Wald statistics of the because the calculated F-statistic is far greater than, I(1), the
lagged explanatory variables, while the Wald statistics on the 5 % upper critical bound. Baek (2015) suggests that the
coefficient of the lagged error correction term (ECTt−1) indi- negative and statistically significant lagged error correction
cates the significance of the long-run causal effect. term (ECTt−1) can be used as an alternative method to pin
down the cointegration relationship between the variables.
As shown in Table 3, the coefficient estimate of ECTt−1 is
Empirical results −0.76, which is negative and statistically significant at 1 %
level. Therefore, we can claim the existence of cointegration
Unit root test and, thus, the long-run relationship between CO2, GDP,
GDP2, EC, URB, TR, and FD. Henceforth, the estimation
This study applies two different unit root tests to the time results that we estimate in the next section are assumed to be
series data on carbon dioxide (CO2) emission, real output economically meaningful, accurate, and consistent.
Environ Sci Pollut Res (2016) 23:1203–1213 1209

Table 1 Unit root analysis

CO2 GDP (GDP2) EC URB TR FD

ADF-test
Level −3.20 −2.81 −3.17 −3.58b −2.30 −2.32
Δ −4.71a −5.28a −4.48a – −6.53a −8.16a
ZA-test
Level −3.62 (1969) −4.53 (2005) −3.71 (1979) −5.04c (1971) −5.15a (1973) −4.61 (1983)
Δ −6.31a (1983) −6.01a (1983) −6.28a (1984) – −7.89a (1975) −9.82a (1982)
Decision I(1) I(1) I(1) I(0) I(1) I(1)

Δ is the first difference term. Years in the parenthesis are the structural break dates. Lag lengths are selected based on the Akaike information criterion
(AIC)
a
Statistical significance at 1 % levels
b
Statistical significance at 5 % levels
c
Statistical significance at 10 % levels

Short-run and long-run estimates Table 3 Estimated coefficients from ARDL model

Regressors Coefficient t ratio

The short-run and long-run estimates of GDP, the square of
GDP, energy consumption, trade, urbanization, and financial (A) Long-run estimates (dependent variable CO2)
development are reported in Table 3. Moreover, the coeffi- GDP −2.13b −2.49
cient estimates of the analyzed variables are economically GDP 2
0.22b
2.23
equal to the elasticity of CO2 with respect to GDP, GDP2, EC 1.16a 32.17
EC, TR, URB, and FD, respectively, because the time series URB 0.43c 1.71
data are transformed into their logarithmic forms. Both the TR −0.08a −4.26
short-run and the long-run elasticity of carbon dioxide emis- FD 0.04 0.80
sions with respective to GDP are negative and statistically Constant 1.04 0.51
significant, while the coefficient estimates of the square of (B) Short-run estimates (dependent variable ΔCO2)
GDP are statistically significant and positive both in the ΔGDP −1.63b −2.41
short run and long run. In other words, the short-run and ΔGDP 2
0.18b 2.37
long-run estimates of GDP are −1.63 and −2.13, respective-
ΔEC 0.89a 14.76
ly, and the short-run and long-run coefficient estimates of
ΔURB 0.33c 1.71
the square of GDP are +0.18 and +0.22, respectively. In the
ΔTR 0.02 0.83
existence of the EKC hypothesis, the effect of GDP and
ΔFD 0.03 0.81
GDP2 on carbon dioxide emissions is expected to be posi-
ECTt−1 −0.76a −15.62
tive and negative, respectively. Because it is not the case for
Diagnostic tests
the analyzed variables, the EKC hypothesis is not present in
Serial correlation (0.87)
the USA. On the other hand, there is a U-shaped relation-
Functional form (0.63)
ship between the level of income and gas emissions. There-
Normality (0.33)
fore, the increase in the level of GDP leads to environmen-
Heteroscedasticity (0.18)
tal improvements until a certain level but then the increase
R2 0.96
DW 1.94
Table 2 Cointegration test results F - test 129.8a

