Ecological total-factor energy efficiency of regions in China

(1)

Ecological total-factor energy efﬁciency of regions in China

Lan-Bing Li

a

, Jin-Li Hu

b,n

a

Institute of Urban and Regional Economics, Nankai University, China

b

Institute of Business and Management, National Chiao Tung University, 118, Sec. 1, Chung-Hsiao W. Rd., Taipei City 10044, Taiwan

a r t i c l e

i n f o

Article history:

Received 10 November 2011 Accepted 20 March 2012 Available online 18 April 2012 Keywords:

Slack-based measure (SBM) Ecological total-factor energy Undesirable outputs

a b s t r a c t

Most existing energy efﬁciency indices are computed without taking into account undesirable outputs such as CO2and SO2 emissions. This paper computes the ecological total-factor energy efﬁciency

(ETFEE) of 30 regions in China for the period 2005–2009 through the slack-based model (SBM) with undesirable outputs. We calculate the ETFEE index by comparing the target energy input obtained from SBM with undesirable outputs to the actual energy input. Findings show that China’s regional ETFEE still remains a low level of around 0.600 and regional energy efficiency is overestimated by more than 0.100 when not looking at environmental impacts. China’s regional energy efficiency is extremely unbalanced: the east area ranks first with the highest ETFEE of above 0.700, the northeast and central areas follow, and the west area has the lowest ETFEE of less than 0.500. A monotone increasing relation exists between the area’s ETFEE and China’s per capita GDP. The truncated regression model shows that the ratio of R&D expenditure to GDP and the degree of foreign dependence have positive impacts, whereas the ratio of the secondary industry to GDP and the ratio of government subsidies for industrial pollution treatment to GDP have negative effects, on the ETFEE.

1. Introduction

China has created an economic growth miracle in the past 30 odd years, as its GDP has grown by almost 110 times from 1978 to 2010. In 2010, China surpassed Japan to rank second in the world with a GDP of 39.8 trillion RMB (National Bureau of Statistics of China, 2011). China in 2010 also became the world’s biggest energy consumer with a whopping 20.3% share of global energy use (BP, 2011). Obviously, its economic growth is driven by huge energy consumption, which is not sustainable. It is very impor-tant for China to now improve energy efﬁciency without harming economic performance due to the following three reasons.

First, China’s energy supply is becoming increasingly insufﬁ-cient. Taking crude oil as example, its import dependence went from a low of 6% in 1993 to a high of 55.2% in 2011 (Ministry of Industry and Information Technology of the People’s Republic of China, 2011). Coal imports increased rapidly as well from 2.18 million tons in 2000 to 125.84 million tons in 2009 (China Energy Statistical Yearbook, 2010). China’s energy demand continues to grow and is showing signs of a shortage in the long term (Liu, 2011).

Second, there is a big gap between China and developed countries on energy consumption per unit of GDP. Although the gap is gradually narrowing, China’s level is 4.5 times that of Japan and almost 2.8 times that of the U.S. in 2008. An important cause for such a gap is its low energy efﬁciency. China’s transformation of its economic growth model and an upgrade to its industrial structure are becoming quite imperative.

Third, pollution is a severe constraint to economic growth in China. The economic loss caused by pollution has reached 1.4 trillion RMB, which accounted for 3.9% of GDP in 2008 (China Environmental and Economic Accounting Report, 2010). Accord-ing to China’s commitments in 2009, its CO2emissions per unit of GDP should decrease 40%–45% by 2020 compared to the level in 2005. However, total energy consumption will not fall due to continued stable economic growth. Energy efﬁciency improve-ment is the major solution to the energy problem and is a major goal that China is pursuing for the sake of current and future economic growth (Shi, 2007).

Energy efficiency plays an important role in economic devel-opment, attracting more and more academic evaluations that use two different methods: one is partial-factor energy efficiency, and the other is total-factor energy efficiency. Based on the partial-factor evaluation framework, the most popular indices include the energy intensity and energy productivity ratios. Energy intensity refers to the ratio of energy input to GDP, while energy produc-tivity is the reciprocal of the energy input: the GDP ratio. Both indices only take energy into account as a single input to produce Contents lists available atSciVerse ScienceDirect

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

Energy Policy

n

Corresponding author. Tel.: þ886 2 23812386x57641; fax: þ 886 2 23494922.

E-mail address: jinlihu@mail.nctu.edu.tw (J.-L. Hu). URL: http://jinlihu.tripod.com (J.-L. Hu).

(2)

GDP while neglecting other key inputs such as capital and labor. There has been widespread criticism of using energy intensity for measuring energy efﬁciency (Patterson, 1996). The main problem with energy/GDP, as pointed out byWilson et al. (1994), is that it does not measure the underlying technical energy efﬁciency, which can present misleading conclusions.

Proposing a total-factor framework, Hu and Wang (2006)

initially put forward the TFEE (total-factor energy efficiency) index, which is constructed as the ratio of the target energy input suggested by DEA to the actual energy input.Hu and Kao (2007)calculated energy-saving target ratios for 17 APEC econo-mies during 1991–2000 by the TFEE index and pointed out that China has the largest energy-saving target at up to almost half its current usage. They further found that the energy-saving target ratio has a positive relation with the percentage of the added value in the industry sector to GDP and a negative relation with that of the service sector.Hu et al. (2006)established an index containing a water adjustment target ratio based on the TFEE, discovering a U-shape relation between the total-factor water efficiency and per capita real income among areas in China with the central area having the worst water efficiency ranking.

