Energy Policy 155 (2021) 112345

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

The nexuses between energy investments, technological innovations,

emission taxes, and carbon emissions in China
Qiang Ma a, Muntasir Murshed b, Zeeshan Khan c, *
a
School of Economics and Management, Liaoning Petrochemical University, 113001, China
b
School of Business and Economics, North South University, Dhaka, Bangladesh
c
School of Economics and Management (SEM), Tsinghua University, Beijing, China

A R T I C L E I N F O A B S T R A C T

Keywords: This paper aims to analyze the impacts of emission taxes, investments in the energy sector, expenditure on
CO2 emissions research and development, technological innovation, and tertiary sector development on the Chinese provincial
Carbon-neutrality carbon dioxide emission figures between 1995 and 2019. The econometric analysis involved the application of
Technological innovation
the latest methods which can simultaneously account for cross-sectional dependency, slope heterogeneity, and
Research and development
China
structural break issues in the data. The overall findings revealed that provincial growth and tertiary sector
Renewable energy development were responsible for the aggravation of the carbon dioxide emission trends in China. In contrast,
higher energy investments, technological innovation, renewable energy use, expenditure on research and
development, and carbon emission taxes are found to abate carbon dioxide emissions and, therefore, facilitate
the carbon-abatement agenda of China. Besides, the findings also revealed emission taxes, investment in research
and development, technological innovation, and renewable energy use jointly reduce carbon dioxide emissions
further. In line with these abovementioned findings, several policy-level recommendations are put forward.

1. Introduction the pre-industrial level (NOAA, 2017). As a result, several countries have
set out plans to become carbon-neutral to which China is no exception.
The aggravation of the global greenhouse gas emission levels has China, by virtue of being the largest manufacturer of industrial
necessitated the implementation of emission-inhibiting policies world­ commodities, accounted for the highest level of CO2 emissions across the
wide (Khan et al., 2021; Murshed and Tanha, 2021). Accordingly, the globe in 2017 (Li et al., 2021). Between 2000 and 2016, China’s
world economies have ratified the Paris Agreement (or COP21) and have year-on-year CO2 emission figures have grown at an average rate of
pledged to control the surge in their respective Greenhouse Gas (GHG) more than 11%, while the per capita CO2 emission figures have almost
emissions in order to collectively keep the global temperature rise in this tripled (World Bank, 2021). In 2016, almost 30% of the global CO2
century to below 2 ◦ C above the pre-industrial level (PAC, 2015; Dimi­ emissions were originated from China (World Bank, 2021). Besides, in
trov, 2016). Among the various GHG, carbon dioxide (CO2) accounted 2014, China’s CO2 emissions intensity was recorded at 3.37 kg per ki­
for almost three-quarters of total GHG emissions in 2018 (World Bank, logram of oil equivalent energy use which was almost 14% higher than
2021). Hence, such a high degree of CO2 emissions is often claimed to be the corresponding CO2 emissions intensity figure in 2000 (World Bank,
a key contributor to global warming which, therefore, abating CO2 2021). Therefore, keep into consideration such sustained aggravation of
emissions is assumed to be the major challenge for the global economies the nation’s carbon emissions scenario, President Xi Jingping recently
in respect of their compliance with the Paris Accord. Besides, it is to be addressed the General Assembly of the United Nations and declared
noted that following the ratification of the Paris Agreement, the global China’s national goal of turning carbon-neutral by 2060. Among the
CO2 emission levels, as opposed to the objectives of the COP21, have several means of achieving the carbon-abatement agenda, President Xi
rather increased. As per the predictions, it is claimed that if appropriate Jingping emphasized on eliminating China’s traditional dependency on
measures are not undertaken to cease the growth in the CO2 emission combustion for non-renewable resources for energy; thus replacing
levels, it would lead to a rise in the earth’s temperature by 3–4 ◦ C above non-renewable energy use with renewable alternatives (Liu et al., 2021;

* Corresponding author.
E-mail addresses: maqiang@lnpu.edu.cn (Q. Ma), muntasir.murshed@northsouth.edu (M. Murshed), Zeeshan.17@sem.tsinghua.edu.cn, zee8447@gmail.com
(Z. Khan).

https://doi.org/10.1016/j.enpol.2021.112345
Received 3 December 2020; Received in revised form 7 April 2021; Accepted 26 April 2021
Available online 11 May 2021
0301-4215/© 2021 Elsevier Ltd. All rights reserved.
Q. Ma et al. Energy Policy 155 (2021) 112345

Dong et al., 2021). strategies can be enhanced further through the enactment of stringent
The persistent deterioration in the levels of China’s CO2 is largely environmental laws. It has been acknowledged in the literature that
attributable to the nation’s predominant reliance on non-renewable environmental laws tend to not only stimulate a renewable energy
energy for electricity generation. In 2015, a little less than 24% of transition within an economy but can also safeguard the environmental
China’s aggregate electricity output was produced by utilizing renew­ attributes by transforming the production processes in a more environ­
able resources (World Bank, 2021). Fig. 1 portrays the contrasting shares mentally friendly manner (Murshed, 2021a,b; Murshed et al., 2021a;
of fossil and non-fossil fuels used for electricity generation purposes in Nathaniel et al., 2021a). In this regard, the imposition of emission taxes
China. It is can be seen that although the share of non-fossil fuels has can be hypothesized to complement China’s 2060 zero-carbon agenda
slightly increased over the years, fossil fuels have remained as the major further. Furthermore, the environmental woes of China are accredited to
source of energy input for electricity generation purposes. Hence, the industrial development of the Chinese economy (Zhu et al., 2017;
achieving a low carbon economy requires the share of clean (non-fossil Liu and Bae, 2018). However, this claim has often been a topic of debate
or renewable) energy in the total electricity output to be significantly since, in recent times, the tertiary (or services) sector has also been
enhanced. labeled as a major source of CO2 emissions in China (Wang et al., 2020).
However, this fuel transition, from unclean to cleaner alternatives, Hence, it is pertinent to assess the effects of tertiary sector development
requires hefty investments for the development of renewable energy on China’s objective of turning carbon-neutral by 2060.
technologies, in particular. Accordingly, the Chinese government has Against this background, this study specifically attempts to evaluate
undertaken investments in renewable energy development in order to the role of energy investments, technological innovation, R&D invest­
facilitate its carbon-abatement agenda. Public investment in the ment, emission taxes, and tertiary sector in achieving China’s 2060
renewable energy sector has increased by almost 286 billion United carbon-abatement target using annual data of 30 Chinese provinces
States dollars in the last one and a half decades (World Bank, 2021). from 1995 to 2019. This study contributes to the existing literature in
Besides, the overall investments in the energy sector have also gone up several aspects. Firstly, this is the only study that links emission taxes to
in China. These investments are anticipated to increase in the upcoming the carbon-neutrality goal of China using provincial-level data. This is
years and assist China in achieving the carbon-neutrality tag. Apart from one of the few studies which uses emission taxes to examine the impacts
such investments, it has also been acknowledged in the literature that of the provincial CO2 emission figures. The role of such environmental
both renewable energy development and carbon emission reduction protection instrument in the context of the carbon-neutrality agenda is
require technological innovation which can ultimately promote eco­ key from policy perspectives.The preceding studies have primarily
nomic growth without marginalizing the environmental attributes assessed the energy consumption impacts instead. However, it is
(Balsalobre-Lorente et al., 2018). On the other hand, technological important to explore the energy investment-CO2 emission nexus since
innovation is said to be conditional on the level of investments for such investments also cover investments made for renewable energy
Research and Development (R&D); therefore, expenditure on R&D development which is crucial for abating the emissions. Secondly, not
projects can also be hypothesized to portray a crucial role in enabling many of the existing studies have also shed light on the nexus between
China to achieve its carbon-mitigation objective. Fig. 2 illustrates the R&D and energy investment and CO2 emissions across the Chinese
trends in energy investments and R&D expenditure alongside the GDP of provinces. Once again, the relevance of R&D for renewable electricity
China between 1995 and 2019. It can be seen that although both the generation cannot be overlooked keeping China’s emission reduction
energy investment and R&D expenditure figures in China have increased obectives at the forefront of macroeconomic policymaking. Fourthly, as
over the years, the growth in these figures has not quite matched the opposed to the majority of the existing studies that have highlighted the
corresponding trends in the GDP values. Therefore, anticipating higher impacts of industrialization on China’s CO2 emission problems, this
energy investments and R&D expenditure to simultaneously ensure study evaluates the impacts of tertiary sector development on the pro­
economic growth and CO2 emissions abatement in China, these in­ vincial CO2 emission figures in China. This is relevant particularly
vestments should also be enhanced. relevant for understanding whether the Chinese tertiary sector is equally
Although reducing the use of non-renewable energy while simulta­ carbon-intensive as its industrial sector or not. Lastly, this study uses the
neously boosting renewable energy use can be a means of curbing the latest econometric techniques which simultaneously control for cross-
CO2 emissions figures in China, such a fuel diversification measure may sectional dependency, slope heterogeneity, and structural break issues
not be sufficient to facilitate the nation’s carbon-abatement agenda. in the provincial data to generate efficient and unbiased outcomes.
Hence, it can be argued that the efficacies of fuel diversification The rest of the paper is organized as follows: Section 2 reviews the

Fig. 1. Trends in fossil and non-fossil fuel-based electricity generation in China.

