Industry-level total-factor energy efficiency in developed countries: A Japan-centered analysis

(1)

Industry-level total-factor energy efﬁciency in developed countries:

A Japan-centered analysis

Satoshi Honma

a

_{, Jin-Li Hu}

b,⇑ a

Faculty of Economics, Kyushu Sangyo University, Japan b

Institute of Business and Management, National Chiao Tung University, Taiwan

h i g h l i g h t s

This study compares Japan with other developed countries for energy efﬁciency at the industry level. We compute the total-factor energy efﬁciency (TFEE) for industries in 14 developed countries in 1995–2005. Energy conservation can be further optimized in Japan’s industry sector.

Japan experienced a slight decrease in the weighted TFEE from 0.986 in 1995 to 0.927 in 2005.

Japan should adapt energy conservation technologies from the primary benchmark countries: Germany, UK, and USA.

a r t i c l e

i n f o

Article history: Received 6 October 2012

Received in revised form 26 November 2013 Accepted 23 December 2013

Available online 25 January 2014 Keywords:

Data envelopment analysis (DEA) Total-factor energy efﬁciency (TFEE) Industry-level analysis

Japan

a b s t r a c t

Japan’s energy security is more vulnerable today than it was before the Fukushima Daiichi nuclear power plant accident in March 2011. To alleviate its energy vulnerability, Japan has no choice but to improve energy efficiency. To aid in this improvement, this study compares Japan’s energy efficiency at the indus-try level with that of other developed countries. We compute the total-factor energy efficiency (TFEE) of industries in 14 developed countries for 1995–2005 using data envelopment analysis. We use four inputs: labor, capital stock, energy, and non-energy intermediate inputs. Value added is the only relevant output. Results indicate that Japan can further optimize energy conservation because it experienced only a marginal decrease in the weighted TFEE, from 0.986 in 1995 to 0.927 in 2005. To improve inefficient industries, Japan should adapt energy conservation technologies from benchmark countries such as Ger-many, the United Kingdom, and the United States.

1. Introduction

Efficient energy consumption is a top priority in the environ-mental field in terms of both resource conservation and combating climate change. In general, accepting declining economic growth as a consequence of decreased energy consumption is not acceptable. Therefore, improving energy efficiency without impairing eco-nomic performance is important for every economy.

The Fukushima Daiichi nuclear power plant accident following the Tohoku earthquake and tsunami in March 2011 prompted a reformation in Japan’s energy policy. Before the Fukushima inci-dent, nuclear energy produced approximately 30% of Japan’s elec-tricity; however, the enormous radioactive release and ensuing evacuation spurred an anti-nuclear energy movement in Japan. Even with periodically regulated monitoring after the earthquake,

no Japanese nuclear power plant was allowed to resume opera-tions. This was because the national government was unable to promptly revise its nuclear power plant safety standards and both mayors and citizens residing near nuclear power plants opposed resuming operations. All of Japan’s 54 nuclear power plants ceased operations on May 5th, 2012. Only the Ohi nuclear power plant re-started in July 2012, because of severe electricity shortages in the Kansai region, but it ceased again for periodic inspection in September 2013.

The Japanese government has promoted nuclear energy for two reasons. First, to improve Japan’s energy security. In 2010, Japan imported 96% of its primary energy supply and relied on imported oil for 99.6% of domestic demand. Moreover, 86.6% of Japan’s im-ported crude oil comes from the politically unstable Middle East. Second, Japan targeted nuclear power as a primary means to re-duce carbon dioxide emissions. In fact, prior to the Fukushima inci-dent, the Japanese government planned to build 14 more reactors and increase the share of nuclear power in the nation’s electricity supply to 50% by 2030.

http://dx.doi.org/10.1016/j.apenergy.2013.12.049

⇑ Corresponding author. Address: 118, Sec. 1, Chung-Hsiao W. Rd., Taipei City 10044, Taiwan. Fax: +886 2 23494922.

E-mail address:jinlihu@mail.nctu.edu.tw(J.-L. Hu).

URL:http://jinlihu.tripod.com(J.-L. Hu).

Contents lists available atScienceDirect

Applied Energy

(2)

The Fukushima incident and ensuing public opposition prompted drastic changes in Japan’s energy policies. Japan no long-er depends on nuclear enlong-ergy for electricity genlong-eration[1]. The Japanese government is currently promoting renewable energy to compensate for the loss of nuclear power and has implemented a feed-in tariff scheme for it. However, even high-level penetration of renewable energy cannot fully replace nuclear power[2].

Japan’s energy security is more vulnerable today than it was be-fore the accident. Japan has no choice but to improve its energy efﬁciency to alleviate its energy vulnerability. The nation re-sponded to the two oil crises in the 1970s by enhancing energy-saving technology; however, little progress was made during the 1980s and 1990s. Japan still has potential to improve energy sav-ings[3]. To assist policymakers in this regard, this study compares Japan’s energy efﬁciency at the industry level with that of other developed countries.

Since the ﬁrst oil crisis in 1973, many major developed coun-tries have implemented measures to improve energy efﬁciency [4]. Recently, the European Council advocated ambitious targets, known as the 20/20/20 goals[5]:

Reduce greenhouse gas (GHG) emissions 20% from 1990 levels by 2020.

Increase energy efﬁciency to reduce EU energy consumption 20% by 2020.

Ensure that 20% of all EU energy consumption comprises renewable energy by 2020.

Energy efficiency appears to be the only item among these goals that will reduce GHG emissions, improve energy stability, cut en-ergy costs, and enhance economic competitiveness[6]. Therefore, energy efficiency can be portrayed as Europe’s biggest energy source[7]. It is important to note that improving energy efficiency can aid in the reduction of GHGs and boost the share of renewable energy without new investment[8]. One driver of improved energy efficiency in the industrial sector is technological change, which is critically affected by the political framework and stringent stan-dards of carbon dioxide reduction[9]. Thus, the importance of en-ergy efficiency targets in policymaking cannot be overemphasized. Unfortunately, the EU’s current 20/20/20 policy may be naïve and suboptimal. Uniform application of goals for all EU members is neither fair nor equitable because energy efficiency among coun-tries varies[10]. The simulation by Capros et al.[11]shows that the EU energy policy is likely to cause an undesirable distributional im-pact; therefore, targets should be set with consideration for fair-ness. A country’s energy consumption savings should be differentiated on the basis of each country’s current efficiency.