Honma and Hu (2008)computed the regional TFEE in Japan for the period 1993–2003 and found that the inland regions and most regions along the Sea of Japan are efficient in energy use, whereas most of the inefficient prefectures that were developing mainly upon energy-intensive industries are located along the Pacific Belt Zone.Lee et al. (2011)computed the three major types of efficient electricity, coal, and gasoline oil savings for 27 regions in China during 2000–2003 by a total factor framework and presented that the east area contains most of the efficient regions with respect to the three major types of energy in every year during the research period. Zhang et al. (2011) employed TFEE to investigate energy efficiency in 23 developing countries during the period 1980–2005 and indicated that Botswana, Mexico, and Panama perform the best in terms of energy efficiency, whereas Kenya, Sri Lanka, Syria, and the Philippines perform the worst during the entire research period. The TFEE is capable of measuring energy efficiency in a total-factor framework, but only takes GDP into account as the single output while neglecting undesirable outputs. However, GDP cannot be produced alone from the use of energy with other inputs, and environmental pollution is an undesirable and unavoidable by-product of GDP output. Global warming is a serious environmental problem in the world and has been a growing concern in recent years (e.g.,Radhi, 2009; Cha et al., 2008; Fearnside, 2002). As such, a sustainable framework should be proposed to assess energy effi-ciency much more accurately, and environmental impacts should be incorporated — that is, not only the desirable GDP, but also the undesirable CO2and SO2should be taken as outputs. Following the sustainable framework, this paper puts forward a new index of energy efficiency and names it the ecological total-factor energy efficiency (ETFEE). The ETFEE is constructed as the ratio of the target energy input suggested from the SBM model with undesirable outputs to the actual energy input in a region.

The purposes of this paper are as follows. The first is to innovatively construct the index of ETFEE based on an SBM model with undesirable outputs. The second is to distinguish the difference between ecological total-factor energy efficiency and traditional total-factor energy efficiency and to comprehen-sively evaluate regional energy efficiency in China by using the ETFEE index. The third is to clarify the discrepancy of ETFEE among different areas (including east, central, west, and northeast areas) in China. The final purpose is to identify the influential factors of regional ETFEE.

This paper is organized as follows.Section 2reviews the SBM and an undesirable output model based on SBM and then describes how the index of ETFEE is constructed.Section 3presents the data source,

variable definitions, and descriptive statistics.Section 4provides an empirical study for the energy efficiency of regions in China based on ETFEE. The final section concludes the paper.

2. Methodology: Ecological total-factor energy efﬁciency (ETFEE) Under the background of energy scarcity and environment pollution, it is preferable for a region to increase GDP while reducing pollution and energy consumption. This section pro-poses the ETFEE index on the viewpoint of sustainability, which is calculated by an undesirable model based on SBM. This model takes energy in conjunction with labor and capital stock as inputs, while taking not only GDP, but also undesirable CO2and SO2as outputs.

2.1. Slacks-based measure of efﬁciency (SBM)

Built upon the earlier work of Farrell (1957), DEA is a well established methodology to evaluate the relative efﬁciencies of a set of comparable entities by some speciﬁc mathematical pro-gramming models (Zhou et al., 2008). The SBM was introduced by

Tone (2001) from the basic CCR-DEA (Charnes et al, 1978) and BCC-DEA (Banker et al., 1984) models. Cooper et al. (2007)

pointed out that SBM has two important properties. First, the measure is invariant with respect to the unit of measurement of each input and output item. Second, the measure is monotone decreasing in each input and output slack. Moreover, the measure generally has no strict requirements with the input and output prices, does not impose any particular functional form on the data, and creates a more ﬂexible piecewise linear function, and so it can be used to analyze the complex production process with multi-inputs and multi-outputs.

The fractional programming problem of the constant-returns to scale (CRS) SBM model is expressed as follows:

min l,sþ_,s

r

¼ 1ð1=mÞPmi ¼ 1si=xio 1 þ ð1=nÞPnr ¼ 1s þ r =yro subject to xo¼X

l

þS yo¼Y

l

S þ

l

Z0,sZ0, sþZ0, ð1Þ

where each region has m inputs and n outputs; xo, yo, si, and srþ

represent the input, output, input slack, and the output slack for the oth region, respectively; X, Y, S_{, and S}þ _{are the} correspond-ing matrices of the input, output, input slack, and output slack; and

l

is a constant vector. The obtained value of

r

is the overall technical efﬁciency score for the oth DMU.

2.2. Undesirable output model: Based on SBM

Consider a production process in which desirable and undesir-able outputs are jointly produced. Assume that X A Rmþ, Y

g

ARn_þ1, and Yb

ARn2

þ are the vectors of inputs, desirable outputs, and

undesirable outputs, respectively. The production technology can be described as:

T ¼ fðx,yg_,yb

Þ: x can produceðyg,yb_Þg: _ð2Þ

In order to reasonably model a production process in which both desirable and undesirable outputs are jointly produced, F ¨are et al. (1989) proposed two assumptions on the production technology. The ﬁrst is that outputs are weakly disposable; i.e., if ðx,yg_,yb_{Þ AT and 0}_r

_y

_{r1, then ðx,}

_y

_yg_,

_y

_yb_{Þ AT. The second}

is that desirable and undesirable outputs are null-joint; i.e., if ðx,yg_,yb_{Þ A}_{T and y}b_¼_{0, then y}g_¼_{0. The ﬁrst assumption implies}

that the reduction of undesirable outputs is not free, and the proportional reduction in desirable and undesirable outputs at

(3)

the same time is feasible. This second assumption implies that some undesirable outputs must also be produced when desirable outputs are produced.

The production possibility set that satisﬁes the above-men-tioned assumptions can be expressed as follows:

PðxÞ ¼ fðyg_,yb_Þ_{: X Z x, y}g_rYg_l_{, y}b_ZYb_l_, _l

Z0, X 4 0, Yg40, Yb40g, ð3Þ where

l

is a constant vector representing the weight of each DMU. The inequalities of the input and desirable output imply that the input and desirable output are strongly disposable. With the inequality of undesirable output considered, the desirable and undesirable outputs are weakly disposable. The inequalities of Yg40 and Yb40 ﬁt the assumption that desirable and undesir-able outputs are null-joint. Notice that the above deﬁnition corresponds to the constant returns to scale technology.