Source: World Development Indicators (World Bank, 2021)

2
Q. Ma et al. Energy Policy 155 (2021) 112345

Fig. 2. The trends in energy investments, R&D expenditure, and GDP of China.
Note: The investment in energy and expenditure on R&D figures are presented along the primary (left) axis while the GDP figures are presented along the secondary
(right) axis.
Source: World Development Indicators (World Bank, 2021)

relevant studies documented in the literature to highlight the existing However, renewable energy development, particularly for control­
gaps this study aims to fill. Section 3 presents the theoretical framework ling CO2 emissions, requires investments within the energy industry
and the methodology used in this study. Section 4 reports the results and (Hojnik and Ruzzier, 2016). It is believed that as the world is enduring
puts forward the discussions. Finally, the concluding remarks and policy the GHG emission-induced global warming adversities, it is high time to
suggestions are presented in section 5. scale up the levels of green energy investments to phase-out the tradi­
tional fossil fuel dependencies (Khan et al., 2019b). Therefore, envi­
2. Literature review ronmental planners often suggest the fossil fuel-dependent economies to
finance renewable energy projects (Mazzucato, and Semieniuk, 2018;
The various macroeconomic determinants of environmental degra­ Taghizadeh-Hesary and Yoshino, 2020) which would not only facilitate
dation, especially CO2 emissions, have been well-documented in the the clean energy transition but would also assist in significantly curbing
literature (Leal and Marques, 2020; Ali et al., 2020; Cheikh et al., 2021; carbon emissions. In line with this notion, Chaiyapa et al. (2018) rec­
Isik et al., 2021). Among these factors, economic growth the most ommended the oil and gas companies in Thailand to invest in renewable
commonly used determinant of CO2 emissions. The environmental im­ energy development technologies to ensure a low-carbon economy.
pacts of economic growth have been mixed. While several studies Moreover, in recent times, technological innovation, especially in
asserted growth to be detrimental for the environment due to the asso­ respect of technological advancement-related environmental welfare, is
ciated CO2 and other GHG emissions (Baloch, 2018; Khan et al., 2020a), strongly considered as a key determinant of CO2 emissions (Chen and
many studies have also recognized growth as the ultimate facilitator of Lee, 2020; Su et al., 2020). In a relevant study on Organization for
environmental development through the reduction in the emissions Economic Cooperation and Development (OECD) countries, Hashmi and
(Özokcu and Özdemir, 2017). Similarly, a plethora of the existing Alam (2019) found that a rise in the number of applications for
studies has also examined the non-linearity of the economic growth-CO2 environmentally-friendly technological patents reduces the CO2 emis­
emissions nexus through the prism of the Environmental Kuznets Curve sion levels within these countries. Similarly in the context of the Group
(EKC) hypothesis (Chen et al., 2019; Xue et al., 2021; Ahmad et al., of Seven (G7) economies, Khan et al. (2020b) argued that
2021). However, the authenticity of the EKC hypothesis could not be eco-innovation reduces the consumption-based CO2 emission levels. The
universally affirmed from the statistical evidence documented within authors claimed that enhancing environmentally-friendly technologies
these studies. However, apart from economic growth, several other key can assist the majority of the G7 nations to comply with their
macroeconomic variables have also been identified to influence CO2 emission-inhibiting commitments under the COP21. Recently,
emissions across the globe. Töbelmann and Wendler (2020), in the case of selected European na­
Among these, the preceding studies have probed into the dynamic tions, stated that compared to technological innovation in general,
impacts of energy consumption, both at aggregate and disaggregate environmental-related innovation is more likely to mitigate CO2 emis­
levels, on CO2 emissions (Inglesi-Lotz and Dogan, 2018; Khan et al., sions. On the other hand, environmental innovation can also be expected
2019a; Gu et al., 2019; Valadkhani et al., 2019; Dong et al., 2020). to facilitate renewable energy use as a mechanism for abating the energy
However, the existing studies in the literature have mostly claimed consumption-related CO2 emissions (Alvarez-Herranz et al., 2017; Bal­
non-renewable energy use to boost CO2 emissions (Gokmenoglu and salobre-Lorente et al., 2018). Similarly, several existing studies have
Sadeghieh, 2019; Joshua and Bekun, 2020). In contrast, renewable en­ scrcutinized the impacts of eco-innovation on CO2 emissions and found
ergy consumption was asserted to be effective in abating CO2 emissions eco-innovation to be effective in abating CO2 emissions (Khan et al.,
(Koengkan et al., 2019; Dong et al., 2020). Accordingly, many studies 2020c,d).
have suggested argued that a transformation from a Much like technological innovation, several studies have also high­
non-renewable-intensive to a renewable energy-intensive energy system lighted the importance of R&D investment for CO2 emission reduction
can help a nation to become carbon-neutral in the future (Dorotić et al., across the globe (Apergis and Ozturk, 2015; Lee and Min, 2015; Alvar­
2019; Murshed and Alam, 2021; Magazzino et al., 2021). ez-Herranz et al., 2017; Mensah et al., 2018). It is believed that sizeable

3
Q. Ma et al. Energy Policy 155 (2021) 112345

investments in R&D can help to boost environmental performances by consumption level), expenditure on R&D projects (RD) (measured in
reducing CO2 emissions through enhancing the efficiency levels of en­ terms of 10,000 Chinese Yuans), charges on emitted pollutants (EMC)
ergy and other resource use (Aragón-Correa et al., 2008). In recent (measured in terms of 10,000 Chinese Yuans), and tertiary value-added
studies on OECD countries, Alam et al. (2020) and Petrović and Lobanov (VAT) (measured in terms of 100 million Chinese Yuans). Since the
(2020) found that higher R&D investments reduce the growth of CO2 provincial CO2 emissions data is unavailable, this study follows Du et al.
emissions and also facilitate the consumption of cleaner energy re­ (2012) and Chen (2018) to compute the CO2 emissions data (measured
sources. Similarly, Paramati et al. (2020) concluded that growth in R&D in terms of kilotons) for the respective Chinese provinces. The emission
investments reduces CO2 emissions in selected European countries. Be­ figures are estimated using the emissions generated from the combus­
sides, the role of R&D investment for technological innovation is a tion of seven major fossil fuels including coal, fuel oil, coke, natural gas,
well-established conjecture within the endogeneous growth model diesel, kerosene, gasoline, and also emissions generated from cement
(Romer, 1990; Aghion and Howitt. 1992). production. The coefficients for emissions are followed from Intergov­
Several studies have also assessed the impacts of environmental ernmental Panel on Climate Change (IPCC). Similarly, the data for the
regulations on CO2 emissions. Among these, Lin and Jia (2019) showed fossil fuels consumed are also collected from the National Bureau of
that taxing fossil fuel energy industries can be an effective means of Statistics (NBS, 2020).
reducing CO2 emissions in China. Similarly, in the context of G7 coun­
tries, Hao et al. (2021) claimed that the imposition of environmental
taxes curbs the CO2 emission levels. On the other hand, Mardones and 3.2. Empirical model
Flores (2018) asserted that low carbon taxes may not be effective in
inducing a fuel change to reduce industrial CO2 emissions in Chile. In this section, the econometric models used in this study are intro­
However, higher levels of carbon taxes were found to be effective in duced. Keeping the core objective of this study to model the macro­
mitigating the emissions. Similarly, Farajzadeh (2018) stated that the economic determinants of CO2 emissions across the Chinese provinces,
imposition of emission taxes in Iran is efficient in reducing CO2 emis­ we first use a baseline model which can be specified as follows:
sions without contracting the national output level. Model (1):
Furthermore, the existing studies have predominantly explored the
CO2, ​ it = δ1 GDPit + δ2 IEit + δ3 TIit + δ4 RECit + εit (1)
impacts of industrialization on CO2 emissions and other forms of envi­
ronmental pollution (Ahmad and Zhao, 2018; Khan et al., 2019c; where the subscripts ‘i’ (i = 1, 2, …, 30) refers to the 30 Chinese
Mahmood et al., 2020). These studies claimed that as an economy gears provinces considered in this study and ‘t’ (t = 1995, 1996, …, 2019) is
up to industrialize its agriculture-based economy, the scale and the period of analysis. The parameters δk (k = 1, 2, …, 4) are the co­
composition effects enhance the industrial output levels which, in turn, efficients to be predicted. Since a rise in the national output level is
increase the level of CO2 emissions as well (Pata, 2018; Liobikienė and anticipated to enhance the demand for energy resources, it is likely to
Butkus, 2019). Similar studies in the context of China revealed that boost the energy consumption-induced CO2 emission levels as well
industrialization of the Chinese economy stimulated the CO2 emission (Nathaniel et al., 2020b). From this understanding, the sign of the co­
levels (Zhu et al., 2017). However, relatively few studies have explored CO2
efficient parameter δ1 can be expected to be positive (i.e., ∂∂GDP > 0). On
the tertiary sector development-CO2 emission nexus. Samargandi (2017)
the other hand, the impact of energy investments on CO2 emissions can
found statistical evidence regarding higher service sector value added
be ambiguous due to it being conditional on the fact that whether the
resulting in higher CO2 emissions in Saudi Arabia. The author further
investment is being undertaken for renewable or non-renewable ener­
claimed that both the industrial and the services sectors are to be equally
gy-oriented projects. For instance, in the context of funds being invested
blamed for the aggravation of Saudi Arabia’s CO2 emission figures.
for scaling up fossil fuel-based electricity generations, the sign of the
Therefore, the aforementioned review of the existing studies leads to
coefficient parameter δ2 can be expected to be positive (i.e., ∂CO ∂IE > 0).
2
the identification of several gaps in the literature. It is clear that not
many empirical studies have been conducted to explore the impacts of Conversely, investments for renewable energy development can make
investments in energy on CO2 emissions. To the best of the authors’ sure that the predicted coefficient parameter δ2 depicts a negative sign
CO2
knowledge, no prior study has attempted to evaluate the energy in­ (i.e., ∂∂GDP < 0). As far as the impacts of technological innovation are
vestment-CO2 emissions nexus using Chinese provincial data. Besides, concerned, it can be hypothesized that technological innovation, espe­
very few studies have also examined the effects of R&D investment in cially those related to environmental development, can effectively
respect of China’s carbon-abatement agenda. Moreover, the impacts of reduce the CO2 emissions in China (Balsalobre-Lorente et al., 2018).
emission taxes on the Chinese provincial CO2 emission figures have also Thus, the coefficient parameter δ3 can be expected to be negative (i.e.,
received nominal mentions in the literature. Furthermore, not many of ∂CO2
)
∂TI < 0 . Lastly, since renewable resources are considered as sources of
the existing studies have highlighted the impacts of tertiary sector
development on CO2 emissions. Therefore, this current study aims to fill clean energy, it is believed that amplifying the volumes of renewable
these existing gaps in the literature. energy consumption can be effective in curbing CO2 emissions (Saidi
and Omri, 2020; Nathaniel et al., 2021c). Hence, the sign of the coeffi­
)
3. Data and model specification CO2
cient parameter δ4 can be expected to be negative (i.e., ∂∂REC < 0 as well.
Based on the objective of evaluating the impacts of R&D investment
3.1. Data
on CO2 emissions in China, the baseline model is augmented with the
expenditure on R&D projects variable. This can be shown in Model (2):
This current study uses annual frequency data for 30 Chinese prov­
Model (2)
inces between 1995 and 2019 which is retrieved from the National
Bureau of Statistics (NBS, 2020) of China. Table 1 in the appendix list the CO2, ​ it = δ1 GDPit + δ2 IEit + δ3 TIit + δ4 RECit + δ5 RDit + εit (2)
names and regional distribution of the Chinese provinces considered in
Since R&D investments are acknowledged to facilitate renewable
this study. Among the determinants of CO2 emissions, this study con­
energy transition (Alam et al., 2020) and also help in the development of
siders provincial GDP (measured in terms of million Chinese Yuans),
environmental protection technologies (Petrović and Lobanov, 2020), it
investment in the energy industry (IE) (measured in terms of in million
can also be expected to reduce CO2 emissions (Apergis and Ozturk,
Chinese Yuans), technological innovation (TI) (measured in terms of the
2015). Consequently, the sign of the coefficient parameter δ5 can be
number of patents applied for), renewable energy consumption (REC)
)
(measured in terms of its share in the total provincial energy expected to be negative (i.e., ∂∂CO
RD
2
< 0 as well.

