You are on page 1of 56

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324117682

Measuring unified efficiency of fossil fuel power plants across provinces in


China: An analysis based on non-radial directional distance functions

Article · March 2018


DOI: 10.1016/j.energy.2018.03.164

CITATIONS READS

4 128

4 authors, including:

Aijun Li Aizhen Zhang


Shandong University University of Jinan (Jinan, China)
20 PUBLICATIONS   398 CITATIONS    10 PUBLICATIONS   164 CITATIONS   

SEE PROFILE SEE PROFILE

Xin Yao
Xiamen University
9 PUBLICATIONS   218 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Intermediate approach View project

All content following this page was uploaded by Aijun Li on 06 April 2018.

The user has requested enhancement of the downloaded file.


Measuring unified efficiency of fossil fuel power plants across provinces in

China: An analysis based on non-radial directional distance functions

Aijun Li a, Aizhen Zhangb , Huijie Huang c,d, Xin Yao c,d*

a
Shandong School of Development, The Center for Economic Research, Shandong

University, Jinan, 250100, China.


b
School of Foreign Languages, University of Jinan, Jinan, Shandong, 250022, China.
c
China Center for Energy Economics Research, School of Economics, Xiamen

University, Xiamen, 361005, China.


d
Collaborative Innovation Center for Energy Economics and Energy Policy, Xiamen

University, Xiamen 361005, China.

This paper can be cited in the following form:

Li, A., Zhang, A., Zhou, Y., Yao, X., 2018. Measuring unified efficiency of

fossil fuel power plants across provinces in China: An analysis based on non-radial

directional distance functions. Energy. 152C, 549-561.

https://doi.org/10.1016/j.energy.2018.03.164

1
Measuring unified efficiency of fossil fuel power plants across provinces in

China: An analysis based on non-radial directional distance functions

Abstract: The static unified efficiency indices are not good at tracking unified

efficiency changes across time. To this end, this study adopts Meta-frontier Malmquist

unified efficiency indices (MMUEI). These indices have three important features: (1)

They provide us with accurate and credible results of efficiency changes originated

from a time effect. (2) They can measure unified efficiency resulting from different

benchmark frontiers (i.e. group and meta frontiers). (3) Through there decomposed

indicators, this study can uncover and identify the major sources of unified efficiency

changes. As an empirical application, the above indices are utilized for examining fossil

fuel power plants in 30 provinces in China from 2004 to 2012, where all provinces are

grouped into two groups. The main findings are summarized as follows. Firstly,

MMUEI provide new valuable information relative to UEIs. The time paths of MMUEI

form M-shaped curve for both coastal and inland regions. Secondly, different UEIs

measure unified efficiency from different perspectives and thus have differentiated time

paths. Finally, there are substantial provincial differences in terms of either MMUEI or

UEIs, resulting from provincial variations. In the short run, the provinces with large

potential should improve unified efficiency measures by “catching-up”.

Keywords: Unified efficiency; Data Envelopment Analysis (DEA); Meta-frontier non-

radial directional distance functions; Group heterogeneity; Environmental assessment

2
1. Introduction

It is important to evaluate unified efficiency changes of power plants in China and

identify the underlying driving forces, because of their contributions to CO2 emissions.

As a final energy, electricity is an important contributor to CO2 emissions. According

to IEA [1], China’s fossil fuel power plants were large energy consumers and consumed

40.49% of China’s primary energy, or equivalently 9.08% of the world’s total primary

energy in 2014. Furthermore, they were large CO2 emitters, since they contributed to

48.15% of China’s CO2 emissions, or equivalently 13.62% of the world’s CO2

emissions in 2014. Under such a recent situation, it is urgently necessary to evaluate

unified efficiency changes of power plants across time periods.

This study examines unified efficiency of fossil fuel power plants across provinces

in China. Then, the following three groups of research were highly related to this study.

The first group of research was related to the DEA (Data Envelopment Analysis)

studies on energy and environmental assessment. Methodologically, this study adopted

DEA, which has established high academic reputation as a tool for relative efficiency

assessment and environmental evaluation. Charne et al. [2] made original contributions.

Many studies made further contributions to DEA, such as Sueyoshi [3], Färe et al. [4],

Chambers et al. [5]. Recent applications can be found in Zhou et al. [6], Li et al. [7],

Wang and Sueyoshi [8], Sueyoshi and Goto [9-12], Sueyoshi and Wang [13], Sueyoshi

et al. [14, 15, 16], Zhang et al. [17] and Sueyoshi and Yuan [18, 19].

Among the DEA studies, some papers proposed several indicators measuring
3
unified efficiency and thus were closely related to this paper. Oh and Lee [20] proposed

an approach of decomposing Malmquist productivity index within the concept of meta-

frontier, which could measure technical and efficiency changes. Sueyoshi and Goto [11]

proposed a DEA approach to measure the unified efficiency of energy firms. Zhou et

al. [6] developed energy performance index (EPI) defined as the ratio of actual energy

efficiency to potential energy efficiency. Zhang et al. [21] proposed the unified

efficiency index (UEI) to measure unified efficiency of fossil fuel power plants.

The second group of research was related to group-heterogeneity. In this regard,

group-heterogeneity has to be considered, since neglecting this issue may produce

biased results. China was a large transitional and developing economy with significant

regional gaps. Thus, many studies considered group heterogeneity across regions in

China. Battese et al. (2004) [22] argued that the results would be biased, if neglecting

the potential group-heterogeneity in production technologies among decision units. Yao

et al. [20] incorporated group-heterogeneity across regions in China and investigated

the driving factors affecting energy efficiency and CO2 emissions performance in China.

Meanwhile, meta-frontier method was frequently adopted to deal with heterogeneity

issue, e.g. Wang et al. [24], Munisamy and Arabi [25] and Li and Lin [26].

The last group of research was related to the DEA studies on Chinese power plants,

as summarized in Table 1. Table 1 summarizes the following concerns. First, there have

been quite limited DEA studies about power plants in China. From 2010 to 2017, there

are only 8 articles in total. Second, for all these studies, environmental evaluation was
4
combined with operational efficiency assessment. Among them, 7 papers considered

only a single pollutant and only paper discussed three pollutants. These studies had

three types of pollutants, where most concerns focused on SO2 (6 articles), CO2 (2

articles). In this respective, SO2 are acid rain gases with serious environmental damages,

while CO2 belongs to greenhouse gases. Finally, regarding the selection of inputs, 7

articles used adopting installed capacity, 5 articles used labor, 3 articles used

expenditures, 2 articles used coal and 2 articles used fossil fuels.

To our knowledge, the static unified efficiency indices suffer from one important

shortcoming, since they are not good at tracking dynamic efficiency changes. Under

such a circumstance, this paper adopts Meta-frontier Malmquist unified efficiency

indices (MMUEI). The MMUEI-related indices have three important features: (1) These

indices can track dynamic efficiency changes and provide us with accurate and credible

results of efficiency changes originated from a time effect. (2) These indices can

distinguish unified efficiency resulting from different benchmark frontiers (i.e. group

and meta frontiers), and thus measure unified efficiency changes originated from

different sources. (3) This study uncover the potential underlying sources affecting

unified efficiency changes and identifies the major sources by comparing the three

decomposed indicators.

Additionally, this study makes empirical contributions by utilizing the above

performance indices to evaluate fossil fuel power plants across provinces in China from

2004 to 2012. So far, there has been limited existing research on environmental
5
assessment of power plants in China (See Table 1). Here, this study provides important

information on unified efficiency changes of fossil fuel power plants in China and their

underlying driving sources. Such information is quite valuable and important to policy

makers. Such an empirical investigation will be the purpose of this study.