In accordance with the above-mentioned production technology,

Cooper et al. (2007)proposed the CRS-SBM model with undesirable outputs to calculate the technical efﬁciency of a production system in which desirable and undesirable outputs are jointly produced. The fractional programming problem solved by the CRS-SBM model with undesirable outputs for region i is as follows:

min l,s_,sg,sb

r

¼ 1ð1=mÞPmi ¼ 1si =xio 1 þ ð1=n1þn2ÞðPnr ¼ 11 ðs g r=y g roÞ þP n2 r ¼ 1ðsbr=ybroÞÞ ð4Þ subject to xo¼X

l

þS ð5Þ yg o¼Y g

l

Sg ð6Þ yb o¼Y b

_l

_þSb _ð7Þ

l

Z0, s Z0, sgZ0, sbZ0, X 40, Y 4 0, ð8Þ

where each region has m inputs, n1 good outputs, and n2 bad outputs; X, Yg_{, and Y}b_{are the matrices of the input, good output, and} bad output, respectively, while all of X, Yg_{, and Y}b_{are strictly larger} than zero; S

, Sg, and Sbare the matrices of the input, good output, and bad output slacks, respectively; and

l

is a constant vector.

Eq. (5) imposes strong disposability of inputs. Eq. (6) ensures the desirable output satisﬁes strong disposability. Eq. (7) depicts that undesirable outputs are weakly disposable. The inequalities X 4 0 and Y 4 0 imply that desirable and undesirable outputs are null-joint and the undesirable output is a byproduct of the desirable output. The computed value of

r

is the overall technical efﬁciency score for the oth region with the inclusion of undesirable outputs. 2.3. Regional ETFEE with undesirable outputs considered

Energy efficiency for region i at time t can be defined as below, which is called ecological total-factor energy efficiency (ETFEE) since it is established on the viewpoint of total factor productivity and sustainable development under the consideration of undesirable outputs.

ETFEEði,tÞ ¼ Target energy inputði,tÞ

Actual energy inputði,tÞ: ð9Þ

The target energy input for each region is obtained from SBM with the undesirable outputs accounted for, which is deﬁned as: Target energy inputði,tÞ ¼ Actual energy inputði,tÞ

2Total energy input slackði,tÞ ð10Þ In other words, the target energy input is the projection on the energy axis when a DMU improves and reaches the efﬁcient frontier. The gap between the target level and the actual level is named total energy input slack, which is regarded as the inefﬁ-cient portion of actual energy consumption. The ETFEE score is

always between zero and one. If the target energy input is equal to the actual level, then the ETFEE is one, indicating the highest efﬁciency level of energy consumption. If the actual energy input is much higher than the target, then the index approaches zero, indicating very low efﬁciency.

2.4. ETFEE for an area

The ETFEE index can be used to evaluate energy efﬁciency not only in a region, but also in an area or for a country consisting of several regions. The ETFEE in an area is equal to the total target energy inputs divided by the total actual energy input of the area. Assuming area a covers q regions, the ETFEE of area a at time t can be calculated as:

ETFEEða,tÞ ¼ P

q A aTarget energy inputðq,tÞ

P

q A aActual energy inputðq,tÞ

: ð11Þ

3. Data source, variable deﬁnitions, and descriptive statistics There are 31 provinces, autonomous regions, and municipalities in mainland China: Beijing, Tianjin, Hebei, Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Hai-nan, Chongqing, Sichuan, Guizhou, YunHai-nan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, and Tibet. Due to serious absence of data, Tibet is excluded. The SBM model with undesirable outputs is applied to the other 30 provinces, autonomous regions, and municipalities in China from 2005 to 2009. This paper employs total energy consumption, capital stock, and labor as inputs, and desirable GDP is accompanied by undesirable CO2and SO2as outputs.

All the data for GDP, labor, and SO2emission come from China Statistical Yearbook. GDP as a monetary output is transformed into 2005 prices with a GDP deﬂator. The total energy consumption is collected from China Energy Statistical Yearbook.

Capital stock is not available in any China statistical data, but it can be calculated as follows. First, we get the capital stock in 2000 fromZhang et al. (2004)and capital formation and capital price indices from China Statistical Yearbook. Second, the capital stock in 2000 and capital formation from 2001 to 2009 should be transformed into 2005 prices with capital price indices. Third, we consider a capital depreciation rate. The capital stock from 2001 to 2009 with 2005 prices can then be calculated, taking 2000 as the starting year. Finally, the capital stock from 2005 to 2009 is taken as the sample data.

Capital Stock in Current Year ¼ Capital Stock in Previous Year ð1-Depreciation rateÞ þCapital Formation in Current Year:

ð12Þ There is no available ofﬁcial data on regional CO2 emission in China. However, we can obtain detailed energy consumptions (including fuel oil, coke coal, gasoline, kerosene, diesel oil, and natural gas liquids) from China Energy Statistical Yearbook and net caloriﬁc value and the effective CO2 emission factor for each kind of energy from 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC, 2006). The CO2 emission of region i at time t is estimated as:

Regional CO2emissioni,t

¼X

n

j ¼ 1

ðEnergy consumptionit,jNet calorific valuej

Effective CO2 emission factorjÞ ð13Þ

(4)

This paper investigates the relations between four inﬂuential factors, including the ratio of intramural expenditure on R&D to GDP, the ratio of the secondary industry to GDP, the foreign dependence degree which is the ratio of total imports and exports to GDP, and the ratio of government subsidies for industrial pollution treatment to GDP and the ETFEE scores. The data of intramural expenditure on R&D are from China Statistical Yearbook on Science and Technology. The added value of the secondary industry, the total imports and exports, and the government sub-sidies for industrial pollution treatment are from China Statistical Yearbook. To sum up, there are 150 observations in all and the descriptive statistics of the original data are portrayed inTable 1.

