Skip to main content

Full text of "Examining the synergistic effect of CO 2 emissions on PM 2.5 emissions reduction: Evidence from China"

See other formats


Journal of Cleaner Production 223 (2019) 759-771 



ELSEVIER 


Contents lists available at ScienceDirect 

Journal of Cleaner Production 

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



Examining the synergistic effect of CO2 emissions on PM2.5 emissions r 
reduction: Evidence from China sag 

Feng Dong , Bolin Yu , Yuling Pan 

School of Management, China University of Mining and Technology, Xuzhou, 221116, PR China 


ARTICLE INFO 


ABSTRACT 


Article history: 

Received 10 January 2019 
Received in revised form 
18 February 2019 
Accepted 12 March 2019 
Available online 15 March 2019 


Keywords: 

PM 2 .5 emissions 
LMDI 

C0 2 emissions 

Synergistic emissions reduction 
China 


Under the background of global climate change, China has been confronted with the dual pressure of CO2 
emissions reduction and PM2.5 pollution control. This research aims to explore the mechanism of 
changes in PM2.5 emissions, which are the key airborne pollutants causing haze. Furthermore, it quan¬ 
tifies the impacts of CO2 emissions reduction activities on PM2.5 emissions reduction. This study takes 
aggregate PM2.5 emissions instead of PM2.5 concentration index as the research object. Based on an 
extended kaya identity, LMDI approach is first performed to decompose the changes of PM2.5 emissions 
during 1998 - 2014 , taking into consideration the synergistic effect of carbon emissions on PM2.5 emis¬ 
sions. Furthermore, following LMDI decomposition, this study adopts the econometric methods to 
quantify the synergistic effect of CO2 emissions reduction on PM2.5 emissions reduction over the period 
1999 - 2014 . The empirical results are as follows: ( 1 ) the LMDI decomposition results specify that the 
synergistic emissions reduction accounts for the most of the reduction in PM2.5 emissions. In addition, 
energy intensity changes also contribute to the reduction in PM2.5 emissions; ( 2 ) by contrast, it is found 
that the economic development effect is the main factor resulting in the increase of PM2.5 emissions, 
while the contributions of the energy emission intensity effect and population effect to PM2.5 emissions 
changes are relatively little; ( 3 ) all the models show CO2 emissions reduction activities will significantly 
contribute to PM2.5 emissions reduction; ( 4 ) for every 10 , 000 1 increase in CO2 emissions reduction, PM2.5 
emissions reduction will increase by 3.3 t, and the potential for synergistic emissions reduction of PM2.5 
differs distinctly among different provinces; ( 5 ) technological progress and population density positively 
influence PM2.5 emissions reduction, while coal consumption rate has a negative impact on PM2.5 
emissions reduction, in addition, there is an inverted U-shaped curve relationship between per capita 
GDP and PM2.5 emissions reduction. 

© 2019 Elsevier Ltd. All rights reserved. 


1. Introduction 

With the acceleration of industrialization and urbanization in 
China, the issue of air pollution has become increasingly serious. 
Specially, haze pollution has frequently occurred in recent years, 
and the airborne PM 2.5 particulates not only lead to environmental 
health damage, but also cause many negative impacts on economic 
development, including affecting foreign investment, introduction 
of talented people and tourism development. PM 2.5 has become the 
primary pollutant worsening the quality of China’s atmospheric 


* Corresponding author. 

** Corresponding author. 

E-mail addresses: dongfeng2008@126.com (F. Dong), yubolinadw@foxmail.com 
(B. Yu). 

https://doi.org/10.1016/jjclepro.2019.03.152 

0959-6526/© 2019 Elsevier Ltd. All rights reserved. 


environment, and it has the characteristics of wide range of influ¬ 
ence, high frequency of occurrence, and difficulty in tackling. To 
address the increasingly prominent haze pollution, it is necessary 
to scrupulously and comprehensively investigate the internal 
changing mechanism of haze pollution in China, so as to provide a 
scientific basis for the formulation of haze mitigation policies. 

Air pollution and climate change are two challenges faced by the 
atmospheric environment today. Developed countries have basi¬ 
cally completed air pollution control at the end of the 20th century, 
and climate change issues have begun to arouse the global atten¬ 
tion in the early 21 st century (Dong et al., 2019b; Li and Su, 2017). At 
present, developed countries mainly bear the international re¬ 
sponsibility for greenhouse gas emissions reduction (Li et al., 2019). 
As a developing country, China is still in the period of industriali¬ 
zation. Due to long-term extensive economic development, China is 
faced with not only the domestic pressure of air pollution control, 














760 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


but also the international pressure of carbon emissions reduction 
(Xian et al., 2018). In 2007, China surpassed the United States and 
became the largest carbon emitter in the world (Dong et al., 2018a). 
Accordingly, China has made lots of efforts in carbon emissions 
reduction, and made many emissions reduction commitments. For 
example, the carbon intensity in 2020 is expected to be decreased 
by 40%—45% compared with the 2005 level (Dong et al., 2018b), and 
the carbon intensity in 2030 will be lower than that in 2005 by 
60%—65% (Dong et al., 2018c). The emissions of greenhouse gas and 
atmospheric pollutants are homologous, and both of them are 
mainly caused by the combustion of fossil fuels. The reductions of 
C0 2 and PM 2.5 emissions are consistent in action, and the realiza¬ 
tion of collaborative control over atmospheric pollutants and 
greenhouse gas emissions has a realistic basis. Air quality can be 
effectively improved in the process of reducing greenhouse gas 
emissions, and the resulting environmental health gain will reduce 
the cost of emissions reduction and increase the cost efficiency of 
emissions reduction technologies (Yang et al., 2013). If China’s 
carbon emissions reduction targets can be achieved, the haze 
pollution will be alleviated to some extent. In 2015, the revised law 
on the prevention and control of atmospheric pollution (NPC, 
2015), known as the “most stringent law on prevention and con¬ 
trol of atmospheric pollution in the history of China", first proposed 
“collaborative control of atmospheric pollutants and greenhouse 
gas emissions”. The work plan for the control of greenhouse gas 
emissions during the 13th Five-Year Plan period and the 13th Five- 
Year Plan for the protection of ecological environment (State 
Council of China, 2016a, 2016b) have clearly identified “strength¬ 
ening the collaborative control of air pollutant emissions and car¬ 
bon emissions” as an important means of low-carbon 
transformation. As China’s economy enters a new normal, the 
quality of ecological environment has become an important indi¬ 
cator for the performance appraisal of local government officials. 
For China, which is in a critical period of economic transformation, 
the collaborative control of carbon emissions and air pollution is an 
important policy path. There are still many challenges in how to 
achieve collaborative control of air pollution and carbon emissions, 
which deserve our in-depth study. 

In response to carbon emissions, the Chinese government has 
not only made a series of emissions reduction commitments, but 
also formulated a number of specific measures, including orders 
and control measures, and carbon emissions trading (Dong et al., 
2019a), with some remarkable results achieved. The govern¬ 
mental measures for haze mitigation are introduced late in China. 
In 2012, the Ministry of Environmental Protection in China passed 
the Ambient Air Quality Standards (DEP, 2012), and began to carry 
out monitors on haze pollution in various places. The Action Plan 
for Air Pollution Prevention and Control (State Council of China, 
2013) clearly proposed 35 specific haze control measures, and set 
clear air pollution control target for each region. For instance, the 
PM 2.5 concentration in the Beijing-Tianjin-Hebei region, Yangtze 
River Delta and Pearl River Delta should be reduced by 25%, 20% and 
15%, respectively. As the research on haze pollution has started late 
in China, compared with carbon emissions reduction, the haze 
pollution control policy is still not mature, and the understanding 
of the socio-economic factors of haze pollution is still insufficient. 
At present, most regions in China have still not achieved effective 
PM 2.5 control, meanwhile, the regional haze pollution dominated 
by industrial source emissions is still serious. The quantitative 
study on the impact of carbon emissions reduction activities on 
PM 2.5 emissions reduction is helpful in formulating practical 
emissions mitigation measures for policy makers. 

Synergistic emissions reduction includes two aspects: one is the 
co-benefits of carbon emissions reduction caused by pollutant 
emissions reduction, and the other is the co-benefits of pollutant 


emissions reduction caused by carbon emissions reduction, which 
is the focus of this study. As China’s carbon emissions regulation 
tightens and the carbon emissions trading mechanism becomes 
more mature, enterprises are faced with tremendous pressure from 
the increase in the cost of carbon emissions reduction. The pressure 
can be partly relieved through the incentive mechanism under 
which carbon emissions reduction activities bring about pollutant 
emissions reduction. The quantitative analysis of the impacts of 
carbon emissions reduction activities on PM 2.5 emissions reduction 
has important reference value for policy makers, which can 
encourage enterprises to achieve carbon emissions reduction tar¬ 
gets, help reduce PM 2.5 emissions, and ultimately achieve a win- 
win situation of economic growth and environmental improve¬ 
ment. Based on the actual conditions of China, it is of great signif¬ 
icance to explore the synergistic effect of C0 2 reduction activities 
on PM 2.5 emissions reduction. 

This paper is intended to resolve the following three questions. 
What factors have led to changes in China’s PM 2.5 emissions? How 
much reduction in PM 2.5 emissions will result from the reduction in 
per unit C0 2 ? Are there differences in synergistic emissions 
reduction in different areas? This paper focuses on the co-benefits 
of PM 2.5 emissions reduction caused by carbon emissions reduc¬ 
tion. Firstly, the synergistic effect of carbon emissions is introduced 
into the Kaya identity of PM 2.5 emissions, and the influencing fac¬ 
tors of PM 2.5 emissions changes are studied by LMDI decomposition 
method. Then, the econometric analysis is applied to quantify the 
impacts of C0 2 emissions reduction activities on PM 2.5 emissions 
reduction, specifically, the estimation methods used in this paper 
include Fixed Effect estimation (FE), Feasible Generalized Least 
Squares (FGLS), comprehensive FGLS, and Generalized Method of 
Moments (GMM) estimation for the robustness test. 

