International Journal of Environmental Studies

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/genv20

The impact of renewable energy, trade, economic

growth on CO2 emissions in China

Xuyi Liu

To cite this article: Xuyi Liu (2021) The impact of renewable energy, trade, economic growth
on CO2 emissions in China, International Journal of Environmental Studies, 78:4, 588-607, DOI:
10.1080/00207233.2020.1834686

To link to this article: https://doi.org/10.1080/00207233.2020.1834686

Published online: 22 Oct 2020.

Submit your article to this journal

Article views: 357

View related articles

View Crossmark data

Citing articles: 11 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=genv20
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES
2021, VOL. 78, NO. 4, 588–607
https://doi.org/10.1080/00207233.2020.1834686

ARTICLE

The impact of renewable energy, trade, economic growth on

CO2 emissions in China
Xuyi Liu
School of Business, Luoyang Normal University, Luoyang, Henan, PR China

ABSTRACT KEYWORDS
This study examines the linkage of per capita carbon dioxide (CO2) Emissions; renewable; trade;
emissions, real GDP, renewable and non-renewable energy use and economy; Granger; Kuznets
trade (exports and imports) in China covering 1965–2016. Long-run
estimates present the invalidation of inverted U-shaped hypothe­
tical environmental Kuznets curve (EKC) in three models based on
the equations of import, export, and trade openness (% of trade in
GDP), respectively. Long-run Granger causalities exist in the equa­
tions of emissions, real GDP, and non-renewable energy, respec­
tively. Non-renewable energy conservation policy not only affects
economic activity, but also influences CO2 emissions. The effect of
renewable energy policy on CO2 emissions reduction is a long-term
process.

Introduction
Global heating and climate change are worldwide problems, accompanied by the devel­
opment of world economy and international trade, and increase of energy use [1].
Environmental economists have been attentive to these relationships [2–5]. Several
international conventions exist in order to reduce the adverse impacts of global heating,
by restricting the emission of greenhouse gases (GHG). The Kyoto Protocol adopted in
1997 is a major agreement of the United Nations Framework Convention on Climate
Change (UNFCCC). China signed the UNFCCC in 1992 and subsequently the Kyoto
Protocol in 1998, ratifying it in 2002 and thus becoming the 37th signatory country. In
2009, the State Council of China announced that it would achieve a target of reducing
55–60% of CO2 emissions per unit of GDP in 2020 compared with the level of 2005 [6].
As the biggest CO2 emission country in the world, China contributed about 27% of the
world’s total GHG emissions in 2014. But the growth rate of emission slowed for the first
time with 1.2% in 2016, compared to the average rate of 6.7% in the past decade [7]. In
2015, China’s per capita CO2 emissions from energy use declined by 0.1% compared to
those of 2014, the first decline in emissions since 1998 [8].
International trade contributes to global heating, from production in one region via
transport to consumption elsewhere [9–12]. The contribution of globalisation, particu­
larly trade openness, towards greenhouse gas (GHG) emissions has been an important

CONTACT Xuyi Liu [email protected] School of Business, Luoyang Normal University, Luoyang, Henan, PR
China
© 2020 Informa UK Limited, trading as Taylor & Francis Group

Published online 22 Oct 2020

 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 589

issue within the context of climate change [13]. Numerous trade economists and envir­
onmentalists consider that international trade may affect environmental enhancement by
using resources efficiently and developing the economy sustainably. Besides, with eco­
nomic development and the great abundance of living material, some scholars argue that
trade liberalisation is the primary cause of local environmental degradation by emissions
from production of exports and emissions from consumption of imports [14–16]. There
are many linkages between international trade and GHG emissions, including direct
effects from transportation, producing exports, and consuming imports [17]. Pan et al.
[17] also reported that China’s international trade contributed more than 30% of CO2
emissions in 2006. Many researchers have found that trade openness can improve the
eco-environment by efficient use and sustainable development of resources [13,18–20].
The purposes of this paper are to explore whether the inverted U-shaped hypothetical
environmental Kuznets curve (EKC) exists or not in China, and also to examine the
causality among per capita CO2 emissions, real GDP (constant 2010 US$), renewable
(including hydropower, wind, solar, etc.) and non-renewable (coal, oil, and natural gas)
energy use, and international trade (trade openness, exports, and imports). The con­
tributions of this paper are in three directions. First, we attempt to investigate whether
the EKC hypothesis exists, combining additional determinants, such as energy use and
trade to explain CO2 emissions, which may fill the gap of the studies in China’s case.
Second, consumptions of renewable and non-renewable energy are used instead of total
primary energy. Second, the model uses renewable and non-renewable energies, instead
of total primary energy, to obtain a more objective and accurate appraisal of emissions.
Third, compared with most previous papers with one international trade variable (total
trade or exports), this paper focuses on not only exports, but also imports, as well as trade
openness (% of GDP). The aim is to provide a better understanding on the impact of each
trade variable, and the impact of energy consumption on the environment in China.
Some suggestions and recommendations are proposed for trade and energy conservation
policies, as well as CO2 emissions mitigation.
The rest of this paper is structured as below: The second section reviews literature. The
third section states the used data, models, and methodologies. The fourth section is
devoted to the empirical results and discussions. The fifth section offers conclusions and
implications.

Literature review
The hypothetical environmental Kuznets curve (EKC) is put forward by Grossman and
Krueger [21] based on Kuznets [22]. The EKC hypothesis is that with the growth of
national or regional average income, the quality of environment deteriorates until reach­
ing a certain point, and then the impact turns into environmental enhancement. The idea
is widely used to investigate the linkage of environment and economy based on an
individual country or a group of countries.
Many analyses of the EKC hypothesis exist, as well as the causality linkages between
environmental pollutants and economic growth [23–31]. Yang et al. [32] examine the
economy-related GHG emissions in Russia covering 1998–2013 based on several types of
emissions. The EKC hypothesis is supported under a business-as-usual scenario. The
turning point will appear in a decade’s time in the current economy. Based on
590 X. LIU

disaggregated and aggregated energy emissions, the findings of Ahmad et al. [33] indicate
the existence of the inverted U-shaped EKC in the long-run estimate in India. Such
results are supported by Adebola Solarin et al. [34] in China and India based on the
autoregressive-distributed lag (ARDL, proposed by Pesaran and Shin [35]), the same
methodology used by Pegkas [36]. Long-run Granger causality is found among the
selected time series in both countries, bidirectionally. By using samples of major nuclear-
generating countries, Baek [37] rejects the so-called hypothetical EKC by using fully
modified ordinary least squares (FMOLS) [38] and dynamic ordinary least squares
(DOLS) [39]. Similarly, the hypothetical EKC is rejected in Liu et al. [40] and Baek
[41] for ASEAN countries, by Pablo-Romero and Jesús [42] for Latin America and the
Caribbean, and by Sirag et al. [43] for developing countries.
Recently, trade openness has received more attention via the extended multivariable
EKC hypothesis [15,44–48]. The results of Le et al. [49] show that there is no empirical
support for the inverted EKC hypothesis with global samples. Moreover, expansion of
trade can improve the environment in high-income countries, but can result in environ­
mental degradation in middle- and low-income countries. Conversely, by using data of
selected Central and Eastern European Countries, Destek et al. [50] find empirical
support for the hypothetical EKC. Besides, trade openness can improve the environment
based on group mean FMOLS and DOLS. Based on ARDL methodology, Ozatac et al
[51] support the hypothetical EKC and they further point out that trade openness is
harmful to Turkey’s environment.
This paper surveys only a few existing studies which explore the existence of the
hypothetical EKC by considering trade openness, and renewable and non-renewable
energy. Ben Jebli and Ben Youssef [52] employ the ARDL and vector error correction
model (VECM) approaches to explore linkages among emissions, output, renewable and
non-renewable energy use, and globalisation in Tunisia. These researchers conclude that
the hypothetical EKC is not found graphically, which means the country has not yet
reached the inverted U-shaped EKC. Jebli et al. [53] employ FMOLS and DOLS to explore
the long-run coefficients among CO2 emissions, economic growth, renewable and non-
renewable energy use, as well as trade with samples of 25 OECD countries covering
1980–2010. These authors find some evidence of two-ways Granger causalities among
selected time series in the short-run; and also some long-run bidirectional causalities
among selected variables. The result supports the existence of the EKC. In addition,
enlarging the use of non-renewable energy accelerates CO2 emissions, and international
trade and renewable energy can decrease it. By employing DOLS and FMOLS test, Al-
mulali et al. [54] indicate that renewable energy decreases CO2 emissions in four regions of
the world significantly, but there is no significant effect on the other two regions.
Moreover, the EKC hypothesis appears to be supported in those regions where renewable
energy and CO2 emissions have close relationships both in the short- and long-run.