4
Q. Ma et al. Energy Policy 155 (2021) 112345

On the other hand, since the imposition of emission taxes is also p p

∑ ∑
hypothesized to be effective in influencing CO2 emissions, the baseline ΔZit = δi + πi Zi,t− 1 + θi Z t− 1 + θil nZt− l + γil nZi,t− l + εit (6)
model is augmented with the charges on emitted pollutant figures for l=0 l=1

each of the Chinese provinces. This can be specified as:

Model (3): whereas, Zt− 1 and ΔZt− l are both the cross-section averages for lagged
and first difference of each series. CPS written below is obtained from
CO2, ​ it = δ1 GDPit + δ2 IEit + δ3 TIit + δ4 RECit + δ6 EMCit + εit (3) CDF as:
It is said emission taxes tend to limit the use of fossil fuels whereby a
1 ∑n
drop in the CO2 emission figures can be anticipated following a rise in CPS = CDFi (7)
N i=1
the degree of the tax imposed (Mardones and Flores, 2018). Under such
a circumstance, the predicted sign of the coefficient parameter δ6 can be CDF is estimated from equation (7). Similarly, along with Pesaran
CO2
expected to be negative (i.e., ∂∂EMC <0 .
) (2007), we have also applied Bai and Carrion-i-Silvestre (2009) 1 which
is useful to overcome not the slope heterogeneity problem, cross-section
Lastly, the baseline model is also augmented with the share of the dependence but also provides structural breaks in each series using
tertiary sector value-added variable to ascertain the impacts of tertiary common factors from Bai and Ng (2004). This approach is useful to
sector development on the provincial CO2 emission figures in China identify unknown number and dates of the structural breaks in the data.
which can be specified as: Modified Sargan-Bhargava (MSB) is used for individual series and also to
Model (4): augment its power. Moreover, three pooling approaches including Z, P,
CO2, ​ it = δ1 GDPit + δ2 IEit + δ3 TIit + δ4 RECit + δ7 VATit + εit (4) Pm are used. P, Pm is computed from the average of p-values, based on
the simplified MSB statistics such as P* and P*m , respectively and, Z is the
Much like the industrial sector, the tertiary (or services) sector is also standardized individual statistics (Dobnik, 2011; Piton, 2016).
believed to be a major contributor to CO2 emissions (Samargandi, 2017).
Hence, the coefficient parameter δ7 can be expected to depict a positive 4.3. Panel cointegration approach
CO2
)
(i.e., ∂∂VAT >0 .
Prior to long-run estimations, it is pertinent to check whether there
are long-run relationships between the variables of concern. Therefore,
4. Econometric methodology
the cointegration analysis acts as a pre-requisite to estimating the long-
run regression coefficients. This study uses the Westerlund (2007) panel
4.1. Cross-section dependence and slope homogeneity tests
cointegration test since simple panel cointegration tests such as
McCoskey and Kao (1998), Larsson et al. (2001), Westerlund (2005),
In order to deal with panel data, it is imperative to check slope ho­
and Pedroni (2004) cannot tackle slope heterogeneity and
mogeneity and cross-sectional independence assumptions. The issue of
cross-sectional dependence issues. This method predicts four test sta­
cross-section dependence mainly arises due to global or local economic
tistics (Gt, Ga, Pt, and Pa) under the null hypothesis of no cointegrating
common shocks, globalization, economic or financial integration, oil
associations between the variables in the respective models. The coin­
price shocks, and unobserved components (Murshed and Dao, 2020).
tegration analysis is followed by the panel regression analysis.
The cross-section dependence issues if not dealt properly shall cause
inefficient and biased results (Murshed et al., 2020, 2021b). The
4.4. The panel regression analysis
cross-section dependence is tested by applying Pesaran (2015) test
which is useful when N > T i.e. cross-sections are more than time period.
Moreover, for long-run estimation, we apply a robust approach
The Pesaran (2015) CSD statistic equation is written as:
called cross-sectionally augmented autoregressive distributed lags (CS-
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
TN(N − 1) ̂ ARDL) method of Chudik and Pesaran (2015). This technique is superior
CSDNT = *ρN (5) to other conventionally used methods such as the pooled mean group
2
(PMG) approach due to its capacity to handle CSD, slope heterogeneity,
In equation (5), the pair-wise correlation coefficient is denoted by
unobserved common factors, and endogeneity issues in the data
ρ N . Similarly, slope homogeneity assumption is tested by using Pesaran
̂
(Pesaran and Tosetti, 2011). The issue of CSD is tackled through a
and Yamagata (2008) test. In the case of large (N) and small (T) cross-sectional averaging method within the model (Chudik and
assuming homogeneity shall provide misleading results. Slope hetero­ Pesaran, 2015). The general specification for the CS-ARDL model can be
geneity is caused by differences in demographic and economic struc­ shown as:
tures of different provinces.
p
∑ p
∑ ∑
3
COi,t o
2 =δ + δit COi,t− j
+ άit Wt− j + ΰit Z t− j + εit (8)
4.2. Panel unit root test j=1
2
j=0 j=0

Unit root tests are pertinent for assessing the order of integration
where Zt = (ΔCOi,t 2 Wt’)’contain cross-section averages and explanatory
among the variables (Murshed et al., 2021c,d; Rehman et al., 2021).
variables are included in Wit = (GDPit , IEit , TIit , RECit , RDit , EMCit ,
Hence, the second-generation panel unit root analysis is conducted in
this study. Unlike first-generation unit root tests Levin-Lin-Chu, Fish­ VAT it ). The robustness analysis is performed by using the augmented
mean group (AMG) test initiated by Eberhardt (2012). The AMG esti­
er-ADF test, Im, Pesaran and Shin, Choi test and, Maddala and Wu are
mator is useful especially when there is cross-section dependence,
not robust to slope heterogeneity and cross-section dependence.
endogeneity, unobserved common factors, and slope heterogeneity
Therefore, a more robust approach developed by Pesaran (2007) is used
problem.
in this study. This second-generation panel unit root test uses
cross-section averages of the first difference and lagged levels for each
observation utilizes augmented dickey fuller (ADF) regression (Dong
et al., 2018). The equation form of cross-section augmented
Dickey-Fuller (CDF) is written as:

1
For further details, please see Bai and Carrion-I-Silvestre (2009).

5
Q. Ma et al. Energy Policy 155 (2021) 112345

5. Results and their discussions Table 3

Pesaran and yamagata slope heterogeneity analysis.
Table 2 reports the estimates from the Pesaran (2015) CSD test. The Models ̃
Δ ̃Adjusted
Δ
statistical significance of all the variables, at 1% significance level, re­
jects the null hypothesis of cross-sectional independence to confirm the Model 1: CO2,it = f(GDPit , IEit , TIit , RECit ) 21.601*** 24.778***

panel data set used in this study to be cross-sectionally dependent. On Model 2: CO2,it = f(GDPit , IEit , TIit , RECit , RDit ) 19.528*** 23.013***

the other hand, the results from the Pesaran and Yamagata (2008) slope Model 3: CO2,it = f(GDPit , IEit , TIit , RECit , EMCit ) 19.223*** 22.654***
heterogeneity analysis, as shown in Table 3, reveal the existence of Model 4: CO2,it = f(GDPit , IEit , TIit , RECit , VATit ) 18.897*** 22.270***
heterogeneous slope coefficient issues for all four models. The statistical Note: *** denotes significance level at 1%.
significance of both the test statistics rejects the null hypothesis of ho­
mogeneous slope coefficients to affirm these claims.
The CSD and slope heterogeneity analyses are followed by the panel Table 4
unit root analysis. Table 4 reports the findings from the CDF (Pesaran, The cross-sectionally augmented dickey-fuller panel unit root analysis.
2007) analysis. The statistical significance of the test statistics, at 1% Variables Considering Trend Order
level, rejects the null hypothesis of non-stationarity to affirm the sta­
Level First Difference
tionarity of the variables at the first difference. This study also applies
the Bai and Carrion-I-Silvestre (2009) panel unit root analysis. The CO2 − 2.281 − 4.021*** I(1)
GDP 1.931 4.259*** I(1)
corresponding results, as reported in Table 5, reveal a mixed order of
− −
IE − 2.233 − 5.236*** I(1)
integration among the variables. It is evidenced that apart from the in­ TI − 2.285 − 4.574*** I(1)
vestment in energy sector variable which is stationary at the first dif­ REC − 2.159 − 5.403*** I(1)
ference, all the other variables considered in this study are stationary at RD − 2.511 − 5.586*** I(1)
EMC 2.182 5.262*** I(1)
level. The mixed order of integration among the variables justifies the − −
VAT − 1.782 − 3.942*** I(1)
choice of applying the CS-ARDL model which can handle variables that
are either integrated at level or at first difference or both. Note: *** denotes significance level at 1%.
The Westerlud (2007) panel cointegration analysis is then conducted
to ascertain the cointegrating properties among the variables of concern. the environment in China, but the negative impacts also tend to escalate
The estimates, as reported in Table 6, confirm the presence of cointe­ as the provincial economies tend to persistently go up. A reason behind
grating equations in all four models. The statistical significance of the this phenomenon could be because the Chinese economy, as a whole, is
predicted test statistics, at 1% significance level, affirm these claims. predominantly fossil fuel dependent, especially coal. Thus, as the
Hence, it can be said that CO2 emissions have long-run associations with provinces grow, it can be expected that the fossil fuel demand would
economic growth, energy investments, technological innovation, follow the same trend. Consequently, the CO2 emission levels are likely
renewable energy consumption, R&D investments, charges on emitted to surge as well. This is a crucial finding from the perspective of China’s
pollutants, and tertiary sector value added at the provincial level in the 2060 carbon-neutrality target since the nation is yet to ensure comple­
context of China. mentarity between its economic and environmental welfare. The posi­
The short- and long-run coefficient parameters are predicted using tive nexus between economic growth and CO2 emissions was also
the CS-ARDL method and the corresponding estimates are reported in confirmed in the studies by Baloch (2018) and Khan et al. (2020a) for
Table 7. The lagged error-correction terms for all four models are seen to Pakistan.
be negative and statistically significant. These confirm the existence of Besides, the coefficient estimates also reveal that investments in the
long-run equilibrium relationships between the variables of concern. energy industry can partially help China become carbon-neutral by
The high values of the error-correction terms imply that long-run 2060. The negative signs of the corresponding coefficient estimates
convergence from a short-run disequilibrium takes place at a rate of suggest that a 1 million Yuan worth rise in the level of energy in­
71.2%–80.5%. Besides, the predicted signs of the short- and long-run vestments contribute to reductions in the short- and long-run CO2
coefficients are as expected. emission levels by 0.026–0.071 kilotons and 0.027–0.064 kilotons,
As far as the coefficient estimates are concerned, it can be seen that respectively, ceteris paribus. Therefore, it can be hypothesized that the
higher levels of provincial GDP boost the CO2 emission levels both in the investments made for the development of the Chinese energy sector are
short- and long-run. A rise in the provincial GDP level by 1 million Yuans being used to develop relatively cleaner energy production capacities
increases the CO2 emissions levels by 0.587–0.694 kilotons in the short- whereby a carbon-inhibiting impact can be justified. These findings
run and by 0.732–0.921 kilotons in the long-run, ceteris paribus. Hence, it resonate with the assertions put forward in the studies by Mazzucato and
can be said that the growth policies pursued by the Chinese provinces Semieniuk (2018) and Taghizadeh-Hesary and Yoshino (2020) in which
are not aligned with the carbon-abatement agenda of China. Besides, the the authors recommended financing of renewable energy projects to be
findings also reveal that the adverse environmental impacts associated necessary for abating CO2 emissions.
with the higher levels of provincial growth are relatively larges in the Similarly, technological innovation is also evidenced to exert
long-run. This implies that not only is provincial growth detrimental to favorable environmental impacts in China. The corresponding coeffi­
cient estimates reveal that higher degrees of technological innovation
help to curb CO2 emissions both in the short- and long-run. The esti­
Table 2
mates show that if the number of patents applied for increases by 1 then
Pesaran cross-sectional dependence analysis.
CO2 emissions reduce by 0.029–0.051 kilotons in the short-run and by
Variables CSD-Statistic 0.032–0.048 kilotons in the long-run, ceteris paribus. This could be due to
CO2 88.581*** the understanding that technological innovations are believed to be
IE 101.786*** effective in facilitating the clean energy adoption processes and also
REC 77.311***
help to enhance the energy efficiency levels. In both cases, technological
TI 103.749***
GDP 96.164*** innovation can be expected to reduce CO2 emissions. Similar findings
RD 98.334*** were reported by Hashmi and Alam (2019) in the context of the OECD
EMC 94.049*** countries. Hence, it can be said that the advancement of technology can
VAT 103.935*** complement China’s other policies that are aimed at achieving the
Note: *** denotes significance level at 1%. carbon-abatement objective. This is an important finding since in the

6
Q. Ma et al. Energy Policy 155 (2021) 112345

Table 5
Bai and Carrion-i-Silvestre panel unit root analysis.
Variables Level First Difference Integration Order for Variables

Z Pm P Z Pm P

CO2 − 2.502*** 3.521*** 79.820*** – – – I(0)

IE − 0.360 − 1.981** 26.921 − 4.010*** 69.530*** 696.210*** I(1)
TI − 2.481*** 1.871*** 63.971*** – – – I(0)
REC − 1.551* 0.090 46.862* – – – I(0)
GDP − 1.922** − 0.017 45.831* – – – I(0)
RD − 2.380*** − 1.710** 48.740** – – – I(0)
EMC − 2.602*** − 1.210 45.990* – – – I(0)
VAT − 1.918** 0.980 49.500** – – – I(0)

Note: For Bai and Carrion-I-Silvestre (2009) test, 1, 5 and 10% critical values (CV) for Z and Pm statistics are 2.326, 1.645 and 1.282 while the critical values (CV) for P
are 56.06, 48.60 and 44.90, separately. ***, **, * denote significance level at 1%, 5% and 10%.