The remainder of this paper is arranged as follows. Section 2 
presents a brief review of the related studies. Section 3 provides 
the methodology and data. Section 4 presents and discusses the 
results of LMDI decomposition. Section 5 shows the results and 
corresponding discussion of econometric model. The final section 
concludes this research and provides some policy implications. 

2. Literature review 

Scholars have conducted a lot of studies on the factors affecting 
PM 2 . 5 , and found that PM 2.5 is mainly affected by meteorological 
conditions and human activities. Although the frequent occurrence 
of haze pollution is affected by climatic factors to some extent, it is 
ultimately caused by extensive economic development, industrial 
structure imbalance, low energy efficiency and inefficient envi¬ 
ronmental governance (Shao et al., 2016). Therefore, the socio¬ 
economic factors affecting haze pollution have attracted more 
attention. The existing studies regarding the influencing factors of 
haze pollution mainly adopt econometric analysis methods (Ji et al., 
2018; Lin et al., 2013). Based on the STIRPAT model, Xu and Lin 
(2016) analyze the effects of economic development, urbaniza¬ 
tion, private car ownership, coal consumption and energy efficiency 
on China’s PM 2.5 emissions in 2001—2012, and find there are 
distinct differences in the impacts of various factors among 
different regions. At present, the research on haze pollution mainly 
adopts the concentration index due to the lack of aggregate PM 2.5 
emissions data. Therefore, the decomposition analysis method is 
rarely found in PM 2.5 related studies. For instance, Guan et al. 
(2014) employ the Structural Decomposition Analysis (SDA) to 
study the socio-economic drivers of PM 2.5 emissions changes in 
China from 1997 to 2010, and find that efficiency gains can offset 
emissions growth caused by economic growth and other factors. In 
addition, capital formation is the most important factor for PM 2.5 
emissions in the final demand side, but the resulting emissions is 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


761 


declining, and export is the only positive driver in the final demand 
side. In the studies with respect to air pollution control, the 
Computable General Equilibrium (CGE) model has been widely 
utilized, the commonly used policy tools include resource tax 
(Sancho, 2010), sulfur tax (Xu and Masui, 2009), and carbon tax 
(Allan et al„ 2014). Wei and Ma (2015) combine the haze control 
policies (including sulfur tax and carbon tax) with energy mix 
adjustment and technological progress, and adopt CGE model to 
conduct scenario analysis, the results show that energy mix opti¬ 
mization and technological progress are the fundamental means to 
alleviate haze pollution. However, the CGE model focuses on 
simulating the implementation effect of the expected policy com¬ 
bination, and it is difficult to simultaneously examine the mecha¬ 
nism of the effects of multiple factors on haze pollution. In addition, 
the CGE model has difficulty in capturing the dynamic changes of 
policy effects over time. 

Synergistic effect was first proposed by the Intergovernmental 
Panel on Climate Change (IPCC) in 2001, and the IPCC defined the 
synergistic effect as other socio-economic benefits (in addition to 
climate improvements) brought about by policy actions to achieve 
greenhouse gas emissions mitigation. Subsequently, the Organiza¬ 
tion for Economic Cooperation and Development (OECD), the US 
Environmental Protection Agency, and the European Environment 
Agency define the synergistic effect from different perspectives. In 
fact, the source of synergistic effect is not limited to energy con¬ 
servation policies, greenhouse gas emissions reduction policies, or 
air pollution control policies. Synergistic effect has attracted more 
and more attention from scholars in the field of environmental 
research. Xue et al. (2015) utilize the life cycle analysis method to 
quantitatively evaluate the co-benefits of wind power generation. 
They find that, compared with coal-fired power plants, wind power 
plants emit lower carbon dioxide by 97.48%, and discharge lower 
atmospheric pollutants SO 2 , NOx and PM 10 by 80.38%, 57.31% and 
30.91%, respectively. The similar findings are found by the study of 
Ma et al. (2013) on wind power in Xinjiang. Hasanbeigi et al. (2013) 
analyze the synergistic effect caused by the energy-saving policies 
of the cement sector in Shandong province, including the re¬ 
ductions in PM 10 and SO 2 emissions, and find the resulting health 
benefits reduce the costs of energy saving. 

The synergistic effect of greenhouse gas emissions reduction on 
air pollution mitigation has been confirmed by a large number of 
studies. Wagner and Amann (2009) employ the GAINS model to 
evaluate the implementation effect of greenhouse gas emissions 
reduction measures in Annex I of the Kyoto Protocol, and conclude 
that the CO 2 emissions reduction target can be achieved with an 
additional 5% reduction in SO 2 , NO x , and PM emissions. Vennemo 
et al. (2009) investigate the benefits and costs of three different 
CO 2 mitigation strategies (including intensity control, total emis¬ 
sions control, and sectoral intensity control), and conclude that 
intensity control has the greatest environmental co-benefits for 
China, while it has a negative impact on rural residents. Shrestha 
and Pradhan (2010) adopt the bottom-up minimum cost optimi¬ 
zation energy system model to study the synergistic effect of 
Thailand’s carbon emissions reduction policies. It is concluded that 
under the 30% carbon reduction target, SO 2 will be reduced by 43%, 
what’s more, the constraints of carbon emissions reduction targets 
are conducive to promoting the optimization and upgrading of 
energy mix. Many studies suggest that greenhouse gas emissions 
reduction strategies will lead to improvement in air pollution and 
resulting public health benefits (Groosman et al., 2011; Haines 
et al., 2009; Nemet et al., 2010), specially, this kind of synergistic 
effect can work out to the largest extent in developing countries 
(Nemet et al., 2010). He et al. (2010) simulate the synergistic effects 
under different combined scenarios of energy policies (greenhouse 
gases or air pollutants oriented), including greenhouse gas 


emissions reduction, air pollutant mitigation and health benefits. 
The existing literature about the collaborative management of air 
pollutants and greenhouse gas emissions mostly employs complex 
model to conduct simulation analysis considering a single policy or 
a combination of multiple emissions reduction measures. However, 
there are too many assumptions and the quantitative results can 
only be regarded as predictions or theoretical values, without the 
retrospective analysis of historical data. Based on historical data of 
CO 2 and PM 2.5 emissions, this paper utilizes the combination of 
index decomposition analysis and econometric analysis methods to 
study the changing mechanism of PM 2.5 emissions, and quantify 
the impacts of C0 2 reduction activities on PM 2.5 emissions 
reduction. 

Through the review of the existing literature, we find the 
following deficiencies. First, with regard to the research on the 
influencing factors of haze pollution, PM 2.5 or PM 10 concentration 
index is commonly used rather than aggregate emissions index, 
moreover, the decomposition analysis method is rarely adopted to 
study the changes in PM 2.5 emissions. Second, the existing litera¬ 
ture has not taken into consideration the synergistic effect of C0 2 
emissions reduction on PM 2.5 emissions reduction. China is faced 
with the dual pressure of C0 2 and PM 2.5 emissions reductions. It is 
necessary to quantify the impacts of CO 2 emissions reduction ac¬ 
tivities on PM 2.5 emissions reduction from a macro policy level, 
thereby providing a scientific basis for enterprises and policy 
makers to formulate emissions reduction strategies. Third, when it 
comes to haze pollution related studies, few research has combined 
the decomposition analysis with the econometric analysis. This 
paper takes advantages of the two methods to analyze the driving 
factors of PM 2.5 emissions and the ways to promote synergistic 
emissions reduction of PM 2 . 5 . The above-mentioned deficiencies 
motivate this research. Apart from previous studies, this study 
contributes to the existing literature in the following ways. (1) This 
paper solves the problem that provincial PM 2.5 emissions data are 
not available, and incorporates the synergistic effect of carbon 
emissions into the Kaya identity of PM 2.5 emissions. Accordingly, 
the LMDI method is utilized to decompose PM 2.5 emissions changes 
into the synergistic effect of carbon emissions, energy emission 
intensity effect, energy intensity effect, economic development 
effect and population effect. (2) Based on LMDI decomposition re¬ 
sults, the econometric analysis method is employed to quantita¬ 
tively analyze the impact of carbon emissions reduction activities 
on PM 2.5 emissions reduction, and examine whether the synergistic 
emissions reduction is affected by energy mix and technological 
progress. (3) This paper employs several rigorous econometric 
techniques to estimate the static panel data model, including FE, 
FGLS and comprehensive FGLS estimations. Furthermore, the GMM 
estimation is performed to conduct robustness test given possible 
endogeneity problems. All models prove the existence of syner¬ 
gistic PM 2.5 emissions reduction resulting from CO 2 emissions 
reduction. (4) In addition to the synergistic effect of C0 2 emissions 
reduction on PM 2.5 emissions reduction, this paper also takes into 
consideration such socio-economic variables as energy mix, tech¬ 
nological progress, per capita GDP and population density. A 
comprehensive empirical identification of the influencing factors of 
PM 2.5 emissions reduction reveals that there is a significant inver¬ 
ted U-shaped relationship between per capita GDP and PM 2.5 
emissions reduction. (5) In view of considerable heterogeneity 
among different regions, this paper divides China’s 30 provinces 
into three major economic regions: the western, central and 
eastern regions. Accordingly, the application of LMDI is to investi¬ 
gate PM 2.5 emissions changes at total economy, regions and prov¬ 
inces levels. In addition, we also compare the potential synergistic 
emissions reduction of PM 2.5 in 30 provinces. 


762 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759—771 


3. Methodology and data 

3.1. LMD1 decomposition 

The logarithmic mean divisia index (LMDI) method is first 
proposed by Ang et al. (1998). As an important branch of index 
decomposition analysis (IDA), this approach can address zero 
values and does not contain residuals in its decomposition results, 
which is widely utilized in energy-related research. This paper 
employs the LMDI method to analyze the influencing factors of 
PM 2.5 emissions changes in China. 