Model, data, and methodology

Model and data
Following the empirical model applied by Ang [55] and Ben Jebli and Ben Youssef [52],
this study uses the following formula based on the hypothetical EKC:
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 591

�
COt ¼ f Yt ; Yt2 ; REt ; NREt ; TRt (1)
where CO stands for per capita carbon emissions, Y and Y2 are per capita real GDP and
the square of per capita real GDP, RE and NRE stand for per capita renewable and non-
renewable energy use, and TR is international trade. In this paper, the variable of
international trade is described by trade openness1 (TR, % of trade in GDP), real exports
(EX, constant 2010 US$), and real imports (IM, constant 2010 US$).
Here the natural logarithmic linear quadratic EKC equation is written as follows:
cot ¼ β0 þ β1 yt þ β2 yt2 þ β3 ret þ β4 nret þ β5 trt þ εt (2)
where ε and t are residual error and time span. βis the long-run elasticity impact of each
explanatory variable on emissions. According to the typical hypothetical EKC, the sign of
economic growth elasticity looks forward to be positive (β1 > 0), whereas the economic
growth elasticity of its square is to be negative (β2 < 0), which is to say the validity of
inverted U-shaped connection of variables, or else results would not support the hypothe­
tical EKC. The expected sign of renewable energy consumption elasticity is negative
(β3 < 0), because an increase use of renewable energy would decrease CO2 emissions.
The sign of non-renewable energy elasticity is expected to be positive (β4 > 0), which
denotes that the augmentation of non-renewable energy use would increase CO2 emissions.
The expected sign of β5 relies on the economic status of the research countries. In general,
for developed countries, β5 looks forward to be negative, as they specialise in products with
no pollution or less pollution, instead of exports or imports of pollution-intensive products
[21]. For example, Germany exports renewable electricity to other countries, which may
make the sign of β5 negative.
In this paper, we apply annual time series data in China covering 1965–2016. The
variables include per capita CO2 emissions, real GDP, renewable and non-renewable
energy use, and trade (exports, imports, and trade openness). Data on renewable and
non-renewable energy use are selected from the BP Statistical Review of World Energy
[56], and others are selected from the World Development Indicator [57]. Figure 1 shows
that the trends of per capita CO2 emissions (metric tons), real GDP (GDP, constant 2010
US$), renewable energy consumption (mainly including hydroelectric, solar, wind,
geothermal, biomass, and others, million tons oil equivalent (MTOE)), non-renewable
energy consumption (mainly comprising oil, gas, and coal, MTOE), real exports of goods
and services, real imports of goods and services (constant 2010 US$) and trade openness
(% of GDP) from 1965 to 2016 in China. Since China entered the WTO in 2001, the
growth rates of each variable have been increasing fast in China.
Figure 1 shows that the growth rates of per capita CO2 emissions and non-renewable
energy use increase slowly. Meanwhile, the growth rates of real GDP and renewable
energy develop very rapidly. In addition, with the global financial crisis in 2009, China’s
international trade decreased nearly 16.6%, the biggest setback on the selected timeline.
Fortunately, with government intervention and the world’s economic recovery, the
proportion and total amount of trade in China’s economy revived in 2010 and the
average growth rate achieved about 24.7% compared to 2009. China’s rapid economic
expansion, affecting international imports and exports of goods and services exchanges,
renewable and non-renewable energy and CO2 emissions, makes studying the linkage of
selected variables very interesting.
592 X. LIU

Figure 1. Trends of analysis variables from 1965 to 2016 in China.

Table 1 presents the summary and correlation for selected variables. From the growth
rate matrix, the correlation between CO2 emissions and non-renewable energy is the
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 593

Table 1. Summary and correlation for selected variables.

co y re nre tra ex im
Mean 0.820 6.804 −3.554 −0.492 3.088 4.608 4.559
Median 0.791 6.632 −3.693 −0.530 3.351 4.834 4.700
Maximum 2.022 8.840 −1.392 0.670 4.184 7.246 7.269
Minimum −0.555 5.153 −5.148 −1.828 1.593 1.739 1.554
Std. Dev. 0.720 1.164 1.070 0.726 0.805 1.937 1.925
Skewness 0.057 0.237 0.406 0.050 −0.493 −0.080 −0.141
Kurtosis 2.236 1.732 2.168 2.134 1.872 1.558 1.648
Jarque-Bera 1.293 3.969 2.925 1.645 4.857 4.559 4.136
Probability 0.524 0.137 0.232 0.439 0.088 0.102 0.126
Sum 42.658 353.826 −184.804 −25.561 160.589 239.608 237.060
Sum Sq. Dev. 26.434 69.116 58.409 26.909 33.033 191.332 189.007
Observations 52 52 52 52 52 52 52
co 1.000
y 0.977 1.000
re 0.983 0.990 1.000
nre 0.998 0.981 0.982 1.000
tra 0.903 0.912 0.887 0.911 1.000
ex 0.965 0.986 0.966 0.971 0.964 1.000
im 0.969 0.983 0.968 0.973 0.967 0.997 1.000

highest. Real GDP correlates with renewable energy. Trade openness correlates with
imports. Exports and imports have the highest correlation.

Econometric methodology
This paper mainly uses three steps to explore the long-run estimates and Granger
causality among emissions, real GDP, renewable and non-renewable energy, and
international trade (openness, exports, and imports), as well as testing the
hypothetical EKC for China over the selected period. The first step is to confirm
the stationary properties of all variables. The second step is to examine the co-
integrating linkages of the variables by using ARDL bounding test and Johansen
co-integration test. Short- and long-run estimations based on ARDL method are
calculated, as well as the models’ stability by the cumulative sum (CUSUM) and
the cumulative sum of squares (CUSUMSQ). The third step is to establish the
directional causality through VECM Granger causal relationship test.
Before starting the co-integration approach, the first step is to employ the unit-
root test to determine whether there is a unit root in the selected variables.
Numerous studies have employed the traditional unit-root tests to confirm the
stationarity of variables, including Augmented Dicky Fuller (ADF) [58,59] and
Phillips–Perron (PP) [60]. Perron [61] argues that traditional tests do not supply
precise outcome without structural breaks in the variables. The results would be
biased and spurious without considering structural break, especially calculating
the data of short-time span. For the purpose of solving the shortcoming, we use
the Zivot and Andrews (ZA) [62] unit-root test, which has the ability to ascertain
the breakpoint.
594 X. LIU