Disaggregated information about energy efﬁciency is essential in establishing well-designed energy efﬁciency targets. Because not all energy sources are perfectly substitutable in every region, the quality of energy problem [12,13] should be considered. Although substitutability and price differences among sources must be examined in terms of energy aggregation, the traditional energy intensity (EI) indicator—energy consumption per unit of GDP—is used in formal statistics[6]. Most EI studies show that lev-els tend to converge [14–19]; however, certain studies indicate that a convergence of EI appears only in some regions [20,21]. Mendiluce et al.[22]claim that while EI in Spain increases owing to growth in transportation, it decreases in other EU countries.

Government programs and academic research use energy pro-ductivity (EP), deﬁned as output per unit of energy consumption, alongside EI. However, EI or EP frameworks are not included, wherein other inputs such as labor and capital can be substituted with energy[23]. As Patterson[24]notes, the EP ratio can be re-duced simply by substituting labor with energy. Therefore, energy efﬁciency should be evaluated using a multiple input–output

model. The data envelopment analysis (DEA) approach, a non-parametric method of linear programming, suits this purpose.

The purpose of this study evaluates industry-level total-factor energy efficiency of 14 developed countries and compares Japan’s energy efficiency with that of other countries. Using the DEA ap-proach, we calculate the total-factor energy efficiency (TFEE) index proposed in Hu and Wang[25]. We regard DEA as the best tool for this purpose, as it explicitly indicates the potential saving of inputs through efficiency calculation.

Few studies of industry-level energy efficiency exist because, even for developed countries, no industry-level, internationally compatible, credible data are derived using a uniform method regarding capital stock. However, the EU KLEMS[26] project, fi-nanced by the European Commission, has developed a comprehen-sive database for developed countries, allowing researchers to internationally compare industry-level efficiency.

This study is organized as follows: Section2reviews relevant literature. Section3presents our methodology and data. Section4 compares energy efﬁciency between Japan and other developed countries, provides sensitivity analyses, and compares the results with traditional EP. Section5concludes this paper.

2. Literature review1

When modeling industrial energy efficiency evaluation, researchers must conciliate data availability, the level of disaggre-gation, and modeling efforts for adequate sectoral representation [27]. Numerous studies address improvements in industrial energy efficiency, commonly through case studies that explore energy effi-ciency improvements in selected industries. Jochem and Gruber [28]analyze the effect of local leaning networks on energy effi-ciency in Germany, identifying preconditions and factors of suc-cessful networks. Klugman et al.[29] investigate a Scandinavian chemical wood-pulp mill using an energy audit and identify en-ergy-saving points. Usón et al. [30] analyze energy efficiency assessment and improvement in a coal-fired plant by using a ther-mo-economic diagnosis system, demonstrating its commendable accuracy for sources of inefficiency.

Ammar et al.[31]examine low-grade heat recovery in process industries, identifying low-grade heat sources and their potential markets in the United Kingdom. Investigating the paper industry in the Netherlands, Poland, and Sweden, Laurijssen et al.[32]find that the natural gas combined cycle prevalent in the Netherlands uses the least energy. Cagno and Trianni[33]investigate 71 small-and medium-sized Italian manufacturing enterprises in multiple case studies, finding that the crucial motivators to adopting en-ergy-efficient technologies are allowances or public financing, external pressures, and long-term benefits. Analyzing 65 foundries in seven European countries, Thollander et al. [34] find that Table 1

Industry list.

Industry classiﬁcation Description

Chemical Chemical and petrochemical

Construction Construction

Food Food and tobacco

Machinery Machinery

Metal Iron and steel, non-ferrous metals Non-metallic minerals Non-metallic minerals

Paper Paper, pulp, and printing

Textile Textile and leather

Transport Transport equipment

Wood Wood and wood products

1

(3)

ﬁnancial and organizational issues are perceived as the most rele-vant forces for driving improved energy efﬁciency. Seck et al.[27] develop a bottom-up energy model for non-energy-intensive industries in France and document the impact of heat recovery with heat pumps in the food and drink industry.

On the other hand, industry-speciﬁc models quantitatively present potential for conserving energy and reducing carbon diox-ide emissions. Using physical production data, Farla et al. [35] apply an index composition approach to the pulp and paper industry in eight countries from the Organisation for Economic Co-operation and Development (OECD). They show that growth of primary energy consumption in this industry was limited to 16% between 1973 and 1991 because of energy efﬁciency improve-ments, whereas production increased by 42% in the sampled countries. Siitonen et al.[36]shed light on differences in process heat conservation from the mill site and national levels in Finland’s pulp and paper industry.

Although studies of industry-specific improvements in energy efficiency involve case studies and industry-specific models, scholars usually employ a unified framework when investigating energy efficiency at national, regional, and industrial levels. DEA is among the most suitable methods for measuring energy efficiency.2

Numerous studies evaluate economy-wide aggregate energy efﬁciency using the DEA approach.3_{Hu and Kao}_[41]_measure

en-ergy efficiency of 17 Asia–Pacific Economic Cooperation (APEC) economies, and Zhou and Ang [42]do so for 21 OECD countries. Moreover, Sözen and Alp [43] use the DEA method to evaluate energy consumption, GHG emissions, and local pollutants in Turkey and 28 EU countries including Switzerland. Lozano and Gutiérrez[44]propose three models for evaluating efficiency using

population, GDP, energy consumption, and GHG emissions and employed them to study 28 Annex B countries from the Kyoto Protocol. Wang et al.[45]apply multidirectional efﬁciency analysis to Chinese regional energy and emission efﬁciency.

The above-mentioned papers analyze national energy effi-ciency; other studies apply DEA to specific industries. Oggioni et al. [46] provide different eco-efficiency measurements of the cement industry in 21 countries, taking carbon dioxide emissions as inputs or undesirable outputs. Wang et al.[47]use a Malmquist– Luenberger index to measure cost efficiency of China’s thermal power industry.

Although information regarding aggregate energy efficiency is useful, it provides only an approximate estimation of nationwide energy consumption. Countries generally have efficient and ineffi-cient industries, and aggregate efficiency scores cannot determine which of these need improvement. Furthermore, more in-depth analysis requires disaggregated data for energy efficiency across countries.