The rest of this study is structured as follows. Section 2 provides the background

of this study. Section 3 presents the DEA model adopted in this study and the data

structure to be evaluated. Section 4 discusses the empirical results obtained from this

study. Section 5 summarizes the concluding remarks.

2. Provincial gaps in some major indicators in China

China is a large developing country with significant regional gaps. Under such

backgrounds, this section compares some important indicators across provinces.

2.1 Provincial gaps in economic and energy indicators

This subsection makes an illustration of provincial gaps in China by considering

the following important economic and energy indicators.

Firstly, this subsection considers provincial disparities in China, where disposable

income per capita across provinces in 2014 is considered as an illustration. The results

are demonstrated in Figure 1a. It can be seen that most provinces with high disposable

income per capita belong to coastal region, while those with low disposable income per

capita are located in inland region. Among all provinces, disposable income per capita

is relatively high in Shanghai (45966 RMB), Beijing (44489 RMB) and Zhejiang

(32658 RMB). By comparison, disposable income per capita is quite low in Guizhou
6
(12371 RMB), Gansu (12185 RMB) and Tibet (10730 RMB). The above data suggest

that there are significant disparities across provinces in disposable income per capita.

Figure 1

Secondly, this subsection discusses trade indicators, where trade openness across

provinces is considered as an example. Trade openness is calculated as the ratio of trade

values to GDP and is presented in Figure 1b. According to National Bureau of Statistics

of China [35], there are 8 provinces with trade openness less than 10%, including

Qinghai (4.59%), Inner Mongolia (5.04%), Hunan (7.01%), Guizhou (7.15%), Gansu

(7.77%), Shanxi (7.82%), Shaanxi (9.51%) and Hubei (9.66%). These provinces belong

to inland provinces. By contrast, trade openness is quite high in Shanghai (121.67%),

Beijing (119.76%) and Guangdong (97.61%). These provinces belong to coastal

provinces. By comparison, most inland provinces are of low trade openness, while all

coastal provinces are of high trade openness. High trade-openness implies that the

economy is highly dependent on trade, with a close trade link to rest of the world. The

above data show that there are substantial gaps in trade openness across provinces.

Thirdly, this subsection discusses energy indicators, where energy intensity is

taken as an instance. Energy intensity is the ratio of energy consumption to GDP and

the results are demonstrated in Figure 1c. According to National Bureau of Statistics of

China [35, 36], there are 5 provinces with energy intensity less than 50 gce/RMB. The

provinces include Beijing (32.03), Guangdong (43.64), Jiangsu (45.88), Zhejiang

(46.86) and Shanghai (47.03). All these provinces belong to coastal provinces. By
7
comparison, there are 4 provinces with energy intensity higher than 150 gce/RMB.

These provinces include Shanxi (155.65), Xinjiang (160.95), Qinghai (173.30) and

Ningxia (179.72). All these provinces are large energy producers or located close to

energy producers, and their economy is highly dependent on energy consumption. Then,

there are considerable gaps in terms of energy intensity across provinces.

Finally, this subsection considers an indicator on power plants, where installed

capacity of thermal power plants is taken as an example. The results are presented in

Figure 1d. According to Wind Database [37], there are substantial differences in

installed capacity across provinces in China. Some provinces have small-sized installed

capacity (e.g. Tibet, Qinghai and Beijing ) and they have either limited economy scale

(such as Tibet and Qinghai) or area scale (such as Beijing). By comparison, several

provinces show large-sized installed capacity and they are large energy consumers (e.g.

Jiangsu and Guangdong) or producers (e.g. Inner Mongolia and Henan).

As a conclusion, there are significant differences in the above indicators across

provinces. Then, provincial-variations may affect unified efficiency measures. So this

study evaluates unified efficiency of power plants at provincial levels.

2.2 Classification of provinces

This study covers 30 provinces or province-equivalents (provinces for short,

hereafter) in China, where Tibet, Hong Kong, Taiwan and Macao are excluded because

of data availability. To consider group-heterogeneity, all provinces are classified into

inland and coastal region based on geographical locations. The geographical location is
8
highly related to provincial gaps in China (see Section 2.1) and thus has been adopted

by many studies, such as Kanbur and Zhang [38] and Whalley and Zhang [39].

Coastal region consists of 12 provinces, i.e. Beijing, Fujian, Guangdong, Guangxi,

Hainan, Hebei, Jiangsu, Liaoning, Shandong, Shanghai, Tianjin and Zhejiang. Inland

region includes 18 provinces, i.e. Anhui, Henan, Heilongjiang, Hubei, Hunan, Jilin,

Jiangxi, Inner Mongolia, Shanxi, Gansu, Guizhou, Ningxia, Qinghai, Shaanxi, Sichuan,

Xinjiang, Yunnan and Chongqing.

3. The model and data

This study adopts a DEA model, after referring to Zhou et al. [6], Zhang et al. [21]

and Zhou et al. [40]. For readability, this section reports the framework of the model.

3.1 The production technology

Suppose that there are N electricity generating units (provinces). Each unit

transforms production inputs into outputs. Specifically, the production technology set

(T) can be specified as follows:

T  {( K , L, FF , EL, CM ) :(
 K , L, FF )can produce EL, CM )} ( 1 )

Where K represents installed capacity. L represents labor. FF is fossil fuels. EL

represents desirable output (electricity). CM represents undesirable output (CO2).

Meanwhile, this section differentiates three types of technology according to the

concepts of group and meta frontier in a time horizon. Suppose there are H groups and

T time periods. Consider the case of time period t and group h. For the observations of

group h, the production technology set at time period t can be specified as follows:
9
TRCh  {( K t , Lt , FF t , ELt , CM t ) :(
 K t , Lt , FF t )can produce ELt , CM t )} (2)

All-period group-frontier is defined over all time periods, covering the

observations of this group. Then, the corresponding production technology set can be

specified as follows:

TRIh  TR1h  TR2h  ...  TRTh (3)

Meta-frontier is defined over all time periods and the observations of all groups.

The production technology set is the union of all production technology sets of all time

periods and all groups.

T M  TRI  TRI2  ...  TRIH (4)

The production technology follows the standard axioms of production theory (Färe

and Grosskopf [41]). Then, T is a closed set, with finite amounts of production inputs

corresponding to finite amounts of outputs. Meanwhile, this study uses the assumptions

of weak-disposability and null-jointness, following Färe and Grosskopf [41].

Following Zhou et al. [6], Zhang et al. [21], Zhou et al. [40], Zhang et al. [42],

Zhang et al. [43], Färe et al [44] and Zhou et al. [45], the environmental production

technology T for N decision units with constant returns to scale is specified as follows.

 N N

 ( K , L , FF , EL , CM ) :  K
n n  K ,   n Ln  L, 
 n 1 n 1

 N N

T     n FFn  FF ,   n ELn  EL,  ( 5 )
 n 1 n 1 
 N 
   nCM n  CM , n  0 
 n 1 

10
3.2 Non-radial directional distance functions

For convenience of expression, D(. ; f ) represents D( K , L, FF , EL, CM ; f ) ,

and other related terms can be interpreted similarly. Following Zhang et al. [21], the

total-factor non-radial directional distance functions are specified as follows:

D(.; f )  sup{ T  : (( K , L, FF , EL, CM )  f  diag ( )) T } ( 6 )

Where  represents normalized weight vector, f represents explicit

directional vector.  represents a vector of scaling factors.

This study can obtain the NDDF value (i.e. D(.; f ) ) from the following model.