The precondition of traditional co-integration studies is that the variables are

integrated under the uniform order, such as Engle and Granger [63] based on
residual approach, Johansen and Juselius [64] based on the maximum likelihood
approach, and Phillips and Hansen [65] based on fully modified and dynamic
ordinary least squares (OLS) procedures. The present study applies the ARDL
bounds test [66] to test the co-integration linkage, which is more appropriate than
traditional methods. The ARDL approach has several econometric advantages. First,
it seems to be more suitable when the selected time series data are stationary at I(1)/
I(0). Second, the ARDL methodology is more appropriate and efficient for small
samples. Third, the short- and long-run estimations can be calculated simulta­
neously. Fourth, the approach avoids endogeneity problems.
This methodology can explore the estimates of short- and long-run parameters simul­
taneously. The ARDL model framework of Equation (2) can be expressed as follows:

p
X p
X p
X p
X
Δcot ¼ γ0 þ γ1i Δcot iþ γ2i Δyt iþ γ3i Δyt2 i þ γ4i Δret i
i¼1 i¼0 i¼0 i¼0
p
X p
X (3)
þ γ5i Δnret i þ γ6i Δtrt i þ γ7 cot 1 þ γ 8 yt 1 þ γ9 yt2 1
i¼0 i¼0
þ γ10 ret 1 þ γ11 nret 1 þ γ12 trt 1 þ μt

where Δ and μt present the first difference term and the error term. F-statistics test
is used to test if the estimations of lagged level variables present significant or not
[66]. Such test can be operated after identifying appropriate lag length by the Akaike
information criterion (AIC) [67]. On the basis of the test, null hypothesis implies no
co-integration between variables (H0: γ7 ＝ γ8 ＝ γ9 ＝ γ10 ＝ γ11 ＝ γ12 ＝ 0),
against the alternative hypothesis H1: γ7≠ γ8≠ γ9≠ γ10≠ γ11≠ γ12 ≠ 0 which indicates
co-integration. The measured values are compared with two terminal critical values
(lower one and upper one) for a given significance level. If the value of the
F-statistic is less than upper critical bound value, the alternative hypothesis of co-
integration would be accepted. Additionally, the residual diagnostic and stability
tests have also been conducted. Residual diagnostic tests examine the problem of
normality, serial correlation, and heteroskedasticity. Moreover, by using CUSUM
and CUSUMSQ techniques [68], the paper also investigates the estimates’ stability of
ARDL test. If the lines of such statistics stay within the critical bounds of the 5%
significant level, the null hypothesis would be accepted, which indicates that all
coefficients in the models are stable.
If all variables are confirmed to be I(1) and models are co-integrated, we move ahead to
investigate the causality linkages among the variables. VECM Granger causality test is
promoted by Engle and Granger [63] to examine causal linkage among variables. They
maintained that if the selected time series data are co-integrated, it is more appropriate to
estimate VECM than to estimate vector auto-regression (VAR) model as in a standard
Granger causality test. The empirical equation of the VECM Granger causal linkage form
can be modelled as follows:
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 595

2 3 2 3 2 32 3
Δcot π1 α11;1 α12;1 α13;1 α14;1 α15;1 α16;1 Δcot 1
6 Δy 7 6 π 7 6 α21;1 α26;1 76 7
6 t 7 6 27 6 α22;1 α23;1 α24;1 α25;1 76 Δyt 1 7
6 2 7 6 7 6 76 2 7
6 Δyt 7 6 π3 7 6 α31;1 α32;1 α33;1 α34;1 α35;1 α36;1 76 Δyt 1 7
6 7 6 7 6 76 7 þ ���
6 Δre 7 ¼ 6 π 7 þ 6 α α43;1 α44;1 α45;1 α46;1 7 6 7
6 t 7 6 4 7 6 41;1 α42;1 76 Δret 1 7
6 7 6 7 6 76 7
4 Δnret 5 4 π5 5 4 α51;1 α52;1 α53;1 α54;1 α55;1 α56;1 54 Δnret 1 5
Δtrt π6 α61;1 α62;1 α63;1 α64;1 α65;1 α66;1 Δtrt 1
2 32 3 2 3 2 3
α11;k α12;k α13;k α14;k α15;k α16;k Δcot k θ1 λ1t
6 α21;k α22;k α23;k α24;k α25;k 7 6 7 6
α26;k 76 Δyt k 7 6 θ2 7 7 6 7
6 6 λ2t 7
6 76 2 7 6 7 6 7
6 α31;k α32;k α33;k α34;k α35;k α36;k 76 Δyt k 7 6 θ3 7 6λ 7
þ6 76 7 þ 6 7ECTt 1 þ 6 3t 7
6α α42;k α43;k α44;k α45;k 7 6 7 6
α46;k 76 Δret k 7 6 θ4 7 7 6λ 7
6 41;k 6 4t 7
6 76 7 6 7 6 7
4 α51;k α52;k α53;k α54;k α55;k α56;k 54 Δnret k 5 4 θ5 5 4 λ5t 5
α61;k α62;k α63;k α64;k α65;k α66;k Δtrt k θ6 λ6t

where Δ and λ denote the difference operator and residual error, and ECTt-1 (the lagged
residual error correction term) denotes the estimation of the long-run linkage and θ, the
coefficient of ECTt-1 term, indicates the convergence speed from short- to long-run
balance relying on its sign. If θ is statistically significant with negative sign, it confirms
that there is long-run causality among selected variables based on t-statistic. To evaluate
short-run relationship, Wald-test is used on differenced and lagged differenced terms of
the independent variables by F-statistic. For example, α12;k �0"k indicates that emissions
are Granger caused by economic growth, and Granger causal relationship may be from
emissions to economic growth if α21;k �0"k .

Empirical results and discussions

Stationary tests
According to ADF and PP unit-root tests, the null hypothesis has a unit root. On the
contrary, the alternative hypothesis is level stationary. Table 2 shows that time series
supports the null hypothesis of a non-stationary nature in the level group; they almost
become stationary at the 5% level after 1st difference. Such results indicate that all the
variables are integrated at I(1).

Table 2. Unit root test without structural break.

ADF test statistic P-P test statistic
Variables Level p-Value 1st Difference p-Value Level p-Value 1st Difference p-Value
co −0.728(1) 0.830 −4.654(0) 0.000 −0.626(4) 0.855 −4.248(10) 0.001
y 1.464(1) 0.999 −6.226(1) 0.000 2.225(5) 1.000 −4.480(8) 0.001
y2 2.251(1) 1.000 −3.624(0) 0.009 3.826(1) 1.000 −3.548(6) 0.011
re 2.029(0) 1.000 −7.021(0) 0.000 2.307(3) 1.000 −7.021(2) 0.000
nre −0.797(1) 0.811 −3.874(0) 0.004 −0.671(2) 0.845 −3.320(11) 0.019
tr −1.336(0) 0.606 −4.781(0) 0.000 −1.312(1) 0.617 −4.695(3) 0.000
ex −0.550(0) 0.872 −5.482(0) 0.000 −0.548(2) 0.873 −5.476(1) 0.000
im −0.849(2) 0.796 −4.960(0) 0.000 −0.448(5) 0.893 −4.627(12) 0.000
The lag length is shown in parenthesis, based on the Schwarz Information Criterion (SIC) automatically.
596 X. LIU

In order to reduce misleading results, this paper also applies the ZA unit-root test with
unknown structural breaks. The results are shown in Table 3, and all the time series data
have a unit-root issue with a structural break in level, which supports the null hypothesis
of non-stationary. They support the alternative hypothesis of stationary after the first
difference at the 5% significant level.