Estimated model F-statistic 5 % critical values The numbers in parenthesis under diagnostic tests are the p - values. DW
is the Durbin–Watson test statistic. The proper lag length of the estimated
I(0) I(1) ARDL model is (1,0,1,0,0,1,0) and selected based on AIC
a
Statistical significance at 1 % level
f(CO2/GDP, GDP2, EC, URB, TR, FD) 9.87a 2.73 4.09 b
Statistical significance at 5 % level
a c
Statistical significance at 5 % level Statistical significance at 10 % level
1210 Environ Sci Pollut Res (2016) 23:1203–1213

Table 4 Granger causality

analysis Dependent variable Short-run analysis Long-run analysis

ΔCO2 ΔGDP (ΔGDP2) ΔEC ΔURB ΔTR ΔFD ECTt−1

ΔCO2 – 5.84b 217.9a 2.94c 0.70 0.65 244.1a

ΔGDP (ΔGDP ) 2
5.91c
– 3.78 7.39b
9.06a
6.60b
2.45
ΔEC 152.6a 7.29a – 0.78 0.11 0.39 18.12a
ΔURB 22.14a 14.40a 3.86 – 5.07c 0.63 11.98a
ΔTR 0.69 10.79a 0.03 3.50c – 5.79c 0.68
ΔFD 0.11 2.65 0.57 3.81c 7.90a – 24.34a

Values are from Wald test based on the chi-square distribution

a
Statistical significance at 1 % level
b
Statistical significance at 5 % level
c
Statistical significance at 10 % level

in the level of income causes environmental degradation. coefficient on trade is ambiguous since it depends on devel-
This finding is consistent with Chandran and Tang (2013), opment stage of an economy. In addition, the effect of trade on
Baek (2015), Farhani and Ozturk (2015), and Al-Mulali gas emissions is usually negative in the developed countries.
et al. (2015a). As shown in Table 4, a 1 % increase in trade openness leads to
The results in Table 3 show that energy consumption has statistically significant decrease in CO2 by 0.08 % in the long
positive impact on gas emissions in the USA both in the short run at 1 % level of significance; however, the short-run elas-
run and long run. More precisely, a 1 % increase in EC stim- ticity estimate of gas emissions with respect to trade is not
ulates CO2 by 0.89 and 1.16 % in the short run and long run, statistically significant at 10 % level of significance. The
respectively, at 1 % level of significance. This outcome is line short-run and long-run findings are in line with those found
with that of most studies in the literature such as Ang (2007, by Halicioglu (2009), Jalil and Mahmud (2009), and Al-
2009), Halicioglu (2009), Jalil and Feridun (2011), Kanjilal mulali et al. (2015b). As it is consistent with the studies of
and Ghosh (2013), Ozturk and Acaravci (2013), Shahbaz et al. Jalil and Feridun (2011) and Ozturk and Acaravci (2013)
(2013a, b), Al-Mulali et al. (2015a, b), Kasman and Duman which show that financial development has no statistically
(2015), and Farhani and Ozturk (2015). The long-run and the significant impact on environmental quality, Table 3 indicates
short-run elasticity estimates of CO2 with respect to urbaniza- that the short-run and long-run elasticity estimates of CO2
tion are expected to be positive in the developed countries with respect to FD are not statistically significant for the
referring to Hossain (2011) and Farhani and Ozturk (2015). USA at 10 % level of significance.
In more detail, a 1 % rise in URB increases gas emissions by The estimated model also passes several diagnostic tests as
0.33 and 0.43 % in the short run and long run, respectively, at given in Table 3. Serial correlation test is based on the La-
10 % level of significance. This is consistent with the study of grange Multiplier test of residual, functional form is based on
Kasman and Duman (2015) and Farhani and Ozturk (2015). Ramsey’s RESET test using the square of the fitted values,
According to Halicioglu (2009), the expected sign of the normality test is based on the test of Skewness and Kurtosis of

