Comparison of technology efficiency for CO2 emissions reduction among European countries based on DEA with decomposed factors

Journal of Cleaner Production 151 (2017) 109e120

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Comparison of technology efficiency for CO2 emissions reduction among European countries based on DEA with decomposed factors Deuk Sin Kwon, Joon Hyung Cho, So Young Sohn* Department of Information and Industrial Engineering, Yonsei University, 134 Shinchonedong, Seoul 120e749, Republic of Korea

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 November 2016 Received in revised form 9 March 2017 Accepted 9 March 2017 Available online 10 March 2017

For consistent and effective CO2 reduction, many countries need to improve not only the technical efficiency of CO2 mitigation, but also environmental responsibility. In this study, we examine both the technical efficiency and voluntary environmental consciousness (VEC) of 12 European countries using a two-stage data envelopment analysis (DEA). In the first stage, we measured the technical efficiency of green energy technologies (GET) associated with fossil fuels, renewable energy, and storage technologies of each country for energy generation with regard to CO2 emissions by surveying GET-related patents. Using the logarithmic mean Divisia index (LMDI), we decomposed CO2 emissions into the following technological factors: energy intensity, fuel mix, and CO2 emission coefficient. In the second stage, we quantified the VEC in each country by investigating GET patent changes via research and development (R&D) investment at given changes in CO2 emissions. The results show different aspects for each country in terms of technical efficiency and VEC, suggesting potential levels of both efficient CO2 reductions and desirable GET development by using reference countries as a benchmark. Our study methodology and results can contribute to establishing effective national technology policy and aid in calls for common responsibility and the active participation of nations in addressing climate change. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Two-stage DEA CO2 reduction Technical efficiency Environmental consciousness Decomposition analysis LMDI

As rising atmospheric CO2 concentrations become an increasingly important global issue, many countries have prioritized reducing CO2 emissions as a main policy goal in national growth plans (Kim et al., 2016). In particular, 15 European countriesdthe EU Bubbledhave strived to mitigate CO2 emissions (Skjærseth et al., 2013), contributing to a 22% reduction in CO2 emissions by the EU in 2015 compared to 1990 levels (European Commission, 2016). For CO2 mitigation, various technologies have been developed so far in the field of energy generation, mainly for electricity (Kim et al., 2017; Sohn et al., 2015). In particular, energy-related technologies related to fossil fuels, renewable energy, and storage technologies have played critical roles in the reduction of CO2 emissions. Fossil fuel combustion for energy generation accounted for most CO2 emissions, and fossil fuels used for electricity generation are implicated in more than 40% of global CO2 emissions

(Quadrelli and Peterson, 2007; Zhang et al., 2013a). Certain renewable energies (e.g., biomass and waste mass) account for a portion of CO2 emissions, and many carbon-free renewable energies have received attention as alternative green energies that can assist in CO2 reduction (Noailly and Shestalova, 2013). Storage technology also contributes to the mitigation of CO2 emissions through improving energy efficiency and managing CO2 (Hadjipaschalis et al., 2009). Therefore, for consistent and effective reduction of CO2 emissions, it is important to improve technical efficiency of these energy-related technologies, which are collectively referred to as green energy technologies (GET).1 CO2 emissions can be influential in investment decisions surrounding GET. In general, it is expected that nations with high levels of CO2 emissions may invest more in GET than other technologies. However, nations with low voluntary environmental consciousness (VEC) would be unwilling to invest in GET despite their large volume of CO2 emissions. On the other hand, nations with high voluntary environmental consciousness (VEC) would

* Corresponding author. E-mail addresses: [email protected] (J.H. Cho), [email protected] (S.Y. Sohn).

1 Non standard abbreviations EC: fossil fuel energy consumption; GET: green energy technologies; VEC: voluntary environmental consciousness.

1. Introduction

(D.S.

http://dx.doi.org/10.1016/j.jclepro.2017.03.065 0959-6526/© 2017 Elsevier Ltd. All rights reserved.

