Energy efficiency of Chinese service sector and its regional differences

Accepted Manuscript Energy efficiency of Chinese service sector and its regional differences Lin Boqiang, Zhang Guanglu PII:

S0959-6526(17)32008-5

DOI:

10.1016/j.jclepro.2017.09.020

Reference:

JCLP 10521

To appear in:

Journal of Cleaner Production

Please cite this article as: Lin Boqiang, Zhang Guanglu, Energy efficiency of Chinese service sector and its regional differences, Journal of Cleaner Production (2017), doi: 10.1016/j.jclepro.2017.09.020 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

The clean version

2

Energy efficiency of Chinese service sector and its regional

3

differences

4

Lin Boqianga,*, Zhang Guanglub

a

SC

School of Management, China Institute for Studies in Energy Policy, Collaborative Innovation Center for Energy Economics and Energy Policy, Xiamen University, Xiamen, Fujian, 361005, PR China. b China Center for Energy Economics Research, The School of Economics, Xiamen University, Xiamen, Fujian, 361005, PR China. *

EP

TE D

M AN U

Corresponding author at: School of Management, China Institute for Studies in Energy Policy, Collaborative Innovation Center for Energy Economics and Energy Policy, Xiamen University, Xiamen, Fujian, 361005, PR China. Tel: +86 5922186076; Fax: +86 5922186075. E-mail addresses: [email protected] ;[email protected] (B. Lin).

AC C

6 7 8 9 10 11 12 13 14 15 16 17

RI PT

5

1

ACCEPTED MANUSCRIPT Abstract: The energy consumption of the service sector in China is larger than the

2

total energy consumption of Japan, great importance should be attached to the energy

3

efficiency of Chinese service sector. Considering undesirable output and regional

4

heterogeneity, a meta-frontier slack-based efficiency measure (MSBM) approach is

5

adopted in this paper to measure the energy efficiency of Chinese service sector using

6

provincial panel data of China during 1995-2013. The empirical results show that, the

7

energy efficiency of the service sector in China is 0.62 and 0.85 under meta-frontier

8

and group-frontier technologies. The eastern regions show the best performance in

9

energy efficiency, the central region follows and the western region shows the poorest

10

performance. Only the eastern region shows an increasing trend in the energy

11

efficiency of Chinese service sector. The potential of energy efficiency improvement

12

for Hainan, Guizhou, Qinghai, Ningxia, Gansu and Shaanxi are relatively high. Some

13

policy recommendations are proposed to improve the energy efficiency of Chinese

14

service sector.

15

Keywords： ：Total-factor energy

16

technology; Slack-based model

SC

M AN U

EP

TE D

efficiency;

AC C

17

RI PT

1

2

Service

sector;

Meta-frontier

ACCEPTED MANUSCRIPT Notation list： ：

RI PT

Full name Meta-frontier slack-based efficiency measure Gross domestic product Total-factor energy efficiency Stochastic frontier analysis Data envelopment analysis Slacks-based measure Carbon dioxide Sulfur dioxide Chemical oxygen demand Ecological total-factor energy efficiency Malmquist Index Decomposition Non-radial directional distance function Directional distance function Group-frontier ecological total-factor energy efficiency Meta-frontier ecological total-factor energy efficiency Decision-making unit Variable returns to scale Meta-technology ratio Technology gap pertaining to energy efficiency Energy efficiency improvement Potential energy efficiency improvement under group-frontier technology Potential energy efficiency improvement under meta-frontier technology Perpetual inventory method

SC

Abbreviation MSBM GDP TFEE SFA DEA SBM CO2 SO2 COD ETFEE MID NDDF DDF GETFEE METFEE DMU VRS MTR TGEE EEI EEIG EEIM PIM

M AN U

1

2 3

1. Introduction

Under the process of economic restructuring and urbanization in recent years, the

5

development of the service sector in China is accelerating. It was clearly stated in the

6

Rules of Clarification of Three Industries (NBSC, 2013) that Chinese service sector is the

7

tertiary industry by definition. The service sector refers to the industries except for the

8

primary industry and the secondary industry. Table 1 presents the classification of three

9

industries in China. As shown in Fig. 1, the proportion of Chinese service sector in gross

10

domestic product (GDP) is increasing, and the role of the service sector in driving

11

economic growth is also getting more important. In 2015, the added value of the service

12

sector accounted for 50.5 % of GDP. The service sector has surpassed the secondary

13

industry, and became the largest among the three industries in China since 2012. From the

14

perspective of employment, in 2011 the employment share of the service sector was

15

35.7 %, surpassing the primary industry and becoming the highest of the three industries.

16

In 2015, the service sector covered 329.4 M employees, accounting for around 42.4 % of

17

total employees in China. From the proportions of both value added and employment in

18

three industries, the service sector has overtaken the secondary industry becoming the

AC C

EP

TE D

4

3

ACCEPTED MANUSCRIPT 1

largest sector in China. Table 1: The classification of three industries in China

2

Industries included Agriculture Forestry Animal husbandry and fishery

The primary industry

The secondary industry

RI PT

Mining industry Manufacturing industry

Electric, heat, gas and water production and supply industry Construction industry Wholesale and retail trade

SC

Transportation, warehousing and postal service Accommodation and catering industry

Information transmission, software and information technology services The tertiary industry

M AN U

Finance, real estate, leasing and business services Scientific research and technical services

Water conservancy, environment and public facilities management Residents service, repair and other services, education, health and social work Culture, sports and entertainment, public management, social security Social organizations, international organizations, Service industries involved in the primary and secondary industry 3

40,000 35,000

EP

30,000 25,000 20,000

AC C

Added Value ( 109 RMB, constant price in 2000)

45,000

TE D

Source: National Bureau of Statistics of China in 2013

15,000 10,000

5,000 0

2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Year The Primary Industry

The Secondary Industry

The Tertiary Industry

4 5

Fig. 1: Composition of GDP by industry in China

6

Although compared with the secondary industry, the service sector has relatively 4

ACCEPTED MANUSCRIPT low energy consumption and low emission, its rapid development also leads to rapid

2

growth in energy consumption and carbon emissions. From 2000-2014, the average

3

annual growth rates of total energy consumption and energy consumed by the service

4

sector in China were 7.9 % and 8.7 %, which indicates the energy consumption of the

5

service sector increased faster than that of the whole country (Fig. 2). The energy

6

consumption of Chinese service sector is larger than the total energy consumption of

7

many other countries. In 2014, the energy consumption of the service sector in China

8

was 672.9 Mtce, larger than the energy consumption of Japan which was 651.4 Mtce.

9

Some sub-sectors in the service sector are also energy-intensive, such as the

10

transportation, which accounts for more than half of Chinese service sector. The same

11

with the energy structure of the whole China, fossil energy occupies a large proportion

12

in energy consumption of Chinese service sector. In 2014, the proportion was over

13

80 %. Huge energy consumption and energy structure given priority to fossil fuels lead

14

to serious environmental pollution and CO2 emissions. The energy consumption and

15

carbon emission of Chinese service sector are not just about China’s energy and

16

environmental problems, they also have influence on global energy market and climate

17

change.

%

TE D

16.0 15.5 15.0 14.5 14.0 13.5 13.0 12.5 12.0

EP

800 700 600 500 400 300 200 100 0

AC C

Mtce

M AN U

SC

RI PT

1

Year Energy consumption of service sector (Unit：Mtce) The proportion to national energy consumption (Unit：%)

18 19 20

Fig. 2: Energy consumption of the service sector and its proportion to total energy consumption of China Energy saving and emission reduction should be on the basis of economic 5

ACCEPTED MANUSCRIPT development, improving energy efficiency is one of the most effective way in solving the

2

energy and environmental problems. According to Shi (2006), researches on energy

3

efficiency can be divided into two categories based on the number of input factors, they

4

are single-factor energy efficiency and total-factor energy efficiency (TFEE). The former

5

is defined as the ratio of the effective output to the total energy input, while is simple and

6

easy to understand, but cannot estimate the substitution effect between factors. The latter

7

is derived from the microeconomic theory of total-factor productivity, which can not only

8

consider the substitution effect between input factors accurately, but can also reflect the

9

overall utilization level of energy under a certain production factor structure of a region.

10

In the analysis of TFEE, how to define the efficiency frontier is the key point, stochastic

11

frontier analysis (SFA) and data envelopment analysis (DEA) approach are two main

12

methods in recent years. SFA is a parameter estimation method and the parameters are

13

estimated by maximum likelihood estimation. Lin and Yang (2013) adopted SFA to

14

estimate the average efficiency of energy inputs and cumulative energy saving potential in

15

China’s thermal power industry over 2005-2010.