4. Empirical analysis

4.1. Comparison of ETFEE and TFEE

The essential difference between total-factor energy efficiency and ecological total-factor energy efficiency is whether to incor-porate the environmental impacts. The ecological total-factor energy efficiency is based on the viewpoint of sustainability and considers not only the good outputs, but also the bad outputs. Compared to traditional total-factor energy efficiency, it evaluates energy efficiency much more accurately. Table 2 shows the regional ETFEE and TFEE of China.

The regional ETFEE in China on average is at a low level of about 60%, which urgently needs to be improved. The actual energy input could be reduced by almost 40%, with output unchanged, through energy efﬁciency improvement. This indicates that energy efﬁciency improvement is an effective way to solve the dilemma between the rising energy consumption demand from rapid economic growth and the increasing pressure on emission reduction.

Without incorporating environmental impacts, regional energy efficiency can be overestimated. AsTable 2 shows, the average ETFEE is always lower than the average TFEE. From 2005 to 2009, the average ETFEE in China is 0.609, 0.607, 0.617, 0.617, and 0.597, while the average TFEE is 0.716, 0.740, 0.715, 0.734, and 0.726, respectively. The Mann-–Whitney U rank test proves that the difference between ETFEE and TFEE presents a statistical significance with a P-value less than 0.001 as Table 3 shows. The comparative result means that the consideration of undesirable outputs has a significant influence on regional energy efficiency.

The energy efﬁciency is much higher with only one dimension of energy savings and becomes much lower with the added dimension of emission reduction. Under the consideration of environmental impacts, most regions are landing much more below the efﬁcient frontier due to poor performance on pollution emission. This indicates that China has achieved much more improvement in energy savings than that of emission reduction.

In order to improve energy efﬁciency and achieve sustainable development, China should concentrate on both energy saving and emission reduction at the same time. The reduction of emissions is not only affected by energy saving, but also by the optimization of energy consumption type. China’s government should be concerned about energy savings, as well as put much more emphasis on the adjustment of energy type. In fact, the government has become aware of this problem, and the read-justment of energy type is one of the main tasks in the Compre-hensive Program on Energy-Saving and Emission Reduction during the Twelfth Five-Year Plan (2011–2015) enacted by the State Council of China.

At the regional level, Table 4shows the gap between ETFEE and TFEE. The regions are divided into three groups. The ﬁrst group includes Beijing, Guangdong, and Shanghai. There are no gaps between ETFEE and TFEE for these three regions, because they always stand on the efﬁcient frontier for both ETFEE and TFEE for each year. The only thing for this group to do is to keep advancing continually in all the regions. The second group consists of Chongqing, Fujian, Guizhou, Hebei, Heilongjiang, Hubei, Liaoning, Qinghai, and Shaanxi. Their gaps between ETFEE and TFEE are becoming narrower, indicating that their ability to manage the undesirable outputs is enhanced. The third group contains Anhui, Gansu, Guangxi, Hainan, Henan, Hunan, Inner Mongolia, Jiangsu, Jiangxi, Jilin, Ningxia, Shandong, Shanxi, Sichuan, Tianjin, Xinjiang, Yunnan, and Zhejiang, whose gaps between ETFEE and TFEE are becoming larger. These regions have paid more attention to energy savings, with less attention towards emission reduction. They should vigorously promote energy savings and emission reduction at the same time. 4.2. ETFEE discrepancy of different areas

According to economic development and geographical location, China can be divided into four areas as in Fig. 1. The east area consists of Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan, which is the most developed Table 1

Descriptive statistics of the data.

Variable Deﬁnition Unit Minimum Maximum Mean Standard

deviation Desirable outputs

y1 Real GDP 100 million RMB in 2005 prices 543.320 36,035.129 8,671.620 7,270.075

Undesirable outputs

y2 SO2Emission Ten thousand tons 2.175 200.200 80.926 47.004

y3 CO2Emission Ten thousand tons 1025.162 74,295.068 21,827.533 15,425.980

Inputs

x1 Energy consumption Ten thousand tons of standard

coal equivalence

822.000 32,420.000 10,452.733 6,747.828

x2 Capital stock 100 million RMB in 2005 prices 1878.801 77,955.769 20,397.321 15,721.950

x3 Labor Ten thousand persons 128.700 5,948.800 2,174.531 1,533.290

Inﬂuential factors

z1 Ratio of intramural expenditure on R&D to GDP % 0.178 5.502 1.190 0.949

z2 Degree of foreign dependence % 4.530 166.816 35.571 40.798

z3 Ratio of the secondary industry to GDP % 23.500 61.500 47.809 7.611

z4 Ratio of government subsidies for industrial

pollution treatment to GDP

(5)

area in China. The west area contains Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, and Tibet, which is the most undeveloped area in China. Because of the absence of data, Tibet is not included in this research. The central area includes Shanxi, Anhui, Jiangxi, Henan, Hubei, and Hunan. The northeast area includes Liaoning, Jilin, and Heilongjiang. Both the northeast and central areas are more devel-oped than the western area, but less develdevel-oped than the east area.

The regional energy efﬁciency of China is very imbalanced, as

Fig. 2presents. There are big gaps among the east, central, west, and northeast areas. The east area has the highest ETFEE, where the ETFEE scores are all larger than 0.7. Three regions, including Beijing, Shanghai, and Guangdong, are found to always have the optimal efﬁciency during the research period, and they are all located in the east area. To put this into more detail, Beijing is the political and cultural center of China and is the only one region whose third industry ratio is more than 50%. Shanghai is the economic and ﬁnancial center of China and has developed service and high-tech industries. Guangdong is an export-oriented region, whose total imports and exports hit 611.09 billion USD in 2009 and occupy more than 25% of the total imports and exports in China. These three regions are the most developed in China. Table 3

Signiﬁcance test between ETFEE and TFEE in China.

Mann-Whitney U Wilcoxon W Z-Value P-value

ETFEE vs. TFEE 7558.000 18883.000 4.923 o0.001

Table 4

Gaps between ETFEE and TFEE in China from 2005 to 2009.