contemporary era of the Fourth Industrial Revolution (4IR) technology

Table 6
is expected to play a massive role in transforming the global production
Westerlund (2007) Cointegration analysis.
and consumption processes and can also be anticipated to ensure envi­
Gt Ga Pt Pa ronmental sustainability while stimulating the growth of the world
Model 1: CO2,it = f(GDPit , IEit , TIit , RECit ) economies in tandem.
− 3.601*** − 6.955 − 21.164*** − 12.630*** As far as the environmental impacts of renewable energy consump­
Model 2: CO2,it = f(GDPit , IEit , TIit , RECit , RDit ) tion are concerned, the coefficient estimates denote that phasing-out
− 6.966*** − 13.231*** − 22.432*** − 13.213*** fossil fuel-dependency is a major means for China to become carbon-
Model 3: CO2,it = f(GDPit , IEit , TIit , RECit , EMCit ) neutral by 2060. The estimates suggest that a 1% rise in the renew­
− 6.503*** − 8.921*** − 20.184*** − 13.42*** able energy share in the total energy consumption level of each province
Model 4: CO2,it = f(GDPit , IEit , TIit , RECit , VATit ) stimulates a reduction in the short-run CO2 emission levels by
− 4.857*** − 10.211*** − 22.760*** − 11.983*** 0.021–0.035 kilotons and also declines the long-run CO2 emission levels
Note: *** denotes significance level at 1%. by 0.027–0.063 kilotons, ceteris paribus. Therefore, it is also clear that
the carbon emission-inhibiting impacts of higher renewable energy use
follow an increasing trend since the long-run coefficients are relatively
Table-7 larger than the short-run coefficients. Since China is predominantly
CS-ARDL short- and long-run analysis. reliant on fossil fuels, it is relevant for the nation to augment renewable
Model Model 1 Model 2 Model 3 Model 4 resources into the energy systems in order to become carbon-neutral at
Variables Coefficient Coefficient Coefficient Coefficient the earliest. The negative renewable energy consumption-CO2 emissions
[Std.Error] [Std.Error] [Std.Error] [Std.Error] nexus was also verified in the studies by Koengkan et al. (2019) and
Short-run analysis Dong et al. (2020) for Latin American countries selected global econo­
mies, respectively.
0.587*** 0.623*** 0.692*** 0.694***
Besides, much like the cases of energy investments, technological
ΔGDPit
[0.164] [0.142] [0.162] [0.172]
ΔIEit − 0.059*** − 0.071*** − 0.053*** − 0.026*** innovation, and renewable energy use, the coefficient estimates also
[0.013] [0.020] [0.017] [0.004] reveal that R&D investments can be effective for China to become
ΔTIit − 0.040*** − 0.051*** − 0.029*** − 0.030*** carbon-neutral. The corresponding coefficient estimates denote that a
[0.009] [0.012] [0.007] [0.005]
10000 Yuan worth rise in R&D expenditure reduces CO2 emissions by
ΔRECit − 0.028* − 0.022*** − 0.021*** − 0.035*
[0.017] [0.005] [0.003] [0.021] 0.012 kilotons and 0.041 kilotons in the short- and long-run, respec­
ΔRDit − 0.012*** – – tively, ceteris paribus. Therefore, these finding complements the earlier
[0.002] findings from the sense that R&D investments, apart from directly
ΔEMCit – − 0.234* – reducing CO2 emissions by the development of carbon emission-
[0.121]
inhibiting technologies, can also indirectly mitigate CO2 emissions by
ΔVATit – – – 0.452***
[0.101] ensuring technological innovation and development of renewable en­
ECM(-1) − 0.712*** − 0.695*** − 0.805*** − 0.802*** ergy generation technologies within China. Besides, since the long-run
[0.046] [0.053] [0.055] [0.049] coefficient is relatively larger than the short-run coefficient, it can be
Long-run analysis
said that the marginal environmental benefits of undertaking R&D in­
GDPit 0.921*** 0.760*** 0.743*** 0.732***
[0.238] [0.162] [0.115] [0.190] vestment tend to increase with time. As a result, a persistent growth in
IEit − 0.064*** − 0.054*** − 0.059*** − 0.037*** the volume of R&D investment in China can be expected to lead the
[0.019] [0.011] [0.019] [0.008] nation to carbon-neutrality by 2060. The finding is parallel to the
TIit − 0.0472*** 0.0482*** − 0.0324*** − 0.039*** findings documented by Alam et al. (2020) and Petrović and Lobanov
[0.010] [0.011] [0.008] [0.008]
(2020) in the context of the OECD countries.
RECit − 0.063*** − 0.041*** − 0.027*** − 0.043***
[0.018] [0.012] [0.006] [0.009] Moreover, the coefficient estimates also justify the decision to
RDit – − 0.041*** – – impose emission taxes to mitigate carbon emissions. The negative signs
[0.013] of corresponding coefficient parameters suggest that if the charges on
EMCit − 0.423***
– – –
emissions increase by 10000 Yuans, it can be associated with a reduction
[0.104]
VATit – – – 0.482*** in CO2 emissions by 0.234 kilotons and 0.423 kilotons in the short- and
[0.103] long-run, respectively. This is a plausible finding from the understand­
ing that environmental protection policies tend to restrict the use of
Note: *** and * denote significance levels at 1% and 10%.
environmentally-unfriendly energy resources and also induce in­
vestments in technologies that can limit the volume of emissions.
Consequently, the emission taxes can indirectly and directly reduce the

7
Q. Ma et al. Energy Policy 155 (2021) 112345

CO2 emission levels in China. Similar conclusions were put forward in between R&D expenditure and renewable energy shares. Once again the
the study by Farajzadeh (2018) for Iran. AMG estimator is used to check the robustness of the long-run estimates.
Lastly, the coefficient estimates portray that the expansion of the Table 10 reports the corresponding coefficient estimates from the
Chinese tertiary sector is detrimental to its environmental attributes. It robustness analysis considering the interaction impacts. It can be seen
can be seen that if the tertiary value-added figures of the Chinese that controlling for the interacted impacts has reduced the adverse
provinces increase by 100 million Yuans, the CO2 emission levels are environmental impact of provincial growth while slightly increasing the
likely to go up by 0.452 kilotons and 0.482 kilotons in the short- and favorable environmental impacts associated with energy investment,
long-run, respectively. This implies that the tertiary sector of China, technological innovation, renewable energy use, and R&D investment.
much like its industry sector, is contributing to the aggravating trends in Besides, the coefficient estimates also reveal joint favorable impacts of
the nation’s CO2 emission figures. As a result, it can be claimed that the R&D investment and technological innovation and of R&D investment
structural transformation policies pursued in China are not aligned with and renewable energy consumption share on the environment in China.
its carbon-abatement objective. This finding corroborates the views of The negative signs, and statistical significance, of both the interaction
Samargandi (2017) where the author also found the tertiary sector to be terms, imply that R&D investment is not only effective in directly
responsible for higher levels of CO2 emissions in Saudi Arabia. mitigating CO2 but it also exerts an indirect emission-inhibiting impact
For a robustness check of the long-run coefficient estimates from the by enhancing the efficacies of technological innovation and renewable
CS-ARDL analysis, the AMG estimator is applied. The results from the energy use in respect of curbing the CO2 emission levels in China. Hence,
robustness analysis, as reported in Table 8, confirm the similarity of the it can be claimed that undertaking R&D investments is extremely
long-run coefficient estimates in respect of their signs across the two important for China to become carbon-neutral.
alternative regression methods considered in this study. However, in
comparison to the CS-ARDL estimates, the EMG estimates are seen to be 6. Conclusion and policy implications
relatively smaller in magnitude. This implies that the CS-ARDL method
corrects for the underestimate bias of the AMG technique. Attainment of China’s target of becoming carbon-neutral by 2060
For further robustness check of the long-run coefficients, we conduct requires proactive measures to be undertaken to limit the nation’s CO2
the AMG analysis across the regional distributions of the Chinese emission levels both at the national and provincial levels. China has
provinces. The results from the regional provincial analysis are reported been recognized as the leading emitter of CO2 across the globe courtesy
in Table 9. Overall, the disaggregated regional estimates are more or less of its monotonic dependency on fossil fuel sources for electricity gen­
similar to the findings from the aggregate analysis shown in Table 8 in eration purposes. Besides, the nation has also ratified the Paris Agree­
respect of the signs of the coefficient estimates. However, the magni­ ment and has pledged to significantly reduce its CO2 and other GHG
tudes of the estimated coefficient parameters are found to differ among emissions to join the global economies in tackling the climate change
the provinces across the different regions in China. It can be seen that adversities worldwide. However, despite having the intention, China has
compared to the other regions, the adverse environmental impacts neither phased-out its traditional fossil fuel-dependency nor has the
associated with provincial growth and tertiary sector development are nation managed to abate its emissions. Against this background, this
relatively higher for the case of the Chinese provinces belonging to the current study aimed to specifically analyze the impacts of energy in­
Northwest region. Conversely, the favorable environmental impacts vestments, technological innovation, R&D investment, emission taxes,
associated with energy investment, technological innovation, renewable and tertiary sector development on the CO2 emission levels in 30 Chi­
energy consumption, R&D expenditure, and emission tax are also found nese provinces between 1995 and 2019.
to be comparatively low for the North region. This implies that the The overall findings revealed that provincial growth and tertiary
provinces in the Northern region in China are the ones that are relatively sector development contribute to higher CO2 emissions, both in the
more polluted. This finding could be because the average volume of CO2 short- and long-run, in China. In contrast, higher investments in the
emissions from the provinces in the Northern region is higher than the energy sector, technological innovation, higher share of renewables in
average volumes of CO2 emissions from the provinces belonging to the the aggregate energy consumption level, R&D investment, and emission
other regions. Hence, keeping into consideration the 2060 carbon- taxes were found to be effective in reducing the CO2 emission levels,
abatement agenda, it is pertinent for the Chinese government to prior­ both in the short- and long-run. Besides, the disaggregated analysis
itize the implementation of CO2 emissions-inhibiting policies across the across the different regions led to the finding that compared to the other
provinces belonging to the Northern region. regions, the Chinese provinces across the Northern region are relatively
Furthermore, since R&D investment is expected to ensure techno­ more polluted. As a result, the adverse environmental impacts in the
logical innovation and also facilitate the renewable energy transition context of these provinces were ascertained to be relatively higher while
phenomenon, we also check the robustness of the long-run coefficient the favorable environmental impacts were evidenced to be compara­
estimates by evaluating the joint impacts of R&D investment, techno­ tively lower. Furthermore, the findings also revealed that greater R&D
logical innovation, and renewable energy use on the provincial CO2 investment not only reduces the CO2 emissions directly, but it also exerts
emission levels. Accordingly, model (2) is augmented with the interac­ an indirect emission-restraining impact through the channels of tech­
tion terms between R&D expenditure and technological innovation and nological innovation and renewable energy use. Therefore, in line with
these findings, several policy-level implications can be put forward to
enable China to achieve its carbon-abatement agenda.
Table 8 Firstly, to reduce emissions, China has to undertake major structural
Robustness Analysis using the AMG estimator. reforms in an environmentally-friendly manner. In this regard, China
Variables Model 1 Model 2 Model 3 Model 4 has to integrate emission-reduction strategies into its economic devel­
opment policies in order to ensure economic and environmental sus­
GDPit 0.818*** 0.597*** 0.679*** 0.633***
tainability in tandem. Secondly, and most importantly, China must
IEit − 0.060*** − 0.048** − 0.051*** − 0.031**
phase-out its predominant reliance on non-renewable energy resources
TIit − 0.025*** − 0.033*** − 0.028*** − 0.016**
and augment renewable energy resources into its national energy-mix.
RECit − 0.056** − 0.038** − 0.192*** − 0.039**
As a result, the nation has to scale up its investments in the energy
RDit − 0.013***
sector, especially for the development of its renewable energy sector.
EMCit − 0.342***
Besides, the government should rethink its decision of building new
VATit 0.402***
coal-fired power plants since enhancing the installed capacities to
Note: *** and ** denote significance level at 1% and 5%. generate electricity from coal would not only enhance the nation’s fossil