The nationwide PM 2.5 emissions can be expressed as: 


= J2 p M OCi • EM i • EI i • CPC, • Pj 


Pi 


( 1 ) 


where i denotes the ith region, PM, is PM 2.5 emissions, Q represents 
CO 2 emissions, E,, CDP, and Pj represent energy consumption, GDP 
and population size, respectively; PMOC is the PM 2.5 emissions per 
unit of CO 2 emissions, i.e., the quantitative measurement of syn¬ 
ergistic emissions reduction, which is the focus of this paper. In 
addition, EM, El, GPC denote energy emission intensity (i.e., energy 
mix), energy intensity and GDP per capita. 

Let dPM refer to the variation in PM 2.5 emissions from year 
0 (the base period) to year t, which can be decomposed by LMDI 
approach into the following effects: 


changes on PM 2.5 emissions. APM n is the energy intensity effect, 
which reflects the impact of energy efficiency changes on PM 2.5 
emissions. APM C pc is the economic development effect, and APM P 
is the population effect in the model. 


3.2. Econometric model 

In the previous section, we make the appropriate trans¬ 
formation on the basic model of Kaya identity (Kaya, 1990), and 
take into account the synergistic effect of carbon emissions on 
PM 2.5 emissions, thereby providing a theoretical basis for empirical 
analysis. Based on the previous factor decomposition process, this 
paper adds into the econometric model various variables repre¬ 
senting the synergistic effect of carbon emissions, energy emission 
intensity effect, energy intensity effect, economic development 
effect and population effect, respectively (i.e., CO 2 emissions 
reduction, energy mix, technological progress, per capita GDP, 
population density). In addition, in order to examine the existence 
of the Environmental Kuznets Curve (EKC) hypothesis, the 
quadratic term of GDP per capita is introduced into the model. 
Finally, a two-way fixed effects model is established in the 
following form: 


PMR it = ft, + ft CRu + P 2 EM it + 0 3 TP it + ftCPC it + /3 5 CPC2 it 
+ PePDit + Tt + + % 

( 8 ) 

where i indicates the ith province, t is time; PMR indicates PM 2.5 


dPM = PM 1 - PM° = J2 PM0C >t' EM it' E, it' CPC it ■ P it - 

i 

y^PMOCjQ • EM i0 • Elj 0 • GPC i0 • P,q = APMpmoc + riPiWgM + APMp/ + APMqpc + APMp 

i 


( 2 ) 


The five effects can be deduced as: 


, n „ v- PM]-PM ° , PMOC f 

APMpMQC \ ---pr In -TT 

PMOC Z lnPM t _ hpM 0 PMOC o 

(3) 

Ann/I V- PM*-PM? , EM\ 

A PM E M = } --- In 0 

V lnpM i - lnFM ? EM ° 

(4) 

PM\-PM° . Elj 

APMp, - > \ 'In ‘ 

V InPMj - lnPM° H? 

(5) 

, nn . v- PM]-PM? , CPC] 

A PMcpc = ^ n In n 

V InPM] - InPM? GPC° 

( 6 ) 

. nn . ^ PM] - PM? . P] 

APMp = > ! 1 n In ‘ 

V InPM] - InPM? Pf 

(7) 


Consequently, the change in PM 2.5 emissions from the base 
period to target period can be decomposed into the contributions of 
five driving factors. APM pm0 c is the synergistic effect of carbon 
emissions, which reflects the impact of energy-related carbon 
emissions activities on PM 2.5 emissions. APM em is the energy 
emission intensity effect, reflecting the impact of energy mix 


emissions reduction, CR, EM, TP, GPC, and PD serve as the proxy 
variables for the five decomposed effects through LMDI decom¬ 
position. CR indicates CO 2 emissions reduction, ft is the estimated 
coefficient of CR, which represents the quantitative measurement 
of synergistic emissions reduction and is the key observation in¬ 
dicator in this paper. GPC refers to GDP per capita, indicating the 
economic development level and residents' income levels, and 
GPC2 is the quadratic term of GPC. EM and PD represent energy mix 
and population density, respectively. As the proxy variable of the 
energy intensity effect, TP indicates technological progress given 
two main reasons. First, if energy intensity is included as an inde¬ 
pendent variable directly, Eq. (8) may have serious multi- 
collinearity. Second, there is evidence that technological progress 
is the main factor leading to the decline of China’s energy intensity 
(Garbaccio et al., 1999). As a result, technological progress is utilized 
to characterize the impact of energy efficiency on PM 2.5 emissions 
reduction. <5, is the non-observed effect in each province that does 
not change over time, which specifies persistent differences among 
provinces, such as consumption habits, natural resources endow¬ 
ments and environmental regulations. The time-fixed effect is 
considered in the model as well, and y t indicates the time non¬ 
observed effect. Considering the limitation of sample observation, 
the degree of freedom will be largely lost and the variance of the 
estimated parameter will increase if introducing (T-l) time dummy 
variables. Therefore, in order to save parameters in the model, this 
paper introduces the time trend item to control the effects of time- 
















F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


763 





2006 


varying factors such as energy prices, energy and environment 
related policies. e jt is a random error term irrelevant to time and 
region. 

3.3. Variable description and data source 

In this study, we focus on two main environmental issues in 
China, i.e„ carbon emissions and PM 2.5 emissions. Since there is no 
officially published provincial carbon emissions data, this paper 
calculates the CO 2 emissions of 30 provinces in China from 1998 to 
2014 using the data of such three energy sources as coal, petroleum 
and natural gas by multiplying their carbon emission coefficients, 
as shown in Eq. (9). Carbon emission coefficients and conversion 
factors from physical units to coal equivalent are derived from 1PCC 
(2006), Xu et al. (2006), and Hu and Huang (2008). 

= (9) 

j 

In Eq. (9), C it is the CO 2 emissions of province i in year t, E it j 
indicates the jth energy consumption (expressed in standard coal 
equivalent), fij is the corresponding carbon emission coefficient. 
Total coal includes raw coal, cleaned coal, briquette, coke, etc. Total 
petroleum products contain crude oil, gasoline, kerosene, diesel, 
etc. 

The PM 2.5 emissions data are obtained from the Surface Process 
Analysis and Simulation Laboratory of Peking University (Huang 
et al., 2014). Specifically, the data for 30 provinces are extracted 
from the monthly average grids (1998—2014) through ArcGIS 
software, then, we get annual-average PM 2.5 emissions for each 
province. The data are calculated according to the method pro¬ 
posed by Huang et al. (2014), and PM 2.5 emissions mainly originate 
from combustion and industrial process sources. The trend of PM 2.5 
emissions (in 10,000 1 ) in 1998, 2006 and 2014 are shown in 
Figs. 1—3. On the whole, haze pollution was concentrated in the 
economically developed and densely populated North China re¬ 
gion. In addition, the PM 2.5 emissions in the eastern region were 
relatively larger than those in the central and western regions. 

In the LMDI decomposition, the sample interval covers 
1998—2014. Energy consumption data come from the China Energy 
Statistical Yearbook (NBSC, 2015a), and the population and GDP 

N 

1998 /\ 


2,000 Kilometers 


Fig. 1. Spatial distribution of PM 2 . 5 emissions in 1998 (10,000 1 ). 


2000 Kilometers 


Fig. 2. Spatial distribution of PM2.5 emissions in 2006 (10,000 1 ). 


2014 


H 44.48 - 58.27 o 500 1.000 2 000 Kilometers 

I 58.28 - 91. 70 1 - 1 - 1 - 1 - 1 - 1 - 1 - 1 - 1 --- 

Fig. 3. Spatial distribution of PM 2.5 emissions in 2014 (10,0001). 

data come from the China Statistical Yearbook (NBSC, 2015b). In 
order to eliminate the impact of price fluctuation, GDP and per 
capita GDP are converted into 2000 constant prices through the 
deflators. 

In the econometric analysis section, this paper utilizes the panel 
data of 30 provinces from 1999 to 2014, and the summary of vari¬ 
ables is shown in Table 1. CO 2 emissions reduction (CR), energy mix 
(EM), technological progress (TP), GDP per capita (GPC) and pop¬ 
ulation density (PD) are adopted to represent the synergistic effect 
of carbon emissions, energy emission intensity effect, energy in¬ 
tensity effect, economic development effect and population effect, 
respectively. In particular, since the original data are used for 
empirical analysis, variable unit adjustments (see Table 1) are 
needed to avoid the occurrence of outliers in the estimated co¬ 
efficients which may result in the difficult explanation for results. 
The data of energy mix come from China Energy Statistical 








764 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


Table 1 

Summary of variables. 


Variables Definition 

Unit of measurement 

Data source 

PMR 

PM 2.5 emissions reduction 

Thousand tons 


CR 

C0 2 emissions reduction 

Ten thousand tons 

China Energy Statistical Yearbook (NBSC, 2015a), IPCC (2006), Xu et al. 

(2006), Hu and Huang (2008) 

EM 

Share of coal consumption in total energy Percent 
consumption 

China Energy Statistical Yearbook (NBSC, 2015a) 

TP 

Number of patent applications granted 

Ten thousand pieces 

China Statistical Yearbook (NBSC, 2015b) 

GPC 

GDP per capita 

Thousand yuan per capita (at 2000 
constant prices) 

China Statistical Yearbook (NBSC, 2015b) 

PD 

The ratio of total population to total area 

Person per square kilometers 

China City Statistical Yearbook (NBSC, 2015c) 


Yearbook (NBSC, 2015a), the data of technological progress and 
GDP per capita are derived from China Statistical Yearbook (NBSC, 
2015b), and data on population density are from China City Sta¬ 
tistical Yearbook (NBSC, 2015c). As for the explained variable, i.e. 
PM 2.5 emissions reduction PMR it , based on the study of Fu and Yuan 
(2017) on SO 2 emissions reduction in China’s power industry, the 
PM 2.5 emissions reduction of province i in year t is defined as: 

PMR it = PM i t - PM it ( 10 ) 

The core explanatory variable, i.e. CO 2 emissions reduction CR it , 
is calculated as follows: 

CR lt = C u , - Q t ( 11 ) 

In Eq. (11), CR it is the CO 2 emissions reduction of province i in 
year t. 