Kwon),

[email protected]

110

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

have a higher propensity to invest in GET, even if they had a smaller volume of CO2 emissions. For this study, VEC was defined as the amount of effort made by a country to develop GET for CO2 reduction, given its level of CO2 emissions, and was measured by the number of GET-related patents. High and low degrees of VEC were estimated via comparison with the GET patenting efforts of other countries given their corresponding CO2 emissions levels. We quantified the VEC of each nation, since this can have a significant role in encouraging active participation in creating and enforcing cooperative plans to mitigate CO2 emissions, when common responsibility and international efforts are required to address climate change. Known factors influencing CO2 emissions include population, industrial structure, and energy intensity. We decomposed CO2 emissions into these factors using the logarithmic mean Divisia index (LMDI). The results of our analysis can be utilized for establishing GET policies for CO2 reduction. In addition, the VEC information can be used as a barometer to pursue common environmental objectives or to enact international treaties related to climate change. The outline of the paper is as follows. In section 2, we clarify our research purpose by reviewing the previous literature. In section 3, we introduce a research framework, explain the selected methodologies of two-stage data envelopment analysis (DEA) and LMDI, and describe the data used. In section 4, we show the DEA results in terms of technical efficiency and VEC for each country. In section 5, we discuss the results along with various CO2 reduction polices in the studied European countries. Finally, in section 6, we conclude with the implications and limitations of this research, presenting areas for further study. 2. Literature review DEA, proposed by Charnes et al. (1978), is a non-parametric frontier efficiency technique using linear programming for efficiency estimation. DEA has drawn increasing attention in various industries and environmental fields (Han and Sohn, 2011; Sohn, 2006; Sohn and Choi, 2006; Sohn and Kim, 2012). In particular, DEA methods that are appropriate to environmental fields and indices for measuring efficiency have been developed (Shi et al., 2010; Zhou and Ang, 2008b; Chang et al., 2013). Because CO2 is an undesirable good that causes a problem of biased efficiency measurements in DEA, it was excluded from the analysis until Zhou and Ang (2008b) used CO2 as an output variable in their DEA model. Subsequently, many studies have used CO2 emissions as input or output in DEA (Sueyoshi and Goto, 2010, 2011; Wu et al., 2012; Zhou et al., 2014). Sueyoshi and Goto (2010, 2011) measured production efficiency in fossil fuel electricity generation using CO2 emissions. Wu et al. (2012) gauged the industrial energy efficiency of CO2 emissions by applying an energy efficiency performance index to a DEA model. Oggioni et al. (2011) measured the ecoefficiency of cement production processes for 21 cement industries in various countries by using DEA with CO2 emissions as input or undesirable output. Zhou et al. (2014) also used CO2 emissions as undesirable output in a DEA model to investigate the optimal allocation of CO2 emissions in several regions of China, the result of which showed that the spatiotemporal allocation strategy could be a good alternative for attaining the optimal control of CO2 emissions. Decomposition analysis has also been applied in environmental fields (Bale
zentis et al., 2016; Miao et al., 2016). In particular, it has been used to quantify the effects of several factors on CO2 emissions (Sun, 1998; Ang and Zhang, 2000; Wang et al., 2005; Lin et al., 2006). Decomposition analysis investigates the impact of each