8
Q. Ma et al. Energy Policy 155 (2021) 112345

Table 9
Robustness analysis at the regional level.
East North

Variables Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4

GDPit 1.707*** 1.745*** 1.651*** 1.559** 2.329*** 2.486*** 2.430*** 2.060***

IEit − 0.050** − 0.073*** − 0.045** − 0.091*** − 0.046** − 0.048*** − 0.014* − 0.034*
TIit − 0.053*** − 0.022** − 0.018*** − 0.016*** − 0.014*** − 0.018*** − 0.016*** − 0.059**
RECit − 0.085* − 0.078** − 0.068* − 0.036** − 0.018* − 0.019* − 0.016** − 0.026***
RDit − 0.028*** − 0.013***
EMCit − 0.496*** − 0.264***
VATit 0.510** 0.819***
Northwest Northeast
Variables Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4
GDPit 1.944*** 1.831*** 1.957*** 2.118*** 0.401*** 0.457*** 0.549*** 0.784***
IEit − 0.035*** − 0.024*** − 0.086*** − 0.017** − 0.039*** − 0.019*** − 0.030*** − 0.026***
TIit − 0.039** − 0.084*** − 0.047** − 0.036** − 0.088** − 0.108** − 0.035** − 0.158***
RECit − 0.094* − 0.069*** − 0.017* − 0.062*** − 0.487 − 0.722*** − 0.447*** − 0.716***
RDit − 0.012*** − 0.021*
EMCit − 0.319*** − 0.324***
VATit 0.489*** 0.472***
South Central Southwest
Variables Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4
GDPit 1.428** 1.498** 1.457** 1.584*** 1. 562*** 1.639*** 1.807*** 1.501**
IEit − 0.074*** − 0.078*** − 0.057*** − 0.119** − 0.093* − 0.092* − 0.097* − 0.089*
TIit − 0.072** − 0.068*** − 0.041* − 0.042* − 0.037*** − 0.047*** − 0.044*** − 0.031***
RECit − 0.123*** − 0.153*** − 0.121** − 0.148*** − 0.016** − 0.014** − 0.017*** − 0.019**
RDit − 0.024*** − 0.038**
EMCit − 0.257*** − 0.121***
VATit 0.671* 0.524***

Note: ***, **, * denote significance level at 1%, 5% and 10%.

for financing green innovation projects across China. Besides, the gov­
Table 10 ernment should also enhance its expenditure budget for R&D projects
Robustness analysis controlling for interaction which can further facilitate technological progress and renewable en­
impacts. ergy transition to help China become carbon-neutral by 2060.
Variables Coefficient Fourthly, environmental pollution taxes should be seriously consid­
ered as a mechanism for the government to reduce the nation’s carbon
GDPit 0.501***
emission woes. It is recommended that the government introduces a
IEit − 0.051**
progressive emission taxation system which can be expected to nudge
TIit − 0.035***
the relatively more pollution-intensive Chinese industries to substitute
− 0.041**
the use of fossil fuels with cleaner alternatives. Besides, stringent
RECit
− 0.022**
emission taxation systems are also likely to obligate the owners of these
RDit
(RDit * TIit ) − 0.029***
industries to invest in technologies that would limit the discharge of CO2
(RDit * RECit ) − 0.078**
and other GHG into the atmosphere. Lastly, it is time that the govern­
Note: *** and ** denote significance level at 1% and ment realizes that not only the Chinese industrial sector is to be blamed
5%. for the nation’s carbon emission adversities but the tertiary section is
equally responsible for aggravating the CO2 emissions trends. Hence, it
fuel dependency but would also bottleneck its prospects of turning is once again necessary for the government to restructure its structural
carbon-neutral by 2060. Rather, the government should redirect in­ transformation strategies by making greater use of greener technologies.
vestment to set up renewable energy-based power plants which can In this regard, the energy-mix within the tertiary sector should be
further be expected to facilitate the carbon-abatement objective of dominated by renewable resources whereby the energy consumption-
China. based emission can be reduced to a large extent. Therefore, these
Thirdly, it is pertinent for China to incentivize investments for abovementioned policy recommendations can collectively function as a
technological development which would not only lead to the discovery roadmap for China to become carbon-neutral by 2060. As a result, it is
of revenant environmental protection technologies but would also help critically important for the government to incorporate these policies
the nation to attain technological innovation for renewable energy within its next five-year plan which could turn out to be the blueprint for
generation. For instance, technological innovation can be a mechanism ensuring environmental sustainability in China.
for China to develop clean coal technology which can be effective in Data limitation was the major constraint faced in conducting this
limiting the adverse environmental impacts associated with coal being study. It would have been ideal if we could extend the analysis to assess
used in the traditional manner. Besides, technological innovation can the impacts of the Chinese transportation sector on the CO2 emission
also be expected to enhance the energy efficiency levels which, in turn, levels. However, due to the unavailability of provincial-level transport
can be hypothesized to reduce the overall demand for energy in China. sector-related data, it could not be incorporated within the analysis. As
Consequently, a lower energy demand is likely to result in a lower level part of the future scope of research, this analysis can be extended to
of CO2 emissions as well. Hence, the government should encourage the reexamine the determinants of China’s CO2 emissions using non-linear
private sector to invest funds in technological innovation-related pro­ modeling approaches.
jects. Additionally, public-private partnerships can also be undertaken

9
Q. Ma et al. Energy Policy 155 (2021) 112345

CRediT authorship contribution statement Declaration of competing interest

Qiang Ma: Conceptualization, Methodology, Software, Data cura­ The authors declare that they have no known competing financial
tion, Formal analysis, Writing – original draft. Muntasir Murshed: interests or personal relationships that could have appeared to influence
Writing – review & editing, Supervision, Project administration. Zee­ the work reported in this paper.
shan Khan: Conceptualization, Software, Formal analysis, Writing –
review & editing, Supervision.

Appendix
Table 1
The Chinese provinces and their regional distributions

Province Region Province Region Province Region

Beijing North Zhejiang East Hainan South Central

Tianjin North Anhui East Chongqing Southwest
Hebei North Fujian East Sichuan Southwest
Shanxi North Jiangxi East Guizhou Southwest
Inner Mongolia North Shandong East Yunnan Southwest
Liaoning Northeast Henan South Central Shaanxi Northwest
Jilin Northeast Hubei South Central Gansu Northwest
Heilongjiang East Hunan South Central Qinghai Northwest
Shanghai East Guangdong South Central Ningxia Northwest
Jiangsu East Guangxi South Central Xinjiang Northwest