Fig. 1 CUSUM and CUSUMQ stability tests

Environ Sci Pollut Res (2016) 23:1203–1213 1211

residuals, and heteroscedasticity test is based on the regression These findings are consistent with those found in the pre-
of squared residuals on squared fitted values. We cannot reject ceding section. More precisely, energy consumption is found
the null hypotheses that there is no serial correlation, no to have impact on environmental degradation both in the short
heteroscedasticity, and no functional form misspecification run and long run. In addition, an increase in the level of energy
and normality of disturbances at 10 % significance level be- use does not cause income. Thus, the USA may decrease
cause the related p values are far greater than 0.10. In addition, energy consumption without harming the GDP for the sake
we have no evidence on serial correlation, heteroscedasticity, of environmental quality. Furthermore, the government may
misspecification, and non-normality. Furthermore, the high encourage and financially support the institutions, universi-
value of R2 (0.96) implies that the adjustment of the model ties, and researchers to propose project on increasing the effi-
in Eq. 2 is fairly perfect. Because the Durbin–Watson statistic ciency of energy and on the application of the methods of
is close to 2, we have enough evidence to reject the null hy- environmental protection. As expected, income (the square
pothesis of autocorrelation between residuals in the estimated of income) has causal relationship with gas emissions. This
model. The statistically significant F- test confirms the joint also supports the presence of the U-shaped relationship be-
significance of explanatory variables in the ARDL model. The tween CO2 and GDP in the USA. Although trade openness
last identification related to the goodness of fit of the model is and financial development do not cause the Granger environ-
stability tests. For this purpose, we perform cumulative sum mental quality in the short run, TR and FD have impact on it in
(CUSUM) and cumulative sum of squares (CUSUMQ) tests. the long run. Also, urbanization is found to be a cause of gas
As seen in Fig. 1, the estimated parameters are stable over time emissions and income. Moreover, trade and financial devel-
since the plot of CUSUM and CUSUMQ test statistics fall opment influence GDP as well. Henceforth, the US govern-
within the 5 % boundaries. ment should take into account the importance of trade open-
ness, urbanization, and financial development in controlling
Granger causality test for the level of GDP and pollution.

In light of the evidence of cointegration relationship among

the analyzed variables, it is an interest for researchers to per- Conclusions
form the Granger causality test so as to pin down appropriate
economic policies, environmental policies, and energy strate- This study examines the relationship between carbon dioxide
gies by understanding the directions of causality between emissions, energy consumption, real output, the square of real
CO2, GDP (GDP2), EC, TR, URB, and FD. Henceforth, the output, trade openness, urbanization, and financial develop-
Granger causality in the vector error correction mechanism is ment in the USA for the period 1960–2010. To analyze this
used to exploit the directions of causality between the afore- relationship, we use unit root tests to find out the stationarity
mentioned variables as well as to decompose the directions of properties of the analyzed variables, the ARDL bounds testing
causality into the short run and long-run effects. approach to explore the possible cointegration between the
The results obtained from the VECM Granger causality test variables, and short-run and long-run estimates, Granger cau-
are reported in Table 4. In the short run, there is bidirectional sality test based on VECM, to reveal the short-run and long-
causality between (1) CO2 and GDP, (2) gas emissions and run causal relationships between analyzed variables.
energy consumption, (3) carbon dioxide emissions and urban- According to the results obtained from augmented Dickey–
ization, (4) GDP and urbanization, (5) trade and urbanization, Fuller and Zivot–Andrews unit root test, we claim that urban-
(6) financial development and trade, and (7) income and trade ization is stationary at level and CO2, energy consumption,
openness. In addition, we have enough evidence to support real output, the square of real output, trade openness, and
one-way causality running (1) from GDP to energy consump- financial development are stationary at first differences. Then,
tion, (2) from financial development to income, and (3) from the ARDL approach to cointegration test indicates that the
urbanization to financial development. Lastly, no causality is analyzed variables are cointegrated at 5 % level of signifi-
determined between (1) CO2 and trade openness, (2) carbon cance. The short-run and long-run estimates show that energy
dioxide emissions and FD, (3) energy consumption and ur- consumption is the main cause of CO2 emissions in the USA.
banization, (4) EC and trade, and (5) energy consumption and In addition, urbanization has positive impact on gas emis-
FD. In the long run, there is an evidence of four causal rela- sions. Furthermore, the estimates of real output and the square
tionships, namely (1) from GDP, EC, URB, TR, and FD to of real output suggest that there is strong evidence against the
CO2; (2) from gas emissions, income, trade, urbanization, and existence of an EKC-type relationship in the USA.
financial development to energy consumption; (3) from CO2, Last, the Granger causality analysis presents that there is
GDP, EC, TR, and FD to URB; and (4) from carbon dioxide strong causal relationship between gas emissions and real out-
emissions, income, energy consumption, TR, and URB to put, energy consumption, and urbanization both in the short
financial development. run and long run. By putting together the results from the
1212 Environ Sci Pollut Res (2016) 23:1203–1213