Group Region 2005 2006 2007 2008 2009 1 Beijing 0.000 0.000 0.000 0.000 0.000 Guangdong 0.000 0.000 0.000 0.000 0.000 Shanghai 0.000 0.000 0.000 0.000 0.000 2 Chongqing 0.255 0.121 0.305 0.318 0.169 Fujian 0.153 0.150 0.085 0.128 0.139 Guizhou 0.189 0.189 0.185 0.186 0.175 Hebei 0.122 0.131 0.000 0.019 0.031 Heilongjiang 0.121 0.079 0.000 0.000 0.000 Hubei 0.155 0.177 0.195 0.095 0.102 Liaoning 0.000 0.182 0.016 0.076 0.075 Qinghai 0.182 0.145 0.149 0.144 0.151 Shaanxi 0.327 0.357 0.158 0.136 0.142 3 Anhui 0.223 0.239 0.251 0.253 0.304 Gansu 0.108 0.125 0.143 0.173 0.173 Guangxi 0.169 0.203 0.241 0.269 0.402 Hainan 0.000 0.172 0.220 0.264 0.211 Henan 0.139 0.172 0.218 0.261 0.190 Hunan 0.125 0.138 0.145 0.155 0.197 Inner Mongolia 0.000 0.014 0.055 0.079 0.102 Jiangsu 0.081 0.117 0.106 0.143 0.146 Jiangxi 0.248 0.246 0.238 0.303 0.312 Jilin 0.183 0.115 0.107 0.188 0.234 Ningxia 0.125 0.166 0.126 0.147 0.166 Shandong 0.053 0.168 0.000 0.042 0.133 Shanxi 0.070 0.086 0.000 0.013 0.046 Sichuan 0.202 0.211 0.223 0.237 0.212 Tianjin 0.009 0.000 0.024 0.000 0.065 Xinjiang 0.134 0.109 0.163 0.156 0.145 Yunnan 0.273 0.294 0.32 0.308 0.291 Zhejiang 0.090 0.108 0.115 0.165 0.172 Table 2

Regional ETFEE and TFEE in China from 2005 to 2009.

ID Region ETFEE TFEE

2005 2006 2007 2008 2009 2005 2006 2007 2008 2009 1 Anhui 0.653 0.657 0.663 0.665 0.600 0.876 0.896 0.914 0.918 0.904 2 Beijing 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 3 Chongqing 0.557 0.560 0.567 0.571 0.512 0.812 0.681 0.872 0.889 0.681 4 Fujian 0.847 0.850 0.819 0.785 0.757 1.000 1.000 0.904 0.913 0.896 5 Gansu 0.352 0.351 0.354 0.356 0.367 0.460 0.476 0.497 0.529 0.540 6 Guangdong 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 7 Guangxi 0.650 0.648 0.649 0.646 0.573 0.819 0.851 0.890 0.915 0.975 8 Guizhou 0.282 0.283 0.285 0.292 0.291 0.471 0.472 0.470 0.478 0.466 9 Hainan 1.000 0.828 0.780 0.736 0.713 1.000 1.000 1.000 1.000 0.924 10 Hebei 0.401 0.402 0.435 0.449 0.444 0.523 0.533 0.435 0.468 0.475 11 Heilongjiang 0.544 0.546 0.578 0.585 0.591 0.665 0.625 0.578 0.585 0.591 12 Henan 0.575 0.575 0.581 0.586 0.529 0.714 0.747 0.799 0.847 0.719 13 Hubei 0.519 0.521 0.526 0.494 0.488 0.674 0.698 0.721 0.589 0.590 14 Hunan 0.540 0.542 0.549 0.564 0.522 0.665 0.680 0.694 0.719 0.719 15 Inner Mongolia 0.355 0.320 0.349 0.359 0.362 0.355 0.334 0.404 0.438 0.464 16 Jiangsu 0.859 0.866 0.833 0.819 0.806 0.940 0.983 0.939 0.962 0.952 17 Jiangxi 0.752 0.754 0.762 0.697 0.688 1.000 1.000 1.000 1.000 1.000 18 Jilin 0.541 0.543 0.594 0.602 0.602 0.724 0.658 0.701 0.790 0.836 19 Liaoning 0.512 0.473 0.515 0.522 0.516 0.512 0.655 0.531 0.598 0.591 20 Ningxia 0.192 0.188 0.204 0.210 0.211 0.317 0.354 0.330 0.357 0.377 21 Qinghai 0.258 0.249 0.269 0.270 0.271 0.440 0.394 0.418 0.414 0.422 22 Shaanxi 0.561 0.564 0.572 0.631 0.620 0.888 0.921 0.730 0.767 0.762 23 Shandong 0.604 0.607 0.663 0.652 0.565 0.657 0.775 0.663 0.694 0.698 24 Shanghai 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 25 Shanxi 0.264 0.261 0.284 0.297 0.295 0.334 0.347 0.284 0.310 0.341 26 Sichuan 0.497 0.498 0.504 0.503 0.511 0.699 0.709 0.727 0.740 0.723 27 Tianjin 0.832 1.000 0.843 0.871 0.870 0.841 1.000 0.867 0.871 0.935 28 Xinjiang 0.421 0.369 0.397 0.395 0.376 0.555 0.478 0.560 0.551 0.521 29 Yunnan 0.457 0.450 0.454 0.456 0.458 0.730 0.744 0.774 0.764 0.749 30 Zhejiang 0.884 0.892 0.855 0.835 0.821 0.974 1.000 0.970 1.000 0.993 Average 0.609 0.607 0.617 0.617 0.597 0.716 0.740 0.715 0.734 0.726

(6)

Compared to the east area, the northeast, central, and west areas obviously fall behind. The ETFEE of the northeast and central areas is between 0.500 and 0.600. The northeast area ranks third in 2006 and second in the other years. The central area ranks second in 2006 and third in the other years. The west area ranks last and all its ETFEE scores are less than 0.500. The Kruskal–Wallis rank test proves ETFEE discrepancy among the different areas, showing statistical signiﬁcance with a p-value less than 0.001.Table 5shows the details.