References Dimitrov, R.S., 2016. The Paris agreement on climate change: behind closed doors.
Global Environ. Polit. 16 (3), 1–11.
Dobnik, F., 2011. Energy Consumption and Economic Growth Revisited: Structural
Aghion, P., Howitt, P., 1992. A model of growth through creative destruction.
Breaks and Cross-Section Dependence, 1981869.
Econometrica 60, 323–351.
Dong, K., Dong, X., Jiang, Q., 2020. How renewable energy consumption lower global
Ahmad, M., Zhao, Z.Y., 2018. Empirics on linkages among industrialization,
CO2 emissions? Evidence from countries with different income levels. World Econ.
urbanization, energy consumption, CO2 emissions and economic growth: a
43 (6), 1665–1698.
heterogeneous panel study of China. Environ. Sci. Pollut. Control Ser. 25 (30),
Dong, K., Hochman, G., Zhang, Y., Sun, R., Li, H., Liao, H., 2018. CO2 emissions,
30617–30632.
economic and population growth, and renewable energy: empirical evidence across
Ahmad, M., Jiang, P., Murshed, M., Shehzad, K., Akram, R., Cui, L., Khan, Z., 2021.
regions. Energy Econ. 75, 180–192.
Modelling the dynamic linkages between eco-innovation, urbanization, economic
Dong, L., Miao, G., Wen, W., 2021. China’s carbon neutrality policy: objectives, impacts
growth, and ecological footprints for G7 countries. Does financial globalization
and paths. East Asian Policy 13 (1), 5–18.
matter? Sustainable Cities and Society 70, 102881. https://doi.org/10.1016/j.
Dorotić, H., Doračić, B., Dobravec, V., Pukšec, T., Krajačić, G., Duić, N., 2019. Integration
scs.2021.102881.
of transport and energy sectors in island communities with 100% intermittent
Alam, M.S., Apergis, N., Paramati, S.R., Fang, J., 2020. The impacts of R&D investment
renewable energy sources. Renew. Sustain. Energy Rev. 99, 109–124.
and stock markets on clean-energy consumption and CO2 emissions in OECD
Du, L., Wei, C., Cai, S., 2012. Economic development and carbon dioxide emissions in
economies. Int. J. Finance Econ.
China: provincial panel data analysis. China Econ. Rev. 23 (2), 371–384.
Ali, M.U., Gong, Z., Ali, M.U., Wu, X., Yao, C., 2020. Fossil energy consumption,
Eberhardt, M., 2012. Estimating panel time-series models with heterogeneous slopes.
economic development, inward FDI impact on CO2 emissions in Pakistan: testing
STATA J. 12 (1), 61–71.
EKC hypothesis through ARDL model. Int. J. Finance Econ.
Farajzadeh, Z., 2018. Emissions tax in Iran: incorporating pollution disutility in a welfare
Alvarez-Herranz, A., Balsalobre-Lorente, D., Shahbaz, M., Cantos, J.M., 2017. Energy
analysis. J. Clean. Prod. 186, 618–631.
innovation and renewable energy consumption in the correction of air pollution
Gokmenoglu, K.K., Sadeghieh, M., 2019. Financial development, CO2 emissions, fossil
levels. Energy Pol. 105, 386–397.
fuel consumption and economic growth: the case of Turkey. Strat. Plann. Energy
Aragon-Correa, A., Hurtado-Torres, N., Sharma, S., García-Morales, J., 2008.
Environ. 38 (4), 7–28.
Environmental strategy and performance in small firms: a resource-based
Gu, W., Zhao, X., Yan, X., Wang, C., Li, Q., 2019. Energy technological progress, energy
perspective. J. Environ. Manag. 86 (1), 88–103.
consumption, and CO2 emissions: empirical evidence from China. J. Clean. Prod.
Bai, J., Carrion-I-Silvestre, J.L., 2009. Structural changes, common stochastic trends, and
236, 117666.
unit roots in panel data. Rev. Econ. Stud. 76 (2), 471–501.
Hao, L.N., Umar, M., Khan, Z., Ali, W., 2021. Green growth and low carbon emission in
Bai, J., Ng, S., 2004. A PANIC attack on unit roots and cointegration. Econometrica 72
G7 countries: how critical the network of environmental taxes, renewable energy
(4), 1127–1177.
and human capital is? Sci. Total Environ. 752, 141853.
Baloch, M.A., 2018. Dynamic linkages between road transport energy consumption,
Hashmi, R., Alam, K., 2019. Dynamic relationship among environmental regulation,
economic growth, and environmental quality: evidence from Pakistan. Environ. Sci.
innovation, CO2 emissions, population, and economic growth in OECD countries: a
Pollut. Control Ser. 25 (8), 7541–7552.
panel investigation. J. Clean. Prod. 231, 1100–1109.
Balsalobre-Lorente, D., Shahbaz, M., Roubaud, D., Farhani, S., 2018. How economic
Hojnik, J., Ruzzier, M., 2016. What drives eco-innovation? A review of an emerging
growth, renewable electricity and natural resources contribute to CO2 emissions?
literature. Environmental Innovation and Societal Transitions 19, 31–41.
Energy Pol. 113, 356–367.
Inglesi-Lotz, R., Dogan, E., 2018. The role of renewable versus non-renewable energy to
Chaiyapa, W., Esteban, M., Kameyama, Y., 2018. Why go green? Discourse analysis of
the level of CO2 emissions a panel analysis of sub-Saharan Africa’s Вig 10 electricity
motivations for Thailand’s oil and gas companies to invest in renewable energy.
generators. Renew. Energy 123, 36–43.
Energy Pol. 120, 448–459.
Isik, C., Ongan, S., Ozdemir, D., Ahmad, M., Irfan, M., Alvarado, R., Ongan, A., 2021. The
Cheikh, N.B., Zaied, Y.B., Chevallier, J., 2021. On the nonlinear relationship between
increases and decreases of the environment Kuznets curve (EKC) for 8 OECD
energy use and CO2 emissions within an EKC framework: evidence from panel
countries. Environ. Sci. Pollut. Control Ser. 1–9.
smooth transition regression in the MENA region. Res. Int. Bus. Finance 55, 101331.
Joshua, U., Bekun, F.V., 2020. The path to achieving environmental sustainability in
Chen, W., 2018. The Effects of Income Inequality on Economic Growth: Evidence from
South Africa: the role of coal consumption, economic expansion, pollutant emission,
China. Doctoral dissertation. University of Bath.
and total natural resources rent. Environ. Sci. Pollut. Control Ser. 27 (9), 9435–9443.
Chen, Y., Lee, C.C., 2020. Does technological innovation reduce CO2 emissions? Cross-
Khan, M.K., Khan, M.I., Rehan, M., 2020a. The relationship between energy
country evidence. J. Clean. Prod. 263, 121550.
consumption, economic growth and carbon dioxide emissions in Pakistan. Financial
Chen, Y., Wang, Z., Zhong, Z., 2019. CO2 emissions, economic growth, renewable and
Innovation 6 (1), 1–13.
non-renewable energy production and foreign trade in China. Renew. Energy 131,
Khan, Z., Ali, S., Dong, K., Li, R.Y.M., 2021. How does fiscal decentralization affect CO2
208–216.
emissions? The roles of institutions and human capital. Energy Econ. 94, 105060
Chudik, A., Pesaran, M.H., 2015. Common correlated effects estimation of heterogeneous
https://doi.org/10.1016/j.eneco.2020.105060.
dynamic panel data models with weakly exogenous regressors. J. Econom. 188 (2),
393–420.