ARDL approach co-integration tests

Because of the limited observations, this paper applies the Akaike information criterion
(AIC) to choose an optimal lag length to the ARDL model. From Figure 2, the graph leads

Table 3. Zivot-Andrews structural break unit root test.

Level 1st Difference
Variables t-Statistic p-Value Time break t-Statistic p-Value Time break
co −2.342(1) 0.939 2002 −5.396(0) <0.01 1970
y −0.437(1) >0.99 1981 −7.431(1) <0.01 1982
y2 0.070(1) >0.99 1991 −5.389(1) <0.01 1994
re −0.414(0) >0.99 2000 −8.125(0) <0.01 1999
nre −2.280(1) 0.950 2001 −4.434(0) 0.051 1969
tra −3.278(0) 0.517 1971 −5.287(0) <0.01 2006
ex −2.678(0) 0.839 1978 −6.122(0) <0.01 2009
im −2.205(0) 0.962 1977 −5.881(1) <0.01 2009
The lag length is shown in parenthesis, based on the Schwarz Information Criterion (SIC) automatically.

Figure 2. The appropriate lag order selection for Trade Openness, Exports, and Imports.
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 597

us to select the correct model, for trade openness model ARDL(1,0,1,3,1,1), exports
model ARDL(1,3,0,0,1,1), and imports model ARDL(1,0,1,3,1,1).
Table 4 presents long-run co-integration among per capita CO2 emissions, real GDP,
the square of real GDP, renewable and non-renewable energy use, and international trade
(trade openness, exports, and imports) at the significance level in China. These results do
not seem to support the null hypothesis of no co-integration. Additionally, Johansen co-
integration test (Table 5) is also employed, which indicates the same results as bounding
test. Such result implies that the used method is effective and robust, which indicates the
following process to be economically significant and consistent.
After establishing that there is co-integration between selected variables, the next step
is to estimate the parameters of short- and long-run terms and to examine the residual
diagnostic tests. Norm and LM tests are based on the residuals’ normality test and
Lagrange multiplier test for serial correlation, respectively.
From Table 6, in trade openness (% of trade in GDP), exports, and imports models,
the long-run estimates of economic growth and the square are expected to be γ1 < 0 and
γ2 > 0 at the 5% significance level under the hypothetical EKC. This means that the
hypothetical inverted U-shaped EKC between emissions and development of economy is
invalid in China. The result is the same as those obtained by Liu et al. [40] and Baek [69],
which do not support the hypothetical EKC for the Southeast Asian and Arctic countries
they studied. Zhang, however, finds that the EKC hypothesis applies in South Korea [45].
Moreover, the long-run elasticities of emissions with respect to economic growth are
0.036gdp-0.244 (trade openness model), 0.046gdp-0.389 (exports model) and 0.028gdp-
0.211 (imports model), respectively.
In the three models, the long-run elasticities of renewable and non-renewable energy
use on CO2 emissions are expected to be γ3 < 0 and γ4 > 0 significantly. These mean that
increasing per capita renewable energy may lead to decreasing CO2 emissions, and an
increase in non-renewable energy use would lead to an increase of emissions. From the
long-run estimates shown in Table 6, we found γ3 is −0.063 in trade openness model

Table 4. Estimated ARDL models and bounds F-test for co-integration.

Critical Value Bounds
ARDL model F-Statistic Significance I(0) Bound I(1) Bound
co|y,y2,re,nre,tr (1,0,1,3,1,3) 4.302 10% 2.08 3.00
co|y,y2,re,nre,ex (1,3,0,0,1,1) 5.583 5% 2.39 3.38
co|y,y2,re,nre,im (1,0,1,3,1,1) 2.845 1% 2.06 4.15

Table 5. Johansen co-integration test.

Hypothesised
Model No. of CE(s) None At most 1 At most 2 At most 3 At most 4 At most 5
Trade Openness Eigenvalue 0.606 0.530 0.364 0.234 0.167 0.010
Trace Statistic 122.094* 78.319* 42.818 21.563 9.062 0.493
Max-Eigen Statistic 43.775* 35.500* 21.255 12.502 8.569 0.493
Exports Eigenvalue 0.713 0.629 0.465 0.406 0.243 0.001
Trace Statistic 168.667* 111.257* 65.636* 36.886* 12.892 0.064
Max-Eigen Statistic 57.410* 45.621* 28.750* 23.995* 12.828 0.064
Imports Eigenvalue 0.732 0.669 0.427 0.398 0.372 0.001
Trace Statistic 181.908* 121.300* 70.458* 44.817* 21.471* 0.058
Max-Eigen Statistic 60.609* 50.841* 25.641* 23.346* 21.413* 0.058
598 X. LIU

Table 6. Long- and short-run estimates (model with trade openness, exports, and imports).
co ¼ f ðy; y2 ; re; nre; trÞ trade co ¼ f ðy; y2 ; re; nre; exÞ co ¼ f ðy; y2 ; re; nre; imÞ
Dependent variable: openness export import
Regressors Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic
Long-run estimates
y −0.244 −4.034*** −0.389 −7.907*** −0.211 −2.827***
y2 0.018 4.036*** 0.023 8.031*** 0.014 2.954***
re −0.063 −2.535** −0.039 −2.871*** −0.057 −1.954*
nre 1.003 56.199*** 0.991 82.598*** 0.998 50.317***
tr/ex/im 0.035 2.355** 0.046 4.955*** 0.021 1.490
constant 1.746 9.433*** 2.445 15.678*** 1.719 7.464***
Short-run estimates
Δ(y) −0.111 −3.163*** −0.136 −2.290** −0.135 −1.929*
Δ(y2) 0.011 2.501** 0.014 3.076*** 0.014 2.640**
Δ(re) −0.002 −0.174 −0.021 −2.340** 0.001 0.736
Δ(nre) 1.004 41.890*** 1.000 69.687*** 1.007 59.998***
Δ(tr)/Δ(ex)/Δ(im) −0.002 −0.208 0.009 1.752* −0.003 −0.601
constant 0.796 3.416*** 1.235 4.258*** 0.628 2.621**
ECTt-1 −0.456 −3.591*** −0.639 −6.854*** −0.425 −5.082***
Residual diagnostics
R2 0.993 0.995 0.992
F-stat 412.996 0.000 563.711 0.000 362.92 0.000
Norm test 0.561 0.755 1.795 0.408 0.362 0.835
LM test 1.705 0.198 1.31 0.284 0.934 0.404
DW 2.350 2.489 2.198
Superscripts ***, **, *denote statistical significance at the 1%, 5%, and 10% levels respectively.
The p-values are shown in parenthesis.