A monotone increasing relation exists between each area’s ETFEE and per capita GDP in China. The per capita GDP represents the economic development level in an area. AsFig. 3shows, the east has the highest level of per capita GDP and also the highest ETFEE score. The northeast and central areas respectively have the second and third highest per capita GDP, and their efﬁciency levels perform the same as with the rank of the economic development level. The west has the lowest per capita GDP and the worst ETFEE score.

The monotone increasing relation reveals that energy effi-ciency tends to improve with economic development in China during the research period. This discovery matches the real condition of regional development in China. The main reasons are as follows. Throughout the thirty-year development after the reform and opening-up in 1978, all areas in China have already crossed the ‘pollution first and treatment after’ stage. In the 21st century, China has especially paid attention to the environmental impacts from regional economic growth. AsFig. 3shows, China is over the inflection point similar to the environmental Kuznets curve and entering the stage of ‘energy efficiency improvement parallel with economic development’. With a higher per capita GDP, an area is inclined to have better industrial structure, better production technology and better environmental protection tech-nology, which are beneficial to improving energy efficiency.

Industrial structure is especially an important interaction carrier between economy development and energy efficiency. Taking the inefficient regions as examples, we describe the three most inefficient regions based on the average scores of the five-year period in order to discover the relationship between indus-trial characteristics and energy efficiency. All three provinces are located in the most undeveloped area of China, i.e., the west area. The province with the lowest energy efficiency score is Ningxia, whose main industries are energy-intensive. Ningxia has five main industries, of which four are the main industries, including metallurgy, coal mining and processing, electricity manufacturing, and building material manufacturing, with the characteristics of high energy consumption and high pollution emission.

Qinghai is the most inefﬁcient province after Ningxia. There are four main industries in Qinghai, including petroleum, elec-tricity, non-ferrous metals, and salt chemical manufacturing industries. These four main industries occupy more than 66% of industry’s total value-added. It should be noted that all of the main industries are energy-intensive in Qinghai.

Shanxi is the third inefﬁcient province and is a typical resource-based region in the west area of China. The output of raw coal in Shanxi occupies more than 20% of China’s total in 2009. Coal mining and processing, coke mining and processing, metallurgy, and elec-tricity manufacturing are the four main industries in Shanxi. All four main industries are energy-intensive in Shanxi.

The above descriptions reveal that energy efﬁciency depends on regions’ industrial structure and the inefﬁcient regions are Fig. 1. Administrative regions and four major areas in China.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 2005 2006 2007 2008 2009 East Central West Northeast

Fig. 2. ETFEE of different areas in China from 2005 to 2009.

Table 5

Signiﬁcance test of ETFEE among different areas in China.

Indicator Chi-square value by the Kruskal–Wallis rank test P-value

(7)

concentrated on energy-intensive industries such as coal, metal-lurgy, electricity manufacturing, and salt chemical manufacturing. In order to improve energy efﬁciency, both the national and local governments must endeavor to change the industrial structure from energy-intensive industries to technology-intensive industries, ser-vice industries, and others. The national government should con-struct a mechanism at reasonably controlling the total energy consumption in regions and strengthen supervision of regional energy consumption, in order to force these energy-inefﬁcient regions to readjust their industry structure. The local government should eliminate outdated production capacity, encourage service industry and high-tech industry development, and actively intro-duce energy conserving equipment and technologies.

4.3. Factors of regional ETFEE

The identification of factors that affect regional ETFEE is very significant for regional energy efficiency improvement, but is always neglected in most existing research studies. In order to distinguish the influential factors of regional ETFEE in China, we employ the truncated regression model based on the truncated characteristics of ETFEE data. The applicability of the truncated regression model in the two-stage procedures to account for exogenous factors that might affect productive efficiency is supported bySimar & Wilson (2007).

This paper investigates four factors. Factor z1 represents the ratio of intramural expenditure on R&D to GDP. Factor z2 is on behalf of the foreign dependence degree, which is the ratio of total imports and exports to GDP. Factor z3refers to the ratio of the secondary industry to GDP. Factor z4is the ratio of govern-ment subsidies for industrial pollution treatgovern-ment to GDP. The model is set as:

ETFEE ¼

b

0þ

b

1z1þ

b

2z2þ

b

3z3þ

b

4z4þ

e

,

where

b

0 is the constant term;

b

1,

b

2,

b

3, and

b

4 are the parameters of the independent variables, respectively; and

e

is the error term.

Taking the ETFEE of 30 regions in China from 2005 to 2009 as the dependent variables, we list the detailed truncated regression results inTable 6. It can be seen that all of the four variables show statistical signiﬁcance in the truncated regression model of ETFEE.

First, the higher ratio of R&D expenditure to GDP contributes to the higher ETFEE. For example, the ratio of R&D expenditure to GDP for Beijing, which is on the efficient frontier, is 5.482%, whereas that of Ningxia, which has the worst energy efficiency, is only 0.571% in 2005. Augmenting R&D expenditure is a very effective way to boost technical progress, and so it can be called the ‘technical improvement effect’. On the one hand, technical progress increases the resources’ usage efficiency and reduces the input per unit of output, hence relieving the environmental impact from the production process. On the other hand, much cleaner technology will be innovated to replace the dirty technol-ogy, hence decreasing the pollution emission per unit of output.

Both the central and local governments should establish a perfect system to encourage R&D activities and technical innovations. For example, the governments can give preferential tax rates of imported R&D equipments and permit the accelerated depreciation of R&D equipments based on fiscal and tax policies. For the financial policy, the governments can support R&D activities through low-interest loans, discount loans, and so on. For the technology input policy, the governments should increase the financial input on R&D activity, optimize financial input usage, and support some major research projects. The government should especially give priority towards supporting R&D activities on energy savings and emission reduction.