M AN U

SC

RI PT

1

DEA is a linear programming method for nonparametric estimation. Compared with

17

SFA, the biggest advantage of DEA is that it does not need to assume an equation form for

18

the efficiency frontier. So it has developed rapidly in recent years. Hu and Wang (2006)

19

first attempted to use DEA to calculate the TFEE of China and made a comparison of

20

different regions. DEA can construct the optimal boundary to measure the relative

21

efficiency of decision unit through measuring the distance between decision-making unit

22

and the optimal boundary. Energy is one of the most important inputs for economic

23

growth, which also leads to a huge number of undesirable outputs such as carbon

24

emissions, waste gas and water, and solid waste during the process of production (Liang

25

et al., 2016). Taking into account the desirable and undesirable outputs is more reliable

26

when measuring the TFEE. Carbon dioxide (CO2), sulfur dioxide (SO2) and chemical

27

oxygen demand (COD) are the most frequently used as undesirable outputs in the energy

28

efficiency analysis, for instance, Li and Hu (2012) considered CO2 and SO2, Rao et al.

29

(2012) considered COD and SO2 and Zhang et al.(2015a) considered all the three types.

30

Tone (2001) first proposed the slacks-based measure (SBM) which was developed by

AC C

EP

TE D

16

6

ACCEPTED MANUSCRIPT Zhou et al. (2006) considering the undesirable output. Li and Hu (2012) combined the

2

method of common frontier DEA by O’Donnell et al. (2008) and the SBM model

3

considering the undesirable output by Cooper et al. (2007) to calculate the ecological

4

total-factor energy efficiency (ETFEE) of 30 regions in China. Zhang et al. (2015a)

5

further developed a method of meta-frontier slack-based efficiency measure (MSBM)

6

incorporating the meta-frontier approach in order to measure the ETFEE of China

7

incorporating regional heterogeneities. Fig. 3 shows the relationship and development of

8

above major methods and energy efficiency indicators developed from DEA.

11 12

Fig. 3: The relationship and development of methods and energy efficiency

EP

10

indicators

There are also other methods on energy efficiency. Wei et al. (2007) used Malmquist

AC C

9

TE D

M AN U

SC

RI PT

1

13

Index Decomposition based on DEA to investigate the energy efficiency of the iron and

14

steel sector in China. Zhou et al. (2012) proposed a non-radial directional distance

15

function (NDDF) approach to modeling energy and CO2 performance and Zhang et al.

16

(2013) extended a meta-frontier NDDF to measure energy efficiency and technology gaps

17

in electricity generation. The NDDF overcomes some disadvantages of conventional

18

directional distance function (DDF), which is another approach to measuring efficiency

19

based on DEA. The directional vector and weight vector are still needed to be assumed.

20

Most researches on energy efficiency of China focused on agriculture, industry (Bi et 7

ACCEPTED MANUSCRIPT al., 2014) and building industry (Liu and Lin, 2016) especially energy-intensive industries

2

(Lin and Tan, 2016). Few researches have looked at the energy and environmental issues

3

of the service sector, because the service sector was thought to be environment-friendly in

4

previous studies. More researches begin to question this traditional view considering the

5

supply chain of the service sector (Alcántara and Padilla, 2009; Butnar and Llop, 2011).

6

They assess the environmental impact of the service sector based on the consumption

7

perspective. Zhang et al. (2015b) used a multi-regional input-output model to estimate

8

consumption-based emissions of the service sector of 41 countries, the results showed that

9

from the perspective of consumption, the service sector consumed much more energy than

10

the direct energy consumption. Schlomann and Schleich (2015) explored factors driving

11

the adoption of energy efficiency improvement measures in the tertiary sector based on

12

data from a survey on energy consumption in the tertiary sector in Germany. Few studies

13

have measured the energy efficiency of the service sector.

M AN U

SC

RI PT

1

In China, as the rapid release of energy conservation and emissions reduction

15

capacity of industrial and agricultural areas, the difficulty of energy conservation and

16

emissions reduction will increase due to the marginal diminishing effect. With the attitude

17

to combat climate change of China being more proactive and the economic structure

18

transformation from the secondary industry to the service sector, the importance of

19

improving energy efficiency should be attached to the service sector.

EP

TE D

14

Following Zhang et al. (2015a), the energy efficiency of the service sector of China’s

21

30 provinces is measured by the approach of MSBM in this paper. The reasons for

22

applying this method are as follows. First, it considers the regional heterogeneities, the

23

undesirable outputs as well as the slack variables in energy efficiency analysis

24

simultaneously. Second, the technology gap pertaining to energy efficiency (TGEE)

25

among regions can also be analyzed. Third, no weight vector or directional vector needs

26

preset. Energy efficiency is the gap between actual and target energy inputs. In order to

27

analyze regional heterogeneities, two kinds of technologies are defined, they are group-

28

frontier technology and meta-frontier technology, and two kinds of ETFEE are calculated,

29

they are group-frontier ecological total-factor energy efficiency (GETFEE) and meta-

30

frontier ecological total-factor energy efficiency (METFEE). The whole sample of this

AC C

20

8

ACCEPTED MANUSCRIPT paper is the 30 provinces in China (see Section 2.2), which are divided into three regions

2

(see Table 2). Each region represents one group. The group-frontier means that the

3

technology frontier is the same within the same group and the meta-frontier means that

4

technology frontier is the same within the whole sample. By decomposition of the

5

METFEE, the technology gap pertaining to energy efficiency can be assessed, which

6

measures the difference between the two kinds of technologies resulting from group

7

limitations. The potential improvement in energy efficiency can also be compared across

8

provinces based on the calculation results of energy efficiency. This paper fills the

9

research gap on the energy efficiency of Chinese service sector based on provincial panel

10

data considering the regional heterogeneities, the undesirable outputs and the slack

11

variables simultaneously. Unlike Zhang et al. (2015a), this study estimates not only the

12

energy efficiency and technology gap, but also the possible energy efficiency

13

improvement under two kinds of technology frontiers. The results can help the

14

government to formulate more targeted and effective policies on energy conservation and

15

emission reduction in Chinese service sector.

M AN U

SC

RI PT

1

The remainder of the paper is structured as follows. Section 2 presents the empirical

17

model and the data used in the paper. Section 3 shows the empirical results and the related

18

discussions until Section 4 concludes and puts forward some policy recommendations.

19

2. Materials and methods

20

2.1 Methodology

EP

TE D

16

Applying the MSBM approach proposed by Zhang et al. (2015a), which combined

22

the SBM model and meta-frontier DEA approach together, this paper investigates the

23

ETFEE of Chinese service sector considering the regional heterogeneity in China, and

24

group-frontier technologies and meta-frontier technologies are defined for this purpose.

25

Two kinds of ETFEE can be derived under the two kinds of technologies, they are group-

26

frontier ecological total-factor energy efficiency (GETFEE) and meta-frontier ecological

27

total-factor energy efficiency (METFEE).

28

2.1.1 Group-frontier ecological total-factor energy efficiency

29

AC C

21

Consider that there are N provinces in China and the service sector in each province 9

ACCEPTED MANUSCRIPT 1

uses M inputs, X ∈ R+m to jointly produce I desirable outputs D ∈ R+i and J undesirable

2

outputs U ∈ R+j . Each province represents a decision-making unit (DMU). Suppose H groups showing technological heterogeneities, there are N h provinces in

4

group h and N1 + N 2 + ⋯ + N h = N . DMUs in different groups are not comparable in

5

efficiency, because it is unreasonable to calculate the efficiency of a province in one group

6

under the technology assumption of another group. The group-frontier technology of

7

group h is defined as: Th = {( X , D, U ) : X can produce ( D, U )} h = 1, 2, ⋯ , H , where Th is

8

often assumed to satisfy the standard axioms of the production theory (Picazo-Tadeo et

9

al., 2005). Inputs and desirable outputs are often assumed to be strongly or freely

10

disposable. A nonparametric DEA piecewise linear production frontier is used to construct

11

Th . Th for the service sector in N h provinces can be expressed as follows:

SC

M AN U

Nh   β n X mn ≤ X m , m = 1, 2,..., M ,  ∑  n =1   Nh   β n Din ≥ Di , i = 1, 2,..., I ,   ∑ Th = ( X , D,U ) : n=1    Nh  β nU jn ≤ U j , j = 1, 2,..., J ,  ∑   n =1   β n ≥ 0, n = 1, 2,..., N h .  