short-run and long-run estimates, and Granger causality tests, Dickey DA, Fuller WA (1979) Distribution of the estimators for
autoregressive time series with a unit root. J Am Stat Assoc
the US government should take into account the importance of
74(366a):427–431
trade openness, urbanization, and financial development in Dogan E (2014) Energy consumption and economic growth: evidence
controlling for the levels of GDP and pollution. While the from low-income countries in Sub-Saharan Africa. Int J Energy
above analysis provides interesting insights, it should be noted Econ Pol 4(2):154–162
that the development of efficient energy policies likely con- Dogan E (2015a) Revisiting the relationship between natural gas con-
sumption and economic growth in Turkey. Energy Sources Part B:
tributes to lower CO2 emissions while preserving real GDP. A Econ Plan Pol 10(4):361–370
promising extension of this work would be to consider the Dogan E (2015b) The relationship between economic growth and elec-
energy supply, rural development, and other environmental tricity consumption from renewable and non-renewable sources: a
variables for the case of the USA. study of Turkey. Renew Sustain Energy Rev 52:534–546
Engle RF, Granger CW (1987) Co-integration and error correction: rep-
resentation, estimation, and testing. Econometrica: J Econo Soc
251–276
Farhani S, Ozturk I (2015) Causal relationship between CO2 emissions,
References real GDP, energy consumption, financial development, trade open-
ness, and urbanization in Tunisia. Environ Sci Pollut Res 1–14
Akbostancı E, Türüt-Aşık S, Tunç G (2009) The relationship between Farhani S, Chaibi A, Rault C (2014) CO2 emissions, output, energy
income and environment in Turkey: is there an environmental consumption, and trade in Tunisia. Econ Model 38:426–434
Kuznets curve? Energy Policy 37(3):861–867 Fodha M, Zaghdoud O (2010) Economic growth and pollutant emissions
Alam MJ, Begum IA, Buysse J, Van Huylenbroeck G (2012) Energy in Tunisia: an empirical analysis of the environmental Kuznets
consumption, carbon emissions and economic growth nexus in curve. Energy Policy 38(2):1150–1156
Bangladesh: cointegration and dynamic causality analysis. Energy Halicioglu F (2009) An econometric study of CO2 emissions, energy
Policy 45:217–225 consumption, income and foreign trade in Turkey. Energy Policy
Alkhathlan K, Javid M (2013) Energy consumption, carbon emissions 37(3):1156–1164
and economic growth in Saudi Arabia: an aggregate and disaggre- Hamit-Haggar M (2012) Greenhouse gas emissions, energy consumption
gate analysis. Energy Policy 62:1525–1532 and economic growth: a panel cointegration analysis from Canadian
Al-mulali U, Lee JY (2013) Estimating the impact of the financial devel- industrial sector perspective. Energy Econ 34(1):358–364
opment on energy consumption: evidence from the GCC (Gulf Hossain MS (2011) Panel estimation for CO2 emissions, energy con-
Cooperation Council) countries. Energy 60:215–221 sumption, economic growth, trade openness and urbanization of
Al-mulali U, Sheau-Ting L (2014) Econometric analysis of trade, exports, newly industrialized countries. Energy Policy 39(11):6991–6999
imports, energy consumption and CO2 emission in six regions. Islam F, Shahbaz M, Ahmed AU, Alam MM (2013) Financial develop-
Renew Sust Energ Rev 33:484–498 ment and energy consumption nexus in Malaysia: a multivariate