Second, the higher foreign dependence degree is beneﬁcial to the higher regional ETFEE, which can be called the ‘internationalization Fig. 3. Relationship between ETFEE and per capita GDP of areas in China.

Table 6

Factors of regional ETFEE scores in China.

Coefﬁcient Signiﬁcance test

Z-Value P-Value

Constant term 71.70917 10.17566 o0.001

Dependent variable

z1 5.858229 4.771630 o0.001

z2 0.345841 12.09995 o0.001

z3 0.579405 4.164364 o0.001

(8)

effect’. Open and frequent communication with the international market is propitious to introducing advanced technology and modern management methods from abroad. Based on imports and exports, Chinese enterprises are affected by the strict environmental regulations of developed countries, and so their environmental protection awareness and technology level have improved.

The east area is much more open than the northeast, central, and west areas. An improvement of the internationalization level in the northeast, central, and west areas will be beneficial to raise the energy efficiency of China. The total imports and exports in the northeast, central, and west areas only occupied 11.5% of China’s total in 2005, and the foreign direct investment in these three areas occupied 20.8% of China’s total at the same time. With the implementation of Western Development Strategy, Central Rise Strategy, and Reviving Northeastern Old Industries Strategy, the proportions of imports and exports and foreign direct invest-ment in the northeast, central, and west areas rose to 12.4% and 22.1% of China’s totals in 2009, respectively. In the future, the central government should keep the preferential policies on the enterprise income tax so as to guide foreign direct investment to invest in the central and west areas. The local governments should make efforts to improve the environments for foreign investment, such as convenient procedures for commercial regis-tration, an efficient governmental system, and financial support.

Third, there is a negative relation between the ratio of the secondary industry to GDP and regional ETFEE in China. This matches the fact that the ratio of the high-tech industry to the secondary industry still stays at a relatively low level, but the energy-intensive industry occupies a high percentage. Compared to the third industry, which is called the service industry, the secondary industry consumes much more energy input and pro-duces much more pollution. To achieve intensive economic growth, more and more regions in China are energetically developing the third industry, while reducing the secondary industrial ratio.

Service industry development and the optimization of the secondary industry are very important for improving China’s energy efﬁciency. The State Council of China put forward Opi-nions on Accelerating the Development of Service Industry and proposed the domestic industrial structure to transform from the secondary industry into the third industry. Most regions in China, such as Beijing, Tianjin, Shanghai, and Zhengjiang, have estab-lished special funds to support service industry development. Except for Beijing whose service industry occupies more 70% of GDP, the secondary industry dominates all the regions in China as of now, and hence the readjustment of the secondary industry is quite imperative in China for energy efﬁciency improvement and sustainable development. China’s government has adopted the following measures to optimize the secondary industry: to inhibit the fast growth of energy-intensive and emission-intensive indus-tries by enhancing the industry access threshold, to accelerate the elimination of outdated production capacity by strict government supervision, to promote the upgrading of traditional industries by high-technology application, etc.

Fourth, a region with a high ratio of government subsidies for industrial pollution treatment to GDP is probably due to having a low ETFEE, which can be called the ‘inverse subsidy effect’. The more pollution a region produces, the more government subsidies for pollution treatment are needed. If a region has a higher ratio of government subsidies for industrial pollution treatment to GDP, then the enterprises of this region prefer to have a lower initiative to reduce pollution and a higher dependence on pollu-tion subsidies.

The government should push enterprises to save energy and reduce emission actively with capital investment. Direct govern-ment subsidies for industrial pollution treatgovern-ment are remedial measures, which do not encourage enterprises to reduce

emission. If the direct government subsidy is changed to prefer-ential ﬁscal, tax, and ﬁnancial policies for enterprises with good performance on energy saving and emission reduction, then such policies will become an invisible hand to encourage enterprises to reduce energy consumption and pollution emission.

5. Conclusions

The ETFEE index is constructed on the viewpoint of sustain-ability by taking the ratio of target energy input from an SBM model with undesirable outputs to the actual energy input. The ETFEE inherits the total-factor framework from the traditional TFEE based on DEA, taking energy consumption with capital and employment as multi-inputs. The essential difference between ETFEE and TFEE is whether undesirable outputs are considered. In the calculation of the ETFEE index, the SBM model with undesir-able outputs takes not only GDP as a desirundesir-able output, but also undesirable CO2and SO2as outputs. Environmental pollution is nowadays a worldwide concern. Therefore, the ETFEE index evaluates energy efﬁciency through an appropriate approach.

Under the framework set up herein, this paper studies regional energy efficiency in China from 2005 to 2009. In the first phase, a comprehensive evaluation of regional energy efficiency in China is accomplished by a comparison between ETFEE and TFEE. In the second phase, the ETFEE discrepancy of different areas is ana-lyzed. In the third phase, the influential factors of regional ETFEE are identified by a truncated regression model.

Regional ETFEE in China is found to be at a low level of about 0.600. Without taking into account environmental impacts, regio-nal energy efficiency can be overestimated by more than 0.100. The ETFEE is always significantly less than TFEE on average. Three regions – Beijing, Shanghai, and Guangdong – are found to always have optimal efficiency during the research period, and all three are located in the east area. Except for Beijing, Shanghai, and Guangdong, there are definite gaps between regional ETFEE and TFEE in the other regions.

The regional energy efficiency of China is extremely unba-lanced: the east area ranks first with the highest ETFEE of above 0.7, the northeast and central areas follow, and the west area falls behind with the lowest ETFEE of less than 0.5. A monotone increasing relation exists between each area’s ETFEE and per capita GDP in China, and energy efficiency tends to improve with economic development. The higher ratio of R&D expenditure to GDP and the higher degree of foreign dependence both contribute to a higher ETFEE. The ratio of the secondary industry to GDP and the ratio of government subsidies for industrial pollution treat-ment to GDP both have negative effects on regional ETFEE.