TE D

12

RI PT

3

(1)

where β n is a nonnegative multiplier for constructing Th through a convex combination.

14

Some constraints such as

15

returns to scale (VRS). The following model can be used to evaluate energy efficiency by

16

adding slack variables:

AC C

EP

13

∑

Nh

n =1

β n = 1 can be imposed on β n for the reason of variable

10

ACCEPTED MANUSCRIPT 1 1− M

θ * = min 1+

S mX0 ∑ m =1 X m 0 M

I J SU SD 1 (∑ i 0 + ∑ j 0 ) I + J i =1 Di 0 j =1 U j 0

s.t.

∑β n =1

n

X mn = X m 0 −S mX0 ,

Nh

∑β D n =1

n

in

= Di 0 +SiD0 ,

jn

=U j 0 − S Uj 0 ,

Nh

∑β U n =1

n

(2)

SC

S mX0 ≥ 0, SiD0 ≥ 0, S Uj 0 ≥ 0, β n ≥ 0.

RI PT

Nh

1

where S mX0 refers to slack variable (potential reduction) of the mth input; S iD0 refers to

3

slack variable (potential increase) of the ith desirable output; S Uj0 refers to slack variable

4

(potential reduction) of the jth undesirable output; and 0 refers to DMU. Eq. (2) is a SBM

5

model. If θ * = 1 , i.e., S 0X = 0, S 0D = 0, S 0U = 0 , then it is believed that the DMU is SBM-

6

efficient in the presence of undesirable outputs. If the slack of energy input is zero

7

( S 0e = 0 ), then the DMU is energy efficient.

10 11

TE D

GETFEEn ,t =

Group − frontier actual energy input n,t

EP

9

The GETFEE for province n at time t can be defined as follows:

Actual energy inputn,t

=

en,t − Sne,t en ,t (3)

AC C

8

M AN U

2

According to the above definition, the GETFEE lies between 0 and 1. If the GETFEE

12

is equal to 1, that means the best energy efficiency on the environmental technology

13

frontier. After solving model (2) using the data of inputs and outputs in group h, we can

14

calculate the GETFEE of the service sector for province n at time t according to Eq. (3).

15

2.1.2 Meta-frontier ecological total-factor energy efficiency

16

The GETFEE values under different group-frontiers are incomparable due to group

17

heterogeneity. In order to make the ETFEE comparable for all observations, a meta-

18

frontier technology is needed to be defined. By enveloping all group-frontier 11

ACCEPTED MANUSCRIPT 1

technologies, a meta-frontier technology could be defined as the intersection of all group-

2

frontier

3

MT = {T1 ∪ T2 ∪ ... ∪ TH } . The nonparametric meta-frontier technology exhibiting constant

4

returns to scale (CRS) can be expressed as follows:

using

all

from

all

the

H

groups,

RI PT

H Nh   β nh X mn ≤ X m , m = 1, 2,..., M ,  ∑∑  h =1 n =1   H Nh   β nh Din ≥ Di , i = 1, 2,..., I ,   ∑∑ MT = ( X , D, U ) : h =1 n =1    H Nh h   U ≤ U , j = 1, 2,..., J , β ∑∑ n jn j   h =1 n =1  h β n ≥ 0, n = 1, 2,..., N h , h = 1, 2,..., H . 

M AN U

6 7

observations

SC

5

technologies

β nh is a nonnegative multiplier and the convexity constraint, i.e.,

(4)

∑ ∑ H

Nh

h =1

n =1

β nh = 1 ,

8

should be imposed on model (4) in order to make the meta-frontier smooth. Combining

9

model (2) and model (4), we can define the MSBM model. The global MSBM can be obtained by solving the following model.

TE D

10

1−

θ * = min

1 M

M

S mX0

∑X m =1

m0

J SU S 1 (∑ + ∑ j0 ) I + J i =1 Di 0 j =1 U j 0 I

1+

D i0

EP

s.t. T

H

Nh

t =1 h =1

n

AC C

∑∑∑ β Nh

T

11

H

∑∑∑ β t =1 h =1 T

H

h n

t X mn = X mt 0 − S mX0 ,

h n

Dint = Dit0 + SiD0 ,

Nh

∑∑∑ β U t =1 h =1 H

h n

t jn

= U tj 0 − S Uj 0 ,

n

Nh

∑∑ β h =1 n =1

(5)

n

h n

=1

S mX0 ≥ 0, SiD0 ≥ 0, S Uj 0 ≥ 0, β nh ≥ 0. 12

After obtaining the optimal solutions for all slack variables under meta-frontier

13

technologies by solving model (5), the METFEE of the service sector for province n at

14

time t can be calculated as follows: 12

ACCEPTED MANUSCRIPT METFEEn ,t =

1

Meta − frontier target energy consumptionn ,t Actual energy consumptionn ,t

=

en ,t − S ne,t en ,t (6)

2 3

The same as the GETFEE index, high (low) METFEE indicates high (low) energy efficiency of the service sector under meta-frontier technology.

5

2.1.3 Decomposition of METFEE

RI PT

4

By comparing the difference between GMTFEE and METFEE, the technology gap

7

pertaining to energy efficiency can be obtained. As shown in Zhang et al. (2015a), energy

8

efficiency under meta-frontier technologies can be divided into within-group energy

9

efficiency and the meta-technology ratio (MTR). The former measures the relative energy

10

efficiency of observations under specific group-frontier technologies, and MTR measures

11

the proximity of a group-frontier technology to a meta-frontier technology. Higher MTR

12

means shorter distance, and MTR=1 means there is no gap in energy efficiency between

13

the two technologies.

METFEE = GETFEE × MTR

15

18 19 20

The technology gap pertaining to energy efficiency for each group can be defined as follows:

TGEE = 1 − MTR

EP

17

(8)

The TGEE measures the technology gap in energy efficiency for the service sector

21

between group-frontier and meta-frontier technologies.

22

2.1.4 Energy efficiency improvement

23

(7)

AC C

16

M AN U

METFEE can be decomposed into the product of GETFEE and MTR:

TE D

14

SC

6

As described in section 2.1.1, slack variables reflect the potential reduction of

24

inputs. In section 2.1.1 and 2.1.2, the energy efficiency under group-frontier and meta-

25

frontier technologies has been evaluated using the calculation results of energy input slack

26

variable, the potential of energy efficiency improvement (EEI) can also be estimated

27

based on the slack variable of energy input. The potential energy efficiency improvement

28

under group-frontier technology (EEIG) and potential energy efficiency improvement 13

ACCEPTED MANUSCRIPT 1

under meta-frontier technology (EEIM) for province n at time t can be measured as

2

follows:

EEI nG,t = 1 − GETFEEn ,t

3

EEI nM,t = 1 − METFEEn,t (9)

RI PT

4 5 6

2.2 Data processing

7

2.2.1 Non-energy input

Capital and labor are the two non-energy inputs. Wu (2015) estimated the capital

9

stock of Chinese service sector of each province form 1995 to 2006. Capital input data in

10

this study before 2007 are from Wu (2015), and then are extrapolated to 2013 with the

11

perpetual inventory method (PIM). PIM is started by Goldsmith in 1951 and now widely

12

used for estimating capital stock based on gross investment or capital formation data in

13

each year. The estimation technique can be expressed as

M AN U

SC

8

K t = K t −1 (1 − δ t ) + I t

14

(10)

δ t is the rate of depreciation in year t, and

Kt

16

It

17

study uses the data of investments in the fixed assets of the service sector to reflect I t ,

18

which is estimated by subtracting the investments in the fixed assets of the primary

19

industry and the secondary industry from the gross investments in the fixed assets of each

20

province. The data of investments in the fixed assets are sourced from the Statistical

21

Yearbook of the Chinese Investment in Fixed Assets (1996-2014)and the provincial

22

depreciation rate data of the service sector are obtained from the calculation results in Wu

23

(2015). Capital stock data have been deflated into constant price in 1995. The data of

24

labor are sourced from the China Statistical Yearbook of the Tertiary Industry (1996-

25

2014).