Hinchy et al.[48]employ DEA to measure energy efﬁciency of 37 industries in 30 countries using 1992 data. They compute potential energy savings and reductions in carbon dioxide emis-sions on the basis of efﬁcient targets; however, they use data from the Global Trade Analysis Project (GTAP) database.4_Because

GTAP operates economic simulation models using data from com-parative studies , it is not recommended and indeed problematic, such as [48]. Mukherjee [49] uses four DEA models to measure energy efficiency of the six US sectors with the highest energy consumption, finding that the paper and allied products sector used energy more efficiently than manufacturing overall. Honma and Hu [50] measure the TFEE of 17 Japanese industries, demonstrating that Japan’s energy-inefficient sectors include Table 2

Summary of statistics.

Variable Value added Labor Capital stock Non-energy intermediate

inputs

Energy Unit 1995 price (million

euros)

Total hours worked by persons engaged (million hours)

1995 price (million euros)

1995 price (million euros) Oil equivalent (million tons) Average 30676.71698 1048.164969 33095.77965 50413.529 4102.372995 Standard deviation 78165.90432 2254.52346 54868.84606 101978.8203 9635.986212 Min 235.2913217 8.871592431 459.6726533 544.922273 10 Max 1371882.94 18764.47287 316462.8206 794361.2266 92500 Observation 1496 1496 1496 1496 1496 Table 3

TFEEs for the chemical industry.

Country 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Australia 0.368 0.335 0.315 0.455 0.240 0.205 0.304 0.627 0.390 0.496 0.530 Austria 0.570 0.605 0.560 0.571 0.673 0.652 0.803 0.645 0.596 0.499 1.000 Czech Republic 0.438 0.529 0.633 0.404 0.274 0.257 0.397 0.575 0.310 0.237 0.346 Denmark 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Finland 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Germany 1.000 1.000 1.000 1.000 1.000 0.987 1.000 1.000 1.000 1.000 1.000 Italy 0.843 0.823 0.823 0.819 0.823 0.818 0.754 0.796 0.777 0.759 0.730 Japan 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.956 0.934 South Korea 0.921 0.863 0.992 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Netherlands 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Portugal 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Sweden 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 United Kingdom 1.000 1.000 0.962 0.897 0.878 0.825 0.827 0.742 0.769 0.775 0.855 United States 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 2

DEA is also widely used in resource and environmental economics[37–39]. 3

Stochastic frontier analysis is an alternative to the DEA approach, but its recent applications in energy efﬁciency studies are few, according to Zhou et al.[40].

4

GTAP is a multiregion, multisector, computable general equilibrium model that computes the impact of a change in trade policy.

(4)

energy-intensive industries as well as agriculture, forestry and ﬁshery, transportation, and communication industries.

Researchers have used the TFEE index proposed in[25]to inves-tigate China’s regional and national economies [51,52], China’s industrial sector [53], APEC economies [41], Japan [54,55], and Taiwan[56]. Furthermore, researchers have adopted the same in-put slack-based approach to comin-pute regional water efﬁciency in China[37].

Apart from the DEA approach, Miketa and Mulder[57]examine EPs of 10 manufacturing sectors in 56 developed and developing countries. They conclude that cross-country differences in EP tend to decline but that the convergence of EI is locally limited. Mulder and de Groot [58] investigate EPs of 14 sectors in 14 OECD

countries along with labor productivity and conclude that conver-gence of EI depends on unspecified country-specific characteristics. Previous country-comparative studies measure energy effi-ciency by country for nationwide energy use[40–44], by country for a particular industry’s energy use[46,47], by region within a country[45,51–56], and by industry within a country[49,50]. Rel-atively fewer studies, such as[48], reveal industry-level energy efficiency and potential energy savings across countries. This study employs the TFEE concept advocated by Hu and Wang[25], defined as the ratio of the target energy input, as suggested by the DEA, to actual energy input. Furthermore, this study is the first to apply the TFEE score to measure industry-level energy efficiency across countries.

Table 4

TFEEs for the construction industry.

Country 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Australia 0.209 0.141 0.156 0.117 1.000 0.293 1.000 1.000 1.000 1.000 1.000 Austria 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Czech Republic 0.454 0.503 0.399 0.276 0.569 0.568 0.519 0.511 0.563 0.560 0.578 Denmark 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Finland 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Germany 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Italy 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Japan 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 South Korea 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.969 1.000 1.000 1.000 Netherlands 0.738 0.558 0.585 0.434 0.837 0.940 0.949 0.937 0.788 0.815 0.952 Portugal 0.323 0.370 0.361 0.302 0.670 0.632 0.609 0.626 0.675 0.638 0.674 Sweden 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 United Kingdom 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 United States 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Table 5

TFEEs for the food industry.

TFEEs for the machinery industry.

Country 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Australia 0.664 0.885 0.885 0.894 0.941 0.914 0.714 0.889 0.864 0.797 0.782 Austria 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.936 0.782 0.772 0.528 Czech Republic 0.294 0.601 0.573 0.581 0.604 0.492 0.352 0.241 0.267 0.304 0.248 Denmark 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Finland 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Germany 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.768 0.559 Italy 1.000 1.000 1.000 0.939 0.827 0.758 0.724 0.606 0.455 0.292 0.246 Japan 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.846 0.708 0.516 South Korea 0.889 0.849 0.803 0.785 0.742 1.000 1.000 1.000 0.587 0.365 0.363 Netherlands 0.517 0.432 0.538 0.484 0.455 0.479 0.412 0.335 0.381 0.399 0.246 Portugal 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Sweden 0.852 0.752 0.636 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 United Kingdom 1.000 1.000 0.973 0.967 0.890 0.786 0.734 0.644 0.404 0.156 0.350 United States 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

(5)

3. Methodology and data 3.1. DEA methodology

DEA is a linear programming method used to assess the com-parative efﬁciency of decision-making units (DMUs) such as coun-tries, regions, ﬁrms, and other organizations. There are K inputs and M outputs for each of the N DMUs. The envelopment of the i-th DMU is derived using the following linear programming prob-lem, assuming variable returns to scale (VRS) proposed by Banker et al.[59]: Minh;kh s:t: yiþ Yk P 0 hxi Xk P 0 ek ¼ 1 k P0; ð1Þ

where h is a scalar that represents the efﬁciency score of the i-th DMU; e is an 1 N vector of ones; k is an N 1 vector of con-stants; yiis an M 1 output vector of DMU i; Y is an M N

out-put matrix composed of all outout-put vectors of the N DMUs; xiis a

K 1 input vector of DMU i; and X is a K N input matrix Table 7

TFEEs for the metal industry.