4. Results and discussion of LMDI decomposition 

4.1. Decomposition results of the whole economy 

The additive decomposition is utilized to study the influencing 
factors of PM 2.5 emissions in China from 1998 to 2014, the 
decomposition results are shown in Fig. 4. Total effect denotes 
China’s PM 2.5 emissions increment compared with last year, and 
other values in Fig. 4 are the contributions of five factors to the 
changes in PM 2.5 emissions. For the total economy, PM 2.5 emissions 
decline during the period from 1998 to 2002, rise during 
2002—2007, and go down and up over the period 2007—2014. It can 
be seen from Fig. 4 that the increase of PM 2.5 emissions is mainly 
caused by economic development. The average annual contribution 
of the economic development effect during 1998—2014 reaches 
1,036,7001. By contrast, the reduction in PM 2.5 emissions is mainly 
affected by the synergistic effect of carbon emissions, which has 
huge potential for synergistic emissions reduction of PM 2 . 5 . The 
average annual contribution of the synergistic effect of carbon 
emissions during 1998—2014 reaches -758,7001, which indicates 
that reducing the PM 2.5 emissions per unit of CO 2 emissions is an 
effective way to reduce PM 2.5 emissions, and the potential for the 
collaborative reductions of CO 2 and PM 2.5 emissions should be 
vigorously explored. It is found that the effect of energy intensity is 
positive during 2001—2002 and 2003—2005, but negative in other 
periods. On the whole, the changes in energy intensity lead to the 
decrease in PM 2.5 emissions, with average annual contribution 
of -317,8001, which indicates the energy efficiency improvements 
can significantly reduce PM 2.5 emissions. The changes in PM 2.5 
emissions caused by the energy emission intensity effect during 
1998—2014 is either positive or negative, with annual average 
contribution of -26,3001, indicating that the energy emission in¬ 
tensity has a weak negative impact on PM 2.5 emissions. Due to the 
characteristics of being “rich in coal, lack of oil and less gas”, China’s 
resource endowment has limited space for energy mix adjustment, 


but its emissions reduction potential cannot be ignored. The PM2.5 
changes caused by the population effect are mostly positive except 
for the period of 2004—2005, with annual average contribution of 
68,3001, indicating that the population changes will promote the 
increase of PM2.5 emissions. To be specific, the expansion of the 
population will inevitably lead to a series of production activities 
and increase the energy demand, thereby increasing pressure on 
the environment and resulting in an increase in PM2.5 emissions. 

On the whole, the economic development effect, energy in¬ 
tensity effect and synergistic effect of carbon emissions signifi¬ 
cantly influence PM2.5 emissions, while the effects of population 
and energy emission intensity are relatively weak. During the 
period from 1998 to 2014, the impact of energy emission intensity 
is either positive or negative, economic development and popula¬ 
tion contribute to increasing PM2.5 emissions, while the energy 
intensity effect and synergistic effect of carbon emissions are 
conducive to reducing PM2.5 emissions. Specially, enhancing the 
synergistic effect of carbon emissions (i.e., promoting the collabo¬ 
rative reductions of CO2 and PM2.5 emissions) is the most effective 
way to reduce PM2.5 emissions. 

4.2. Decomposition results of three economic regions 

As shown in Figs. 1—3, China displays considerable heteroge¬ 
neity in PM2.5 emissions among different regions, which deserves 
our research attention. In this section, based on the traditional di¬ 
vision of Chinese economic regions as well as industrial develop¬ 
ment and geographical proximity, we divide China’s 30 provinces 
into three major economic regions' (Xie et al., 2018), i.e., Eastern 
China, Central China and Western China, and explore the regional 
differences in PM2.5 changes during 1998—2014 in each region. The 
decomposition results are shown in Fig. 5. The total effect indicates 
the variation in PM2.5 emissions from 1998 to 2014, and it can be 
seen that PM2.5 emissions present faint changes in three regions. As 
shown in Fig. 5, the economic development effect is still the most 
important factor for the increase of PM2.5 emissions in three re¬ 
gions. The synergistic effect of carbon emissions and the energy 
intensity effect are still the main reasons for the decline in PM2.5 
emissions. The changes in population promote the increase of PM2.5 
emissions, but the effect in the eastern, central and western regions 
is distinctly different. The contribution of the population effect in 
Eastern China reaches 777,200 1 , much higher than the 109,2001 in 
Central China and 119,5001 in Western China. The eastern region is 
the most economically developed in China, where the population 
size has exploded with the rapid growth of the economy. Thus, the 


1 Eastern region includes Beijing, Shanghai, Tianjin, Fujian, Hebei, Liaoning, 
Jiangsu, Shandong, Hainan, Zhejiang, Guangdong. Central region comprises Henan, 
Shanxi, Hubei, Jilin, Heilongjiang, Anhui, Hunan, Jiangxi. Western region contains 
Inner Mongolia, Sichuan, Chongqing, Xinjiang, Shaanxi, Gansu, Guizhou, Yunnan, 
Guangxi, Qinghai, Ningxia. 






F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


765 



Fig. 4. Decomposition results of PM 2.5 emissions changes in China. 


■ APM(PMOC) 1 APM(EM) lAPM(EI) iAPM(GPC) BAPM(P) ■ Overall effecct 

2000 1 

1500 ■ 

1000 ■ 



-1000 ■ 

-1500 ■ 


Fig. 5. Decomposition results of PM2.5 emissions changes in three regions. 


population effect in the eastern region is significantly greater than 
that in the central and western regions. In Western China, the en¬ 
ergy emission intensity effect results in an increase in PM2.5 
emissions. In Central and Eastern China, it has a negative impact on 
PM2.5 emissions, indicating that energy mix in the central and 
eastern regions shows a low-carbon clean trend and the PM2.5 
emissions can be reduced to some extent through the optimization 
of energy mix, while energy mix in western region has not been 
effectively improved. At the nationwide level, the changes in en¬ 
ergy emission intensity lead to a slight reduction in PM2.5 emis¬ 
sions, and the impacts of other factors on PM2.5 emissions are 
consistent with the regional decomposition results. 

On the whole, the contribution value of each factor to PM2.5 
emissions differs significantly among three economic regions. The 


synergistic effect of carbon emissions and energy intensity effect 
are factors promoting the reduction of PM2.5 emissions. The eco¬ 
nomic development effect and population effect are the reasons for 
the increase in PM2.5 emissions. Specially, the energy emission in¬ 
tensity in Western China leads to an increase in PM2.5 emissions, 
indicating that the energy mix in Western China is urgent to be 
improved. 

4.3. Decomposition results of 30 provinces 

Fig. 6 shows the decomposition results of PM 2.5 emission 
changes in China’s 30 provinces. The total effect denotes the vari¬ 
ation in PM 2.5 emissions from 1998 to 2014 in each province. It can 
be seen that the changes in PM 2.5 emissions in various provinces 

























766 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759—771 


IfiPM(PMOC) ■APM(EM) lAPM(EI) ■ AEMIGPE) BAPMIP) ■ Overall effecct 

150 ■ 


100 



Fig. 6. Decomposition results of PM 2 .s emissions changes in 30 provinces. 


are quite different. The PM 2.5 emissions in 16 provinces, including 
Hebei, Inner Mongolia, Shandong, etc, increase during 1998—2014, 
among which Shandong province presents the largest increment of 
200,6001. On the contrary, PM 2.5 emissions in other 14 provinces 
decrease, specially, Yunnan province shows the largest decline of 
85,5001. 

Among the five effects, the synergistic effect of carbon emissions 
contributes the most to decreasing PM 2.5 emissions in all provinces, 
indicating that the implementation of carbon emissions reduction 
policies has effectively promoted the synergistic emissions reduc¬ 
tion of PM 2 . 5 . However, the co-benefits of PM 2.5 emissions reduc¬ 
tion are largely different among 30 provinces. The five provinces 
with the largest absolute values of contributions are Henan, 
Shandong, Inner Mongolia, Shaanxi and Hebei. The provinces with 
the largest and smallest absolute values are Henan (888,9001) and 
Hainan (63,4001), respectively, indicating that there are huge dif¬ 
ferences in the potential for synergistic emissions reduction of 
PM 2.5 among provinces, which is due to considerable regional 
disparities of socioeconomic development in China. 

Fig. 6 shows energy intensity changes contribute to reducing 
PM 2.5 emissions in most province except Fujian province with 
contribution value of 15,7001. This is because the increase in en¬ 
ergy intensity in Fujian leads to a slight increase in PM 2.5 emissions. 
However, in most provinces, the energy intensity effect leads to the 
reduction in PM 2.5 emissions, indicating that these provinces have 
achieved remarkable results in reducing energy intensity and 
improving energy efficiency. The contribution of the energy in¬ 
tensity effect is significantly different in different provinces. The 
provinces with the largest and smallest absolute values are Shanxi 
(-410,9001) and Hainan (-1,7001), respectively. As a traditional 
coal-producing province, Shanxi’s economic development has been 
relying on coal consumption for a long time, resulting in serious 
resource and environmental problems. The energy intensity in 
Shanxi drops from 4.11 1 of coal equivalent per 10,000 yuan in 2007 
to 2.36 1 of coal equivalent per 10,000 yuan in 2014. Shanxi has 
significantly reduced the energy consumption per unit of GDP by 
improving the energy efficiency and developing low energy¬ 
consuming industries, thereby effectively promoting the reduc¬ 
tion of PM 2.5 emissions. 

The effect of economic development is positive in all provinces. 
It should be noted that the economic development effect is the 
main driving force for the increase of PM 2.5 emissions, which shows 
that China’s provinces develop their economies at the expense of 
the environmental deterioration and public health loss, and they 
are still in the stage of extensive economic development. 