Al-Mulali U, Saboori B, Ozturk I (2015a) Investigating the environmen- time series analysis. Econ Model 30:435–441
tal Kuznets curve hypothesis in Vietnam. Energy Policy 76:123–131 Iwata H, Okada K, Samreth S (2010) Empirical study on the environmen-
Al-mulali U, Tang CF, Ozturk I (2015b) Does financial development tal Kuznets curve for CO2 in France: the role of nuclear energy.
reduce environmental degradation? Evidence from a panel study Energy Policy 38(8):4057–4063
of 129 countries. Environ Sci Pollut Res 1–10 Jalil A, Feridun M (2011) The impact of growth, energy and financial
Al-Mulali U, Weng-Wai C, Sheau-Ting L, Mohammed AH (2015c) development on the environment in China: a cointegration analysis.
Investigating the environmental Kuznets curve (EKC) hypothesis Energy Econ 33(2):284–291
by utilizing the ecological footprint as an indicator of environmental Jalil A, Mahmud SF (2009) Environment Kuznets curve for CO2 emis-
degradation. Ecol Indic 48:315–323 sions: a cointegration analysis for China. Energy Policy 37(12):
Ang JB (2007) CO2 emissions, energy consumption, and output in 5167–5172
France. Energy Policy 35(10):4772–4778 Jayanthakumaran K, Verma R, Liu Y (2012) CO2 emissions, energy
Ang JB (2009) CO2 emissions, research and technology transfer in consumption, trade and income: a comparative analysis of China
China. Ecol Econ 68(10):2658–2665 and India. Energy Policy 42:450–460
Aslan A (2014) Electricity consumption, labor force and GDP in Turkey: Kanjilal K, Ghosh S (2013) Environmental Kuznet’s curve for India:
evidence from multivariate Granger causality. Energy Sources Part evidence from tests for cointegration with unknown structural
B: Econ Plan Pol 9(2):174–182 breaks. Energy Policy 56:509–515
Aslan A, Apergis N, Topcu M (2014) Banking development and energy Kasman A, Duman YS (2015) CO2 emissions, economic growth, energy
consumption: evidence from a panel of Middle Eastern countries. consumption, trade and urbanization in new EU member and can-
Energy 72:427–433 didate countries: a panel data analysis. Econ Model 44:97–103
Baek J (2015) Environmental Kuznets curve for CO2 emissions: the case Komal R, Abbas F (2015) Linking financial development, economic
of Arctic countries. Energy Econ 50:13–17 growth and energy consumption in Pakistan. Renew Sust Energ
Bastola U, Sapkota P (2015) Relationships among energy consumption, Rev 44:211–220
pollution emission, and economic growth in Nepal. Energy 80:254– Lau LS, Choong CK, Eng YK (2014) Investigation of the environmental
262 Kuznets curve for carbon emissions in Malaysia: do foreign direct
Boutabba MA (2014) The impact of financial development, income, en- investment and trade matter? Energy Policy 68:490–497
ergy and trade on carbon emissions: evidence from the Indian econ- Lee CC, Lee JD (2009) Income and CO2 emissions: evidence from panel
omy. Econ Model 40:33–41 unit root and cointegration tests. Energy Policy 37(2):413–423
Chandran VGR, Tang CF (2013) The impacts of transport energy con- Martínez-Zarzoso I, Maruotti A (2011) The impact of urbanization on
sumption, foreign direct investment and income on CO2 emissions CO2 emissions: evidence from developing countries. Ecol Econ
in ASEAN-5 economies. Renew Sust Energ Rev 24:445–453 70(7):1344–1353
Environ Sci Pollut Res (2016) 23:1203–1213 1213