TFEEs for the non-metallic minerals industry.

TFEEs for the paper industry.

Country 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Australia 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.621 0.904 Austria 0.577 0.567 0.666 0.520 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Czech Republic 1.000 1.000 0.474 0.514 0.911 0.619 0.862 0.864 0.400 0.395 0.262 Denmark 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Finland 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Germany 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Italy 0.959 0.886 0.886 0.893 0.886 0.919 0.958 0.959 0.931 0.876 0.847 Japan 1.000 1.000 0.991 0.956 0.864 0.843 1.000 1.000 1.000 1.000 1.000 South Korea 0.662 0.479 0.533 0.489 0.546 0.564 0.568 0.307 0.318 0.214 0.314 Netherlands 0.936 0.918 0.940 0.984 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Portugal 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.132 0.155 0.144 Sweden 0.713 0.655 0.723 0.841 0.784 0.897 0.871 0.920 0.954 0.926 0.918 United Kingdom 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 United States 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

(6)

composed of all input vectors of the N DMUs. The efficiency score satisfies 0 6 h 6 1, which is a radial contraction coefficient for the inputs. If h = 1, DMU i operates on the efficiency frontier and is technically efficient. This is an input-oriented model in which the radial adjustment coefficient, h, is multiplied by the input vector of DMU i. The constraint ek ¼ 1 is the convexity constraint to make the envelope the boundary of a minimum convex hull containing all DMUs within an industry in the same year. To con-trol the industry and annual environment, all efficiency scores and input targets for DMU i in year t are determined by compar-ing them to the industry efficiency frontier in year t. That is, the DEA model uses observations from the same industry in the same year.

Target Energy Input(i,j,t)is deﬁned as follows:

Actual Energy Inputði;j;tÞ ½Radial Adjustmentði;j;tÞ

þ Non radial Slack Adjustmentði;j;tÞ; ð2Þ

where (i, j, t) refers to each value for the j-th industry in the i-th country in the t-th year. The radial adjustment is given by (1 h) x(i,j,t); the non-radial slack is deﬁned as the amount of energy that

can be reduced using the non-radial method. The TFEE index is de-ﬁned as

TFEEði;j;tÞ¼ Target Energy Inputði;j;tÞ=Actual Energy Inputði;j;tÞ: ð3Þ

On the basis of the above definition, the TFEE assumes a value between zero and unity. Higher TFEEs imply greater energy effi-ciency, whereas a TFEE score of unity indicates that an industry is efficient and cannot save energy without reducing its value added. A TFEE score below unity implies that an industry is ineffi-cient and can increase energy savings.

3.2. Data

Our annual dataset includes 10 industries in 14 developed countries for 1995–2005. Economic data are obtained from[26], a comprehensive database ﬁnanced by the European Commission.5

Energy data are obtained from[60], one of the most reliable sources of international energy statistics. Consumption of each energy source—e.g., coal, oil products, and natural gas—are changed from original units to their equivalent in tons of oil using speciﬁc conver-sion factors. The aggregated energy consumption may have the uncertainty of energy data by the conversion factors applied. But they are scrutinized by the experts of the International Energy Agency.

The EU KLEMS project[25]has developed a revolutionary, com-prehensive database comprising European and other developed countries to analyze economic growth and productivity. This data-base facilitates international comparisons of industry-level efﬁciency.

Economic and energy-related data for various industries are then matched using the sources indicated above.6 _{The countries}

in the database include Australia, Austria, the Czech Republic, Denmark, Finland, Germany, Italy, Japan, South Korea, the Nether-lands, Portugal, Sweden, the United Kingdom, and the United States. The industries include construction; chemical and petro-chemical; food and tobacco; iron, steel, and non-ferrous metals; machinery; non-metallic minerals; paper, pulp, and printing; tex-tile and leather; transport equipment; and wood and wood prod-ucts. Thus, the total number of industries is 140 (10 for each country). Table 1 presents details of the 10 industries. Because there are no energy data for four industries,7 consistent annual data are available for 136 industries.

This model includes four inputs: labor, capital stock, energy, and non-energy intermediate inputs. Many international com-parisons that use DEA adopt GDP as their output. However, this study calculates efﬁciency by industry. Since GDP indicates the total value added in each industry, we consider value added as the sole output. Monetary values are measured in euros, with 1997 as the base year, using purchasing power parity, also ta-ken from [26]. Table 2 presents descriptive statistics of all variables.

Table 10

Benchmarks for inefﬁcient Japanese industries. Industry Benchmark (peer ratio)

Chemical 2004: Netherlands (0.491), Germany (0.316), United States (0.192)

2005: Netherlands (0.433), Germany (0.401), United States (0.165)

Machinery 2003: Germany (0.775), United States (0.225) 2004: Sweden (0.707), United States (0.293) 2005: United States (0.789), Sweden (0.211) Non-metallic

Minerals

1995: Germany (0.558), United States (0.248), United Kingdom (0.194)

1998: United Kingdom (0.419), Germany (0.293), United States (0.288)

1999: United States (0.395), United Kingdom (0.379), Austria (0.226)

2000: United Kingdom (0.572), United States (0.34), Germany (0.088)

2001: United Kingdom (0.522), United States (0.277), Germany (0.201)

2002: United Kingdom (0.526), Germany (0.329), United States (0.145)

2003: Germany (0.622), United Kingdom (0.276), United States (0.102)

Paper 1997: Germany (0.839), United States (0.161) 1998: Germany (0.858), United States (0.142) 1999: Germany (0.627), United Kingdom (0.241), United States (0.132)

2000: United Kingdom (0.598), United States (0.223), Denmark (0.179)

Table 11

Countries referencing Japan as a benchmark. Industry Benchmark (Number of Times)

Chemical Italy (7), United Kingdom (3), Germany (1) Construction None

Food Czech Republic (10), Australia (1), Germany (1), Italy (1), Sweden (1), United Kingdom (1)

Machinery None

Metal Germany (2), Italy (2), Austria (1) Non-metallic

Minerals

None

Paper None

5_{Although additional assumptions are inevitably imposed when constructing an} industry-level economic data series, they are based on the standard economic growth theory. For more details, please visit the EU KLEMS website (http://

www.euklems.net).

6

Data for France and Belgium are not released in[26]because conﬁdentiality had to be respected. See footnote 16 in[61].