factor on CO2 emissions by decomposing the changes in CO2 into well-defined identities such as emission intensity, energy intensity, and economic activity effects (Ang and Pandiyan, 1997; Ang and Zhang, 2000; Lin et al., 2006; Sun, 1998; Wang et al., 2005). By applying the Divisia index method, Lin et al. (2006) decomposed CO2 emissions of Taiwan into four factors, including emission coefficient and energy intensity, to identify dominant factors of CO2 emissions, providing some insights for the CO2 reduction strategies. In particular, LMDI is preferred over many available decomposition models for CO2 emissions because it can be applied to small datasets and can measure the exact effects of many factors on CO2 emissions with zero residual terms (Ang et al., 2003; Liu et al., 2007; Lin and Ouyang, 2014). Lin and Ouyang (2014) applied LMDI to the non-metallic mineral industry of China to investigate five factors causing CO2 emissions changes. The results suggested that energy intensity and industrial activity are the main drivers of changing CO2 emissions in the industrial sector. In recent years, methods combining DEA and decomposition analysis have been utilized in many studies (Chen and Duan, 2016; Wang et al., 2015; Zhou and Ang, 2008a; Zhang et al., 2013b). These studies compute the efficiency between input and output variables through a distance function based on DEA or linear programming similar to the principle of DEA. The produced efficiency is combined with identities in decomposition analysis, creating new identities that reflect technological effects in the environmental sphere and in particular their impacts on changes in CO2 emissions. Zhou and Ang (2008a) suggested a production-theoretical decomposition analysis (PDA) that decomposes the change in CO2 emissions into several factors using the Shephard distance function. Chen and Duan (2016) also applied PDA to explore the impact of factors causing CO2 emissions in a specific region of China during 2001e2010. They evaluated the impact of six driving factors on CO2 emissions, revealing technical efficiency and technological progress were the main contributors to CO2 reduction. Zhang et al. (2013b) proposed an alternative decomposition method with a distance function based on DEA to measure the effect of changes in high and low technical efficiencies. They decomposed CO2 emissions in 25 OECD countries and China into ten factors. Kim and Kim (2012) also combined the Shepard distance function with LMDI to assess energy efficiency. The authors found that the dominant contributing factors to CO2 reductions in most of the OECD and non OECD countries are potential energy intensity and energy mix among seven components. However, previous research combining DEA with decomposition analysis considered technological changes and effects to have already been reflected in the decomposed factors. Although the impact of the decomposed factors on CO2 emissions was elucidated with data on energy sources and energy consumption, the technical efficiency of the relationship between CO2 emissions and technology was also indirectly investigated. However, to measure exact technical efficiency, the effect of technology on CO2 emissions changes under the direct relationship between input and output must be investigated. Notably, there have been no studies investigating the effect of technology on CO2 mitigation thus far that use patent data and CO2 emissions as input and output variables. To measure the direct effect of technology on CO2 emissions, patent dataduseful as a proxy of technology developmentdare used in this study. We used LMDI to more accurately investigate the technological performance of CO2 reductions by calculating in detail the amount of CO2 emissions changed by technology. Although patents do not represent all technologies related to CO2 reductions, they can explain many aspects of CO2 reduction when used as input variables for investigating CO2 emissions (Popp,

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

2001; Hall and Helmers, 2010; Weina et al., 2015). The environmental awareness level, defined as the broad academic or political interest in issues related to the environment, has been widely studied since it has a positive influence on solving environmental problems through individual behaviors (Krause, ka et al., 2013). Zso ka et al. 1993; S anchez and Lafuente, 2010; Zso (2013) argued that environmental consciousness is affected by environmental education and contributes to addressing environmental problems by helping people shape pro-environmental attitudes like sustainable consumption. Moreover, research considering both environmental awareness level and environmental policies has been undertaken, as environmental consciousness influences environmental policies (Moraga-Gonzalez and Padron-Fumero, 2002; Sarrica et al., 2016). In other words, a nation’s environmental consciousness, or its sense of responsibility towards and awareness of environmental protection, has a significant role in setting environmental policy direction for a nation. However, there have been few studies that quantify national environmental consciousness whilst considering the environment and technological development levels. In this study, we quantified nations’ VEC levels in terms of technology by using a DEA method, considering the effect of CO2 emissions on patent applications.

111

Fig. 1. Two-stage data envelopment analysis (DEA) model based on a two-way relationship between patents and CO2 emissions.

changes in CO2 emissions decomposed by LMDI as inputs and GETrelated patents as outputs. Considering the time lag between the first and second stages, we interpreted the level of technical efficiency in a country and their consistent effort to address climate change through technology according to the four types of classification based on the results of each stage in the DEA.