D(.; f )  max K  K   L L   FF  FF   EL K  CM  CM


N
s.t.  n K n  K   K f K
n 1
N
  n Ln  L   L f L
n 1
N
  n FFn  FF   FF f FF ( 7 )
n 1
N
  n ELn  EL   EL f EL
n 1
N
  nCM n  CM   CM f CM
n 1

 n  0, K ,  L ,  FF ,  EL ,  CM  0

If D(.; f )  0 , then this decision unit operates with the best-practice frontier in

the f direction. The direction vector f can be set according to the policy goals.

3.3 Unified efficiency index and Meta-frontier Malmquist unified efficiency index

This study develops meta-frontier Malmquist unified efficiency indices (MMUEI),

after referring to Zhang et al. [43] and Zhang et al. [21]. To this end, this study

11
differentiates three sets of production technologies following Oh and Lee [20] and

Tulkens and Vanden Eeckaut [46].

Following Zhang et al. [43], this study adopts the concepts of group- and meta-

frontier in order to explore group-heterogeneity. Suppose that there are H groups. Then,

in time period t, the non-radial directional distance functions (NDDF) within a specific

time period can be written as follows:

C
D (. ; f )  sup{ T  C : (( K , L, FF , EL, CM )  f  diag ( C ))  TRCh } ( 8 )

The NDDF, which covers all time periods, can be written as follows:

I
D (. ; f )  sup{ T  I : (( K , L, FF , EL, CM )  f  diag ( I ))  TRIh } ( 9 )

The NDDF based on meta-frontier technology can be written as follows:

M
D (. ; f )  sup{T  M : (( K , L, FF , EL, CM )  f  diag ( I )) T M } (10)

To calculate and decompose MMUEI, this study needs to solve six NDDFs.

d
D (.s ; f )  max K  Kd   L Ld   FF  FF
d
  EL EL
d
 CM  CM
d

N
s.t.  ns K ns  K   Kd f K
con
N
  ns Lsn  L   Ld f L
con
N
  ns FFns  FF   FF
d
f FF (11)
con
N
  ns ELsn  EL   EL
d
f EL
con
N
  ns CM ns  CM   CM
d
f CM ,
con

d  (C , I , M ), n  0,  K ,  L ,  FF ,  EL ,  CM  0

The con under  corresponds to the conditions of three kinds of NDDFs. If

d=C, then con  {n  Rh } and this indicates that the NDDF is defined within a specific

12
time period. If d=I, then con  {n  Rh , s  1,2,.., T } and this implies that the NDDF

covers all time periods. If d=M, then con  {n  R1  R2  ...  RH , s  [1,2,..., T ]}

and this shows that the NDDF is based on meta-frontier technology.

After referring to Sueyoshi and Goto [11], Zhou et al. [6] and Zhou et al. [40], this

study adopts static unified efficiency indices (UEI) to evaluate performance of decision

1 1 1 1 1
units. Here, the weight vector is set as  T  ( , , , , )T and the directional vector
9 9 9 3 3
is set as f  ( f K ,  f L ,  f FF , f EL ,  fCM ) .

If  s* are the optional solutions to Model (12), then UEI can be obtained as

follows:

1
1  [ Kd *   Ld *   FF
d*
  CM
d*
]
UEI (. ) 
d s 4 ,
1   ELd*

UEI g  UEI M (.s ),UEI i  UEI I (.s ),UEI c  UEI C (.s ), (12)
d  (C , I , M ), s  t , t  1

The UEIs values range from 0 to 1, where higher values correspond to better

unified efficiency. If UEI values are equal to 1, then this decision unit operates with the

best-practice technology.

Then, this study can obtain MMUEI values after the following calculation.

UEI G (.t )
MMUEI (.s )  (13)
UEI G (.t 1 )

Then, MMUEI is similar to Malmquist indices (Oh and Lee [20]).

Following Oh and Lee [20], this study performs further decomposition of MMUEI.

Then, MMUEI can be decomposed as follows:

13
UEI M (.t )
MMUEI (.s ) 
UEI M (.t 1 )
 UEI C (.t )   UEI I (.t ) / UEI C (.t )   UEI G (.t ) / UEI I (.t ) 
 C t 1  
 I t 1

C t 1   G t 1 I t 1 
UEI (. )  UEI (. ) / UEI (. )  UEI (. ) / UEI (. )  (14)
 TE t   BPR t   TGR t 
  t 1    
t 1   t 1 
 TE   BPR   TGR 
 ECI  BPR  TGR

On the right side of Eq. (18), there are three components. The first component

refers to efficiency change index (ECI), which measures changes in technical efficiency

between observations and specific-period group-frontier. Here, specific-period group-

frontier refers to group frontier at the analyzed time period (e.g. t or t-1). Then, ECI

measure efficiency changes relative to specific-period group-frontier.

The second component is named as best-practice gap ratio (BPR). This indicator

is defined as the relative changes in UEI I relative to UEI C . The difference between

UEI I and UEI C relies on the benchmark frontiers. The benchmark frontier of UEI C

is specific-period group-frontier, while the benchmark frontier of UEI C is all-period

group frontier.

The third component is named as technology gap ratio (TGR). This component is

defined as the changes in UEI G relative to UEI I , where their difference relies on

benchmark frontiers, too. Their benchmark frontiers are all-period group frontier and

meta frontier respectively.

In comparison with the existing static indices (such as UEI or EPI), MMUEI-

related indices have three important features: First, MMUEI belong to dynamic indices

and can track dynamic unified efficiency changes. Second, They can measure unified
14
efficiency resulting from different benchmark frontiers (i.e. two types of group frontiers

and one meta frontier). Under such a setting, this study can measure unified efficiency

resulting from different benchmark frontiers. (3) The three decomposed indicators can

measures changes in various unified efficiency indices. Through there decomposed

indicators, this study can identify the major sources of unified efficiency changes.

3.4 The data

This study adopts software Stata 12.0 to perform data processing, where the data

range from 2004 to 2012. Production inputs include capital (K), labor (L) and fossil

fuels (FF). Electricity generated (EL) across provinces is taken as desirable output and

CO2 emissions (CM) are measured as undesirable output.

Specifically, the data source of installed capacity of thermal power across

provinces is Wind Database [37]. Labor input is calculated based on the number of

employed persons at year-end. The data source is National Bureau of Statistics of China

[47]. Fossil fuels (FF) are calculated according to fuel consumption across provinces.

There are 12 types of fuels, including raw coal, cleaned coal, briquette, coke, coke oven

gas, crude oil, gasoline, diesel oil, fuel oil, petroleum coke, refinery gas and natural gas.

The data of fuel consumption come from National Bureau of Statistics of China [36].

Desirable output is calculated based on thermoelectricity across provinces. The

data source is Wind Database [37]. Undesirable output is derived according to the

following methods in Du [48]. Here, CO2 emissions are assumed to be the product of

fossil fuel consumption and emission factor. Table 2 reports CO2 emission factor of
15
different fossil fuels. Technically, CO2 emissions are calculated as follows:

EC   ECi   ( FFi  MFi ) (15)


i i

Where Subscript i represents fuel type. EC represents CO2 emissions. FF

represents consumption of fossil fuels. MF represents emission factor.

Table 2 and Table 3

Table 3 reports descriptive statistics of production variables. Consider mean as an

example. By comparison, installed capacity in coastal region is 1.37 times of that in

inland region, labor is 1.11 times, fossil fuels are 1.45 times, electricity generated is

1.61 times and CO2 emissions are 1.46 times. Therefore, there are considerable gaps in

production variables between coastal and inland regions. In this regard, it is necessary

and meaningful to compare unified efficiency measures of power plants across regions.