Furthermore, there are huge differences in the contributions of the 
economic development effect among provinces. The provinces with 
the largest and smallest contributions are Shandong (1,229,3001) 
and Hainan (48,1001), respectively. China is in the critical period of 
economic transformation, promoting the collaborative manage¬ 
ment of CO 2 and PM 2.5 emissions is an important way to curb 
environmental issues. 

Among the five effects, the energy emission intensity effect has 
relatively little impact on PM 2.5 emissions. The effect of energy 
emission intensity in such 20 provinces as Hebei, Shanxi, Sichuan, 
etc, is negative, and the accumulated effect is up to -1,013,5001, 
indicating that the transformation into the low-carbon clean en¬ 
ergy mix in these provinces has effectively promoted the reduction 
of PM 2.5 emissions. The effect of energy emission intensity in other 
10 provinces is positive because these provinces have long relied on 
traditional fossil energy consumption and have not effectively 
improved their energy mix. 

In most provinces, the effect of population is positive. However, 
only five provinces such as Anhui, Hubei, Chongqing, Sichuan and 
Guizhou present weak negative effects. The main reason is that 
these provinces are known for their large labor outputs in consid¬ 
eration of population migration, therefore, the changes in popula¬ 
tion contribute to decreasing PM 2.5 emissions. 

On the whole, for all provinces, economic development is esti¬ 
mated to be the most important factor contributing to the increase 
of PM 2.5 emissions, while the synergistic effect of carbon emissions 
is the largest contributor to the drop of PM 2.5 emissions. 
Strengthening the synergistic effect of carbon emissions is the most 
effective way to reduce PM 2.5 emissions. In addition, it is important 
to reduce energy intensity and improve energy efficiency, thereby 
reducing PM 2.5 emissions. The population effect causes slight in¬ 
crease in PM 2.5 emissions in most provinces, which indicates its 
impact on PM 2.5 emissions is limited. In most provinces, the 
contribution of energy emission intensity is negative. Although 
energy emission intensity effect contributes little to PM 2.5 emis¬ 
sions changes in most provinces, its potential influence on PM 2.5 
emissions reduction is great for the whole country. For most 
provinces, there is still large space for the optimization of energy 
consumption structure, and it is of great significance to reduce coal 
consumption and increase utilization of quality and cleaner energy, 
thereby effectively reducing PM 2.5 emissions. 

5. Results and discussion of econometric model 

5.J. Panel unit root test and co-integration test 

The application of LMDI is to investigate the contributions of the 
synergistic effect of carbon emissions, energy emission intensity 
effect, energy intensity effect, economic development effect and 
population effect to PM2.5 emissions changes during 1998—2014 at 
total economy, regions and provinces levels. In addition to the 
changing mechanism of PM2.5 emissions, an understanding of the 
channels through which each factor affects PM2.5 emissions is 
useful when making specific emissions reduction measures. Based 
on the results obtained by LMDI decomposition, this paper adopts 
econometric analysis methods to study the impacts of proxy vari¬ 
ables (i.e., CO2 emissions reduction, energy mix, technological 
progress, per capita GDP and population density) on PM2.5 emis¬ 
sions reduction, using the historical data of CO2 and PM 2 .s emis¬ 
sions during 1999—2014. More importantly, this paper focuses on 
quantifying the synergistic effect of CO 2 emissions reduction ac¬ 
tivities on PM2.5 emissions reduction. Generally speaking, most 
economic variables are non-stationary sequences. In order to avoid 
spurious regression, it is necessary to test the stability of variable 
sequences before the regression analysis, that is, whether there is a 
























F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


767 


Table 2 

Results of panel unit root tests. 



Series 

Fisher ADF 


Fisher PP 


LLC 


constant 

Trend and intercept 

constant 

Trend and intercept 

constant 

Trend and intercept 

Levels 

PMR 

-2.7503 

1.0468 

38.1443*** 

35.4452*** 

-8.3050*** 

-8.8124*** 


CR 

4.7708*** 

3.1674*** 

18.8407*** 

15.3854*** 

-7.1415*** 

-10.1597*** 


EM 

1.7657** 

4.9579*** 

2.7139*** 

9.4018*** 

-2.3376*** 

-5.0459*** 


TP 

-2.2016 

5.2633*** 

-4.8685 

-3.9649 

4.9814 

-3.8802*** 


GPC 

1.3177* 

11.1270*** 

-3.8037 

1.6822** 

1.1666 

-4.6839*** 


PD 

-0.2808 

11.9281*** 

23.2684*** 

34.7018*** 

-7.2329*** 

-7.5807*** 

First difference 

PMR 

2.3603*** 

2.2548** 

122.9618*** 

135.2901*** 

-22.1514*** 

-30.4643*** 


CR 

1.8191** 

10.4396*** 

83.5477*** 

91.8090*** 

-20.7025*** 

-23.1599*** 


EM 

3.6326*** 

16.5003*** 

70.7638*** 

69.7636*** 

-12.2646*** 

-24.0260*** 


TP 

8.9224*** 

36.4778*** 

6.3209*** 

5.2060*** 

-2.8647*** 

-2.4822*** 


GPC 

5.1944*** 

9.5448*** 

25.9438*** 

28.0357*** 

-8.5161*** 

-13.7832*** 


PD 

6.8390*** 

12.1078*** 

111.8903*** 

103.4200*** 

-11.3041*** 

-25.6590*** 


Note: ***p < 0.01, **p < 0.05, *p < 0.1; the number of lags of the series is chosen according to the Akaike information criterion (AIC). 


unit root. The unit root test for the panel data needs to consider the 
heterogeneity of cross-section sequences, and the panel unit root 
test consists of two major categories. The one assumes that each 
cross-section sequence has the same unit root, including LLC test, 
Breitung test and Hadri test. The other assumes that different cross- 
section sequences have different unit roots, such as IPS test, Fisher- 
ADF test and Fisher-PP test. In this paper, the commonly used LLC, 
Fisher-ADF and Fisher-PP tests are utilized to perform panel unit 
root tests. The possible cross-section correlation is alleviated, and 
the results of panel unit root tests are presented in Table 2. It is 
shown that some variables are non-stationary, but their first dif¬ 
ference series significantly reject the null hypothesis of containing 
the unit root. On the whole, all the test results are combined to 
determine that every variable is a first-difference stationary 
sequence. Furthermore, it is necessary to examine whether there 
exists any co-integration relationship between PMR and the inde¬ 
pendent variables. The Kao co-integration test yields an ADF sta¬ 
tistic of-11.7147, which rejects the null hypothesis that there is no 


significant co-integration relationship between the explanatory 
variables and PMR at confidence level of 1%. In addition, consid¬ 
ering that there may exist multi-collinearity between explanatory 
variables, this paper utilizes VIF statistic to test whether there is 
multi-collinearity in the model. The test results show that the VIF 
values of all variables are less than 2, indicating there is no multi- 
collinearity problem in the data set. 

5.2. Results and discussion 

This paper establishes a two-way fixed effects model consid¬ 
ering time fixed effect and individual fixed effect, and adopts 
several methods to estimate the model, including Fixed Effect 
estimation (FE), FGLS (only considering the intra-group autocor¬ 
relation) and CFGLS (considering both intra-group autocorrelation, 
inter-group heteroscedasticity and cross-sectional correlation), 
among which the CFGLS estimation results are more effective. 
Table 3 presents the regression results obtained by various 


Table 3 

Estimates results through different estimation methods. 


Variables 

Model (1) 

FE 

Model (2) 

FE 

Model (3) 

FGLS 

Model (4) 

CFGLS 

Model (5) 

CFGLS 

Model (6) 

CFGLS 

Model (7) 

CFGLS 

CR 

0.0037*** 

0.0037*** 

0.0037*** 

0.0033*** 

0.0033*** 

0.0005 

0.0036*** 


(0.0008) 

(0.0008) 

(0.0013) 

(0.0003) 

(0.0003) 

(0.0010) 

(0.0004) 

EM 

-0.1041 

-0.1806 

-0.1834 

-0.1776** 

-0.1363*** 

-0.2631*** 

-0.1703** 


(0.1325) 

(0.1517) 

(0.1870) 

(0.0776) 

(0.0504) 

(0.0748) 

(0.0767) 

TP 

0.4956* 

0.5825** 

0.5855 

0.7554** 

0.2594 

1.1234** 

-0.3772 


(0.3050) 

(0.2705) 

(0.6121) 

(0.3165) 

(0.4591) 

(0.5317) 

(0.3827) 

GPC 

0.5019* 

1.8877** 

1.8891* 

1.9101*** 

1.1802** 

1.7302** 

2.8683*** 


(0.3031) 

(0.8126) 

(0.9790) 

(0.5837) 

(0.5937) 

(0.7338) 

(0.8009) 

GPC2 


-0.0175* 

-0.0175* 

-0.0142* 

-0.0128 

-0.0202* 

-0.0275*** 



(0.0086) 

(0.0095) 

(0.0086) 

(0.0094) 

(0.0108) 

(0.0106) 

Eastern*CR 





-0.0014*** 








(0.0005) 



Central*CR 





0.0018*** 








(0.0005) 



EM*CR 






0.0000** 








(0.0000) 


TP*CR 







-0.0002*** 








(0.0001) 

PD 

0.0308 

0.0464 

0.0459* 

0.0429** 

0.0281 

0.0932*** 

0.0345 


(0.0257) 

(0.0358) 

(0.0235) 

(0.0218) 

(0.0489) 

(0.0294) 

(0.0414) 

T 

-1.2702** 

-2.3399*** 

-2.3431** 

-2.2184*** 

-1.6082*** 

-1.9235*** 

-3.0931*** 


(0.5865) 

(0.8423) 

(1.0271) 

(0.4940) 

(0.5975) 

(0.5980) 

(0.7164) 

Constant 

1.2946 

-6.9258 

-50.4177* 

8.0823 

14.1354 

10.3895 

5.9582 


(13.6704) 

(16.3388) 

(27.8036) 

(24.1401) 

(23.9543) 

(24.3771) 

(23.7614) 

Observations 

480 

480 

480 

480 

480 

480 

480 

Provincial fixed effects 

/ 

/ 

/ 

y 

y 

y 

y 

Time fixed effects 

/ 

/ 

/ 

y 

y 

y 

y 


Note: Robust standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1, # p < 0.15. 


