3.2. Methodology 3. Material and methods 3.1. Research framework We measured the technical efficiency and VEC of European countries in each stage of a two-stage DEA, considering the interactive relationship between technology and CO2 emissions. Based on the results, we classified the countries into four types as displayed in Table 1. Fig. 1 presents our two-stage DEA model for technical efficiency and VEC. In the first stage (technical efficiency), we measured how effectively GET reduces CO2 emissions using patent data. We used GET-related patents in several fieldsdincluding fossil fuels, renewable energy, and storagedas inputs, and changes in CO2 emissions as outputs. To elucidate CO2 emissions for each driving factor, we used LMDI, which is a useful decomposition method for quantifying the impact of several factors on CO2 emissions. With LMDI, we decomposed CO2 using individual factors. In the second stage, we quantified VEC by measuring country efforts to develop GET. For measuring the environmental consciousness of a nation, it is necessary to consider more technology fields associated with CO2 reduction than those included in the first stage. We extended the range of GET-related patents, adding solar energy, nuclear energy, and CO2 capture and storage (CCS) technologies. We also used

3.2.1. LMDI In order to analyze changes in CO2 emissions with respect to each factor, a decomposition method was used. The influence of each factor on changes in CO2 emissions may vary. Decomposition is a useful way to assess CO2 emissions trends and effects of the propulsive forces behind emissions. In general, index decomposition analysis (IDA) is divided into two approaches: (1) Laspeyres and (2) Divisia index approaches. LMDI belongs to the latter group. There have been many studies examining the effects of factors on energy consumption or CO2 emissions using LMDI. LMDI can obtain exact decomposition results without residual terms. In addition, it can be separated into multiplicative and additive decomposition models. The multiplicative decomposition model decomposes ratio changes, while the additive decomposition model decomposes the different amounts of change (Ang et al., 1998; Ang and Liu, 2001). We used LMDI since it is effective for accurately analyzing the effects of factors on CO2 emissions, in addition to the previously mentioned advantages. We used an additive decomposition model to gauge the different amounts of change in CO2 emissions. The extended Kaya identity, developed by Jung et al. (2012), was used to investigate the effect of energy consumption from fossil fuels. Since fossil fuels have significant effects on energy consumption, many studies have used the Kaya identity. We decomposed CO2 emissions

Table 1 Classification of countries based on technical efficiency and voluntary environmental consciousness (VEC) by two-stage data envelopment analysis (DEA). First stage

Second stage

Technical efficiency

VEC

High

Low

Case

Description

High

A

Low

B

High

C

Low

D

Countries with relatively more CO2 reduction due to high technical efficiency in the early stage and with high VEC thereafter, striving for GET-related patent applications Countries with relatively more CO2 reduction due to high technical efficiency in the early stage but with low VEC thereafter, with relatively minimal effort in making GETrelated patent applications Countries with relatively less CO2 reduction due to low technical efficiency in the early stage but with high VEC thereafter, striving to apply for GET-related patents Countries with relatively less CO2 reduction due to low technical efficiency in the early stage and with low VEC thereafter, with relatively minimal effort in making GET-related patent applications

112

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

DCmix ¼ changes in CO2 emissions due to changes in the relative

into five factors used by Line et al. (2013) as follows:

C ¼

P

Pi 

ij

¼

P

ECij GDPi TECi    Pi GDPi TECi

share of fossil fuel energy consumption to total energy consumption, DCeff ¼ changes in CO2 emissions due to changes in the carbon emissions per unit of fossil fuel energy consumed.