In addition, this study considers average annual growth rates of main variables

from 2004 to 2012, as illustrated in Figure 2. It can be found that, for all production

inputs and outputs, average annual growth rates in inland region are markedly higher

than those in coastal region. Consider capital input as an example. Average annual

growth rate in coastal region is 10.22%, while that in inland region is 13.50%. Thus,

power plants in inland region grow relatively rapidly, in comparison with coastal region.

Figure 2

Meanwhile, different production inputs grow unevenly. For coastal region,

average annual growth rate of capital input reaches 10.22%, while that of labor input is

1.20%. Likewise, there are similar results in inland region. Average annual growth rate
16
of capital input reaches 13.50%, while that of labor input is 2.17%. These data indicate

that there may exist significant substitution among production factors.

In addition, special attention should be given to undesirable outputs, because of

rapid growth. For coastal provinces, average annual growth rate of CO2 emissions was

12.20%, higher than that of all other variables. For inland provinces, average annual

growth rate of CO2 emissions was 13.06%. This growth rate was higher than that of

labor, fossil fuels and electricity, but slightly less than that of installed capacity.

4. Results and discussion

This section considers the results. Section 4.1 and Section 4.2 consider three

indicators of static unified efficiency. Section 4.3 and Section 4.4 discuss dynamic

indicators of unified efficiency and its decomposed indicators.

4.1 UEIs index at regional levels

Figure 3 reports the results of three unified efficiency indices (UEIs), i.e. UEIg,

UEIi and UEIc. The difference between UEIi and UEIc relies on time coverage, since

UEIi is an all-period index and UEIc is a specific-period index. Their similarity is that

both of them are calculated according to the efficiency distances between DMUs and

group-frontier. Differently, UEIg is a meta-frontier unified efficiency index and thus

calculated according to the efficiency distances between DMUs and meta-frontier. All

these three indices are bounded by zero and unity, where higher values are associated

with better performance.

According to Figure 3a, UEIg values of coastal region are significantly higher than
17
those of inland region from 2004 to 2012, implying substantial regional differences

between coastal and inland regions in terms of UEIg. Meanwhile, between coastal and

inland regions show slightly upward trends in terms of UEIg. Thus, unified efficiency

measures improve over time. Moreover, there are similar results in terms of UEIi, as

illustrated in Figure 3b. Concerning UEIc, there are mixed results across regions, as

presented in Figure 3c. UEIc values of inland region are higher than coastal region in

2004 and 2005, but become less than those in coastal region since 2007.

Figure 3

4.2 UEIs index at provincial levels

This subsection discusses UEIs at provincial level, in order to investigate the time

trends of unified efficiency measures across provinces.

4.2.1 UEIg at provincial levels

Figures 4a and 4b presents UEIg values of coastal and inland provinces

respectively.

For most coastal provinces, UEIg values range from 0.2 to 0.5, as illustrated in

Figure 4a. For these provinces, there are no clear time trends in terms of UEIg values.

In addition, three provinces (i.e. Jiangsu, Hebei and Shanghai) show large variation

across time periods. These results are not surprising, since there are large changes in

installed capacity in these provinces. Consider Jiangsu as an example. According to

Wind Database [37], there is sharp growth in installed capacity of thermal power plants

from 2004 to 2007. However, the installed capacity declines in 2008 and begin
18
increasing since 2009. Then, substantial changes in installed capacity affect unified

efficiency measures considerably. Changes in UEIg values of Hebei and Shanghai can

be explained partly in a similar way. Moreover, Shanghai and Jiangsu are among the

best-developed provinces in China and large-sized capital input can contribute to

unified efficiency improvements. In addition, Hebei is close to large coal producers

(Inner Mongolia and Shanxi) and low transportation costs help improve unified

efficiency measures.

Figure 4

As for inland provinces, UEIg values of most provinces range from 0.2 and 0.4. It

is noteworthy that there are sharp increases in UEIg values of Ningxia. These results

are caused by sharp increases in installed capacity of thermal power plants. According

to Wind Database [37], average annual growth rates reach 20.90% from 2004 to 2012

for Ningxia’s installed capacity of thermal power plants beyond 6000 kW.

4.2.2 UEIi at provincial levels

Figures 5a and 5b presents UEIi values of coastal and inland provinces

respectively.

For most coastal provinces, UEIi values range from 0.6 to 0.9. Among them,

Guangxi shows the lowest UEIi values, mainly caused by small-sized installed capacity

of thermal power plants. Meanwhile, Beijing shows large variation from 2004 to 2012,

due to small-sized installed capacity and its comparatively limited growth rate. By

comparison, Shanghai and Hebei show relatively high UEIi values.


19
Figure 5

Concerning inland provinces, UEIi values of most provinces range from 0.2 to 0.9.

Clearly, there are considerable differences across provinces. Among inland provinces,

Shanxi and Inner Mongolia show high UEIi values, since they are large coal producers

with rapid growth in installed capacity of thermal power plants. By contrast, Sichuan,

Yunnan, Chongqing and Xinjiang show low UEIi values. These results are not

surprising, since these provinces are of low GDP per capita with limited electricity

demand. Furthermore, Sichuan, Yunnan and Chongqing belong to southwest China

with limited coal resources. As a consequence, high transportation costs curb demand

for coal-fired electricity. In addition, UEIi values of Qinghai and Ningxia vary

substantially across time periods, since these two provinces have small-sized GDP with

limited installed capacity and electricity demand.

Meanwhile, it can be found that there are considerable differences in terms of UEIi

values across provinces in China, caused by province-variations. These province-

variations mainly include many factors, such as scale of economy, structure of economy,

geographical locations, energy endowments and demands. In this respective, such

information is quite important for policy makers in local governments.

In general, UEIi values are of large-variation across time periods, in comparison

with UEIg values in Figure 4. For most provinces, UEIg values remain relatively stable

across time periods. However, things are different in UEIi values, since UEIi values of

many provinces are of significant variation across time periods. Furthermore, these two
20
indicators exhibit different time trends. These results are quite interesting, since these

two indices convey different information.

4.2.3 UEIc at provincial levels

Figures 6a and 6b illustrate UEIc values of coastal and inland provinces

respectively.

From Figure 6a, it can be found that there are considerable differences in UEIc

values across provinces. Among coastal provinces, Shanghai shows the highest UEIc

values, since it is actually a well-developed populous city. By contrast, Guangxi shows

the lowest UEIc, since it is a comparatively poor province with limited urbanization. In

addition, UEIc values vary significantly for most provinces across time periods.

Meanwhile, there are similar results in Figure 6b.

Figure 6

From Figures 4-6, it can be found that there are significant differences between

three UEIs (UEIc, UEIi and UEIg) for any province, since different indicators collect

information from different perspectives. By comparison, UEIc values are of large-

variation across time periods, while UEIg values are relatively stable in time trends.

4.3 MMUEI and the decomposed components at regional levels

This subsection begins to explore MMUEI and the decomposed components at

regional level to investigate dynamic unified efficiency changes over time.

4.3.1 MMUEI values at regional levels

Figure 7a reports MMUEI values. China’s average MMUEI values are 1.0199
21
during 2004 to 2012, implying that UEI values annually increase by 1.99% on average.

Figure 7

It is interesting to note that coastal region shows similar time trends of MMUEI

values to inland region. Furthermore, the time paths of MMUEI values form M-shaped

curve. In 2005 and 2006, MMUEI values are less than unity. In 2008, MMUEI values

are less than unity, mainly caused by the subprime crisis. In 2012, MMUEI values are

less than unity, mainly caused by sharp oil price changes. By contrast, there are

considerable increases in MMUEI values in 2007, 2009, 2010 and 2011, implying

improvements in unified efficiency measures. In general, there are slow improvements

in unified efficiency measures, as illustrated in Figure 7a.