(−0.039 in exports model and −0.057 in imports model), which indicates that an increase
of 1% renewable energy consumption may lead to a decrease 0.063% reduction in CO2
emissions (0.039% in exports model and 0.057% in imports model).
Meanwhile, in both models, the γ4 is nearly 1.00 (trade openness model: 1.003;
export model: 0.991; imports model: 0.997) at the 1% significance level separately,
suggesting the contribution of non-renewable energy to CO2 emissions is certainly
tremendous during the estimation period. An increase of 1% non-renewable energy
use improves CO2 emissions by 1% in both functions. In addition, the coefficients of
openness and exports are positive and statistically significant at the 1% level, γ5 > 0.
This shows that an increase by 1% of trade openness or exports would enhance
0.035% or 0.046% of per capita CO2 emissions. Moreover, the coefficient of imports
is insignificant to the CO2 emissions, which may depend on importing high-tech
and low-pollution products. In 2015, China’s main import products were electro­
mechanical products and high-tech products.
From the second group of Table 6, the short-run estimated coefficients for trade
openness, exports, and imports models are shown as the first differenced variables. All
variables are shown to be significant, except renewable energy and openness in openness
model and import model. Concretely, in the trade openness model, the coefficient of
estimated ECTt-1, −0.456 (exports, −0.639; imports, −0.425) are negative at the 1% level
significance, confirming each deviation of variables changing from short- to long-run
equilibrium by 45.6% (exports 63.9%; imports 42.5%) for each period.
In the last part of Table 6, the null hypotheses of the absence of normality and serial
correlation could not be rejected. R2 (0.993, 0.995, and 0.992) are so high that the
adjustments of each model (trade openness, exports, and imports equations) are fairly
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 599

perfect. With Durbin-Watson (DW) at about 2, the models can reject the null hypothesis
of autocorrelation among residuals. Additionally, F-statistic is significantly based on
ARDL analysis.
The findings of long-run linkage among selected variables do not display the stability of
estimations. On account of the changes in China’s economic structure, the stabilities of the
short- and long-run coefficients are examined by CUSUM and CUSUMSQ tests. Figures 3–5
display the plots of CUSUM and CUSUMSQ test statistics falling into the upper and lower
critical bounds at the 5% significance in the trade openness, exports, and imports models,
which indicate that all estimated coefficients are stable over the sample period. Thus, the CO2
emissions model estimated based on trade openness, exports, and imports could be reliable.

VECM Granger causality tests

Since there is co-integration among analysed variables, the directional causal relationship
must be examined at least from one side. Therefore, short- and long-run causal linkages are
also explored between the selected time series data by traditional VECM Granger causality
(Table 7). In order to make the short-run causalities to be more intuitive, Figure 6 shows

Figure 3. Plots of CUSUM and CUSUMSQ of recursive residual (model with Trade Openness, % of GDP).
The straight lines represent critical bounds at 5% significance level.

Figure 4. Plots of CUSUM and CUSUMSQ of recursive residual (model with exports). The straight lines
represent critical bounds at 5% significance level.
 600 X. LIU

Figure 5. Plots of CUSUM and CUSUMSQ of recursive residual (model with imports). The straight lines
represent critical bounds at 5% significance level.

Table 7. VECM Granger causality tests.

Short-run Long-run
Dependent
variable Δco Δy/Δy2 Δre Δnre Δtr/Δex/Δim ECT
Trade Openness
Δco - 7.837*** 0.554 2.221 2.316* −1.690***[−3.413]
(0.001) (0.674) (0.111) (0.100)
Δy/Δy2 5.009*** - 1.410 5.718*** 3.382** −1.387***[−3.183]
(0.007) (0.263) (0.004) (0.034)
Δre 0.331 1.304(0.295) - 0.264 2.691* −0.683[−0.551]
(0.803) (0.851) (0.068)
Δnre 3.463** 7.857*** 0.484 - 3.157** −1.639***[−3.548]
(0.031) (0.001) (0.670) (0.042)
Δtr 1.714 3.498** 1.115 1.858 - 2.518[1.515]
(0.190) (0.030) (0.362) (0.163)
Exports
Δco - 7.027*** 0.137 1.523 1.472 −1.690***[−3.435]
(0.001) (0.937) (0.233) (0.246)
Δy/Δy2 7.053*** - 1.656 7.855*** 4.419** −1.436***[−3.744]
(0.001) (0.202) (0.001) (0.013)
Δre 0.525 0.504 - 0.586 1.908 0.619[0.502]
(0.669) (0.683) (0.630) (0.154)
Δnre 2.773* 6.759*** 0.087 - 2.123 −1.641***[−3.541]
(0.062) (0.002) (0.966) (0.123)
Δex 1.226 1.880 0.499 1.139 - 2.066[0.977]
(0.321) (0.159) (0.686) (0.353)
Imports
Δco - 6.552*** 0.581 1.728 1.645 −1.379***[−3.102]
(0.002) (0.633) (0.187) (0.204)
Δy/Δy2 3.121** - 1.097 3.647 2.116 −0.969**[−2.404]
(0.044) (0.369) (0.026) (0.124)
Δre 0.388 1.755 - 0.310 2.608* −1.074[−1.013]
(0.763) (0.182) (0.818) (0.074)
Δnre 2.784* 6.892*** 0.576 - 2.315* −1.392***[−3.378]
(0.062) (0.002) (0.636) (0.100)
Δim 1.282 2.615* 2.030 1.734 - 0.640[0.366]
(0.302) (0.073) (0.135) (0.186)
Superscripts ***, **, *denote statistical significance at the 1%, 5%, and 10% levels respectively.
The p-values are shown in parenthesis and t-statistics are shown in brackets.

the short-run Granger causality results based on the analysis of trade openness, exports,
and imports models. On the basis of Figure 6 and Table 7, the main Granger causal
relationships between variables can be outlined.
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 601

Figure 6. Short-run Granger causality results.

In both short- and long-run, bidirectional Granger there are causal linkages of
emissions per capita, real GDP, and non-renewable energy. This indicates that they affect
each other. Any change in the use of non-renewable energy affects the development of the
economy directly and carbon dioxide emissions indirectly through real GDP. Non-
renewable energy conservation policy not only affects economic development, but also
influences climate change and global heating. Such results suggest that Chinese policy­
makers should consider the relation of energy conservation policy to economic develop­
ment and ecological protection. The results are in accord with Kumar et al. [70]
confirming the feedback linkage between real income and emissions both in the short-
and long-run in India during the period of 1966–2011. Nevertheless, the results are
different from Ben Jebli and Ben Youssef [52] indicating only long-run one-way causal
linkage from non-renewable energy to emissions in Tunisia.
Unidirectional causalities from trade openness, real exports, and real imports to
emissions are found in the long-run directly. This indicates that trade openness policy
and the process of globalisation of goods and services would affect the emissions of CO2.
Furthermore, Granger causality from renewable energy to CO2 emissions is discovered in
the long-run, so the influences of renewable energy policy on emission reductions are
a long-term process. From the long-run estimates from ARDL test, the non-renewable
energy is the major source of emissions, so decreasing the use of non-renewable energy
can reduce CO2 emissions, as well as increasing the share of renewable energy in total
energy consumption.
Bidirectional causal linkage is found between real GDP and non-renewable energy
both in the short- and long-run. Any changes in non-renewable energy influence
China’s economic development, and vice versa. Such results are similar to Jebli et al.
[53] for OECD countries and Koengkan and Fuinhas for Mercosur countries [71], but
different from Ben Jebli and Ben Youssef [52] for Tunisia. For several decades, China’s
economic growth has relied greatly on non-renewable energy use. According to BP, the
proportion of non-renewable energy in Chinese primary energy use is extremely high,
especially coal, nearly 68% in 2014 [56]. In the long-run, one-way causality is found
from trade openness, exports, and imports to economic development. In the short-run,
exports can boost the local economy and economic growth can promote importing
other countries’ advanced technology and products to meet social and individual
needs.
602 X. LIU

In the long-run, there is a Granger causal linkage from renewable to non-renewable

energy. Any changes in renewable energy use may impact non-renewable energy in the
long-run. Our result differs from Apergis and Payne [72], who indicate the bidirectional
causal linkage of renewable and non-renewable energy for 80 countries. Moreover, there
is evidence from trade openness and imports to the renewable energy consumption in the
short-run. One-way causal linkages from trade openness, exports, and imports to non-
renewable energy exist, for both short- and long-run. The result differs from the results of
Jebli et al. [53], finding feedback causal linkage of trade and non-renewable energy in
both short- and long-run.