Cij ECij

(1)

Pi  Gi  Ii  Mij  Cij

ij

where C denotes carbon emissions from energy use, P is the population size, GDP is the gross domestic product, TEC is the total energy consumption, EC is the fossil fuel energy consumption, the subscripts i and j represent the year and fuel type. G (formulated as GDP/P) denotes the GDP per capita where GDP is measured in terms of current USD, I (formulated as TEC/GDP) denotes energy intensity, M (formulated as EC/TEC) denotes fuel mix where TEC and EC are measured in terms of TJ, and C (formulated as CO2/EC) denotes the CO2 emissions coefficient where CO2 is measured in terms of Mt. However, for the measurement of technical efficiency, we used only three factors: energy intensity, fuel mix, and the CO2 emissions coefficient, since these factors include energy consumption as a structural element and are closely related to the development of energy efficiency technologies. Energy intensity represents the ratio of total energy consumption to GDP in a year. High energy intensity indicates high costs associated with converting energy into GDT, or low energy efficiency. Fuel mix refers to the relative share of fossil fuel energy consumption to total energy consumption in the country, meaning a country with high fuel mix mainly depends on energy from fossil fuel. The CO2 emissions coefficient is the ratio of carbon emissions to fossil fuel energy consumption. A high CO2 emissions coefficient means that much of the carbon is emitted by fossil fuel energy consumption. Factor values used for measuring technology efficiency related to CO2 reduction were obtained through the additive LMDI formulae described in Table 2. By the LMDI, the total change in CO2 emissions between the reference year 0 and a target year T (DCtot ) can be decomposed into the following five factors:

DCpop ¼ changes in CO2 emissions due to changes in the population, DCpdn ¼ changes in CO2 emissions due to changes in the GDP per capita, DCint ¼ changes in CO2 emissions due to changes in the total energy consumption per unit of GDP, Table 2 Additive logarithmic mean Divisia index (LMDI) formulae for decomposing changes in CO2 emissions. CO2 emission factor

C¼

P j

EC

j TEC P  GRDP P  GRDP  TEC 

CO2j ECj

Total Effect

DCtot ¼ C T  C 0 ¼ DCpop þ DCpdn þ DCint þ DCmix þ DCeff Effect by Factor

DCpop ¼ DCpdn ¼ DCint ¼ DCmix ¼ DCeff ¼

 

P

CTj C0j

j

ln CTj ln C0j

P

CTj C0j

j

ln CTj ln C0j

P

CTj C0j

j

ln CTj ln C0j CTj C0j

j

ln CTj ln C0j

P

CTj C0j

j

CTj ln

ln

C0j

  T ln G0 G

 

ln

P

PT P0

ln

ln

ln

T

I I0 MTj M0j ETj E0j

!

!

3.2.2. Two-stage DEA DEA has been extensively used for measurement of efficiency in a variety of areas, including global environmental problems. It includes two models: the CCR proposed by Charnes et al. (1978) (named after its deviser Chames, Cooper and Rhodes), and the BCC developed by Banker et al. (1984) (named after its deviser Banker, Chames and Cooper). The CCR model provides an efficiency score between input and output under constant return to scale (CRS). The BCC model uses the assumption of variable return to scale (VRS); it alleviates the previous CRS assumption of convexity conditions, and can calculate scale efficiency (SE). We selected the BCC model under the practical assumption that the rate of CO2 reduction would differ at different levels of related patent applications. Moreover, we undertook output-oriented DEA in both stages, where output efficiency is measured when inputs are fixed at given levels. We obtained an efficiency score equating to technical efficiency (first stage) and VEC (second stage). We used multiple inputs and outputs with three types of patents related to CO2 reduction and the LMDI-decomposed factors, since the DEA method is useful for measuring efficiency between multiple inputs or outputs (Charnes et al., 1982). Through DEA analysis, we identified the effect of GET-related patents on CO2 emissions in detail in relation to a specific factor, and those fields where inputs or outputs must be increased to improve efficiency. Two-stage DEA measures the efficiency between inputs and outputs in the first stage and uses the outputs of the first stage as intermediates. In the second stage, the intermediates are used as inputs and the efficiency of the second stage is calculated by intermediates and final outputs. Previous research has utilized twostage DEA for measuring the efficiency of sub-processes and comparing it with that of the whole process (Seiford and Zhu, 1999; Kao and Hwang, 2008). Seiford and Zhu (1999) divided a production process into two-stage processes and measured performances of profitability (first stage) and marketability (second stage). However, unlike the general framework of two-stage DEA that separates one process into sub-processes, we operated in terms of technical efficiency and VEC at the country level, based on the interactive and circular relationship between CO2 emissions and technologies related to CO2 reduction. In the first stage of DEA, we measured technical efficiency: how effectively GET reduces CO2 emissions with multiple inputs and outputs. GET-related patents and LMDI factors influenced by technical factors were used as inputs and outputs. Although countries reduced CO2 emissions equally with respect to each factor, the effect of CO2 reductions could be different for each country depending on their total CO2 emissions. Therefore, we divided CO2 emissions changes calculated by LMDI by the total CO2 emissions for each country, so that the relative proportion of CO2 changes could be used as output values in the first stage. In order to measure the technical efficiency of CO2 reduction through DEA, an adjustment to the amount of CO2 change calculated by LMDI was necessary. Since CO2 is an undesired good that has an increasingly negative impact as its change in emissions becomes larger, we changed CO2 emissions into CO2 reductions by multiplying the CO2 emissions change produced by LMDI by 1. This solved the problem of undesired goods, thereby measuring the technical efficiency of GET. However, where the original value of the input or output was positive, the converted value became negative,