4.3.2 The decomposed indicators at regional levels

This subsection begins with ECI values, which measure unified efficiency changes

obtained according to the efficiency distances between DMUs and specific-period

group-frontier within a given time period.

Figure 7b demonstrates ECI values of coastal and inland regions from 2004 to

2012. China’s average ECI values are 0.9919, showing that its unified efficiency

measures (obtained according to specific-period group-frontier) annually decrease by

0.81%. Interesting, it can be found that ECI values remain stable before 2011, but drop

substantially in 2011 and 2012. Coastal and inland regions do not show clear downward

trends of ECI values over time. By comparison, inland region is of large-variation In

2012, there is a substantial drop in ECI values of inland region, which may be caused
22
by sharp increases in oil prices and energy prices. High energy price changes make an

impact on unified efficiency measures.

Then, this subsection considers BPR, which is calculated as the ratio of changes in

UEI I to changes in UEI C . In such a setting, BPR values incorporate two different

benchmark frontiers (i.e. specific-period and all-period group frontiers). In this way, the

changes in efficiency distances between specific-period and all-period group frontiers

(i.e. changes in group frontiers across time) can be expressed in BPR values.

Figure 7c displays BPR values of coastal and inland regions from 2004 to 2012.

China shows BPR values of 1.0471, thus implying that the relative unified efficiency

improves over time. For coastal region, BPR values vary around unity, without clear

time trends. There are similar results for inland region. By comparison, BPR values of

inland region are of large-variation across time periods.

Finally, this subsection analyzes TGR, which measures the relative changes in

UEI G relative to UEI I . TGR incorporates two benchmark frontiers (i.e. all-period

group frontier and meta frontier). So, this index can incorporate the changes in

efficiency distances between these two types of frontiers.

Figure 7d demonstrates TGR values of coastal and inland regions from 2004 to

2012. China shows average TGR values of 1.0323, thus showing there are UEI G

values increases more rapidly in comparison with UEI I . Meanwhile, there are almost

opposite time trends between coastal and inland regions.

From Figures 7b, 7c and 7d, it can be found that there are different time trends
23
between ECI values and BPR values. In the meantime, different factors affect MMUEI

disproportionally. In this regard, it is quite meaningful to decompose MMUEI, since

different components provide information from different perspectives. Furthermore,

one factor may have opposite effects on MMUEI values in different regions.

4.4 MMUEI and the decomposed components at provincial levels

This subsection considers MMUEI-related indices at provincial levels.

4.4.1 MMUEI values at provincial levels

According to Figure 8a, MMUEI values range from 0.8 to 1.2 for most coastal

provinces, with significant variation across periods. Among them, Jiangsu shows the

largest-variation across periods. These results are not surprising, since there are rapid

increases in installed capacity of thermal power plants. By contrast, MMUEI values of

Shandong are relatively stable, due to small changes in installed capacity. Concerning

inland provinces, MMUEI values are reported in Figure 8b. It can be found that MMUEI

values of most inland provinces range from 0.8 to 1.2, which is similar to those of

coastal provinces. Among all inland provinces, MMUEI values of Ningxia are of large-

variation, since this province is of small-sized economy and installed capacity. In this

respect, it can be seen that all provinces show considerable variation in MMUEI values.

Thus, it is quite useful and meaningful to capture annual changes in unified efficiency

measures and investigate the driving forces affecting their time paths across periods.

Figure 8 and Table 4

According to Table 4, it can be seen that MMUEI values of 19 provinces are greater
24
than unity and those of 11 provinces are less than unity. Among all provinces, Ningxia

shows the highest MMUEI (1.1419), indicating that its UEIg values annually grow by

14.19%. By contrast, Qinghai shows the lowest MMUEI (0.9508), implying that UEIg

values annually reduce by 4.92%. The above results show that there are substantial

differences in average MMUEI values across provinces.

4.4.2 ECI values at provincial levels

This subsection considers ECI values, which are presented in Figures 9a and 9b.

In terms of coastal provinces, ECI values of most provinces range from 0.8 to 1.3.

For all coastal provinces, there are considerable changes in ECI values across time

periods. However, there are not clear upward or downward time trends of ECI values

for any province. These results are not surprising, given that ECI values measure unified

efficiency changes according to the efficiency distances between DMU and specific-

period group-frontier. During the analysis period, there are substantial changes in many

factors, such as installed capacity of thermal power. As a result, changes in these driving

factors affect group-frontier and decision units simultaneously. Consequently, there are

considerable changes in ECI values for all coastal provinces. Moreover, there are

similar results concerning inland provinces, as illustrated in Figure 9b.

Figure 9

According to Table 4, there are 14 provinces with average ECI values greater than

unity, implying increases in unified efficiency for the observed provinces. By contrast,

ECI values of 14 provinces are less than unity, implying increasing efficiency gaps
25
between observed provinces and specific-period group-frontier. Among all provinces,

Beijing shows the highest ECI values (1.0948), indicating that its technical efficiency

measures annually rise by 9.48% relative to specific-period group-frontier. By contrast,

Yunnan shows the lowest average ECI (0.8573), thus meaning that its unified efficiency

measures annually drop by 14.27%.

4.4.3 BPR values at provincial levels

This subsection considers BPR values, which are presented in Figures 10a and 10b.

According to Figure 10a, most coastal provinces have similar time trends in BPR

values. These results imply that there are relatively small changes in group-frontiers

over time. Meanwhile, there are no clearly upward or downward time trends of BPR

values. In addition, things are similar in terms of inland provinces, as illustrated in

Figure 10b. In this regard, it is very interesting to investigate the potential driving

factors of changes in group-frontier over time.

Figure 10

According to Table 4, average BPR values of 19 provinces are greater than unity,

indicating that there are reducing efficiency gaps between specific-period and all-period

group-frontiers. By contrast, BPR values of 11 provinces are less than unity, implying

increasing efficiency gaps between specific-period and all-period group-frontiers.

Among all provinces, Chongqing shows the highest BPR values (1.2412), while Beijing

shows the lowest BPR values (0.9762). Therefore, their group-frontiers move in

opposite directions.
26
4.4.3 TGR values at provincial levels

Now, this subsection considers TGR values, as depicted in Figures 11a and 11b.

From Figure 11a, it can be seen that TGR values range from 0.8 to 1.3 for most

coastal provinces and change considerable for any province. Among them, Jiangsu and

Shanghai show large-variation in TGR values. However, there are not clearly upward

or downward time trends in TGR values for any province. In the meantime, most

provinces share similar time paths of TGR values across time periods. There are similar

results in inland provinces, as presented in Figure 11b. By comparison, TGR values of

inland provinces are of comparatively large-variation.

Figure 11

From Table 3, it can be found that TGR values of 23 provinces are greater than

unity, while those of 4 provinces are less than unity. Among all provinces, Jilin shows

the lowest TGR (0.9769) and while Ningxia shows the highest TGR (1.1291). Thus,

there are considerable differences in TGR values across provinces.

5. The concluding remarks

Currently, fossil fuel power plants consume about 40% of China’s primary energy

and emit half of China’s total CO2 from fossil fuel combustion. Against such

backgrounds, this study adopts several static unified efficiency indices (UEIs) and

dynamic unified efficiency indices (MMUEI). These indices are utilized for unified

efficiency measurement in power plants in China. This study provides important

information on unified efficiency of power plants at provincial levels. Such information


27
is valuable and useful for policy makers.

The main findings are as follows.

Firstly, the most striking finding is that MMUEI provide more valuable

information relative to UEIs. Importantly, MMUEI capture time trends of unified

efficiency. Our results show that the time paths of MMUEI form M-shaped curve for

both coastal and inland regions. By comparison, UEIs remain comparatively stable.