26

2.2.2 Energy input

27

TE D

15

is the real value of capital stock in year t,

AC C

EP

is the real value of incremental capital stock or gross capital formation in year t. This

The energy input of the service sector in each province over 1995-2013 is obtained by 14

ACCEPTED MANUSCRIPT aggregating the terminal energy consumption of “Transportation, warehousing and postal

2

service”, “Wholesale and retail trade, Accommodation and catering industry” and “other

3

services” from the Chinese Energy Statistical Yearbook (1996-2014). The energy

4

consumption is estimated by the summation of coal, oil product, liquefied petroleum gas,

5

natural gas and electricity consumed by each subsector after conversion from physical

6

units to coal equivalent using the conversion coefficients of different energy varieties

7

provided by the Chinese Energy Statistical Yearbook 2014.

8

2.2.3 Desirable output and undesirable output

RI PT

1

The desirable output is measured by the added value of the service sector of each

10

province. The added value data have been deflated into constant price in 1995 and are

11

sourced from the China Statistical Yearbook of the Tertiary Industry (1996-2014).

M AN U

SC

9

The undesirable output taken into consideration in this study is CO2 emissions. Other

13

undesirable outputs like SO2 are not incorporated since the emission data about the service

14

sector are not available and not easy to estimate. The CO2 emissions of the service sector

15

can be estimated by the summation of carbon emissions of each kind of fossil energy. The

16

CO2 emission coefficients are from IPCC (2006) (Eggleston et al., 2006), and are are

17

assumed to remain the same during 1995-2013.

TE D

12

The data of 30 provincial administrative districts (except for Tibet) in mainland China

19

over 1995-2013 are used in this paper, and the 30 provinces are divided into three regions

20

according to their location. The detailed division is shown in Table 2.

EP

18

Table 2: The division of different regions in China

21

Provinces included Liaoning, Hebei, Shandong, Jiangsu, Zhejiang, Fujian, Guangdong ,Hainan, Beijing, Tianjin, Shanghai Heilongjiang, Jilin, Shanxi, Henan, Anhui, Jiangxi, Hubei , Hunan Gansu, Qinghai, Shaanxi, Sichuan, Guizhou, Yunnan, Xinjiang, Ningxia, Inner Mongolia , Guangxi, Chongqing

AC C

Regions

Eastern region Central region

Western region 22

The eastern region is the most developed, while the central region has achieved fast

23

development in the past decades. Compared with the other two regions, the western

24

region is the least developed one, but is the most abundant region in natural resources, and

25

it also grows fast under the background of implementing the west development strategy.

26

In recent years, the environmental constraints eastern region faces are more serious, which 15

ACCEPTED MANUSCRIPT 1

promote the industrial sectors transferring to the western region objectively and then

2

speed up the development of the western region in a certain extent. Table 3 describes some statistical features of the service sector in three regions. It can

4

be seen that in terms of the mean value, eastern region is the most developed one,

5

followed by the central region, while the western region is relatively backward. But in

6

terms of the growth rate, all indicators in the western region are larger than the national

7

averages, and larger than most indicators in the eastern and central regions. That means

8

although the western region has low level of development, it is catching up rapidly.

RI PT

3

Growth

10

3. Results and discussion

12

3.1 Results of GETFEE

Eastern

Central

Western

2190.2 8.3 12.3

1339.5 8.1 7.4

878.2 4.8 5.9

215.2 17.0

351.5 23.0

178.4 15.1

105.7 12.4

17.4 4.3 11.7

15.9 4.6 10.4

18.0 3.6 11.9

18.4 4.4 13.0

11.7 11.3

11.83 9.7

11.4 11.9

11.8 12.5

M AN U

1482.3 7.0 8.7

EP

11

There exist big differences in the geographical, economic and cultural aspects among

AC C

13

National

TE D

Mean

Variable (Unit) Inputs: Capital stock (B RMB) Labor (M persons) Energy (Mtce) Outputs: Value-added (B RMB) CO2 emissions (Mt) Inputs: Capital stock (%) Labor (%) Energy (%) Outputs: Value-added (%) CO2 emissions (%)

SC

Table 3: Regional differences of Chinese service sector

9

14

different regions since China is a big country. The characteristics of provinces within the

15

same region are more similar, so comparison of energy efficiency is firstly made within

16

the same group. The GETFEE is estimated under group technologies using region-specific

17

data only, which is calculated for each province. The result is showed in Table A1.

18

Table 4: The average GETFEE performance of each province（ （1995-2013） ） Eastern Province Shanghai Jiangsu

GETFEE 1.000 1.000

Central Province Heilongjiang Anhui

Western GETFEE 1.000 0.993

16

Province Guangxi Xinjiang

GETFEE 1.000 0.990

ACCEPTED MANUSCRIPT 1.000 0.913

Hunan Henan

0.929 0.917

Inner Mongolia Sichuan

0.986 0.946

Shandong Zhejiang Tianjin Liaoning

0.888 0.846 0.823 0.776

Shanxi Jilin Hubei Jiangxi

0.912 0.853 0.789 0.770

Chongqing Yunnan Shaanxi Qinghai

0.909 0.851 0.809 0.742

Beijing Hebei

0.734 0.704

Gansu Ningxia

0.741 0.646

Hainan

0.543

Guizhou

0.501

Average

0.839

Average

0.829

Average

0.895

RI PT

Guangdong Fujian

Table 4 shows the average GETFEE performance of the service sector in three

2

regions of China from 1995 to 2013. For East China, Shanghai, Jiangsu and Guangdong

3

have the highest energy efficiency, and Hainan performs the poorest. As Shanghai,

4

Jiangsu and Guangdong are the most developed provinces and have the most advanced

5

service sector in China, the energy efficiency of the three provinces is also highest. For

6

Central China, Heilongjiang and Anhui have the first and second highest GETFEE of the

7

service sector, while Jiangxi has the lowest GETFEE. For West China, Guangxi shows the

8

best performance of energy efficiency, and Guizhou performs the poorest among the

9

eleven western provinces. All the results from the DEA technique can only reflect a

10

relative level, which can only be compared within the model since different models have

11

different frontiers. Shanghai, Jiangsu, Guangdong, Heilongjiang and Guangxi all reach 1

12

in GETFEE, but it does not mean that they reach the absolutely energy-efficient level or

13

they perform the same level of energy efficiency. These provinces only perform best in

14

energy efficiency within the regions they belong to.

15

3.2 Results of METFEE

M AN U

TE D

EP

AC C

16

SC

1

The energy efficiency calculated under group-frontier technologies cannot be directly

17

compared among different groups because the existence of heterogeneous technologies.

18

So the energy efficiency of Chinese service sector under meta-frontier technologies is

19

expected to be calculated. The result is showed in Table A2. The average energy

20

efficiency under meta-frontier technologies is lower than that under group-frontier

21

technologies, indicating the existence of a technology gap between the group-frontier and

22

the meta-frontier.

23

The METFEE of the service sector can be estimated under meta-frontier 17

ACCEPTED MANUSCRIPT technologies using all data. According to the average METFEE of each province in Table

2

5, most of the provinces show weak performance in energy efficiency in the service

3

sector, indicating a gap in energy efficiency from the efficiency frontier. Comparing the

4

METFEE by region, the eastern region is the most efficient, followed by the central

5

region. The western region shows relatively poor METFEE performance. Guangdong,

6

Heilongjiang and Guangxi are the three provinces which perform best in their specific

7

regions while Hainan, Jiangxi and Guizhou perform the worst in East, Central and West

8

China, which coincides with the results of GETFEE. Unlike the GETFEE, the METFEE

9

can be compared among different regions. In terms of nationwide, Guangdong, Fujian,

10

Jiangsu, Shanghai, Shandong, Zhejiang and Liaoning have the highest levels of METFEE,

11

all of these provinces are located in the eastern region, and their METFEEs of the service

12

sector are higher than the highest level of that in central and western regions. Jiangsu and

13

Guangdong have METFEE unity in five and four years, indicating that these two

14

provinces have the most advanced production technology due to lying on the efficiency

15

frontier.