7_{The following four industries were eliminated because energy consumption data} are unavailable: transport equipment in Australia; transport equipment in Japan; wood and wood products in Japan; and textile and leather in Japan.

(7)

4. Empirical results

4.1. Total-factor energy efﬁciency (TFEE)

To accommodate the varied structures of each industry exam-ined, we calculate individual TFEEs.Tables 3–9show the TFEE of seven industries.8,9

Let us examine TFEE results by industry. The chemical industry in six countries—Denmark, Finland, the Netherlands, Portugal, Sweden, and the United States—displays unvarying unity scores during the period and merits best practices among the sampled countries (Table 3). Australia, Austria, and the Czech Republic exhi-bit low efﬁciency TFEE scores for most years, implying inefﬁcient operation.

The construction industry in Austria, Denmark, Finland, Germany, Italy, Japan, Sweden, the United Kingdom, and the United States operates efﬁciently (Table 4). South Korea consistently exhibits TFEE scores of unity except for in 2002. The Czech Republic, Portugal, and Australia present low TFEE scores in the initial surveyed years.

Total-factor energy efficiency (TFEE)

Country

Fig. 1. Average TFEE per industry by country.

Fig. 2. Energy-consumption weighted total-factor energy.

8 _{We calculate data for 10 industries; however, owing to space limitations and the} purpose of this study, we report results for only the seven industries for which Japanese data are available.

9 _{Four countries show similar results in our study and in}_[41]_{: Australia, Japan,} South Korea, and the United States. Efﬁciency rankings for these countries are similar in our results and in[41].

(8)

The food industry in Austria, Finland, Japan, the Netherlands, and the United States exhibits TFEE scores at unity throughout the period. Australia, the Czech Republic, Portugal, and South Korea show low TFEE scores, below 0.5 (Table 5).

The machinery industry in Denmark, Finland, Portugal, and the United States exhibits TFEE scores of unity for all years. Austria, Germany, Japan, and Sweden show TFEE scores of unity for most sampled years (Table 6). The Czech Republic and the Netherlands exhibit low TFEE scores, some below 0.5. Italy’s TFEE scores deteri-orate from unity in the ﬁrst three years to 0.246 at the end of the period.

The metal industry in Denmark, Japan, the Netherlands, Portugal, and the United States exhibits TFEE scores of unity during the

period (Table 7). Australia, the Czech Republic, Finland, Germany, Italy, South Korea, Sweden, and the United Kingdom present mixed results, indicating efﬁcient and inefﬁcient years. Australia (except in 1995) and Austria present consistently low TFEE scores.

The non-metallic minerals industry in Denmark, Finland, Germany, Italy, the United Kingdom, and the United States displays consistent unity (Table 8). South Korea and Portugal register con-sistently low TFEE scores.

The paper industry in Denmark, Finland, Germany, the United Kingdom, and the United States exhibits TFEE scores of unity throughout the period (Table 9). TFEE scores for Austria and the Netherlands rise from 0.557 and 0.936, respectively, in 1995 to unity in 2005. In contrast, TFEE scores for the Czech Republic and

M

illio

n

to

n

Fig. 3. Potential energy savings by country 1995–2005.

TFEE

Energy productivity

Fig. 4a. Average TFEE and energy productivity in the chemical industry 1995–2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

TFEE

Energy productivity

Fig. 4b. Average TFEE and energy productivity in the construction industry 1995– 2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

TFEE

Energy productivity

Fig. 4c. Average TFEE and energy productivity in the food industry 1995–2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

TFEE

Energy productivity

Fig. 4d. Average TFEE and energy productivity in the machinery industry 1995– 2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

(9)

Portugal deteriorate drastically from unity in 1995 to 0.262 and 0.114, respectively, in 2005.

Now we examine Japan’s energy efficiency for the seven industries on the basis of results shown in Tables 3–9. Japan’s construction, food, and metal industries maintain TFEEs at unity, indicating that they operated at the efficiency frontier through-out the period. However, Japan’s chemical, machinery, non-metallic minerals, and paper industries display inefficient TFEEs below unity for several years within the period. Japan’s non-metallic minerals industry became efficient only in the final two years studied. However, efficiency in its chemical and machinery industries worsened after 2004 and 2003, respectively.

TFEE

Energy productivity

Fig. 4e. Average TFEE and energy productivity in the metal industry 1995–2005. Note: Eenergy productivity in euros of value added per ton of oil equivalent.

TFEE

Energy productivity

Fig. 4f. Average TFEE and energy productivity in the non-metallic minerals industry 1995–2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

TFEE

Energy productivity

Fig. 4g. Average TFEE and energy productivity in the paper industry 1995–2005. Note: Energy productivity in euros of value added per ton of oil equivalent.

0 0.2 0.4 0.6 0.8 1 1.2 1995 2000 2005 Ener gy productivit y TFEE

TFEE without labor TFEE without capital

TFEE without non-energy intermediate inputs TFEE with all four inputs

Energy productivity

Fig. 5a. TFEEs in Japan’s chemical industry 1995–2005.

0 0.2 0.4 0.6 0.8 1 1.2 1995 2000 2005 Energy productivity TFEE

Energy productivity

Fig. 5b. TFEEs in Japan’s construction industry 1995–2005.

0 0.2 0.4 0.6 0.8 1 1.2 1995 2000 2005 Ener gy p roductivit y TFEE

Energy productivity

Fig. 5c. TFEEs in Japan’s food industry 1995–2005.

0 0.2 0.4 0.6 0.8 1 1.2 1995 2000 2005 Ener gy p roductivit y TFEE

Energy productivity

(10)

Japan’s paper industry was inefﬁcient only from 1997 to 2000. In sum, Japan’s chemical, machinery, non-metallic minerals, and paper industries used energy inefﬁciently in some sampled years. Apart from in the machinery industry in 2004 and 2005, however, potential for energy-saving in these industries was at most 20%.

In general, a small number of efficient DMUs can drastically affect the efficiency frontier in DEA. Slight variation in a specific DMU’s position relative to other DMUs may change its status from inefficient to efficient and vice versa[62]. Fluctuations in the metal industry, for example, are observed in some countries during some years (e.g., Australia in 1995–1996, the Czech Republic in 1995–1997, Finland in 1995–1996, and South Korea in 1997–1998). Because data for each industry are available for 14 countries, the annual industry efficiency frontier position can change significantly with a change in the status of a specific coun-try. Similar fluctuations appear in other industries, albeit less fre-quently than in the metal industry.