Secondly, it is quite meaningful to decompose MMUEI, since different

decomposed indices affect unified efficiency measures disproportionally. Furthermore,

different provinces or region show different time trends. According to the results,

technical gap ratio index is of large variation across time periods in coastal region.

Meanwhile, all three decomposed indicators are of large variation across time periods.

Thirdly, there are significant differences in terms of various UEIs across provinces

or regions. Quantitatively, there are considerable differences in their values and time

trends across provinces or regions. Thus, it is quite meaningful to differentiate these

indices, since different indices measure unified efficiency from different perspectives.

Finally, different provinces should adopt differentiated policies in order to improve

unified efficiency. According to the results, there are substantial provincial or regional

differences in terms of UEIs or MMUEI, caused by provincial or regional variations. In

terms of UEIs, fossil fuel power plants in coastal provinces perform considerably better

than those in inland provinces. However, there are mixed results in terms of MMUEI.

In this regard, this study provides important information at provincial levels, regarding
28
the driving factors affecting annual changes in unified efficiency measures. In the short

run, it is urgent for the provinces with large potential to improve unified efficiency

measures by “catching-up”.

It is true that this study is not perfect and there are some limitations that can be

overcome in future research. Firstly, there is arbitrariness in group-classification, given

that there are considerable group-heterogeneity. Secondly, there is only one undesirable

output (CO2 emissions). Other undesirable outputs can be added into modeling, such as

sulfur dioxide emissions.

Acknowledgements

The authors are grateful to the anonymous referees for their valuable suggestions.

This paper is supported by the National Natural Foundation of China (Grant No.

71403147), the Ministry of Education Research of Social Sciences Youth Funded

Projects (Grant No. 13YJC790065), Shandong Social Science Planning Fund Program

(Grant No. 12DJJJ12) and Young Scholars Program of Shandong University (Grant No.

2016WLJH02).

29
References

[1] IEA. World energy outlook 2016, Paris. Accessed from:

http://www.worldenergyoutlook.org/ publications/ weo-2016/; 2016.

[2] Charnes A, Cooper WW, Rhodes E. Measuring the efficiency of decision making

units. European Journal of Operational Research 1978; 6: 429-444.

[3] Sueyoshi T. DEA Duality on Returns to Scale (RTS) in Production and Cost

Analyses: An Occurrence of Multiple Solutions and Differences between

Production-Based and Cost-Based RTS Estimates. Management Science 1999; 45:

1593-1608.

[4] Färe R, Grosskopf S, Norris M, Zhang Z. Productivity Growth, Technical Progress,

and Efficiency Change in Industrialized Countries. American Economic Review

1994; 84: 66-83.

[5] Chambers RG, Chung Y, Färe R. Benefit and distance functions. Journal of

economic theory 1996; 70(2): 407-419.

[6] Zhou P, Ang BW, Wang H. Energy and CO2 emission performance in electricity

generation: a non-radial directional distance function approach. European Journal

of Operational Research 2012a; 221(3): 625-635.

[7] Li A, Zhang A, Zhou, Y, Yao, X. Decomposition analysis of factors affecting

carbon dioxide emissions across provinces in China. Journal of Cleaner Production

2017; 141: 1428-1444.

30
[8] Wang D, Sueyoshi T. Assessment of large commercial rooftop photovoltaic system

installations: Evidence from California. Applied Energy 2017; 188: 45-55.

[9] Sueyoshi T, Goto M. Measurement of returns to scale on large photovoltaic power

stations in the United States and Germany. Energy Economics 2017; 64: 306-320.

[10] Sueyoshi T, Goto M. Undesirable congestion under natural disposability and

desirable congestion under managerial disposability in U.S. electric power industry

measured by DEA environmental assessment. Energy Economics 2016; 55: 173-

188.

[11] Sueyoshi T, Goto M. DEA approach for unified efficiency measurement:

Assessment of Japanese fossil fuel power generation. Energy Economics 2011; 33:

292-303.

[12] Sueyoshi T, Goto M. Environmental Assessment on Energy and Sustainability by

Data Envelopment Analysis. 1-704. London: John Wiley & Sons; 2018.

[13] Sueyoshi T, Wang D. Measuring scale efficiency and returns to scale on large

commercial rooftop photovoltaic systems in California. Energy Economics 2017;

65: 389-398.

[14] Sueyoshi T, Yuan Y, Goto M. A literature study for DEA applied to energy and

environment. Energy Economics 2017; 62: 104-124.

[15] Sueyoshi T, Yuan Y, Li A, Wang D. Social Sustainability of Provinces in China: A

Data Envelopment Analysis (DEA) Window Analysis under the Concepts of

Natural and Managerial Disposability. Sustainability 2017b; 9: 2078.

Doi:10.3390/su9112078.
31
[16] Sueyoshi T, Yuan Y, Li A, Wang D. Methodological Comparison among Radial,

Non-radial and Intermediate Approaches for DEA Environmental Assessment.

Energy Economics 2017a; 67: 439-453.

[17] Zhang A, Li A, Gao Y. Social Sustainability Assessment across Provinces in China:

An Analysis of Combining Intermediate Approach with Data Envelopment

Analysis (DEA) Window Analysis. Sustainability 2018; 10(3): 732.

Doi:10.3390/su10030732.

[18] Sueyoshi T, Yuan Y. Returns to damage under undesirable congestion and

damages to return under desirable congestion measured by DEA environmental

assessment with multiplier restriction: Economic and energy planning for social

sustainability in China. Energy Economics 2016; 56: 288-309.

[19] Sueyoshi T, Yuan Y. Social Sustainability Measured by Intermediate Approach for

DEA Environmental Assessment: Chinese Regional Planning for Economic

Development and Pollution Prevention. Energy Economics 2017; 66: 154-166

[20] Oh D, Lee J. A metafrontier approach for measuring Malmquist productivity index.

Empirical economics 2010; 38(1): 47-64.

[21] Zhang N, Kong F, Choi Y, Zhou P. The effect of size-control policy on unified

energy and carbon efficiency for Chinese fossil fuel power plants. Energy Policy

2014; 70:193–200.

32
[22] Battese G, Rao D, O'Donnell C. A metafrontier production function for estimation

of technical efficiencies and technology gaps for firms operating under different

technologies. Journal of Productivity Analysis 2004; 21(1):91–103.

[23] Yao X, Zhou HC, Zhang AZ, Li AJ. Regional energy efficiency, carbon emission

performance and technology gaps in China. Energy Policy 2015; 84:142-154.

[24] Wang QW, Zhao ZY, Zhou P, Zhou DQ. Energy efficiency and production

technology heterogeneity in China: A meta-frontier DEA approach. Economic

Modelling 2013; 35: 283-289.

[25] Munisamy S, Arabi B. Eco-efficiency change in power plants: using a slacks-based

measure for the meta-frontier Malmquist–Luenberger productivity index. Journal

of Cleaner Production 2015; 105: 218-232.

[26] Li JL, Lin BQ. Ecological total-factor energy efficiency of China's heavy and light

industries: Which performs better? Renewable and Sustainable Energy Reviews

2017; 72: 83-94.

[27] Zhou Y, Xing X, Fang K, Liang D, Xu C. Environmental efficiency analysis of

power industry in China based on an entropy SBM model. Energy Policy 2013; 57:

68-75.

[28] Zhang N, Choi Y. Total-factor carbon emission performance of fossil fuel power

plants in China: A metafrontier non-radial Malmquist index analysis. Energy

Economics 2013; 40: 549-559.