SC

M AN U

Table 5: The average METFEE performance of each province（ （1995-2013） ）

TE D

16

RI PT

1

Central Province METFEE Heilongjiang 0.694 Hunan 0.639 Anhui 0.629 Shanxi 0.615 Jilin 0.574 Henan 0.564 Hubei 0.554 Jiangxi 0.533

AC C

EP

Eastern Province METFEE Guangdong 0.917 Fujian 0.862 Jiangsu 0.861 Shanghai 0.836 Shandong 0.826 Zhejiang 0.784 Liaoning 0.709 Tianjin 0.673 Beijing 0.647 Hebei 0.598 Hainan 0.475 Average 0.744

Average

0.600

Western Province METFEE Guangxi 0.708 Chongqing 0.613 Sichuan 0.589 Yunnan 0.550 Xinjiang 0.523 Inner Mongolia 0.515 Shaanxi 0.492 Qinghai 0.470 Gansu 0.415 Ningxia 0.400 Guizhou 0.334 Average 0.510

17

Fig. 4 shows the METFEE trends of the service sector in three regions of China

18

under meta-frontier technologies. The METFEE line of the eastern region is highest, the

19

METFEE line of the western region is lowest, and METFEE line of the central region is

20

between the other two lines. It also can be seen from the figure that there exist obvious

21

gaps among the three regions, and the gaps seem to show increasing trends. From 1995 to 18

ACCEPTED MANUSCRIPT 2013, the mean values of METFEE in each region are all above or around 0.5 and the

2

average METFEE of all 30 provinces is 0.618. Zhang et al. (2015a) found most regions in

3

China show weak efficiency in ecological energy use and the average METFEE of China

4

in 2010 was 0.418. The METFEE results of energy intensive industries (Lin and Tan,

5

2016) and building industry (Liu and Lin, 2016) in most provinces are less than 0.5. It

6

cannot show that the energy efficiency of Chinese service sector is relatively high

7

compared to other industries in China because the results of these sectors are not derived

8

from the same model. During the research interval, the METFEE of western region

9

increased from 0.648 to 0.801, which means that the energy efficiency in eastern region

10

was increasing over 1995-2013. The METFEE did not show an increasing trend in either

11

central region or western region. The METFEE in central region was 0.572 in 1995, and

12

increased to 0.641 in 2007, then began to decrease from the year of 2008, and fell into

13

0.551 in 2013. The METFEE of western region shows a slightly decreasing trend which

14

decreased from 0.543 to 0.491 and remained around 0.5 in general.

M AN U

SC

RI PT

1

Many studies about China have found that the eastern region performs best in energy

16

efficiency (Zhang et al., 2015a), environmental efficiency (Chen and Jia, 2017) and other

17

efficiency indicators (Fan et al., 2017). It mainly depends on the level of economic

18

development. Firstly, with well-developed economy, the eastern region has stronger

19

economic power to do research in energy conservation and emission reduction and

20

promote the diffusion of energy-efficient technologies. Secondly, with the rapid economic

21

growth of the eastern region, constraints of resources and environment to economic

22

development have become increasingly prominent, which make these provinces have

23

greater incentive to improve energy efficiency. This result also supports the

24

Environmental Kuznets Curve hypothesis, i.e., environmental improvements occur after a

25

certain level or after the turning point of income.

AC C

EP

TE D

15

19

ACCEPTED MANUSCRIPT 0.90 0.85 0.80

0.70

2 3 4

2013

2007

2005

2004

2006

Western

M AN U

1

Year Central

SC

Eastern

2003

2002

2001

2000

1999

1998

1997

1996

1995

0.40

2012

0.45

2011

0.50

2010

0.55

2009

0.60

RI PT

0.65

2008

METFEE

0.75

Fig. 4: METFEE trends at regional level of the service sector in China over 19952013 3.3 Technology gap in energy efficiency

To better understand the regional heterogeneities, the technology gap pertaining to

6

energy efficiency of the service sector is investigated in this section. Table A3 reports the

7

TGEE based on Eq. (11). As shown in Fig. 5, The eastern region generally shows the

8

lowest and decreasing TGEE, indicating that it has the smallest technology gap when

9

comparing with the other two regions. The average TGEE of the eastern region in 2013

10

was 0.043, i.e., the best energy efficiency performance of the service sector in the eastern

11

region under group-frontier technologies is 95.7 % of that under meta-frontier

12

technologies. The average TGEE for the central and western regions are 0.431 and 0.407,

13

much higher than that of the eastern region. As the TGEE index lies between 0 and 1, and

14

a larger TGEE implies a larger gap between group-frontier and meta-frontier technologies,

15

the average TGEE values in Fig. 5 indicate large gaps between group-frontier and meta-

16

frontier technologies for the central and western regions. Technology gap can be reduced

17

by innovation and shift to higher frontier.

AC C

EP

TE D

5

20

1

3

2013

2012

2011

2010

RI PT 2009

2008

2006

2005

2007

Western

Fig. 5: TGEE trends of Chinese service sector at regional level over 1995-2013

M AN U

2

2004

Year Central

SC

Eastern

2003

2002

2001

2000

1999

1998

1997

1996

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 1995

TGEE

ACCEPTED MANUSCRIPT

3.4 The potential of energy efficiency improvement

The EEI potential of the service sector in each province can be derived through

5

Eq.(9). It is noteworthy that the environmental constraint (CO2 emissions) is taken into

6

consideration when discussing EEI in this paper, which is often ignored in previous

7

studies.

TE D

4

Fig. 6 classifies the average EEI potential of the service sector in each province

9

during 1995-2013 into five degrees. It should be pointed out that Tibet and Taiwan are

10

included in the figure for geographical integrity, which are not included in our

11

calculations, so the color in the two provinces can be ignored. The Fig.(6-a) and Fig.(6-b)

12

are under group-frontier technologies and under meta-frontier technologies. The darker

13

color means larger potential of energy efficiency improvement.

AC C

14

EP

8

It can be seen from Fig. 6 that the average EEI potential of Hainan, Guizhou,

15

Qinghai, Ningxia, Gansu and Shaanxi are relatively large under both frontiers due to their

16

dark color. The color turns from the light to the dark as the provinces proceed from the

17

east to the west in Fig. (6-b), indicating that the less developed areas tend to have larger

18

energy efficiency improvement potential of the service sector. The average EEI potentials

19

of Hainan, Guizhou, Qinghai, Ningxia, Gansu and Shaanxi are the top six in nationwide,

20

which are all above 0.5 under meta-frontier technologies, so these provinces still have

21

relative large potential to improve the energy efficiency. 21

1 2

RI PT

ACCEPTED MANUSCRIPT

(6-a) Under group-frontier technologies

(6-b) Under meta-frontier technologies

6

The MSBM approach is used in this paper to present a quantitative assessment of

7

energy efficiency of Chinese service sector over the period 1995-2013. The main findings

8

of this paper can be summarized as follows.

M AN U

SC

5

Fig. 6: The average potential of energy efficiency improvement under group-frontier and meta-frontier technologies during 1995-2013 4. Conclusion and policy implications

3 4

First, the eastern region shows the best energy efficiency performance for the service

10

sector followed by the central region, and the western region shows the poorest

11

performance. Only the eastern region shows an obviously increasing trend in energy

12

efficiency and an obviously decreasing trend in technology gap. The central and western

13

regions should make more efforts to improve their energy efficiencies. The results support

14

the Environmental Kuznets Curve hypothesis, indicating that developing the service

15

sector vigorously can be an effective way for less developed areas.to improve energy

16

efficiency.

EP

AC C

17

TE D

9

Second, the average TGEE for the central and western regions are much higher than

18

that of the eastern region. There exist relatively large technology gaps pertaining to energy

19

efficiency of Chinese service sector. In the light of the regional heterogeneity of Chinese

20

service sector, the central government should provide more political incentives to

21

popularize advanced technology, strengthen technical cooperation, and encourage

22

technology diffusion across regions in order to eliminate the technology gap and

23

development imbalances in Chinese service sector.

24

Third, there is still room for energy efficiency improvement in Chinese service sector 22

ACCEPTED MANUSCRIPT since the average potential of EEI under meta-frontier and group-frontier technologies are

2

38 % and 15 %. The efficiency value calculated by DEA model is a relative value, so the

3

provinces that achieve efficient status have not necessarily reached absolutely efficient

4

levels of energy efficiency. All the provinces should optimize their inner structures of

5

services in order to make it cleaner and more efficient. For instance, raising the proportion

6

of modern service industries with low energy consumption and emission such as

7

information transmission, computer service and software industry can improve the overall

8

energy efficiency of the service sector. It is of great importance to promote energy

9

conservation and emission reduction in traditional service industries with high energy

SC

TE D

M AN U

.

EP

11

consumption and emission such as transportation rather than focusing on industry only.