768 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


estimation methods, and all models demonstrate the existence of 
synergistic PM 2.5 emissions reduction caused by carbon emissions 
reduction. As shown in Table 3, as a baseline reference, Model (1) is 
the regression result based on FE estimation without the quadratic 
term of GPC. Models (2)—(4) report the estimation results through 
FE, FGLS and CFGLS, respectively. Models (5)—(7) give the estima¬ 
tion results through CFGLS with the interaction terms of CR. It can 
be seen that the estimated coefficients in Model (4) are more sig¬ 
nificant than Model (2) and Model (3). Since the CFGLS estimation 
is highly effective and can resolve the autocorrelation and hetero- 
scedasticity problems, the following discussion is mainly based on 
the results of Model (4). 

The results show that the estimated coefficient of the core 
explanatory variable CR is 0.0033 and significant at the 1% level, 
which indicates that the CO 2 emissions reduction activities will 
significantly affect PM 2.5 emissions reduction. This means for every 
10,000 1 increase in C0 2 emissions reduction, PM 2.5 emissions 
reduction will increase by 3.31, confirming the feasibility of 
collaborative reductions of CO 2 and PM 2.5 emissions. China is 
currently in the stage of industrialization and faced with the dual 
pressure of carbon emissions and haze pollution, the ecological 
environment predicament needs to be resolved. China has formu¬ 
lated a number of specific policies on carbon emissions reduction, 
and made a series of emissions reduction commitments to the in¬ 
ternational community. However, the mitigation policy for haze 
pollution is still insufficient. The quantitative analysis on the im¬ 
pacts of carbon emissions reduction activities on PM 2.5 emissions 
reduction can provide important information for policy makers. 

Among the control variables, the coefficient of energy mix is 
significantly negative at the 5% level, indicating that the proportion 
of coal consumption hinders the reduction in PM 2.5 emissions. Coal 
contains a large amount of nitrogen and sulfur, and its combustion 
will produce the main ingredients of PM 2 . 5 , i.e. sulfur dioxide and 
nitrogen oxides. The coefficient of technological progress is positive 
and significant at the 5% level. Technological progress will enhance 
the energy efficiency, in addition, the substitution of new power 
generation technology for coal-fired power generation makes great 
potential for reductions in sulfur, nitrogen and carbon, thereby 
leading to PM 2.5 emissions reduction. The coefficient of population 
density is significantly positive, which indicates population density 
contributes to promoting PM 2.5 emissions reduction. This is 
because population agglomeration can promote intensive energy 
utilization and improve the energy utilization efficiency, thereby 


reducing PM 2.5 emissions. As shown in Table 3, the coefficient of per 
capita GDP is significantly positive, and the coefficient of its 
quadratic term is significantly negative, indicating that there is a 
nonlinear inverted U-shaped relationship between per capita GDP 
and PM 2.5 emissions reduction. In the early stage of economic 
development, economic growth contributes to promoting PM 2.5 
emissions reduction. As the economy develops to a certain level, 
PM 2.5 emissions reduction potential becomes smaller and emis¬ 
sions reduction decreases accordingly. 

We estimate the potential synergistic emissions reduction of 
PM 2.5 using historical CO 2 emissions reduction by multiplying the 
coefficient of CR 0.0033 obtained in Model (4), and the annual 
average during the observation period (1999—2014) is taken. Since 
CO 2 emissions in all provinces increase during the observation 
period (i.e., the annual average historical CO 2 emissions reduction 
are negative), this paper defines the absolute value of synergistic 
emissions reduction of PM 2.5 |/3 1 -zlO?| as the potential synergistic 
emissions reduction of PM 2.5 in each province. Fig. 7 shows the 
potential synergistic emissions reduction of PM 2 . 5 . Since we have 
made some adjustments to variables units in econometric analysis 
(see Table 1), the unit of emissions reduction is kiloton in Fig. 7. It 
can be seen that there are huge differences in the potential syner¬ 
gistic emissions reduction of PM 2.5 in different provinces. In 
particular, in Inner Mongolia and Shandong, the increase in PM 2.5 
emissions indirectly caused by the increase in C0 2 emissions is 
more than 13,0001, while the least is reported in Hainan, Beijing 
and Qinghai. Comparing the potential synergistic PM 2.5 emissions 
reduction and historical PM 2.5 emissions reduction, Zhejiang has a 
potential synergistic emissions reduction 28 times higher than the 
historical emissions reduction. Guangdong has a potential syner¬ 
gistic emissions reduction 22 times higher than the historical 
emissions reduction. Shanxi has a potential synergistic emissions 
reduction 19 times higher than the historical emissions reduction. 
By contrast, the gaps in other provinces are relatively smaller. 

In order to study whether there is any difference in the syner¬ 
gistic effect of CO 2 emissions reduction on PM 2.5 emissions reduc¬ 
tion among the eastern, central and western regions, this paper 
adds the interaction terms between regional dummy variables and 
CR in Model (5). The results show that the synergistic effect in the 
central region is significantly larger than that in the western and 
eastern regions. According to the estimation results of Model (4), 
the economically developed eastern region has crossed the inflec¬ 
tion point of the inverted U-shaped curve between economic 



Fig. 7. Potential synergistic emissions reduction of PM2.5. 








Table 4 

Results of robustness tests. 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


769 


Variables 

Model (8) 

Model (9) 

Model (10) 


LSDV 

GMM 

GMM 

CR 

0.0037*** 

0.0071*** 

0.0072** 


(0.0008) 

(0.0025) 

(0.0028) 

EM 

-0.1806 


-0.0826 


(0.1566) 


(0.1997) 

TP 

0.5825** 


0.2297 


(0.2792) 


(0.6493) 

GPC 

1.8877** 


2.0830* 


(0.8388) 


(1.1616) 

GPC2 

-0.0175* 


-0.0217** 


(0.0088) 


(0.0103) 

PD 

0.0464 


0.0559 


(0.0369) 


(0.0427) 

T 

-2.3399** 

0.0526 

-2.0407* 


(0.8694) 

(0.3139) 

(1.2782) 

Observations 

480 

450 

450 

Provincial fixed effects 

/ 

/ 

/ 

Time fixed effects 

/ 

/ 

/ 

Underidentification test (Kleibergen-Paap rk LM statistic) 


28.416*** 

23.790*** 

Weak identification test (Cragg-Donald Wald F statistic/Kleibergen-Paap rk Wald F statistic) 


49.497/27.610 

36.862/23.185 

Endogeneity test 


2.070 

1.919 

Hansen J statistic 


Exactly identified 

Exactly identified 


Note: Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1, # p < 0.15. 


development and PM 2.5 emissions reduction. Compared with cen¬ 
tral and western regions, PM 2.5 emissions in eastern region are less 
sensitive to CO 2 emissions reduction. For some developed prov¬ 
inces in eastern region, pollutant emissions have become the rigid 
need to stabilize local economic development, which partly offsets 
the negative impacts of carbon emissions reduction activities to 
some extent. In Model ( 6 ) and Model (7), the interaction terms 
between energy mix, technological progress and CR are introduced 
to study the expansion mechanism of synergistic effect. As shown 
in Model (4) and Model ( 6 ), after controlling the interaction term 
between energy mix and CR, the coefficient of CR decreases and is 
no longer significant, the negative impact of energy mix increases, 
and the coefficient of interaction term approaches to zero. Although 
the energy mix has a direct negative impact on PM 2.5 emissions 
reduction, it is not ideal to promote the synergistic effect by 
improving energy mix. Therefore, in order to enhance synergistic 
emissions reduction of PM 2 . 5 , in addition to reducing the propor¬ 
tion of coal consumption, it is crucial to improve the efficiency of 
energy processing conversion and utilization in CO 2 emissions 
reduction activities, such as promoting the clean utilization of fossil 
energy especially coal and improving the cleanliness of the pro¬ 
duction process, thereby strengthening synergistic emissions 
reduction of PM 2 . 5 . According to Model (4) and Model (7), after 
controlling the interaction term between the technological prog¬ 
ress and CR, the coefficient of CR slightly increases, the coefficient of 
technological progress is no longer significant, and the coefficient 
of the interaction term is significantly negative, indicating that 
technological progress will weaken the synergistic effect. At pre¬ 
sent, the control of CO 2 emissions mainly depends on energy¬ 
saving measures. There is no feasible end-of-pipe control technol¬ 
ogy for CO 2 emissions reduction. For example, carbon capture and 
storage (CCS) may increase electricity consumption and lead to 
more pollutant emissions due to higher power consumption, 
therefore, there should be a trade-off between reducing green¬ 
house gas emissions and local air pollutants (Yang et al., 2013). 
Although technological progress (or energy efficiency improve¬ 
ment) contributes to mitigating CO 2 emissions to a certain extent, it 
should be noted that as the energy utilization efficiency increases, 
the marginal cost of energy services decreases, which may result in 
the rebound effect of energy resources (Gillingham et al., 2016; 


Greening et al., 2000). Thus, energy efficiency improvement may 
lead to more energy consumption and not be conducive to PM 2.5 
emissions reduction. 