We evaluate an inefficient TFEE score on the basis of efficient DMUs as benchmarks (reference sets), which are useful refer-ences for inefficient DMUs. Table 10 presents the benchmark countries for Japan’s inefficient industries. The weighted input combination from the ratios in parentheses indicates the point used to evaluate the radial efficiency of Japanese industries. Inef-ficiency removal is accomplished by contracting the actual in-puts to a linear combination by the ratios whereby input slacks reduce to zero (if they occur). Thus, the benchmarks and ratios provide the corresponding inefficient industry with an indication of improved energy efficiency. Germany, the United Kingdom, and the United States frequently appear as

bench-marks in Table 9; especially, the United States is a benchmark for inefﬁcient countries. Japan can beneﬁt from implementing energy-saving technologies utilized by these countries.

Japan is known to have efficient DMUs by the standards of inefficient countries. Table 11 shows countries with inefficient industries that consider Japan as one of their benchmarks for the years in which the corresponding Japanese industry is effi-cient. As per the table, Japan’s food industry is the benchmark 10 times for the Czech Republic and once for each of the five other countries. Furthermore, Japan’s chemical and metal indus-tries are benchmarks for certain counindus-tries. However, Japan’s con-struction, machinery, non-metallic minerals, and paper industries are never benchmarks for any country during the period, imply-ing that these four industries hold simply-ingular positions along effi-ciency frontiers in their respective industries.

4.2. International comparison of energy efﬁciency and energy-savings potential

Fig. 1 presents average TFEEs of each industry in the 14 countries for 1995–2005. First, average TFEEs of all industries in the United States stand at unity, implying that the United States operated efﬁciently in all industries during the period and had the best technology and production processes. Denmark (except for its food industry) and Finland (except for its metal industry) display perfect average TFEE scores. The Czech Republic, South Korea, and Australia (except for its wood industry) display no industries with average TFEE scores of unity. Many industries in these three countries exhibit lower average TFEEs, implying that signiﬁcant amounts of energy can be conserved.

Tables 3–9show mixed results for Japan. Even though Japan exhibits perfect average TFEEs in the construction, food, and metal industries, average TFEEs are below unity but above 0.9 for the chemical, machinery, non-metallic minerals, and paper industries. Next, we examine energy efﬁciency by country.Fig. 2shows energy-consumption-weighted TFEEs. Because energy consump-tion in each industry varies widely within a country, we disre-gard simple average TFEEs for the countries and present energy-consumption-weighted TFEEs.10_{Japan’s weighted TFEE}

de-creases marginally from 0.986 in 1995 to 0.927 in 2005. AsFig. 2 shows, the three lowest-ranking countries in 2005 (the Czech Republic, Australia, and Portugal) suffer from falling energy efﬁciency during the period. In contrast, the relatively efﬁcient 0.97 0.975 0.98 0.985 0.99 0.995 1 1.005 1995 2000 2005 Ener gy p roductivit y TFEE

Energy productivity

Fig. 5e. TFEEs in Japan’s metal industry 1995–2005.

Energy productivity

Fig. 5f. TFEEs in Japan’s non-metallic minerals industry 1995–2005.

Energy productivity

Fig. 5g. TFEEs in Japan’s paper industry 1995–2005.

10

Note that the energy-consumption-weighted TFEE is obtained by using the available industry TFEE scores. Japan’s score is computed using the seven available industries, whereas other countries’ scores, except Australia’s, are calculated using all 10 available industries.

(11)

countries with average values exceeding 0.9 in 2005 include Fin-land, Germany, Japan, the Netherlands, Sweden, and the United States. It is noteworthy that the United States and Finland exhibit efﬁciency across all industries during the period (except in 1995). How much energy consumption can each country reduce with-out simultaneous reductions in economic with-output? Fig. 3 shows potential energy savings by country. Note that potential energy savings of each industry are given by the sum of radial and non-radial slack adjustments. Potential energy savings of each country inFig. 3are calculated by summing all potential energy savings of each industry in that country.

In 1995, South Korea had the greatest potential for energy sav-ings, followed by Australia and the Czech Republic. In 2005, three countries—South Korea, Australia, and Italy—shared the greatest potential for energy savings. Japan accounted for 5.4% of the poten-tial energy savings of the 14 countries in 1995 and witnessed a growth of 11.7% in 2005.

4.3. TFEE comparisons with EP

As indicated, scholars have traditionally used EP—energy con-sumption per output—as an index of energy efﬁciency. Neverthe-less, that measure disregards the substitutability of inputs. Figs. 4a–4gcompare Japan’s average TFEE position to average EP during the period. Differences in rankings between the two indexes arise from whether other inputs are considered when measuring efﬁciency.

Correlation coefficients of the two indexes in the seven indus-tries range from 0.348 for the construction industry to 0.593 for the metal industry. Only Japan’s food industry attains first place in both indexes. When other inputs are ignored in measuring effi-ciency, the EPs of Japan’s construction and metal industries are moderate, but they attain full TFEE scores when other inputs are considered. Although Japan attains high average TFEE scores exceeding 0.9 for the machinery, non-metallic minerals, and paper industries, average EPs are low.

4.4. Sensitivity analyses

We conduct sensitivity analyses to ensure the robustness of these TFEE results. Because DEA is a non-parametric method, we cannot statistically verify whether a variable should be included in the analysis. We recalculate the annual TFEE of each industry using the remaining observations in which one input except en-ergy is removed.11_{Figs. 5a–5g}_{compare the results with the original}

TFEE. Except for the construction industry, all TFEE results obtained by dropping one input exhibit the same direction as the ordinary TFEEs.12

We also compare Japan’s TFEE results with a conventional par-tial factor EP index. EP, a traditional energy efficiency index, is de-fined as value added divided by energy consumption. Whereas all inputs are taken into account in TFEE, energy is the sole input in EP. Each industry’s EP is indicated along the right axis ofFigs. 4a–4g. The same tendency as that between the original TFEE and EP is ob-served in the chemical, non-metallic minerals, and paper indus-tries. On the other hand, EP values in the construction, food, machinery, and metal industries significantly diverge from their TFEEs. The divergence stems from the difference between total-factor and partial-total-factor frameworks.