AC C

10

RI PT

1

23

ACCEPTED MANUSCRIPT

Appendix:

1998 0.851 1.000 1.000 0.547 0.628 1.000 0.845 1.000 1.000 0.755 0.748 0.900 0.867 1.000 0.885 1.000 1.000 1.000 0.868 0.531 1.000 0.509 0.753 0.741 1.000

1999 0.712 1.000 1.000 0.509 0.617 1.000 0.792 1.000 1.000 0.710 0.740 0.903 1.000 1.000 0.884 1.000 0.825 0.882 0.897 0.548 1.000 0.462 0.868 0.763 0.867

24

2000 0.756 0.916 1.000 0.511 0.598 1.000 0.733 1.000 1.000 0.700 0.694 1.000 1.000 1.000 0.873 1.000 0.878 0.879 1.000 0.571 1.000 0.442 0.984 0.735 0.929

SC

1997 0.887 1.000 1.000 0.531 0.684 1.000 0.774 1.000 1.000 0.698 1.000 0.929 0.874 1.000 0.906 1.000 1.000 1.000 1.000 0.518 1.000 0.513 0.643 0.713 0.827

M AN U

1996 0.799 1.000 1.000 0.578 0.673 1.000 0.806 0.956 1.000 0.765 0.888 0.983 0.785 1.000 0.801 1.000 1.000 1.000 0.762 0.442 1.000 0.484 0.640 0.684 0.830

TE D

1995 0.745 1.000 0.961 0.585 0.676 1.000 0.788 0.856 1.000 0.658 0.943 1.000 0.905 1.000 1.000 1.000 1.000 1.000 0.888 0.415 1.000 0.476 0.626 0.665 0.850

EP

Region E E E E E E E E E E E C C C C C C C C W W W W W W

AC C

Province Beijing Fujian Guangdong Hainan Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai

RI PT

Table A1: The GETFEE of Chinese service sector over 1995-2013

2001 0.787 0.947 1.000 0.529 0.674 1.000 0.739 1.000 1.000 0.735 0.779 1.000 1.000 1.000 0.859 1.000 0.891 0.750 1.000 0.587 1.000 0.451 1.000 0.711 0.928

2002 0.723 0.877 1.000 0.509 0.639 1.000 0.743 1.000 1.000 0.783 0.708 1.000 1.000 1.000 0.849 1.000 0.817 0.731 0.962 0.596 1.000 0.431 1.000 0.630 0.723

2003 0.749 0.906 1.000 0.508 0.788 1.000 0.832 1.000 1.000 0.802 0.798 1.000 1.000 1.000 0.768 1.000 0.819 0.681 0.959 0.583 1.000 0.410 1.000 0.590 0.687

2004 0.705 0.855 1.000 0.497 0.756 1.000 0.829 1.000 1.000 0.840 0.867 1.000 0.786 1.000 0.795 1.000 0.801 0.713 0.980 0.619 1.000 0.426 1.000 0.696 0.696

ACCEPTED MANUSCRIPT

0.753 0.819 1.000 1.000 1.000 2009 0.692 0.932 1.000 0.552 0.636 1.000 0.746 0.837 1.000 0.934 0.885 1.000 0.853 1.000 0.686 0.778 0.804 0.738 0.762 0.824 1.000 0.457 1.000 0.565

25

0.861 0.885 1.000 1.000 1.000 2010 0.687 0.900 1.000 0.582 0.670 1.000 0.718 0.822 1.000 1.000 0.865 1.000 0.783 1.000 0.654 0.757 0.780 0.633 0.745 0.870 1.000 0.480 1.000 0.599

0.921 0.909 1.000 0.948 1.000 2011 0.747 0.801 1.000 0.577 0.651 1.000 0.678 0.746 1.000 0.788 0.870 1.000 0.958 1.000 0.744 1.000 1.000 0.836 0.824 1.000 1.000 0.684 1.000 0.607

RI PT

0.703 0.842 1.000 1.000 1.000 2008 0.719 0.962 1.000 0.565 0.687 1.000 0.787 0.856 1.000 0.919 0.906 1.000 0.852 1.000 0.685 0.820 0.794 0.728 0.776 0.742 1.000 0.440 1.000 0.615

SC

0.715 0.951 1.000 1.000 1.000 2007 0.717 1.000 1.000 0.550 0.706 1.000 0.771 0.877 1.000 0.924 0.908 1.000 0.841 1.000 0.751 0.838 0.793 0.706 1.000 0.742 1.000 0.472 1.000 0.613

M AN U

0.643 1.000 1.000 1.000 1.000 2006 0.778 1.000 1.000 0.540 0.698 1.000 0.856 0.794 1.000 0.757 0.984 1.000 0.889 1.000 0.781 0.870 0.785 0.739 1.000 0.899 1.000 0.610 1.000 0.682

TE D

0.672 0.688 1.000 1.000 1.000 2005 0.758 1.000 1.000 0.601 0.713 1.000 0.892 0.857 1.000 0.880 0.944 1.000 0.806 1.000 0.752 0.851 0.780 0.727 1.000 0.742 1.000 0.456 1.000 0.716

EP

W W W W W Region E E E E E E E E E E E C C C C C C C C W W W W W

AC C

Shaanxi Sichuan Xinjiang Yunnan Chongqing Province Beijing Fujian Guangdong Hainan Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia

0.874 0.934 1.000 0.862 1.000 2012 0.734 0.790 1.000 0.560 0.749 1.000 0.814 0.719 1.000 0.775 0.862 1.000 1.000 1.000 0.820 1.000 1.000 0.858 0.854 1.000 1.000 0.656 1.000 0.599

0.720 0.941 1.000 0.818 1.000 2013 0.733 0.790 1.000 0.546 0.990 1.000 0.711 0.800 1.000 0.817 0.871 1.000 1.000 1.000 0.912 1.000 1.000 0.905 0.932 0.844 1.000 0.639 1.000 0.560

0.702 0.927 0.884 1.000 0.766

ACCEPTED MANUSCRIPT

0.658 0.756 1.000 1.000 0.747 0.826

0.682 0.741 1.000 1.000 0.723 0.757

0.679 0.789 1.000 1.000 0.750 0.889

0.707 0.821 1.000 1.000 0.737 0.785

M AN U TE D 26

0.698 0.825 0.904 1.000 0.765 0.884

RI PT

0.725 0.749 1.000 0.963 0.880 1.000

SC

0.760 0.809 1.000 1.000 1.000 1.000

EP

W W W W W W

AC C

Qinghai Shaanxi Sichuan Xinjiang Yunnan Chongqing

0.698 0.833 1.000 1.000 0.778 0.869

0.659 1.000 0.888 1.000 0.725 0.842

ACCEPTED MANUSCRIPT

Table A2: The METFEE of Chinese service sector over 1995-2013 1999 0.599 0.831 0.828 0.425 0.543 0.813 0.701 0.912 0.831 0.585 0.651 0.599 0.576 0.632 0.549 0.673 0.535 0.590 0.562 0.352 0.722 0.328 0.493 0.456 0.511 0.492 0.554 0.496

2000 0.637 0.830 0.870 0.428 0.558 0.810 0.685 0.939 0.841 0.607 0.651 0.625 0.601 0.630 0.549 0.673 0.553 0.582 0.593 0.350 0.700 0.312 0.542 0.441 0.524 0.560 0.578 0.517

27

2001 0.658 0.856 0.913 0.439 0.611 0.793 0.671 0.910 0.826 0.632 0.671 0.648 0.622 0.644 0.553 0.708 0.570 0.530 0.612 0.355 0.689 0.315 0.589 0.423 0.553 0.579 0.581 0.523

RI PT

1998 0.636 0.879 0.799 0.427 0.549 0.794 0.736 0.849 0.786 0.585 0.644 0.573 0.540 0.616 0.533 0.648 0.523 0.598 0.536 0.339 0.712 0.363 0.420 0.436 0.531 0.457 0.577 0.435

SC

1997 0.626 0.791 0.787 0.396 0.533 0.754 0.626 0.782 0.741 0.522 1.000 0.577 0.523 0.586 0.522 0.618 0.516 0.664 0.539 0.328 0.801 0.360 0.378 0.422 0.510 0.456 0.620 0.419

M AN U

1996 0.557 0.795 0.742 0.402 0.497 0.703 0.613 0.724 0.698 0.538 0.636 0.599 0.512 0.648 0.526 0.597 0.474 0.658 0.496 0.284 0.858 0.368 0.386 0.409 0.499 0.429 0.683 0.410

TE D

1995 0.554 1.000 0.743 0.429 0.509 0.688 0.612 0.712 0.739 0.460 0.676 0.708 0.513 0.553 0.550 0.606 0.505 0.645 0.492 0.274 1.000 0.384 0.392 0.384 0.499 0.460 0.509 0.432