5.3. Robustness test 

There are three sources of endogeneity problems: measurement 
bias, missing variables, and reverse causality, which may reduce the 
robustness of model and lead to biased estimation results. In order 
to address possible endogeneity problems, this paper assumes the 
core explanatory variable CR is an endogenous variable, adopts the 
lagged term of CR as instrument variable, and estimates the model 
by Generalized Method of Moments (GMM). Since the results of 
2SLS (Two stage least squares) and GMM are exactly the same, the 
estimation result of 2SLS is not reported in Table 4. The first stage 
estimation result of 2SLS shows that the lagged term of CR has a 
good explanatory power for CR (the estimated coefficient is 0.2854, 
P value is 0.000). As shown in Table 4, the endogenous test shows 
that CR is not related to the disturbance term and can be considered 
as an exogenous explanatory variable because the null hypothesis 
of CR as the exogenous variable cannot be rejected even at the 15% 
significant level. Since the number of endogenous variables is equal 
to that of instrument variables, the over-identification test fails in 
the case of exact identification, and the exogenous nature of the 
instrument variables can only be qualitatively analyzed. China has 
made many carbon emissions reduction commitments, including 
achieving its carbon emissions peak by 2030 and lowering carbon 
intensity by 40%—45% from the 2005 level by 2020 (Dong et al., 
2018c). Under the binding targets of carbon emissions reduction, 
enterprises will decide on future emissions reduction measures 
based on historical emissions reduction. Therefore, the previous 
carbon emissions reduction will affect current PM 2.5 emissions 
reduction by influencing the current carbon emissions reduction, 
while the current PM 2.5 emissions reduction has no impact on the 
previous carbon emissions reduction. Accordingly, the lagged term 
of CR is selected as the exogenous instrument variable in this paper. 
The weak instrument variable test (i.e„ weak identification test) 
reports two statistics. The Cragg-Donald statistic is obtained under 
the assumption of spheroidal disturbance term, and the 
Kleibergen-Paap statistic relaxes this assumption. For the Model 








770 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


(9), the Cragg-Donald statistic and the Kleibergen-Paap statistic are 
much larger than the critical value of 16.38 under the 10% bias, and 
the similar test result is reported in the Model (10). Thus, weak 
identification test can reject the null hypothesis of the lagged term 
of CR as weak instrument variable, and the instrument variable is 
considered to be strongly related with CR, indicating there is no 
weak instrument variable problem considering the lagged term of 
CR as instrument variable. 

Table 4 reports the GMM estimation results, and the least square 
dummy variable (LSDV) estimation results are provided as a base¬ 
line reference. In all models, the coefficients of CR are significantly 
positive, supporting the existence of the synergistic effect of C0 2 
emissions reduction on PM 2.5 emissions reduction. In addition, the 
coefficient signs of the control variables are consistent with the 
results in Table 3. On the whole, the previous conclusions have been 
further confirmed. Using the lagged term of CR as instrument var¬ 
iable, ordinary least squares (OLS) estimation is more effective than 
instrument variable estimation in the absence of endogeneity 
problems. As shown in Table 4, the standard errors in Model (9) and 
Model (10) through CMM estimation are higher than those ob¬ 
tained from LSDV, and the coefficient of the core explanatory var¬ 
iable CR is distinctly improved, but the significance is slightly 
weaker. 

6. Conclusions and policy implications 

Faced with the severe situations of air pollution control and 
greenhouse gas emissions reduction domestically and interna¬ 
tionally, the collaborative control of air pollutants and greenhouse 
gas emissions is an important policy choice for China’s environ¬ 
mental improvement. This research on synergistic PM2.5 emissions 
reduction caused by carbon emissions reduction can provide a 
quantitative basis for policy makers in the collaborative control of 
greenhouse gas emissions and air pollution. In this paper, the basic 
model of Kaya identity is appropriately extended, considering the 
synergistic effect of C0 2 emissions on PM2.5 emissions. The varia¬ 
tion in PM2.5 emissions is decomposed by the LMDI decomposition 
into the synergistic effect of carbon emissions, energy emission 
intensity effect, energy intensity effect, economic development 
effect and population effect. Then, this paper analyzes the impacts 
of five decomposed factors on PM2.5 emissions changes from na¬ 
tional, regional and provincial levels. Based on the LMDI decom¬ 
position results, various proxy variables comprising C0 2 emissions 
reduction, coal consumption rate, technological progress, GDP per 
capita and its quadratic term, and population density are intro¬ 
duced into the econometric model. By using several rigorous 
econometric techniques, we further quantify the co-benefits of 
PM2.5 emissions reduction generated by C0 2 emissions reduction 
activities. 

The LMDI decomposition method is employed to analyze the 
influencing factors of PM2.5 emissions changes from 1998 to 2014. 
The main conclusions are as follows. (1) From the nationwide 
decomposition results, the synergistic effect of carbon emissions 
and energy intensity effect are the main factors resulting in the 
reduction of PM2.5 emissions, and the economic development effect 
is the main reason for the increase of PM2.5 emissions, while the 
impacts of population and energy emission intensity are relatively 
weak. (2) From the perspective of three economic regions, the 
synergistic effect of carbon emissions and energy intensity effect 
are conducive to the reduction of PM2.5 emissions, by contrast, the 
economic development effect and population effect are the factors 
leading to the increase of PM2.5 emissions. In the western region, 
the energy emission intensity effect leads to the increase of PM2.5 
emissions. However, it has a negative impact on PM2.5 emissions in 
the central and eastern regions. From the national perspective, 


energy emission intensity changes have a small negative effect on 
PM2.5 emissions during 1998 — 2014 . ( 3 ) According to the provincial 
decomposition results, the contribution of each factor differs 
distinctly in different provinces. For all provinces, the economic 
development effect is the most important factor resulting in the 
increase of PM2.5 emissions. The synergistic effect of carbon emis¬ 
sions and energy intensity effect are the main factors for the decline 
in PM2.5 emissions. The population changes make little contribu¬ 
tion to PM2.5 emissions changes. For most provinces, there is large 
room for the optimization of energy consumption structure. 
Accordingly, decreasing coal consumption and promoting the use 
of renewable energy can effectively reduce PM2.5 emissions for the 
whole country. 

Based on the LMDI decomposition, the impacts of various proxy 
variables on PM 2.5 emissions reduction are investigated by econo¬ 
metric analysis methods, specially, we focus on the impact of C0 2 
emissions reduction on PM 2.5 emissions reduction. The empirical 
results are complementary to the results of LMDI decomposition. 
All models indicate that there is a significant synergistic effect of 
CO 2 emissions reduction on PM 2.5 emissions reduction, and the 
synergistic effect in the central region is significantly larger than 
that in the western and eastern regions, indicating that PM 2.5 
emissions are more sensitive to CO 2 emissions reduction in the 
central region. There are large differences in the potential syner¬ 
gistic emissions reduction of PM 2.5 in 30 provinces. Among them, 
Inner Mongolia, Shanxi, Jiangsu and Shandong have the largest 
potential synergistic emissions reduction. In addition, it is found 
that technological progress and population density have positive 
impacts on PM 2.5 emissions reduction, and there exists a significant 
inverted U-shaped relationship between economic development 
and PM 2.5 emissions reduction, while the increase in coal con¬ 
sumption is not conducive to PM 2.5 emissions reduction. 

To sum up, this paper proposes the following policy implica¬ 
tions. First, regardless of national, regional or provincial level, the 
synergistic effect of carbon emissions is the main reason for the 
reduction of PM2.5 emissions. Therefore, it is necessary to vigor¬ 
ously bring into play the role of the synergistic effect of carbon 
emissions, which is the most effective way to reduce PM2.5 emis¬ 
sions. More specifically, coordinating the carbon emissions reduc¬ 
tion measures with haze pollution control policies, carrying out the 
comprehensive and systematic management, thereby reducing the 
costs of environmental policies. Second, economic development is 
still the most important factor contributing to haze pollution. As 
China has started late in PM2.5 control, haze pollution has not been 
effectively mitigated with rapid economic development. Specially, 
there is an inverted U-shaped relationship between economic 
development and PM2.5 emissions reduction. Therefore, in the 
process of economic development, it is necessary to rationally 
adjust the intensity of environmental regulation, moderately rise 
the weight of environmental indicator in the index system of local 
performance evaluation, and coordinate the relationship between 
economic development and environmental protection. Third, given 
China’s resource endowment, energy mix has relatively little 
impact on PM2.5 emissions than other factors. The potential 
contribution of energy mix optimization to PM2.5 emissions 
reduction cannot be ignored in most provinces. More importantly, 
it is necessary to improve the efficiency of energy processing con¬ 
version and utilization in C0 2 emissions reduction activities, such 
as developing clean coal technology and promoting new energy 
power generation. Fourth, technological progress is conducive to 
the reduction of PM2.5 emissions, but technological progress may 
cause an increase in PM2.5 emissions through C0 2 emissions 
reduction activities, i.e„ synergistic emissions increase. Accord¬ 
ingly, it is required to discriminatively develop the haze pollution 
mitigation technology and low-carbon energy saving technology, 


F. Dong et al. / Journal of Cleaner Production 223 (2019) 759-771 


771 


strengthen the investment in the research and development of 
pollution control technology, and adopt more direct end-of-pipe 
treatment measures against haze pollution. Fifth, the develop¬ 
ment of compact cities and population agglomeration are condu¬ 
cive to promoting the intensive energy utilization, reducing the 
energy waste and improving the heating efficiency, thereby 
contributing to alleviating haze pollution. 

Acknowledgments 

This work was supported by the National Natural Science 
Foundation of China (Grant No. 71573254), Jiangsu Funds for Social 
Science (Grant No. 17JDB004), Jiangsu Education Science Project 
(Grant No. B-b/2015/01/027), and Key Project of Postgraduate Ed¬ 
ucation and Teaching Reform in Jiangsu Province (JGZZ18_047). The 
authors also would like to thank the anonymous reviewers for their 
valuable suggestions on the earlier draft of this paper. 

References 

Allan, G., Lecca, P., McGregor, P., Swales, K., 2014. The economic and environmental 
impact of a carbon tax for Scotland: a computable general equilibrium analysis. 
Ecol. Econ. 100, 40-50. 

Ang, B., Zhang, F., Choi, K., 1998. Factorizing changes in energy and environmental 
indicators through decomposition. Energy 23, 489—495. 

DEP (Department of Environmental Protection), 2012. Ambient Air Quality Stan¬ 
dards (GB3095-2012). Available at. http://img.jingbian.gov.cn/upload/ 
CMSjingbian/201806/201806210853050.pdf. (Accessed 10 February 2019). 
Dong, F., Bian, Z., Yu, B., Wang, Y., Zhang, S., Li, J., Su, B., Long, R., 2018a. Can land 
urbanization help to achieve CO 2 intensity reduction target or hinder it? Evi¬ 
dence from China. Resour. Conserv. Recycl. 134, 206—215. 