Conclusions and implications

This paper has examined the dynamic causal linkage of per capita CO2 emissions, real
GDP, renewable and non-renewable energy use, and international trade (trade openness,
exports, and imports) in China from 1965 to 2016. According to the econometric results,
all the time series data are stationary at first difference. Moreover, the variables are co-
integrating under the ARDL bounding test. The coefficients of real GDP and the square
of real GDP (β1 < 0 and β2 > 0) show the invalidity of the inverted U-shaped hypothetical
EKC between CO2 emissions and real GDP in trade openness, exports, and imports
models in the Chinese economy. Furthermore, the estimates indicate that an increase of
non-renewable energy use would accelerate CO2 emission in these three models, but an
increase of renewable energy consumption would lead to CO2 emission reduction
keeping other economic components constant. Long-run Granger causal linkages exist
in the equations of CO2 emissions, real GDP, and non-renewable energy, respectively.
Bidirectional short-run causalities are found between CO2 emissions and GDP, between
GDP and non-renewable energy, and between GDP and trade openness, respectively.
One-way causal linkages from emissions to non-renewable energy, from openness,
exports, and imports to non-renewable energy, from openness and imports to renewable
energy, from exports to GDP, from GDP to imports are found, respectively.
Accordingly, there are some policy implications. It is necessary to accelerate the
development of renewable energy and use of renewable energy, and augment the
quota of renewable energy in energy consumption in the long-run policy. The use of
renewable energy not only decreases CO2 emissions, but also leads to economic
prosperity. Moreover, policymakers should build more renewable energy infrastruc­
ture to raise the efficient use of renewable energy. According to Chinese National
Energy Administration, with the weak grid infrastructure, nearly 12.6% of China’s
solar capacity and 15% of wind capacity remains unused in 2015 [73]. Germany’s
efficient use of renewable energy depends on its reliable electricity grids [74]. China
should both invest in renewable energy and in improvements to infrastructure,
especially a strong smart grid, to accommodate the intermittent electricity (except
hydropower) by renewable energy. The renewable energy laws and the feed-in tariff
(FIT) need to be implemented everywhere. Coal power requires up to 200 times more
water compared to solar or wind power for the same amount of electricity [75].
A recent study of China’s power sector finds that a combination of renewable energy
and better plant cooling can reduce the water-intensity of China’s power generation
by up to 42% [76]. It is necessary to import clean non-renewable energy (natural gas
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 603

and nuclear power) to change the current energy mix. China has imported much non-
renewable energy from other countries to optimise the structure. In 2014, the share of
energy trade in total trade accounts for more than 10%, mainly by oil pipeline, natural
gas pipeline and liquefied natural gas sea transport. China should also use nuclear
energy to meet the general energy demand in preference to coal. There should be
a long-run policy not only to reduce CO2 emissions, but also to increase China’s
competitiveness in international trade under sustained economic growth. The recent
economic growth has depended on exports, with low-efficiency energy use, high CO2
emissions, and high pollution. Development of good quality is possible with efficient
use of renewable energy. That can keep growth sustainable and improve the
environment.

Note
1. Trade openness denotes percentage of trade in GDP, used by some authors [13,51,77–79].

Disclosure statement
No potential conflict of interest was reported by the author.

Funding
This study is supported by Department of Science and Technology of HeNan Province (No.
192400410095/202400410287) and Luoyang Normal University (No. 2018-PYJJ-019).

ORCID
Xuyi Liu http://orcid.org/0000-0003-3664-6543

References
[1] Liu, X., Zhang, S., and Bae, J., 2017, The nexus of renewable energy-agriculture- environ­
ment in BRICS. Applied Energy 204, 489–496. doi:10.1016/j.apenergy.2017.07.077
[2] Obersteiner, M., Bednar, J., Wagner, F., Gasser, T., Ciais, P., Forsell, N., et al., 2018, How to
spend a dwindling greenhouse gas budget. Nature Climate Change 8, 7–10. doi:10.1038/
s41558-017-0045-1
[3] Lu, X., Yao, T., Fung, J.C.H., and Lin, C., 2016, Estimation of health and economic costs of
air pollution over the Pearl River Delta region in China. Science of the Total Environment
566, 134–143. doi:10.1016/j.scitotenv.2016.05.060
[4] Bakhsh, K., Rose, S., Ali, M.F., Ahmad, N., and Shahbaz, M., 2017, Economic growth, CO2
emissions, renewable waste and FDI relation in Pakistan: New evidences from 3SLS. Journal
of Environmental Management 196, 627–632. doi:10.1016/j.jenvman.2017.03.029
[5] Jia, P., Li, K., and Shao, S., 2018, Choice of technological change for China’s low-carbon
development: Evidence from three urban agglomerations. Journal of Environmental
Management 206, 1308–1319. doi:10.1016/j.jenvman.2017.08.052
[6] Liu, Z., Guan, D., Crawford-Brown, D., Zhang, Q., He, K., and Liu, J., 2013, Energy policy: A
low-carbon road map for China. Nature 500, 143–145. doi:10.1038/500143a
604 X. LIU

[7] Liu, X. and Bae, J., 2018, Urbanization and industrialization impact of CO2 emissions in
China. Journal of Cleaner Production 172, 178–186. doi:10.1016/j.jclepro.2017.10.156
[8] PBL Netherlands Environmental Assessment Agency, 2016, Trends in Global CO2
Emissions 2016 Report. Available online at: https://edgar.jrc.ec.europa.eu/news_docs/jrc-
2016-trends-in-global-co2-emissions-2016-report-103425.pdf (accessed August 2020)
[9] Zhang, Q., Jiang, X., Tong, D., Davis, S.J., Zhao, H., Geng, G., et al., 2017, Transboundary
health impacts of transported global air pollution and international trade. Nature 543,
705–709. doi:10.1038/nature21712
[10] Zhang, Z., Zhao, Y., Su, B., Zhang, Y., Wang, S., Liu, Y., et al., 2017, Embodied carbon in
China’s foreign trade: An online SCI-E and SSCI based literature review. Renewable and
Sustainable Energy Reviews 68, 492–510. doi:10.1016/j.rser.2016.10.009
[11] Arce, G., López, L.A., and Guan, D., 2016, Carbon emissions embodied in international
trade: The post-China era. Applied Energy 184, 1063–1072. doi:10.1016/j.
apenergy.2016.05.084
[12] Long, R., Li, J., Chen, H., Zhang, L., and Li, Q., 2018, Embodied carbon dioxide flow in
international trade: A comparative analysis based on China and Japan. Journal of
Environmental Management 209, 371–381. doi:10.1016/j.jenvman.2017.12.067
[13] Naranpanawa, A., 2011, Does trade openness promote carbon emissions ? Empirical evi­
dence from Sri Lanka. The Empirical Economics Letters 10, 973–986.
[14] Ahmed, K., Shahbaz, M., and Kyophilavong, P., 2016, Revisiting the emissions-energy- trade
nexus: Evidence from the newly industrializing countries. Environmental Science and
Pollution Research 23, 7676–7691. doi:10.1007/s11356-015-6018-x
[15] Zhang, S., Liu, X., and Bae, J., 2017, Does trade openness affect CO2 emissions: Evidence
from ten newly industrialized countries? Environmental Science and Pollution Research 24,
17616–17625. doi:10.1007/s11356-017-9392-8
[16] Liddle, B., 2017, Consumption-based accounting and the trade-carbon emissions nexus.
Energy Economics 69, 71–78. doi:10.1016/j.eneco.2017.11.004
[17] Pan, J., Phillips, J., and Chen, Y., 2008, China’s balance of emissions embodied in trade:
Approaches to measurement and allocating international responsibility. Oxford Review of
Economic Policy 24, 354–376. doi:10.1093/oxrep/grn016
[18] Baek, J., Cho, Y., and Koo, W.W., 2009, The environmental consequences of globalization: A
country-specific time-series analysis. Ecological Economics 68, 2255–2264. doi:10.1016/j.
ecolecon.2009.02.021
[19] Shahbaz, M., Nasreen, S., Ahmed, K., and Hammoudeh, S., 2017, Trade openness–carbon
emissions nexus: The importance of turning points of trade openness for country panels.
Energy Economics 61, 221–232. doi:10.1016/j.eneco.2016.11.008
[20] Tan, H., Sun, A., and Lau, H., 2013, CO2 embodiment in China-Australia trade: The drivers
and implications. Energy Policy 61, 1212–1220. doi:10.1016/j.enpol.2013.06.048
[21] Grossman, G.M. and Krueger, A.B., 1991, Environmental Impacts of a North American Free
Trade Agreement. National Bureau of Economic Research (NBER Working Paper No. 3914;
issued in November 1991).
[22] Kuznets, S., 1955, Economic growth and income inequality. The American Economic Review
45, 1–28.
[23] Pablo-Romero, M.P., Cruz, L., and Barata, E., 2017, Testing the transport
energy-environmental Kuznets curve hypothesis in the EU27 countries. Energy Economics
62, 257–269. doi:10.1016/j.eneco.2017.01.003
[24] Özokcu, S. and Özdemir, Ö., 2017, Economic growth, energy, and environmental Kuznets
curve. Renewable and Sustainable Energy Reviews 72, 639–647. doi:10.1016/j.
rser.2017.01.059
[25] Zaman, K., Shahbaz, M., Loganathan, N., and Raza, S.A., 2016, Tourism development,
energy consumption and Environmental Kuznets Curve: Trivariate analysis in the panel
of developed and developing countries. Tourism Management 54, 275–283. doi:10.1016/j.
tourman.2015.12.001
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 605