5. Conclusion

This study compares Japan’s industry-level energy efﬁciency with that of other developed countries. We analyze TFEE and po-tential energy savings of 10 industries in 14 developed countries for 1995–2005 using the DEA approach. For robustness, we con-ducted sensitivity analyses and comparisons with EP.

Even though several Japanese industries were benchmarks for less energy-efficient countries and DMUs during 1995–2005, our in-depth analysis indicates further potential for energy consump-tion savings within Japan’s industrial sectors. Japan’s construcconsump-tion, food, and metal industries display efficient TFEE scores throughout the period. However, its chemical, machinery, non-metallic miner-als, and paper industries show inefficient TFEE scores in some years. The non-metallic minerals industry in particular became efficient within the final two years sampled. Overall, Japan’s weighted TFEE declines slightly from 0.986 in 1995 to 0.927 in 2005.

Benchmarking countries provides useful information about improving energy efficiency among inefficient industries. Germany, the United Kingdom, and the United States frequently appear as benchmarks for inefficient Japanese industries. The United States consistently appears as a benchmark for each Japanese industry examined in this study.

Our study presents several policy implications. First, to improve inefficient industries, Japan should adopt energy conservation technologies employed in benchmark countries. We also find that Japan’s efficient industries are benchmarks for other countries, such as Italy’s chemical industry and the Czech Republic’s food industry. Japan can provide energy-saving technologies to these countries.

Three suggestions for future research emerge from these find-ings. First, this study compares energy efficiency across countries by industry but does not analyze factors contributing to ineffi-ciency. Because disaggregated industry-level economic and energy data are insufficient for this analysis, future research can regress energy efficiency scores on control variables of aggregated country-level data. Second, because this study considers only manufactur-ing, future research can extend the analysis to non-manufacturing industries. Third, we treat energy measured in tons of oil as one equivalent input; however, aggregating energy sources inevitably raises questions about how differences in substitutability and cost among energy sources are managed [12]. Therefore, improving energy input measurement will be a constant goal for the future.

Acknowledgements

The authors thank two anonymous referees for their valuable comments. We also thank the participants at a conference held by the Society for Environmental Economics and Policy Studies in Japan, September 2010, and at the International Atlantic Economic Conference in Greece, March 2011. The lead author received a Grant-in-Aid (22530253) from the Ministry of Education, Science, Sports and Culture in Japan. The second author received partial ﬁnancial support from Taiwan’s National Science Council (NSC-101-2410-H-009-044).

References

[1]Vivoda V. Japan’s energy security predicament post-Fukushima. Energy Policy

2012;46:135–43.

[2]Zhang Q, Ishihara KN, Benjamin CM, Tezuka T. Scenario analysis on future

electricity supply and demand in Japan. Energy 2012;38:376–85.

[3]Moe E. Vested interests, energy efﬁciency and renewables in Japan. Energy

Policy 2012;40:260–73.

[4]Geller H, Harrington P, Rosenfeld AH, Tanishima S, Unander F. Policies for

increasing energy efﬁciency: thirty years of experience in OECD countries.

Energy Policy 2006;34:556–73.

11

We thank anonymous referees for the suggestion on this point.

12 _{Divergence between TFEE without non-energy intermediate inputs and others is} observed in other countries. We surmise that this is a result of peculiarities in the construction industry.

(12)

[5] Council of the European Union. Brussels European Council 8/9 March 2007. Presidency Conclusions, (7224/1/07), Brussels; 2007.

[6] European Communities (EC). Panorama of energy: energy statistics to support EU policies and solutions. Eurostat, Statistical Books. Office for Official Publications of the European Communities, Luxembourg; 2009, 156p. [7] European Commission. Energy Efficiency Plan 2011. COM (2011) 109/4; 2011.

[8]Harmsen R, Wesselink B, Eichhammer W, Worrell E. The unrecognized

contribution of renewable energy to Europe’s energy savings target. Energy

Policy 2011;39:3425–33.

[9]Blesl M, Kober T, Bruchof D, Kuder R. Effects of climate and energy policy

related measures and targets on the future structure of the European energy

system in 2020 and beyond. Energy Policy 2010;38:6278–92.

[10]Tolón-Becerra A, Lastra-Bravo X, Botta GF. Methodological proposal for

territorial distribution of the percentage reduction in gross inland energy consumption according to the EU energy policy strategic goal. Energy Policy

2010;38:7093–105.

[11]Capros P, Mantzos L, Parousos L, Tasios N, Klaassen G, van Ierland T. Analysis of

the EU policy package on climate change and renewables. Energy Policy

2011;39:1476–85.

[12]Berndt E. Aggregate energy, efﬁciency, and productivity measurement. Ann

Rev Energy 1978;3:225–73.

[13]Stern D. Energy and economic growth in the USA: a multivariate approach.

Energy Econ 1993;15:137–50.

[14]Nilsson LJ. Energy intensity trends in 31 industrial and developing countries

1950–1988. Energy 1993;18:309–22.

[15]Mielnik O, Goldemberg J. Converging to a common pattern of energy use in

developing and industrialized countries. Energy Policy 2000;28:503–8.

[16]Sun JW. The decrease in the difference of energy intensities between OECD

countries from 1971 to 1998. Energy Policy 2002;30:631–5.

[17]Alcántara V, Duro JA. Inequality of energy intensities across OECD countries: a

note. Energy Policy 2004;32:1257–60.

[18]Markandya A, Pedroso-Galinato S, Streimikiene D. Energy intensity in

transition economies: is there convergence towards the EU average? Energy

Econ 2006;28:121–45.

[19]Ezcurra R. Distribution dynamics of energy intensities: a cross-country

analysis. Energy Policy 2007;35:5254–9.

[20]Le Pen Y, Sévi B. On the non-convergence of energy intensities: evidence from

a pair-wise econometric approach. Ecol Econ 2010;69:641–50.

[21]Liddle B. Revisiting world energy intensity convergence for regional

differences. Appl Energy 2010;87:3218–25.

[22]Mendiluce M, Pérez-Arriaga I, Ocaña C. Comparison of the evolution of energy

intensity in Spain and in the EU15. Why is Spain different? Energy Policy

2010;38:639–45.

[23]Wilson B, Trieu LH, Bowen B. Energy efﬁciency trends in Australia. Energy

Policy 1994;22:287–95.

[24]Patterson MG. What is energy efﬁciency? Concepts, indicators, and

methodological issues. Energy Policy 1996;24:377–90.