Dong, F., Dai, Y., Zhang, S., Zhang, X., Long, R., 2019a. Can a carbon emission trading 
scheme generate the Porter effect? Evidence from pilot areas in China. Sci. Total 
Environ. 653, 565-577. 

Dong, F., Long, R., Yu, B., Wang, Y., Li, J., Wang, Y., Dai, Y., Yang, Q., Chen, H., 2018b. 
How can China allocate CO 2 , reduction targets at the provincial level consid¬ 
ering both equity and efficiency? Evidence from its Copenhagen accord pledge. 
Resour. Conserv. Recycl. 130, 31—43. 

Dong, F., Wang, Y., Su, B., Hua, Y., Zhang, Y., 2019b. The process of peak CO 2 emis¬ 
sions in developed economies: a perspective of industrialization and urbani¬ 
zation. Resour. Conserv. Recycl. 141, 61-75. 

Dong, F., Yu, B., Hadachin, T., Dai, Y., Wang, Y., Zhang, S., Long, R., 2018c. Drivers of 
carbon emission intensity change in China. Resour. Conserv. Recycl. 129, 
187-201. 

Fu, J., Yuan, Z., 2017. Evaluation of effect and analysis of expansion mechanism of 
synergic emission abatement in China’s power industry. China Ind. Econ. 2, 
43—59 (in Chinese ;. 

Garbaccio, R.F., Ho, M.S., Jorgenson, D.W., 1999. Why has the energy-output ratio 
fallen in China? Energy J. 20, 63—92. 

Gillingham, K., Rapson, D., Wagner, G., 2016. The rebound effect and energy effi¬ 
ciency policy. Rev. Environ. Econ. Pol. 10, 68 — 88 . 

Greening, L.A., Greene, D.L., Difiglio, C., 2000. Energy efficiency and consumption- 
the rebound effect-a survey. Energy Policy 28, 389—401. 

Groosman, B., Muller, N.Z., O’Neill-Toy, E., 2011. The ancillary benefits from climate 
policy in the United States. Environ. Resour. Econ. 50, 585—603. 

Guan, D., Su, X., Zhang, Q., Peters, G., Liu, Z., Lei, Y., He, K., 2014. The socioeconomic 
drivers of China’s primary PM 2.5 emissions. Environ. Res. Lett. 9, 024010. 
Haines, A., McMichael, A.J., Smith, K.R., Roberts, I., Woodcock, J., Markandya, A., 
Armstrong, B.G., Campbell-Lendrum, D., Dangour, A.D., Davies, M., Bruce, N., 
Tonne, C., Barrett, M., Wilkinson, P., 2009. Public health benefits of strategies to 
reduce greenhouse-gas emissions: overview and implications for policy 
makers. Lancet 374, 2104—2114. 

Hasanbeigi, A., Lobscheid, A., Lu, H., Price, L., Dai, Y., 2013. Quantifying the co¬ 
benefits of energy-efficiency policies: a case study of the cement industry in 
Shandong Province, China. Sci. Total Environ. 458, 624-636. 

He, K., Lei, Y., Pan, X., Zhang, Y., Zhang, Q., Chen, D., 2010. Co-benefits from energy 
policies in China. Energy 35, 4265—4272. 

Hu, C., Huang, X., 2008. Characteristics of carbon emission in China and analysis on 
its cause. Chin. J. Popul. Resour. Environ. 18, 38—42. 

Huang, Y., Shen, H., Chen, H., Wang, R., Zhang, Y., Su, S., Chen, Y., Lin, N., Zhuo, S., 
Zhong, Q., Wang, X., Liu, J., Li, B., Liu, W., Tao, S., 2014. Quantification of global 


primary emissions of PM 2 . 5 , PMjo, and TSP from combustion and industrial 
process sources. Environ. Sci. Technol. 48,13834-13843. 

IPCC, 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories. 

Ji, X., Yao, Y., Long, X., 2018. What causes PM 2.5 pollution? Cross-economy empirical 
analysis from socioeconomic perspective. Energy Policy 119, 458—472. 

Kaya, Y., 1990. Impact of carbon dioxide emission control on GNP growth: inter¬ 
pretation of proposed scenarios. In: Paper Presented to the IPCC Energy and 
Industry Subgroup. Response Strategies Working Group, Paris. 

Li, Y., Su, B., 2017. The impacts of carbon pricing on coastal megacities: a CGE 
analysis of Singapore. J. Clean. Prod. 165,1239—1248. 

Li, Z., Shao, S., Shi, X., Sun, Y., Zhang, X., 2019. Structural transformation of 
manufacturing, natural resource dependence, and carbon emissions reduction: 
evidence of a threshold effect from China. J. Clean. Prod. 206, 920—927. 

Lin, G., Fu, J., Jiang, D., Hu, W., Dong, D., Huang, Y., Zhao, M., 2013. Spatio-temporal 
variation of PM 2.5 concentrations and their relationship with geographic and 
socioeconomic factors in China. Int. J. Environ. Res. Public Health 11,173—186. 

Ma, Z., Xue, B., Geng, Y., Ren, W., Fujita, T., Zhang, Z., Puppim de Oliveira, J.A., 
Jacques, D.A., Xi, F., 2013. Co-benefits analysis on climate change and environ¬ 
mental effects of wind-power: a case study from Xinjiang, China. Renew. En¬ 
ergy 57, 35-42. 

National People’s Congress (NPC), 2015. Law of the people’s Republic of China on 
the prevention and control of atmospheric pollution. Available at. http://www. 
npc.gov.cn/npc/xinwen/2015-08/31 /content_1945589.htm. (Accessed 10 

February 2019). 

NBSC (National Bureau of Statistics of China), 2015a. China Energy Statistical 
Yearbook. China Statistics Press, Beijing. 

NBSC (National Bureau of Statistics of China), 2015b. China Statistical Yearbook. 
China Statistics Press, Beijing. 

NBSC (National Bureau of Statistics of China), 2015c. China City Statistical Yearbook. 
China Statistics Press, Beijing. 

Nemet, G.F., Holloway, T., Meier, P., 2010. Implications of incorporating air-quality 
co-benefits into climate change policymaking. Environ. Res. Lett. 5, 014007. 

Sancho, F., 2010. Double dividend effectiveness of energy tax policies and the 
elasticity of substitution: a CGE appraisal. Energy Policy 38, 2927—2933. 

Shao, S., Li, X., Cao, J., Yang, L., 2016. China’s economic policy choices for governing 
smog pollution based on spatial spillover effects. Econ. Res. J. 9, 73—88 (in 
Chinese). 

Shrestha, R.M., Pradhan, S., 2010. Co-benefits of CO 2 emission reduction in a 
developing country. Energy Policy 38, 2586-2597. 

State Council of China, 2013. The action plan for air pollution prevention and 
control. Available at. http://www.jingbian.gov.cn/gk/zfwj/gwywj/41211.htm. 
(Accessed 10 February 2019). 

State Council of China, 2016a. Work plan for the control of greenhouse gas emis¬ 
sions during the 13th five-year Plan period. Available at. http://www.gov.cn/ 
xinwen/2016-ll/04/content_5128653.htm. (Accessed 10 February 2019). 

State Council of China, 2016b. The 13th Five-Year Plan for the protection of 
ecological environment. Available at. http://www.gov.cn/zhengce/content/ 
2016-12/05/content_5143290.htm. (Accessed 10 February 2019). 

Vennemo, H., Aunan, K., Jianwu, H., Tao, H., Shantong, L., 2009. Benefits and costs to 
China of three different climate treaties. Resour. Energy Econ. 31,139—160. 

Wagner, F., Amann, M., 2009. Analysis of the proposals for GHG reductions in 2020 
made by UNFCCC Annex I Parties: implications of the economic crisis. Inter¬ 
national Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria. 
Available at. http://www.iiasa.ac.at/web/home/research/researchPrograms/air/ 
Annexl.html. (Accessed 10 February 2019). 

Wei, W., Ma, X., 2015. Optimal policy for energy structure adjustment and haze 
governance in China. Chin. J. Popul. Resour. Environ. 25, 6—14 (in Chinese). 

Xian, Y., Wang, K., Shi, X., Zhang, C„ Wei, Y.M., Huang, Z., 2018. Carbon emissions 
intensity reduction target for China’s power industry: an efficiency and pro¬ 
ductivity perspective. J. Clean. Prod. 197,1022—1034. 

Xie, H., Zhai, Q., Wang, W., Yu, J., Lu, F., Chen, Q., 2018. Does intensive land use 
promote a reduction in carbon emissions? Evidence from the Chinese industrial 
sector. Resour. Conserv. Recycl. 137,167—176. 

Xu, B., Lin, B., 2016. Regional differences of pollution emissions in China: contrib¬ 
uting factors and mitigation strategies. J. Clean. Prod. 112,1454—1463. 

Xu, G., Liu, Z., Jiang, Z., 2006. Decomposition model and empirical study of carbon 
emissions for China, 1995-2004 Chin. J. Popul. Resour. Environ. 16,158—161 (in 
Chinese). 

Xu, Y., Masui, T., 2009. Local air pollutant emission reduction and ancillary carbon 
benefits of SO 2 control policies: application of AIM/CGE model to China. Eur. J. 
Oper. Res. 198, 315-325. 

Xue, B., Ma, Z., Geng, Y., Heck, P., Ren, W., Tobias, M., Maas, A., Jiang, P., Puppim de 
Oliveira, J.A., Fujita, T., 2015. A life cycle co-benefits assessment of wind power 
in China. Renew. Sustain. Energy Rev. 41, 338—346. 

Yang, X., Teng, F., Wang, G., 2013. Incorporating environmental co-benefits into 
climate policies: a regional study of the cement industry in China. Appl. Energy 
112,1446-1453.