Nasir M, Rehman FU (2011) Environmental Kuznets curve for carbon Skaza JS, Blais BS (2013) The relationship between economic growth
emissions in Pakistan: an empirical investigation. Energy Policy and environmental degradation: exploring models and questioning
39(3):1857–1864 the existence of an environmental Kuznets curve. Cent Glob Econ
Omri A, Daly S, Rault C, Chaibi A (2015) Financial development, envi- Stud Bryant Univ Work Pap (2013-05)
ronmental quality, trade and economic growth: what causes what in Smyth R, Narayan PK (2014) Applied econometrics and implications for
MENA countries. Energy Econ 48:242–252 energy economics research. Energy Econ
Osabuohien ES, Efobi UR, Gitau CMW (2014) Beyond the environmen- Soytas U, Sari R (2003) Energy consumption and GDP: causality rela-
tal Kuznets Curve in Africa: evidence from panel cointegration. J tionship in G-7 countries and emerging markets. Energy Econ 25(1):
Environ Pol Plan 16(4):517–538 33–37
Ozturk I (2010) A literature survey on energy–growth nexus. Energy Soytas U, Sari R (2009) Energy consumption, economic growth, and
Policy 38(1):340–349 carbon emissions: challenges faced by an EU candidate member.
Ozturk I, Acaravci A (2010) CO2 emissions, energy consumption and Ecol Econ 68(6):1667–1675
economic growth in Turkey. Renew Sust Energ Rev 14(9):3220–3225 Soytas U, Sari R, Ewing BT (2007) Energy consumption, income, and
Ozturk I, Acaravci A (2013) The long-run and causal analysis of energy, carbon emissions in the United States. Ecol Econ 62(3):482–489
growth, openness and financial development on carbon emissions in Tamazian A, Rao BB (2010) Do economic, financial and institutional
Turkey. Energy Econ 36:262–267 developments matter for environmental degradation? Evidence from
Pesaran MH, Shin Y, Smith RJ (2001) Bounds testing approaches to the transitional economies. Energy Econ 32(1):137–145
analysis of level relationships. J Appl Econ 16(3):289–326 Tamazian A, Chousa JP, Vadlamannati KC (2009) Does higher economic
Saboori B, Sulaiman J, Mohd S (2012) Economic growth and CO2 emis- and financial development lead to environmental degradation: evi-
sions in Malaysia: a cointegration analysis of the environmental dence from BRIC countries. Energy Policy 37(1):246–253
Kuznets curve. Energy Policy 51:184–191 Tang CF, Tan BW (2014) The linkages among energy consumption,
Sadorsky P (2010) The impact of financial development on energy con- economic growth, relative price, foreign direct investment, and fi-
sumption in emerging economies. Energy Policy 38(5):2528–2535 nancial development in Malaysia. Qual Quant 48(2):781–797
Salahuddin M, Gow J (2014) Economic growth, energy consumption and Tiwari AK, Shahbaz M, Hye QMA (2013) The environmental Kuznets
CO2 emissions in Gulf Cooperation Council countries. Energy 73: curve and the role of coal consumption in India: cointegration and
44–58 causality analysis in an open economy. Renew Sust Energ Rev 18:
Shahbaz M (2013) Does financial instability increase environmental deg- 519–527
radation? Fresh evidence from Pakistan. Econ Model 33:537–544 Wang SS, Zhou DQ, Zhou P, Wang QW (2011) CO2 emissions, energy
Shahbaz M, Lean HH (2012) The dynamics of electricity consumption consumption and economic growth in China: a panel data analysis.
and economic growth: a revisit study of their causality in Pakistan. Energy Policy 39(9):4870–4875
Energy 39(1):146–153 Wolde-Rufael Y (2005) Energy demand and economic growth: the
Shahbaz M, Tiwari AK, Nasir M (2013a) The effects of financial devel- African experience. J Policy Model 27(8):891–903
opment, economic growth, coal consumption and trade openness on Yavuz N (2014) CO2 emission, energy consumption, and economic
CO2 emissions in South Africa. Energy Policy 61:1452–1459 growth for turkey: evidence from a cointegration test with a struc-
Shahbaz M, Hye QMA, Tiwari AK, Leitão NC (2013b) Economic tural break. Energy Sources Part B: Econ Plan Pol 9(3):229–235
growth, energy consumption, financial development, international Zhang YJ (2011) The impact of financial development on carbon emis-
trade and CO2 emissions in Indonesia. Renew Sust Energ Rev 25: sions: an empirical analysis in China. Energy Policy 39(4):2197–2203
109–121 Zhang XP, Cheng XM (2009) Energy consumption, carbon emissions,
Shahbaz M, Loganathan N, Zeshan M, Zaman K (2015) Does renewable and economic growth in China. Ecol Econ 68(10):2706–2712
energy consumption add in economic growth? An application of Ziaei SM (2015) Effects of financial development indicators on energy
auto-regressive distributed lag model in Pakistan. Renew Sust consumption and CO2 emission of European, East Asian and
Energ Rev 44:576–585 Oceania countries. Renew Sust Energ Rev 42:752–759
Sharma SS (2011) Determinants of carbon dioxide emissions: empirical Zivot E, Andrews DWK (2002) Further evidence on the great crash, the oil-
evidence from 69 countries. Appl Energy 88(1):376–382 price shock, and the unit-root hypothesis. J Bus Econ Stat 20(1):25–44

View publication stats