10
Q. Ma et al. Energy Policy 155 (2021) 112345

Khan, Z., Ali, M., Jinyu, L., Shahbaz, M., Siqun, Y., 2020c. Consumption-based carbon Murshed, M., Nurmakhanova, M., Elheddad, M., Ahmed, R., 2020. Value addition in the
emissions and trade nexus: Evidence from nine oil exporting countries. Energy Econ services sector and its heterogeneous impacts on CO 2 emissions: revisiting the EKC
89, 104806. hypothesis for the OPEC using panel spatial estimation techniques. Environ. Sci.
Khan, Z., Ali, S., Umar, M., Kirikkaleli, D., Jiao, Z., 2020b. Consumption-based carbon Pollut. Control Ser. 27 (31), 38951–38973.
emissions and international trade in G7 countries: the role of environmental Nathaniel, S., Barua, S., Hussain, H., Adeleye, N., 2021b. The determinants and
innovation and renewable energy. Sci. Total Environ. 730, 138945. interrelationship of carbon emissions and economic growth in African economies:
Khan, Z., Malik, M.Y., Latif, K., Jiao, Z., 2020d. Heterogeneous effect of eco-innovation fresh insights from static and dynamic models. J. Publ. Aff. 21 (1), e2141.
and human capital on renewable & non-renewable energy consumption: Murshed, M., Rahman, M.A., Alam, M.S., Ahmad, P., Dagar, V., 2021a. The nexus
Disaggregate analysis for G-7 countries. Energy 209, 118405. between environmental regulations, economic growth, and environmental
Khan, Z., Shahbaz, M., Ahmad, M., Rabbi, F., Siqun, Y., 2019c. Total retail goods sustainability: linking environmental patents to ecological footprint reduction in
consumption, industry structure, urban population growth and pollution intensity: South Asia. Environ. Sci. Pollut. Control Ser. https://doi.org/10.1007/s11356-021-
an application of panel data analysis for China. Environ. Sci. Pollut. Res 26 (31), 13381-z.
32224–32242. Murshed, M., Tanha, M.M., 2021. Oil price shocks and renewable energy transition:
Khan, Z., Sisi, Z., Siqun, Y., 2019b. Environmental regulations an option: Asymmetry Empirical evidence from net oil-importing South Asian economies. Energy Ecol.
effect of environmental regulations on carbon emissions using non-linear ARDL. Environ. 6 (3), 183–203. https://doi.org/10.1007/s40974-020-00168-0.
Energy Sources, Part A Recovery, Util. Environ. Eff. 41 (2), 137–155. Nathaniel, S.P., Alam, M.S., Murshed, M., Mahmood, H., Ahmad, P., 2021c. Is energy
Khan, M.K., Teng, J.Z., Khan, M.I., Khan, M.O., 2019a. Impact of globalization, economic diversification through enhancing nuclear and renewable energy consumptions
factors and energy consumption on CO2 emissions in Pakistan. Sci. Total Environ. effective in abating carbon dioxide emissions in the G7 countries? Environ. Sci.
688, 424–436. Pollut. Control Ser. https://doi.org/10.1007/s11356-021-13728-6.
Koengkan, M., Fuinhas, J.A., Marques, A.C., 2019. The effect of fiscal and financial Nathaniel, S.P., Murshed, M., Bassim, M., 2021a. The nexus between economic growth,
incentive policies for renewable energy on CO2 emissions: the case for the Latin energy use, international trade and ecological footprints: the role of environmental
American region. In: The Extended Energy-Growth Nexus. Academic Press, regulations in N11 countries. Energy, Ecology & Environment. https://doi.org/
pp. 141–172. 10.1007/s40974-020-00205-y.
Larsson, R., Lyhagen, J., Löthgren, M., 2001. Likelihood-based cointegration tests in NBS, 2020. National Bureau of Statistics of China. China Statistical Yearbook.
heterogeneous panels. Econom. J. 4 (1), 109–142. Özokcu, S., Özdemir, Ö., 2017. Economic growth, energy, and environmental Kuznets
Leal, P.H., Marques, A.C., 2020. Rediscovering the EKC hypothesis for the 20 highest curve. Renew. Sustain. Energy Rev. 72, 639–647. https://doi.org/10.1016/j.
CO2 emitters among OECD countries by level of globalization. International rser.2017.01.059.
Economics 164, 36–47. Paramati, S.R., Alam, M.S., Hammoudeh, S., Hafeez, K., 2020. Long-run relationship
Lee, K.H., Min, B., 2015. Green R & D for eco-innovation and its impact on carbon between R&D investment and environmental sustainability: evidence from the
emissions and firm performance. J. Clean. Prod. 108, 534–542. European Union member countries. Int. J. Finance Econ.
Li, Z.Z., Li, R.Y.M., Malik, M.Y., Murshed, M., Khan, Z., Umar, M., 2021. Determinants of Pata, U.K., 2018. The effect of urbanization and industrialization on carbon emissions in
carbon emission in China: how good is green investment? Sustainable Production Turkey: evidence from ARDL bounds testing procedure. Environ. Sci. Pollut. Control
and Consumption 27, 392–401. https://doi.org/10.1016/j.spc.2020.11.008. Ser. 25 (8), 7740–7747.
Lin, B., Jia, Z., 2019. How does tax system on energy industries affect energy demand, Pedroni, P., 2004. Panel cointegration: asymptotic and finite sample properties of pooled
CO2 emissions, and economy in China? Energy Econ. 84, 104496. time series tests with an application to the PPP hypothesis. Econom. Theor. 20 (3),
Liobikienė, G., Butkus, M., 2019. Scale, composition, and technique effects through 597–625.
which the economic growth, foreign direct investment, urbanization, and trade Pesaran, M.H., 2007. A simple panel unit root test in the presence of cross-section
affect greenhouse gas emissions. Renew. Energy 132, 1310–1322. dependence. J. Appl. Econom. 22 (2), 265–312.
Liu, J., Murshed, M., Chen, F., Shahbaz, M., Kirikkaleli, D., Khan, Z., 2021. An empirical Pesaran, M.H., 2015. Testing weak cross-sectional dependence in large panels. Econom.
analysis of the household consumption-induced carbon emissions in China. Rev. 34 (6–10), 1089–1117.
Sustainable Production and Consumption 26, 943–957. https://doi.org/10.1016/j. Pesaran, M.H., Tosetti, E., 2011. Large panels with common factors and spatial
spc.2021.01.006. correlation. J. Econom. 161 (2), 182–202.
Liu, X., Bae, J., 2018. Urbanization and industrialization impact of CO2 emissions in Pesaran, M.H., Yamagata, T., 2008. Testing slope homogeneity in large panels.
China. J. Clean. Prod. 172, 178–186. J. Econom. 142 (1), 50–93.
Magazzino, C., Mele, M., Schneider, N., 2021. A Machine Learning approach on the Petrović, P., Lobanov, M.M., 2020. The impact of R&D expenditures on CO2 emissions:
relationship among solar and wind energy production, coal consumption, GDP, and evidence from sixteen OECD countries. J. Clean. Prod. 248, 119187.
CO2 emissions. Renew. Energy 167, 99–115. Piton, S., 2016. A European Disease? Non-tradable Inflation and Real Interest Rate
Mahmood, H., Alkhateeb, T.T.Y., Furqan, M., 2020. Industrialization, urbanization and Divergence. No. 2016-09.
CO2 emissions in Saudi Arabia: asymmetry analysis. Energy Rep. 6, 1553–1560. Rehman, A., Ma, H., Ozturk, I., Murshed, M., Dagar, V., 2021. The dynamic impacts of
Mardones, C., Flores, B., 2018. Effectiveness of a CO2 tax on industrial emissions. Energy CO2 emissions from different sources on Pakistan’s economic progress: a roadmap to
Econ. 71, 370–382. sustainable environment. Environ. Dev. Sustain. https://doi.org/10.1007/s10668-
Mazzucato, M., Semieniuk, G., 2018. Financing renewable energy: who is financing what 021-01418-9.
and why it matters. Technol. Forecast. Soc. Change 127, 8–22. Romer, Paul M., 1990. Endogenous technological change. J. Polit. Econ. 98 (5).
McCoskey, S., Kao, C., 1998. A residual-based test of the null of cointegration in panel Saidi, K., Omri, A., 2020. Reducing CO2 emissions in OECD countries: do renewable and
data. Econom. Rev. 17 (1), 57–84. nuclear energy matter? Prog. Nucl. Energy 126, 103425.
Mensah, C.N., Long, X., Boamah, K.B., Bediako, I.A., Dauda, L., Salman, M., 2018. The Samargandi, N., 2017. Sector value addition, technology and CO2 emissions in Saudi
effect of innovation on CO2 emissions of OCED countries from 1990 to 2014. Arabia. Renew. Sustain. Energy Rev. 78, 868–877.
Environ. Sci. Pollut. Control Ser. 25 (29), 29678–29698. Su, C.W., Naqvi, B., Shao, X.F., Li, J.P., Jiao, Z., 2020. Trade and technological
Murshed, M., 2021a. Can regional trade integration facilitate renewable energy innovation: the catalysts for climate change and way forward for COP21. J. Environ.
transition to ensure energy sustainability in South Asia? Energy Rep. 7 (C), 808–821. Manag. 269, 110774.
https://doi.org/10.1016/j.egyr.2021.01.038. Taghizadeh-Hesary, F., Yoshino, N., 2020. Sustainable solutions for green financing and
Murshed, M., 2021b. Modeling primary energy and electricity demands in Bangladesh: investment in renewable energy projects. Energies 13 (4), 788.
An autoregressive distributed lag approach. Sustain. Prod. Consum 27 (C), 689–712. Töbelmann, D., Wendler, T., 2020. The impact of environmental innovation on carbon
https://doi.org/10.1016/j.spc.2021.01.035. dioxide emissions. J. Clean. Prod. 244, 118787.
Murshed, M., Alam, S., 2021. An estimation of the macroeconomic determinants total, Valadkhani, A., Smyth, R., Nguyen, J., 2019. Effects of primary energy consumption on
renewable and non-renewable energy demands in Bangladesh: the role of CO2 emissions under optimal thresholds: evidence from sixty countries over the last
technological innovations. Sustainable Production and Consumption. https://doi. half century. Energy Econ. 80, 680–690.
org/10.1007/s11356-021-12516-6. Wang, R., Hao, J.X., Wang, C., Tang, X., Yuan, X., 2020. Embodied CO2 emissions and
Murshed, M., Dao, N.T.T., 2020. Revisiting the CO2 emission-induced EKC hypothesis in efficiency of the service sector: evidence from China. J. Clean. Prod. 247, 119116.
South Asia: the role of Export Quality Improvement. GeoJournal. https://doi.org/ Westerlund, J., 2005. New simple tests for panel cointegration. Econom. Rev. 24 (3),
10.1007/s10708-020-10270-9. 297–316.
Murshed, M., Elheddad, M., Ahmed, R., Bassim, M., Than, E.T., 2021c. Foreign Direct Westerlund, J., 2007. Testing for error correction in panel data. Oxf. Bull. Econ. Stat. 69
Investments, Renewable Electricity output, and Ecological Footprints: Do financial (6), 709–748.
globalization facilitate renewable energy transition and environmental welfare in World Bank, 2021. World Bank. World Dev. Indic. http://databank.worldbank.org/data
Bangladesh? Asia Pac. Financ. Mark. https://doi.org/10.1007/s10690-021-09335-7. /reports.
Murshed, M., Ferdaus, J., Rashid, S., Tanha, M.M., Islam, M.J., 2021d. The Xue, L., Haseeb, M., Mahmood, H., Alkhateeb, T.T.Y., Murshed, M., 2021. Renewable
Environmental Kuznets curve hypothesis for deforestation in Bangladesh: An ARDL energy use and ecological footprints mitigation: evidence from selected South Asian
analysis with multiple structural breaks. Energy Ecol. Environ 6, 111–132. https:// economies. Sustainability 13 (4), 1613. https://doi.org/10.3390/su13041613.
doi.org/10.1007/s40974-020-00188-w. Zhu, Z., Liu, Y., Tian, X., Wang, Y., Zhang, Y., 2017. CO2 emissions from the
Murshed, M., Haseeb, M., Alam, M.S., 2021b. The Environmental Kuznets Curve industrialization and urbanization processes in the manufacturing center Tianjin in
hypothesis for Carbon and Ecological footprints in South Asia: The role of Renewable China. J. Clean. Prod. 168, 867–875.
Energy. GeoJournal. https://doi.org/10.1007/s10708-020-10370-6.

11