[26] Liobikienė, G. and Butkus, M., 2017, Environmental Kuznets Curve of greenhouse gas
emissions including technological progress and substitution effects. Energy 135, 237–248.
doi:10.1016/j.energy.2017.06.120
[27] Apergis, N., 2016, Environmental Kuznets curves: New evidence on both panel and
country-level CO2 emissions. Energy Economics 54, 263–271. doi:10.1016/j.
eneco.2015.12.007
[28] He, Z., Xu, S., Shen, W., Long, R., and Chen, H., 2017, Impact of urbanization on energy
related CO2 emission at different development levels: Regional difference in China based on
panel estimation. Journal of Cleaner Production 140, 1719–1730. doi:10.1016/j.
jclepro.2016.08.155
[29] Zhao, Y., Wang, S., and Zhou, C., 2016, Understanding the relation between urbanization
and the eco-environment in China’s Yangtze River Delta using an improved EKC model
and coupling analysis. Science of the Total Environment 571, 862–875. doi:10.1016/j.
scitotenv.2016.07.067
[30] Dong, C., Dong, X., Jiang, Q., Dong, K., and Liu, G., 2018, What is the probability of
achieving the carbon dioxide emission targets of the Paris Agreement? Evidence from the
top ten emitters. Science of the Total Environment 622, 1294–1303. doi:10.1016/j.
scitotenv.2017.12.093
[31] Shahbaz, M., Mahalik, M.K., Shahzad, S.J.H., and Hammoudeh, S., 2018, Does the environ­
mental Kuznets curve exist between globalization and energy consumption? Global evi­
dence from the cross-correlation method. International Journal of Finance and Economics
24, 540–557. doi:10.1002/ijfe.1678
[32] Yang, X., Lou, F., Sun, M., Wang, R., and Wang, Y., 2017, Study of the relationship between
greenhouse gas emissions and the economic growth of Russia based on the Environmental
Kuznets Curve. Applied Energy 193, 162–173. doi:10.1016/j.apenergy.2017.02.034
[33] Ahmad, A., Zhao, Y.H., Shahbaz, M., Bano, S., Zhang, Z.H., Wang, S., et al., 2016, Carbon
emissions, energy consumption and economic growth: An aggregate and disaggregate
analysis of the Indian economy. Energy Policy 96, 131–143. doi:10.1016/j.
enpol.2016.05.032
[34] Adebola Solarin, S., Al-Mulali, U., and Ozturk, I., 2017, Validating the environmental
Kuznets curve hypothesis in India and China: The role of hydroelectricity consumption.
Renewable and Sustainable Energy Reviews 80, 1578–1587. doi:10.1016/j.rser.2017.07.028
[35] Pesaran, M.H. and Shin, Y., 1998, An autoregressive distributed lag modelling approach to
cointegration analysis. Econometric Society Monographs 31, 371–413.
[36] Pegkas, P., 2019, The impact of renewable and non-renewable energy consumption on
economic growth : The case of Greece. International Journal of Sustainable Energy 39,
1–16.
[37] Baek, J., 2015, A panel cointegration analysis of CO2 emissions, nuclear energy and income
in major nuclear generating countries. Applied Energy 145, 133–138. doi:10.1016/j.
apenergy.2015.01.074
[38] Pedroni, P., 2000, Fully modified Ols for heterogeneous cointegrated panels. Nonstationary
Panels, Panel Cointegration and Dynamic Panels 15, 93–130.
[39] Mark, N.C. and Sul, D., 2003, Cointegration vector estimation by panel DOLS and long-run
money demand. Oxford Bulletin of Economics and Statistics 65, 655–680. doi:10.1111/
j.1468-0084.2003.00066.x
[40] Liu, X., Zhang, S., and Bae, J., 2017, The impact of renewable energy and agriculture on
carbon dioxide emissions: Investigating the environmental Kuznets curve in four selected
ASEAN countries. Journal of Cleaner Production 164, 1239–1247. doi:10.1016/j.
jclepro.2017.07.086
[41] Baek, J., 2016, A new look at the FDI-income-energy-environment nexus: Dynamic panel
data analysis of ASEAN. Energy Policy 91, 22–27. doi:10.1016/j.enpol.2015.12.045
[42] Pablo-Romero, M.D.P. and De Jesús, J., 2016, Economic growth and energy consumption:
The energy-environmental Kuznets curve for Latin America and the Caribbean. Renewable
and Sustainable Energy Reviews 60, 1343–1350. doi:10.1016/j.rser.2016.03.029
606 X. LIU