In China the demand for energy is showing rigid sustained growth with the acceleration of industrialization and urbaniza-tion. Energy consumption is constrained by domestic resource supplies, environmental capacity, and global energy security. China’s energy policy thus faces substantial difﬁculties, and so energy efﬁciency is becoming an important issue.

For energy efﬁciency improvement, China should concentrate on energy savings as well as pay close attention to pollution emission reduction. Emission reduction not only depends on total energy consumption reduction, but also depends on the adjust-ment of energy consumption type. China’s high dependence on coal should be changed, and much more renewable and clean energy should be developed and used. The industrial structure is a decisive factor for energy efﬁciency, and so both the national and local governments must endeavor to change the industrial struc-ture from energy-intensive industries to service industries, tech-nology-intensive industries, and others. In order to save energy as well as reduce emission, China’s government should promote an

(9)

increase in R&D expenditure, a rise in the internationalization level, and the industrial structure transition. Direct subsidies for industrial pollution should be carefully used to avoid any inverse subsidy effect, and they can be changed to preferential ﬁscal, tax, and ﬁnancial policies to push enterprises to save energy and reduce emission.

This paper still has some limitations. With more available statistical data, more factors including the high-technology ratio can be considered, and the estimated CO2 emission can be substituted by statistical data. Global energy efficiency can be evaluated with the same method to find the ETFEE rank of China, which is helpful for energy efficiency improvement in China. As China is now falling into a dilemma between high economic growth, high energy consumption, and high ecological pollution, ETFEE improvement is an effective way to achieve sustainable development for the country.

Acknowledgements

The authors are grateful to the editor-in-chief and three anon-ymous referees of this journal for their valuable comments. The first author thanks financial supports from the Fundamental Research Funds of the Central University in China (NKZXA10010) and National Natural Science Foundation of China (71103099). The second author gratefully acknowledges financial support from Taiwan’s National Science Council (NSC100-2410-H-009-051).

References

Banker, R.D., Charnes, A., Cooper, W.W., 1984. Some models for estimating technical and scale inefﬁciencies in data envelopment analysis. Management Science 30, 1078–1092.

BP, 2011. China Overtakes USA as Top Energy Consumer as World Demand Grows Strongly, Says BP in 60th Year of Global Energy. Downloaded from: /http://www.bp.com/extendedgenericarticle.do?categoryId=2012968&contentId =7069439S.

Charnes, A., Cooper, W.W., Rhodes, E., 1978. Measuring the efﬁciency of decision making units. European Journal of Operational Research 2, 429–444.

Cha, K., Lim, S., Hur, T., 2008. Eco-efﬁciency approach for global warming in the context of Kyoto mechanism. Ecological Economics 67, 274–280.

Cooper, W.W., Seiford, L.M., Tone, K., 2007. Data Envelopment Analysis: A Comprehensive Text with Models, Applications, References and DEA-Solver Software, 2nd edn. , Springer Science þBusiness Media, LLC.

Farrell, M.J., 1957. The measurement of productive efﬁciency. Journal of the Royal Statistical Society, Series A 120, 253–281.

Fearnside, P.M., 2002. Time preference in global warming calculations: a proposal for a uniﬁed index. Ecological Economics 41, 21–31.

Hu, J.L., Wang, S.C., 2006. Total-factor energy efﬁciency of regions in China. Energy Policy 34, 3206–3217.

Hu, J.L., Kao, C.H., 2007. Efﬁcient energy-saving targets for APEC economies. Energy Policy 35, 373–382.

Hu, J.L., Wang, S.C., Yeh, F.Y., 2006. Total-factor water efﬁciency of regions in China. Resources Policy 31, 217–230.

Honma, S., Hu, J.L., 2008. Total-factor energy efﬁciency of regions in Japan. Energy Policy 36, 821–833.

IPCC 2006. IPCC Guidelines for National Greenhouse Gas Inventories. Prepared by Task Force on National Greenhouse Gas Inventories. Downloaded from /http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.htmlS.

Lee, Y.C., Hu, J.L., Kao, C.H., 2011. Efﬁcient saving targets of electricity and energy for regions in China. Electrical Power and Energy Systems 33, 1211–1219. Liu, Y.J., 2011. Analysis of China’s current energy structure and trend of

develop-ment: supply and demand. Energy Procedia 5, 2593–2598.

Ministry of Industry and Information Technology of the People’s Republic of China. The Development of Petroleum and Chemical Industry Economy of China in the First Half of 2011. Beijing.

National Bureau of Statistics of China, 2011. The Bulletin of National Economic and Social Development in 2010 of China. Beijing.

Patterson, M.G., 1996. What is energy efﬁciency? concepts, indicators, and methodological issues. Energy Policy 24, 377–390.

Radhi, H., 2009. Evaluating the potential impact of global warming on the UAE residential building a contribution to reduce the CO2emissions. Building and

Environment 44, 2451–2462.

Shi, D., 2007. Regional differences in China’s energy efﬁciency and conservation potentials. China & World Economy 15 (1), 96–115.

Simar, L., Wilson, P.W., 2007. Estimation and inference in two-stage, semi-parametric models of production processes. Journal of Econometrics 136, 31–64.

Tone, K., 2001. A slacks-based measure of efﬁciency in data envelopment analysis. European Journal of Operational Research 130, 498–509.

Wilson, B., Trieu, L.H., Bowen, B., 1994. Energy efﬁciency trends in Australia. Energy Policy 22, 287–295.

Zhang, J., Wu, G.Y., Zhang, J.P., 2004. The estimation of China’s provincial capital stock: 1952–2000. Economic Study 10, 35–44.

Zhang, X.P., Cheng, X.M., Yuan, J.H., Gao, X.J., 2011. Total-factor energy efﬁciency in developing countries. Energy Policy 39, 644–650.

Zhou, P., Ang, B.W., Poh, K.L., 2008. A survey of data envelopment analysis in energy and environmental studies. European Journal of Operational Research 189, 1–18.