33
[29] Zhang N, Choi Y. A comparative study of dynamic changes in CO2 emission

performance of fossil fuel power plants in China and Korea. Energy Policy 2013;

62: 324-332.

[30] Mou D. Understanding China’s electricity market reform from the perspective of

the coal-fired power disparity. Energy Policy 2014; 74: 224-234.

[31] Liu X, Chu J, Yin P, Sun J. DEA cross-efficiency evaluation considering

undesirable output and ranking priority: a case study of eco-efficiency analysis of

coal-fired power plants. Journal of Cleaner Production 2017; 142: 877-885.

[32] Ji X, Li G, Wang Z. Allocation of emission permits for China’s power plants: A

systemic Pareto optimal method. Applied Energy 2017; 204: 607-619.

[33] Ji X, Li G, Wang ZH. Impact of emission regulation policies on Chinese power

firms’ reusable environmental investments and sustainable operations. Energy

Policy 2017; 108: 163-177.

[34] Bi G, Shao Y, Song W, Yang F, Luo Y. A performance evaluation of China's coal-

fired power generation with pollutant mitigation options. Journal of Cleaner

Production 2018; 171: 867-876.

[35] National Bureau of Statistics of China. China Statistical Yearbook 2015. China

Statistics Press. Beijing. China; 2015a.

[36] National Bureau of Statistics of China. China Energy Statistical Yearbook 2005-

2013. China Statistics Press. Beijing. China; 2005-2015b.

[37] Wind Database. http://www.wind.com.cn/; 2016 [accessed 16/01/06].


34
[38] Kanbur R, Zhang XB. Which regional inequality? The evolution of rural–urban

and inland–coastal inequality in China from 1983 to 1995. Journal of Comparative

Economics 1999; 27: 686–701.

[39] Whalley J, Zhang SM. A numerical simulation analysis of (Hukou) labor mobility

restrictions in China. Journal of Development Economics 2007; 83(2): 392-410.

[40] Zhou P, Ang BW, Zhou DQ. Measuring economy-wide energy efficiency

performance: a parametric frontier approach. Applied Energy 2012b; 90(1): 196-

200.

[41] Färe R, Grosskopf S. New Directions: Efficiency and Productivity. Springer. New

York; 2005.

[42] Zhang M, Liu X, Wang WW, Zhou M. Decomposition analysis of CO2 emissions

from electricity generation in China. Energy Policy 2013a; 52: 159–165.

[43] Zhang N, Zhou P, Choi Y. Energy efficiency, CO2 emission performance and

technology gaps in fossil fuel electricity generation in Korea: A meta-frontier non-

radial directional distance function analysis. Energy policy 2013b, 56: 653-662.

[44] Färe R, Grosskopf S, Pasurka CA. Environmental production functions and

environmental directional distance functions. Energy 2007; 32(7): 1055-1066.

[45] Zhou P, Ang BW, Han JY. Total factor carbon emission performance: a Malmquist

index analysis. Energy Economics 2010; 32(1): 194-201.

35
[46] Tulkens H, Vanden Eeckaut P. How to measure efficiency and productivity with

special reference to banking. Draft Report. CORE. Université Catholique de

Louvain, Louvain-la-Neuve. Belgium; 1995.

[47] National Bureau of Statistics of China. China Labour Statistical Yearbook 2005-

2013. China Statistics Press. Beijing. China; 2005-2013a.

[48] Du, L. M. Impact Factors of China’s Carbon Dioxide Emissions: Provincial Panel

Data Analysis. South China Journal of Economics 2010; 11: 20-33. (in Chinese)

[49] IPCC. IPCC guidelines for national greenhouse gas inventories (2006).

http://wenku.baidu.com/link?url=gcFmt_h2PddckW3nViNAaWYmgFH0xgPKlg

ysRQgzXtDnAH2e1tg8tMWz3iUkAPv46x1f7jMcsA3UEfYBENgdO6fkyyoFKz

s6vHrCZ296K1y; 2006.

36
Figure 1a Disposable income across provinces in China

Figure 1b Ratio of trade values to GDP across provinces in China

37
Figure 1c Energy intensity across provinces in China

Figure 1d Installed capacity of thermal power plants across provinces in China

Figure 1 Provincial gaps in major indicators of China in 2014

Sources: Wind Database [33] and National Bureau of Statistics of China [31, 32].
38
Please note: The figures do not provided detailed information about China’s map, since

several regions are not covered (e.g. Nanhai).

39
15
coastal provinces
inland provinces
Annual growth rate (%)

10

0
K L FF EL CM

Figure 2 Average annual growth rates of production inputs and outputs

Sources: National Bureau of Statistics of China [31, 32, 43] and IPCC [45].

40
a
0.6

UEIg of C-P
UEIg of I-P
0.5
UEIg

0.4

0.3

0.2
2004 2005 2006 2007 2008 2009 2010 2011 2012

b 1.0
UEIi of C-P
0.9 UEIi of I-P

0.8

0.7
UEIi

0.6

0.5

0.4

0.3
2004 2005 2006 2007 2008 2009 2010 2011 2012

0.9

0.8
UEIc

0.7 UEIc of C-P


UEIc of I-P

0.6

2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 3 UEI indices across regions in China

41
a
1.0
Beijing
Fujian
Guangdong
0.8 Guangxi
Hainan
Hebei
UEIg of coastal provinces

Jiangsu
Liaoning
0.6 Shandong
Shanghai
Tianjin
Zhejiang
0.4

0.2

0.0
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
1.0 Anhui
Henan
Heilongjiang
Hubei
Hunan
0.8 Jilin
Jiangxi
UEIg of inland provinces

Inner
Mongolia
Shanxi
0.6 Gansu
Guizhou
Ningxia
Qinghai
Shaanxi
Sichuan
0.4 Xinjiang
Yunnan
Chongqing

0.2

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 4 UEIg of provinces in China from 2004 to 2012

42
a
1.0
Beijing
Fujian
Guangdong
0.9 Guangxi
Hainan
UEIi of Coastal provinces

Hebei
Jiangsu
0.8
Liaoning
Shandong
Shanghai
0.7 Tianjin
Zhejiang

0.6

0.5

0.4
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
1.0 Anhui
Henan
Heilongjiang
Hubei
Hunan
0.8 Jilin
UEIi of inland provinces

Jiangxi
Inner
Mongolia
Shanxi
0.6 Gansu
Guizhou
Ningxia
Qinghai
Shaanxi
0.4 Sichuan
Xinjiang
Yunnan
Chongqing
0.2

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 5 UEIi of coast and inland regions from 2004 to 2012

43
a
1.0
Beijing
Fujian
Guangdong
0.9 Guangxi
Hainan
UEIc of coastal provinces

Hebei
Jiangsu
0.8 Liaoning
Shandong
Shanghai
0.7 Tianjin
Zhejiang

0.6

0.5

0.4
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
1.0 Anhui
Henan
Heilongjiang
Hubei
Hunan
0.8 Jilin
Jiangxi
UEIc of inland provinces

Inner
Mongolia
Shanxi
0.6 Gansu
Guizhou
Ningxia
Qinghai
Shaanxi
0.4 Sichuan
Xinjiang
Yunnan
Chongqing

0.2

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 6 UEIc of coast and inland regions from 2004 to 2012

44
a
1.2
MMUEI of C-P
MMUEI of I-P
1.1
MMUEI

1.0

0.9

0.8
2004 2005 2006 2007 2008 2009 2010 2011 2012

b
1.2

1.1

1.0
ECI

0.9
ECI of C-P
ECI of I-P
0.8

0.7
2004 2005 2006 2007 2008 2009 2010 2011 2012

c
1.6

1.5
BPR of C-P
1.4 BPR of I-P

1.3
BPR

1.2

1.1

1.0

0.9

0.8
2004 2005 2006 2007 2008 2009 2010 2011 2012

45
d
1.3

TGR of C-P
1.2
TGR of I-P
TGR

1.1

1.0

0.9
2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 7 MMUEI and the decomposed indices across regions in China