[43] Sirag, A., Matemilola, B.T., Law, S.H., and Bany-Ariffin, A.N., 2017, Does environmental
Kuznets curve hypothesis exist? Evidence from dynamic panel threshold. Journal of
Environmental Economics and Policy 7, 145–165. doi:10.1080/21606544.2017.1382395
[44] Hu, H., Xie, N., Fang, D., and Zhang, X., 2018, The role of renewable energy consumption
and commercial services trade in carbon dioxide reduction: Evidence from 25 developing
countries. Applied Energy 211, 1229–1244. doi:10.1016/j.apenergy.2017.12.019
[45] Zhang, S., 2018, Is trade openness good for environment in South Korea ? The role of non-
fossil electricity consumption. Environmental Science and Pollution Research, 25, 9510–
9522. doi:10.1007/s11356-018-1264-3
[46] Hübler, M., 2017, The inequality-emissions nexus in the context of trade and development:
A quantile regression approach. Ecological Economics 134, 174–185. doi:10.1016/j.
ecolecon.2016.12.015
[47] Shahbaz, M., Solarin, S.A., Hammoudeh, S., and Shahzad, S.J.H., 2017, Bounds testing
approach to analyzing the environment Kuznets curve hypothesis with structural beaks:
The role of biomass energy consumption in the United States. Energy Economics 68,
548–565. doi:10.1016/j.eneco.2017.10.004
[48] Jalil, A. and Mahmud, S.F., 2009, Environment Kuznets curve for CO2 emissions:
A cointegration analysis for China. Energy Policy 37, 5167–5172. doi:10.1016/j.
enpol.2009.07.044
[49] Le, T.H., Chang, Y., and Park, D., 2016, Trade openness and environmental quality:
International evidence. Energy Policy 92, 45–55. doi:10.1016/j.enpol.2016.01.030
[50] Destek, M.A., Balli, E., and Manga, M., 2016, The relationship between CO2 emission,
energy consumption, urbanization and trade openness for selected CEECs. Research in
World Economy 7, 52–58. doi:10.5430/rwe.v7n1p52
[51] Ozatac, N., Gokmenoglu, K.K., and Taspinar, N., 2017, Testing the EKC hypothesis by
considering trade openness, urbanization, and financial development: The case of Turkey.
Environmental Science and Pollution Research 24, 16690–16701. doi:10.1007/s11356-017-
9317-6
[52] Ben, M. and Ben, S., 2015, The environmental Kuznets curve, economic growth, renewable
and non-renewable energy, and trade in Tunisia. Renewable and Sustainable Energy Reviews
47, 173–185. doi:10.1016/j.rser.2015.02.049
[53] Ben, J.M., Ben, Y.S., and Ozturk, I., 2016, Testing environmental Kuznets curve hypothesis:
The role of renewable and non-renewable energy consumption and trade in OECD
countries. Ecological Indicator 60, 824–831. doi:10.1016/j.ecolind.2015.08.031
[54] Al-mulali, U., Ozturk, I., Solarin, S.A., and Adebola, S., 2016, Investigating the environ­
mental Kuznets curve hypothesis in seven regions: The role of renewable energy. Ecological
Indicator 67, 267–282. doi:10.1016/j.ecolind.2016.02.059
[55] Ang, J.B., 2007, CO2 emissions, energy consumption, and output in France. Energy Policy
35, 4772–4778. doi:10.1016/j.enpol.2007.03.032
[56] BP, 2019, Statistical Review of World Energy. Available online at: https://www.bp.com/en/
global/corporate/energy-economics/statistical-review-of-world-energy/downloads.html
(accessed September 2020).
[57] World Bank, 2019, World Development Indicators. Available online at: https://databank.
worldbank.org/reports.aspx?source=2&series=NY.GDP.MKTP.KD.ZG&country=#
(accessed September 2020).
[58] Dickey, D.A. and Fuller, W.A., 1979, Distribution of the estimators for autoregressive time
series with a unit root. Journal of the American Statistical Association 74, 427–431.
[59] Dickey, D. and Fuller, W., 1981, Likelihood ratio statistics for autoregressive time series with
a unit root. Econometrica 49, 1057–1072. doi:10.2307/1912517
[60] Phillips, P.C.B. and Perron, P., 1988, Testing for a unit root in time series regression.
Biometrika 75, 335–346. doi:10.1093/biomet/75.2.335
[61] Perron, B.Y.P., 1989, The great crash, the oil price shock, and the unit root hypothesis.
Econometrica 57, 1361–1401. doi:10.2307/1913712
 INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 607

[62] Zivot, E. and Andrews, D.W.K., 1992, Further evidence on the great crash, the oil-price
shock, and the unit-root hypothesis. Journal of Business and Economic Statistics 10, 251–270.
[63] Engle, R.F. and Granger, C.W.J., 1987, Co-integration and error correction: Representation,
estimation, and testing. Econometrica 55, 251–276. doi:10.2307/1913236
[64] Johansen, S. and Juselius, K., 1990, Maximum likelihood estimation and inference on
cointegration with applications to the demand for money. Oxford Bulletin of Economics
and Statistics 52, 169–210. doi:10.1111/j.1468-0084.1990.mp52002003.x
[65] Phillips, P.C.B. and Hansen, B.E., 1990, Statistical inference in instrumental variables
regression with I(1) processes. Review of Economic Studies 57, 99–125. doi:10.2307/2297545
[66] Pesaran, M.H., Shin, Y., and Smith, R.J., 2001, Bounds testing approaches to the analysis.
Journal of Applied Econometrics 326, 289–326. doi:10.1002/jae.616
[67] Shrestha, M.B. and Chowdhury, K., 2007, Testing financial liberalization hypothesis with
ARDL modelling approach. Applied Financial Economics 17, 1529–1540. doi:10.1080/
09603100601007123
[68] Brown, R., Durbin, J., and Evans, J., 1975, Techniques for testing the constancy of regression
relationships over time. Journal of the Royal Statistical Society 37, 149–192.
[69] Baek, J., 2015, Environmental Kuznets curve for CO2 emissions: The case of Arctic
countries. Energy Economics 50, 13–17. doi:10.1016/j.eneco.2015.04.010
[70] Kumar, A., Shahbaz, M., Muhammad, Q., and Hye, A., 2013, The environmental Kuznets
curve and the role of coal consumption in India: Cointegration and causality analysis in an
open economy. Renewable and Sustainable Energy Reviews 18, 519–527. doi:10.1016/j.
rser.2012.10.031
[71] Koengkan, M. and Fuinhas, J.A., 2020, The interactions between renewable energy con­
sumption and economic growth in the Mercosur countries. International Journal of
Sustainable Energy 39, 594–614. doi:10.1080/14786451.2020.1732978
[72] Apergis, N. and Payne, J.E., 2012, Renewable and non-renewable energy consumption-
growth nexus : Evidence from a panel error correction model. Energy Economics 34,
733–738. doi:10.1016/j.eneco.2011.04.007
[73] China National Energy Administration, 2015, China’s Photovoltaic Industry Development
Analysis in 2015. Available online at: http://www.xinhuanet.com/fortune/2018-06/12/c_
1122972354.htm (accessed September 2020).
[74] Matschoss, P., Bayer, B., Thomas, H., and Marian, A., 2019, The German incentive regula­
tion and its practical impact on the grid integration of renewable energy systems. Renewable
Energy 134, 727–738. doi:10.1016/j.renene.2018.10.103
[75] International Renewable Energy Agency (IRENA), 2015, Renewable Energy in the Water,
Energy & Food Nexus (Abu Dhabi: IRENA Press).
[76] International Renewable Energy Agency (IRENA), 2016, International Renewable Energy
Agency and China Water Risk. Available online at: www.chinawaterrisk.org/resources/
research-reports (accessed September 2020).
[77] Acaravci, A., Erdogan, S., and Akalin, G., 2015, The electricity consumption, real income,
trade openness and foreign direct investment : The empirical evidence from Turkey.
International Journal of Energy Economics and Policy 5, 1050–1057.
[78] Shahzad, S.J.H., Kumar, R.R., Zakaria, M., and Hurr, M., 2017, Carbon emission, energy
consumption, trade openness and financial development in Pakistan: A revisit. Renewable
and Sustainable Energy Reviews 70, 185–192. doi:10.1016/j.rser.2016.11.042
[79] Dogan, E. and Seker, F., 2016, The influence of real output, renewable and non-renewable
energy, trade and financial development on carbon emissions in the top renewable energy
countries. Renewable and Sustainable Energy Reviews 60, 1074–1085. doi:10.1016/j.
rser.2016.02.006