EP

Region E E E E E E E E E E E C C C C C C C C W W W W W W W W W

AC C

Province Beijing Fujian Guangdong Hainan Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan Xinjiang

2002 0.630 0.863 0.995 0.448 0.635 0.815 0.721 1.000 0.830 0.672 0.714 0.672 0.658 0.684 0.554 0.674 0.561 0.489 0.641 0.386 0.724 0.306 0.661 0.410 0.473 0.571 0.597 0.550

2003 0.651 0.862 1.000 0.447 0.660 0.826 0.810 0.909 0.802 0.666 0.744 0.704 0.679 0.713 0.535 0.676 0.580 0.454 0.659 0.394 0.696 0.290 0.690 0.395 0.457 0.475 0.597 0.546

2004 0.626 0.828 0.984 0.438 0.630 0.800 0.806 0.917 0.761 0.656 0.775 0.704 0.549 0.716 0.553 0.642 0.573 0.482 0.693 0.431 0.647 0.284 0.581 0.476 0.440 0.470 0.587 0.518

ACCEPTED MANUSCRIPT

0.593 0.682 2009 0.673 0.906 0.963 0.520 0.626 1.000 0.719 0.790 1.000 0.821 0.876 0.654 0.577 0.827 0.577 0.658 0.648 0.523 0.627 0.491 0.702 0.308 0.528 0.372 0.461 0.506 0.584

0.603 0.660 2010 0.674 0.884 0.950 0.558 0.648 1.000 0.686 0.776 0.901 0.835 0.859 0.620 0.550 0.847 0.563 0.651 0.639 0.503 0.619 0.486 0.660 0.303 0.512 0.373 0.455 0.499 0.555

28

0.561 0.660 2011 0.717 0.799 0.998 0.562 0.631 1.000 0.663 0.729 0.907 0.761 0.869 0.568 0.531 0.729 0.554 0.612 0.642 0.483 0.595 0.537 0.620 0.405 0.503 0.365 0.425 0.491 0.549

RI PT

0.595 0.693 2008 0.662 0.898 0.956 0.510 0.637 0.940 0.719 0.807 0.868 0.778 0.855 0.658 0.573 0.823 0.570 0.690 0.632 0.516 0.636 0.468 0.712 0.304 0.529 0.389 0.464 0.492 0.608

SC

0.609 0.734 2007 0.647 0.951 0.968 0.509 0.655 0.937 0.728 0.850 0.856 0.758 0.862 0.674 0.573 0.777 0.587 0.638 0.601 0.508 0.769 0.453 0.705 0.329 0.537 0.391 0.447 0.504 0.597

M AN U

0.618 0.772 2006 0.656 0.948 1.000 0.489 0.604 0.854 0.829 0.794 0.767 0.656 0.896 0.662 0.565 0.693 0.561 0.588 0.539 0.484 0.713 0.466 0.631 0.367 0.524 0.378 0.428 0.456 0.621

TE D

0.642 1.000 2005 0.634 0.890 0.928 0.490 0.591 0.828 0.811 0.792 0.763 0.679 0.821 0.705 0.558 0.709 0.563 0.595 0.552 0.493 0.736 0.431 0.633 0.296 0.537 0.378 0.420 0.451 0.589

EP

W W Region E E E E E E E E E E E C C C C C C C C W W W W W W W W

AC C

Yunnan Chongqing Province Beijing Fujian Guangdong Hainan Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan

0.563 0.673 2012 0.717 0.783 1.000 0.554 0.629 1.000 0.656 0.707 0.957 0.761 0.853 0.513 0.528 0.694 0.562 0.623 0.648 0.477 0.590 0.550 0.601 0.384 0.494 0.359 0.424 0.495 0.582

0.558 0.684 2013 0.733 0.790 1.000 0.546 0.621 1.000 0.668 0.800 1.000 0.817 0.836 0.488 0.489 0.660 0.563 0.576 0.612 0.450 0.575 0.513 0.649 0.343 0.489 0.351 0.413 0.509 0.632

0.718 0.526

ACCEPTED MANUSCRIPT anhui heilongjiang henan hubei

0.525 0.528 0.532

0.518 0.503 0.553

0.557 0.503 0.512

0.563 0.489 0.493

EP

W W W

0.586 0.503 0.505

AC C

Xinjiang Yunnan Chongqing

TE D

M AN U

SC

RI PT

hunan

29

0.592 0.475 0.484

0.596 0.469 0.505

0.598 0.461 0.492

0.550 0.465 0.489

ACCEPTED MANUSCRIPT

Table A3: The TGEE of Chinese service sector over 1995-

2013

jiangxi

RI PT

jilin shanxi1 beijing fujian guangdong

jiangsu liaoning

M AN U

shandong

SC

hebei

shanghai tianjin

zhejiang

chongqing gansu

TE D

guangxi guizhou

neimenggu ningxia

Province Beijing Fujian Guangdong Hainan

Region E E E E

1995 0.257 0.000 0.226 0.266

AC C

EP

qinghai

1996 0.302 0.205 0.258 0.304

1997 0.294 0.209 0.213 0.254

shanxi2 sichuang xinjiang yunnan

1998 0.253 0.121 0.201 0.219

1999 0.159 0.169 0.172 0.166

30

2000 0.157 0.094 0.130 0.163

2001 0.164 0.096 0.087 0.169

2002 0.128 0.017 0.005 0.120

2003 0.131 0.049 0.000 0.121

2004 0.112 0.031 0.016 0.119

ACCEPTED MANUSCRIPT

0.120 0.187 0.115 0.088 0.169 0.175 0.120 0.336 0.424 0.368 0.378 0.327 0.351 0.331 0.374 0.357 0.278 0.291 0.432 0.402 0.410 0.347 0.323 0.504 0.407 0.318 2009 0.028 0.029 0.037

0.066 0.190 0.065 0.061 0.159 0.133 0.062 0.375 0.399 0.370 0.372 0.327 0.371 0.339 0.407 0.387 0.300 0.295 0.449 0.400 0.436 0.350 0.347 0.483 0.397 0.340 2010 0.019 0.018 0.050

31

0.092 0.207 0.092 0.090 0.174 0.140 0.138 0.352 0.378 0.356 0.355 0.292 0.360 0.293 0.388 0.395 0.311 0.302 0.411 0.405 0.405 0.372 0.362 0.477 0.409 0.340 2011 0.040 0.002 0.002

RI PT

0.126 0.206 0.129 0.151 0.214 0.225 0.138 0.363 0.377 0.384 0.397 0.352 0.477 0.402 0.383 0.361 0.288 0.287 0.442 0.412 0.469 0.350 0.314 0.565 0.405 0.307 2008 0.080 0.067 0.044

SC

0.221 0.246 0.192 0.218 0.259 0.252 0.000 0.379 0.401 0.414 0.424 0.382 0.484 0.336 0.461 0.368 0.199 0.299 0.412 0.408 0.383 0.363 0.348 0.581 0.391 0.266 2007 0.098 0.049 0.032

M AN U

0.261 0.297 0.239 0.243 0.302 0.297 0.283 0.390 0.347 0.352 0.344 0.403 0.526 0.342 0.349 0.356 0.142 0.240 0.396 0.403 0.398 0.333 0.317 0.590 0.382 0.228 2006 0.157 0.052 0.000

TE D

0.247 0.312 0.223 0.168 0.261 0.300 0.283 0.292 0.433 0.447 0.450 0.394 0.495 0.355 0.446 0.340 0.000 0.193 0.374 0.422 0.413 0.316 0.260 0.568 0.358 0.000 2005 0.163 0.110 0.072

EP

E E E E E E E C C C C C C C C W W W W W W W W W W W Region E E E

AC C

Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan Xinjiang Yunnan Chongqing Province Beijing Fujian Guangdong

0.006 0.185 0.029 0.000 0.170 0.142 -0.008 0.328 0.342 0.316 0.348 0.326 0.313 0.331 0.333 0.352 0.276 0.291 0.339 0.348 0.346 0.348 0.361 0.450 0.348 0.327 2012 0.023 0.009 0.000

0.163 0.174 0.026 0.091 0.198 0.169 0.067 0.296 0.321 0.287 0.303 0.324 0.292 0.333 0.312 0.325 0.304 0.292 0.310 0.330 0.335 0.340 0.366 0.454 0.318 0.316 2013 0.000 0.000 0.000