[25]Hu JL, Wang SC. Total-factor energy efﬁciency of regions in China. Energy

Policy 2006;34:3206–17.

[26] EU KLEMS. EU KLEMS Growth and Productivity Accounts. <http://

www.euklems.net/euk08i.shtml>; 2008.

[27]Seck GS, Guerassimoff G, Maïzi N. Heat recovery with heat pumps in

non-energy intensive industry: a detailed bottom-up model analysis in the French

food & drink industry. Appl Energy 2013;111:489–504.

[28]Jochem E, Gruber E. Local learning-networks on energy efﬁciency in industry –

successful initiative in Germany. Appl Energy 2007;84(7–8):806–16.

[29]Klugman S, Karlsson M, Moshfegh B. A Scandinavian chemical wood pulp mill.

Part 1: Energy audit aiming at efﬁciency measures. Appl Energy

2007;84(3):326–39.

[30]Usón S, Valero A, Correas L. Energy efﬁciency assessment and improvement in

energy intensive systems through thermoeconomic diagnosis of the operation.

Appl Energy 2010;87:1989–95.

[31]Ammar Y, Joyce S, Norman R, Wang Y, Roskilly AP. Low grade thermal energy

sources and uses from the process industry in the UK. Appl Energy

2012;89(1):3–20.

[32]Laurijssen J, Faaij A, Worrell E. Energy conversion strategies in the European

paper industry – a case study in three countries. Appl Energy 2012;98:102–13.

[33]Cagno E, Trianni A. Exploring drivers for energy efﬁciency within small- and

medium-sized enterprises: ﬁrst evidences from Italian manufacturing

enterprises. Appl Energy 2013;104:276–85.

[34]Thollander P, Backlund S, Trianni A, Cagno E. Beyond barriers – a case study on

driving forces for improved energy efﬁciency in the foundry industries in Finland, France, Germany, Italy, Poland, Spain, and Sweden. Appl Energy

2013;111:636–43.

[35]Farla J. Energy efﬁciency developments in the pulp and paper industry: a

cross-country comparison using physical production data. Energy Policy

1997;25:745–58.

[36]Siitonen S, Tuomaala M, Suominen M, Ahtila P. Implications of process energy

efﬁciency improvements for primary energy consumption and CO2emissions

at the national level. Appl Energy 2010;87:2928–37.

[37]Hu JL, Wang SC, Yeh FY. Total-factor water efﬁciency of regions in China.

Resour Policy 2006;31:217–30.

[38]Kim JW, Lee JY, Kim JY, Lee HK. Sources of productive efﬁciency: International

comparison of iron and steel ﬁrms. Resour Policy 2006;31:239–46.

[39]Tsolas I. Performance assessment of mining operations using nonparametric

production analysis: a bootstrapping approach in DEA. Resour Policy

2011;36:159–67.

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

performance: a parametric frontier approach. Appl Energy 2012;90:196–200.

[41]Hu JL, Kao CH. Efﬁcient energy-saving targets for APEC economies. Energy

Policy 2007;35:373–82.

[42]Zhou P, Ang BW. Linear programming models for measuring economy-wide

energy efﬁciency performance. Energy Policy 2008;36:2911–6.

[43]Sözen A, Alp I. Comparison of Turkey’s performance of greenhouse gas

emissions and local/regional pollutants with EU countries. Energy Policy

2009;37:5007–18.

[44]Lozano S, Gutiérrez E. Non-parametric frontier approach to modelling the

relationships among population, GDP, energy consumption and CO2emissions.

Ecol Econ 2008;66:687–99.

[45]Wang K, Wei YM, Zhang X. Energy and emissions efﬁciency patterns of Chinese

regions: a multi-directional efﬁciency analysis. Appl Energy 2013;104:105–16.

[46]Oggioni G. Eco-efﬁciency of the world cement industry: a data envelopment

analysis. Energy Policy 2011;39(5):2842–54.

[47]Wang YS, Xie BC, Shang LF, Li WH. Measures to improve the performance of

China’s thermal power industry in view of cost efﬁciency. Appl Energy

2013;112:1078–86.

[48] Hinchy M, Trieu LH, Hanslow K. Technical efﬁciency and carbon dioxide emissions. ABARE Conference Paper 98.32; 1998. p. 1–18.

[49]Mukherjee K. Energy use efﬁciency in U.S. manufacturing: a nonparametric

analysis. Energy Econ 2008;30:76–96.

[50]Honma S, Hu JL. Total-factor Energy Efﬁciency for Sectors in Japan. Energy Sour

B 2013;8:130–6.

[51]Chang TP, Hu JL. Total-factor energy productivity growth, technical progress,

and efﬁciency change: an empirical study of China. Appl Energy

2010;87:3262–70.

[52]Lee YC, Hu JL, Kao CH. Efﬁcient saving targets of electricity and energy for

regions in China. Int J Elect Power 2011;33:1211–9.

[53]Wang ZH, Zeng HL, Wei YM, Zhang YX. Regional total factor energy efﬁciency:

an empirical analysis of industrial sector in China. Appl Energy

2012;97:115–23.

[54]Honma S, Hu JL. Total-factor energy efﬁciency of regions in Japan. Energy

Policy 2008;36:821–33.

[55]Honma S, Hu JL. Total-factor energy productivity growth of regions in Japan.

Energy Policy 2009;37:3941–50.

[56]Hu JL, Lio MC, Kao CH, Lin YL. Total-factor energy efﬁciency for regions in

Taiwan. Energy Sour. B 2012;7:292–300.

[57]Miketa A, Mulder P. Energy productivity across developed and developing

countries in 10 manufacturing sectors: patterns of growth and convergence.

Energy Econ 2005;27:429–53.

[58]Mulder P, De Groot H. Sectoral energy- and labor-productivity convergence.

Environ Resour Econ 2007;36:85–112.

[59]Banker RD, Charnes A, Cooper WW. Some models for estimating technical and

scale inefﬁciencies in data envelopment analysis. Manage Sci

1984;30:1078–92.

[60] International Energy Agency. Energy balances of OECD Countries. Each year.

[61]O’Mahony, Timmer PM. Output, input and productivity measures at the

industry level: the EU KLEMS database. Econ J 2009;119(538):F374–403.

[62]Cooper WW, Li S, Seiford LM, Tone K, Thrall RM, Zhu J. Sensitivity and stability