46
a
Beijing
Fujian
1.2 Guangdong
Guangxi
Hainan
MMUEI of coastal provinces

Hebei
Jiangsu
1.0 Liaoning
Shandong
Shanghai
Tianjin
Zhejiang
0.8

0.6

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
Anhui
1.6 Henan
Heilongjiang
Hubei
Hunan
Jilin
MMUEI of inland provinces

1.4 Jiangxi
Inner
Mongolia
Shanxi
Gansu
1.2 Guizhou
Ningxia
Qinghai
Shaanxi
1.0 Sichuan
Xinjiang
Yunnan
Chongqing

0.8

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 8 MMUEI across provinces in China

47
a
1.5

Beijing
1.4
Fujian
Guangdong
Guangxi
1.3
Hainan
Hebei
ECI of coastal provinces

1.2 Jiangsu
Liaoning
Shandong
1.1 Shanghai
Tianjin
Zhejiang
1.0

0.9

0.8

0.7
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
1.6

Anhui
1.4 Henan
Heilongjiang
Hubei
Hunan
1.2
Jilin
ECI of inland provinces

Jiangxi
Inner Mongolia
1.0 Shanxi
Gansu
Guizhou
0.8 Ningxia
Qinghai
Shaanxi
Sichuan
0.6
Xinjiang
Yunnan
Chongqing
0.4

0.2
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 9 ECI across provinces in China

48
a
1.3

Beijing
Fujian
1.2 Guangdong
Guangxi
Hainan
Hebei
BPR of coastal provinces

1.1 Jiangsu
Liaoning
Shandong
Shanghai
1.0 Tianjin
Zhejiang

0.9

0.8

0.7
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b
3.0
Anhui
Henan
Heilongjiang
2.5 Hubei
Hunan
BPR of inland provinces

Jilin
Jiangxi
2.0 Inner Mongolia
Shanxi
Gansu
Guizhou
1.5 Ningxia
Qinghai
Shaanxi
Sichuan
1.0 Xinjiang
Yunnan
Chongqing

0.5

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 10 BPR across provinces in China

49
a
1.8

1.6
Beijing
Fujian
Guangdong
1.4 Guangxi
TGR of coastal provinces

Hainan
Hebei
1.2 Jiangsu
Liaoning
Shandong
1.0 Shanghai
Tianjin
Zhejiang

0.8

0.6

0.4
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

b 1.8
Anhui
Henan
1.6 Heilongjiang
Hubei
Hunan
1.4 Jilin
TGR of inland provinces

Jiangxi
Inner Mongolia
Shanxi
1.2 Gansu
Guizhou
Ningxia
Qinghai
1.0
Shaanxi
Sichuan
Xinjiang
0.8 Yunnan
Chongqing

0.6

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Figure 11 TGR across provinces in China

50
Table 1: Recent DEA studies on China’s power plants

Research Coverage Variables of production a

Y B X

Zhou et al. [27] Power industry EL SO2, NOx, CO2 L, Capital,

Energy

Zhang and Fossil fuel power plants EL CO2 K, L, FF

Choi [28]

Zhang and Fossil fuel power plants, EL CO2 K, L, FF

Choi [29] China and Korea

Mou [30] Coal-fired power plants EL SO2 K, L, C

Liu et al. [31] Coal-fired power plants EL SO2 K, OE

Ji et al. [32] Coal-fired power plants EL SO2 K, OE

Ji et al. [33] Coal-fired power plants EL SO2 K, OE, DE

Bi et al. [34] Coal-fired power plant EL SO2 K, L, C

a
Note: EL, K, C, FF, OE and DE stand for electricity, installed capacity, coal,

desulphurization expenditure and operating expenditure respectively. Meanwhile, Y, B,

and X stand for desirable outputs, undesirable outputs and production inputs separately.

51
Table 2 Carbon dioxide emission factor by fuel type (in kg CO2/GJ)

Fuel type raw coal cleaned coal briquette coke

Emission factor 94.6 98.3 97.5 107.07

Fuel type coke oven gas crude oil gasoline diesel oil

Emission factor 44.37 73.3 69.3 74.07

Fuel type fuel oil petroleum coke refinery gas natural gas

Emission factor 77,400 97,500 66.73 56.1

Sources: IPCC [49].

52
Table 3 Descriptive statistics of input and output variables

Variable Unit Groups Mean Standard Maximum Minimum

deviation

K 10 MW C-P a 2320.1 1950.0 7080.2 153.1

I-P 1694.4 1351.7 6015.8 88.9

L Persons C-P 85106.8 48914.0 171614 12426

I-P 76626.9 34889.3 176359 10346

FF 104 tce C-P 4307.5 3443.1 13401.5 141.7

I-P 2975.1 2458.3 15299.4 147.7

EL 108 C-P 1207.8 936.7 3651.4 56.6

kWh I-P 751.8 578.2 2747.9 55.6

CM 104 t C-P 12946.4 11822.7 50685.6 385.2

I-P 8893.8 7500.9 45069.8 86.1

Note: a C-P represents coast region and I-P represents inland region.

Sources: Wind Database [33] and National Bureau of Statistics of China [36, 47].

53
Table 4 Average MMUEI and the decomposed components of provinces in China

Provinces or Regions MMUEI ECI BPR TGR

Beijing 1.0284 1.0948 0.9762 0.9831

Fujian 1.0137 1.0100 0.9969 1.0058

Guangdong 1.0210 1.0211 0.9966 1.0077

Guangxi 1.0378 1.0327 0.9912 1.0304

Hainan 1.0674 0.9900 1.0102 1.0832

Hebei 1.0126 1.0000 1.0125 0.9998

Jiangsu 1.0808 1.0439 1.0025 1.0297

Liaoning 0.9900 0.9821 0.9938 1.0213

Shandong 1.0214 0.9992 0.9996 1.0302

Shanghai 0.9871 1.0090 0.9893 1.0150

Tianjin 1.0062 0.9943 0.9945 1.0215

Zhejiang 1.0530 1.0474 0.9882 1.0193

Anhui 1.0686 1.0118 1.0438 1.0421

Henan 1.0224 1.0176 1.0102 1.0025

Heilongjiang 1.0021 1.0147 1.0135 1.0020

Hubei 0.9921 0.9743 0.9945 1.0468

Hunan 0.9783 0.9963 1.0064 1.0090

Jilin 0.9816 1.0290 1.0428 0.9769

Jiangxi 0.9904 1.0470 0.9781 0.9887


54
Inner Mongolia 1.0358 0.9794 1.0008 1.0677

Shanxi 1.0576 1.0000 1.0017 1.0593

Gansu 0.9913 0.9109 1.1416 1.0885

Guizhou 0.9817 0.9124 1.1351 1.0881

Ningxia 1.1419 1.0020 1.1398 1.1291

Qinghai 0.9508 0.8926 1.0978 1.1284

Shaanxi 1.0381 0.9745 1.1481 1.0334

Sichuan 0.9934 1.0005 1.0507 1.0000

Xinjiang 1.0483 0.9543 1.2217 1.0000

Yunnan 0.9940 0.8573 1.2019 1.0589

Chongqing 1.0098 0.9578 1.2412 1.0000

Coast region 1.0266 1.0187 0.9960 1.0206

Inland region 1.0155 0.9737 1.0814 1.0401

China 1.0199 0.9919 1.0471 1.0323

55

View publication stats

You might also like