0.167 0.200 0.028 0.083 0.239 0.219 0.106 0.296 0.302 0.284 0.304 0.358 0.284 0.324 0.293 0.303 0.353 0.333 0.419 0.316 0.368 0.331 0.367 0.414 0.282 0.313

ACCEPTED MANUSCRIPT

0.059 0.015 0.000 0.037 0.056 0.000 0.121 0.011 0.346 0.324 0.173 0.160 0.154 0.193 0.291 0.177 0.404 0.298 0.325 0.472 0.340 0.321 0.359 0.416 0.414 0.330 0.432

0.040 0.032 0.000 0.044 0.057 0.099 0.165 0.007 0.380 0.297 0.153 0.139 0.140 0.181 0.206 0.170 0.442 0.340 0.369 0.488 0.377 0.357 0.392 0.445 0.408 0.356 0.383

32

0.026 0.031 0.000 0.023 0.022 0.093 0.034 0.002 0.432 0.445 0.271 0.255 0.388 0.358 0.423 0.278 0.463 0.380 0.408 0.497 0.400 0.391 0.405 0.392 0.404 0.387 0.428

RI PT

0.098 0.074 0.060 0.086 0.058 0.132 0.154 0.056 0.342 0.328 0.177 0.169 0.158 0.204 0.291 0.181 0.369 0.288 0.310 0.471 0.367 0.320 0.336 0.392 0.437 0.324 0.349

SC

0.074 0.073 0.063 0.055 0.031 0.144 0.179 0.051 0.326 0.318 0.223 0.218 0.239 0.242 0.280 0.231 0.390 0.295 0.304 0.463 0.361 0.320 0.333 0.403 0.443 0.326 0.380

M AN U

0.094 0.135 0.146 0.032 0.000 0.233 0.134 0.090 0.338 0.364 0.307 0.282 0.325 0.314 0.345 0.287 0.481 0.369 0.397 0.476 0.445 0.410 0.391 0.379 0.463 0.429 0.447

TE D

0.184 0.171 0.172 0.091 0.076 0.237 0.229 0.131 0.295 0.307 0.291 0.251 0.301 0.292 0.322 0.264 0.419 0.367 0.351 0.463 0.472 0.447 0.443 0.411 0.475 0.472 0.468

EP

E E E E E E E E C C C C C C C C W W W W W W W W W W W

AC C

Hainan Hebei Jiangsu Liaoning Shandong Shanghai Tianjin Zhejiang Anhui Henan Heilongjiang Hubei Hunan Jilin Jiangxi Shanxi Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan Xinjiang Yunnan Chongqing

0.011 0.161 0.000 0.193 0.017 0.043 0.018 0.011 0.487 0.472 0.306 0.314 0.377 0.352 0.445 0.309 0.450 0.399 0.415 0.506 0.402 0.392 0.405 0.418 0.402 0.408 0.434

0.000 0.372 0.000 0.061 0.000 0.000 0.000 0.041 0.512 0.511 0.340 0.383 0.424 0.388 0.503 0.383 0.392 0.351 0.464 0.511 0.372 0.374 0.491 0.288 0.450 0.358 0.420

ACCEPTED MANUSCRIPT References

AC C

EP

TE D

M AN U

SC

RI PT

Alcá ntara, V., Padilla, E., 2009. Input–output subsystems and pollution: An application to the service sector and CO2 emissions in Spain. Ecological Economics 68, 905-914. Bi, G.-B., Song, W., Zhou, P., Liang, L., 2014. Does environmental regulation affect energy efficiency in China's thermal power generation? Empirical evidence from a slacks-based DEA model. Energy Policy 66, 537-546. Butnar, I., Llop, M., 2011. Structural decomposition analysis and input–output subsystems: Changes in CO emissions of Spanish service sectors (2000–2005). Ecological Economics 70, 2012-2019. Chen, L., Jia, G., 2017. Environmental efficiency analysis of china's regional industry: a data envelopment analysis (DEA) based approach. Journal of Cleaner Production 142, 846-853. Cooper, W.W., Seiford, L.M., Tone, K., 2007. Data envelopment analysis: a comprehensive text with models, applications, references and DEA-Solver software. Second edition. Boston: Kluwer Academic Publishers. Eggleston, S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K., 2006. 2006 IPCC Guidelines for national greenhouse gas inventories volume 1: general guidance and reporting. Fan, Y., Bai, B., Qiao, Q., Kang, P., Zhang, Y., Guo, J., 2017. Study on eco-efficiency of industrial parks in china based on data envelopment analysis. Journal of Environmental Management 192, 107-115. Hu, J.-L., Wang, S.-C., 2006. Total-factor energy efficiency of regions in China. Energy Policy 34, 3206-3217. Li, L.-B., Hu, J.-L., 2012. Ecological total-factor energy efficiency of regions in China. Energy Policy 46, 216224. Liang, Q.-M., Deng, H.-M., Liu, M., 2016. Co-control of CO2 emissions and local pollutants in China: the perspective of adjusting final use behaviors. Journal of Cleaner Production 131, 198-208. Lin, B., Tan, R., 2016. Ecological total-factor energy efficiency of China’s energy intensive industries. Ecological Indicators 70, 480-497. Lin, B., Yang, L., 2013. The potential estimation and factor analysis of China's energy conservation on thermal power industry. Energy Policy 62, 354-362. Liu, H., Lin, B., 2016. Ecological indicators for green building construction. Ecological Indicators 67, 68-77. National Bureau of Statistics of the People's Republic of China (NBSC),2013. Rules of Clarification of Three Industry. Available at: www.stats.gov.cn/tjsj/tjbz/201301/t20130114_8675.html(last access: 20170801) [dataset] National Bureau of Statistics of the People's Republic of China (NBSC), 1996-2014. Statistical Yearbook of the Chinese Investment in Fixed Assets. China Statistics Press, Beijing, China. [dataset] National Bureau of Statistics of the People's Republic of China (NBSC),1996-2014. China statistical yearbook of the tertiary industry. China Statistics Press, Beijing, China. [dataset] National Bureau of Statistics of the People's Republic of China (NBSC),1996-2014. China energy statistical yearbook. China Statistics Press, Beijing, China. O’Donnell, C.J., Rao, D.S.P., Battese, G.E., 2008. Metafrontier frameworks for the study of firm-level efficiencies and technology ratios. Empirical Economics 34, 231-255. Picazo-Tadeo, A.J., Reig-Martı́nez, E., Herná ndez-Sancho, F., 2005. Directional distance functions and environmental regulation. Resource & Energy Economics 27, 131-142. Rao, X., Wu, J., Zhang, Z., Liu, B., 2012. Energy efficiency and energy saving potential in China: An analysis based on slacks-based measure model. Computers & Industrial Engineering 63, 578-584. Schlomann, B., Schleich, J., 2015. Adoption of low-cost energy efficiency measures in the tertiary sector— An empirical analysis based on energy survey data. Renewable and Sustainable Energy Reviews 43, 11271133. Shi, D., 2006. Regional differences in China's energy efficiency and conservation potentials. China Industrial Economy 10, 49–58. (in Chinese) Tone, K., 2001. A slacks-based measure of efficiency in data envelopment analysis. European Journal of Operational Research 130, 498-509. Wei, Y.M., Liao, H., Fan, Y., 2007. An empirical analysis of energy efficiency in China's iron and steel sector. Energy 32, 2262-2270. Wu, Y., 2015. China's Capital Stock Series by Region and Sector. Economics Discussion. Zhang, N., Kong, F., Yu, Y., 2015a. Measuring ecological total-factor energy efficiency incorporating regional heterogeneities in China. Ecological Indicators 51, 165-172. Zhang, N., Zhou, P., Choi, Y., 2013. Energy efficiency, CO2 emission performance and technology gaps in fossil fuel electricity generation in Korea: A meta-frontier non-radial directional distance functionanalysis. Energy Policy 56, 653-662. Zhang, W., Peng, S., Sun, C., 2015b. CO2 emissions in the global supply chains of services: An analysis based on a multi-regional input–output model. Energy Policy 86, 93-103. Zhou, P., Ang, B.W., Poh, K.L., 2006. Slacks-based efficiency measures for modeling environmental

33

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

performance. Ecological Economics 60, 111-118. Zhou, P., Ang, B.W., Wang, H., 2012. Energy and CO2 emission performance in electricity generation: A non-radial directional distance function approach. European Journal of Operational Research 221, 625635.

34