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

113

Table 3 Data sources. Data

Year

Data source

Notes

Emission Factor ðCO2 Þ

2008e2010

Mt (¼1 000 000 t)

GDP per market prices Population Total energy consumption Fuel energy consumption Fuel energy-related patent

2008e2010 2008e2010 2008e2010 2008e2010 2005e2007 2011e2013

Draft 2006 IPCC Guidelines for Greenhouse Gas Inventories The World Bank The World Bank Eurostat Eurostat Wipson [EPO]

causing a further problem. There are many DEA methods that address the problem of the existence of negative values in input or output variables. One such method converts negative values to positive by adding an amount slightly larger than the most negative value to all values in the dataset (Barua et al., 2004; Deng et al., 2007; Rotela et al., 2015). In the second stage, we used CO2, the first stage output, as an input to measure VEC. We used the unconverted values for changes in CO2 emissions in the second stage in terms of original CO2 emissions. We divided the nationwide CO2 emissions changes from each LMDI factor by the total CO2 emissions in order to reflect the relative extent of CO2 emissions as an output value in the second stage. After reflecting the relative proportion of CO2 reduction by each country, we addressed the negative values by adding a small value to all inputs. The model takes into account the time lag between a patent application and the patent coming into effect. It takes a long time to apply for, grant, and commercialize patents. Popp (2001) argued that it takes three years for a patent application to produce results in reducing energy consumption, which occurs simultaneously with CO2 emissions. According to this, the DEA considers a three-year time lag between a patent application and the related CO2 emissions reduction. GET-related patents applied for in the three years from 2005 to 2007 were used as inputs in terms of ‘stock’ since technological influence accumulated over a long time period. Corresponding to inputs, the changes in CO2 reduction by technological factors in the three years from 2008 to



Current USD Population size TJ TJ The number of patent applications

2010 were used as outputs. We selected 2005 as the reference year because this is when the Kyoto protocol took effect and many countries started to make an intensive effort to reduce CO2 emissions. We also considered a three year time lag between CO2 emissions and the patent applications that are filed in the second stage similarly to the first, since annual changes in CO2 emissions influence the level of research and development (R&D) investment for CO2 reduction technologies the following year, and R&D investment affects the application for patents in related fields with a oneor two-year time lag (Prodan, 2005; Corsatea, 2014). In this context, we used GET-related patents from 2011 to 2013 as output in the second stage, reflecting the three-year time lag with changes in CO2 emissions from 2008 to 2010 being used as input. 3.3. Data The 15 countries that comprise the EU Bubble have made great efforts to reduce CO2 emissions. We selected 12 of these countries for analysis: Austria, Belgium, Denmark, Finland, France, Germany, Italy, Luxembourg, the Netherlands, Spain, Sweden, and the United Kingdom. We excluded three countries: Greece, Portugal, and Ireland, since the total number of GET-related patents applied for in these countries between 2005 and 2007 was less than ten. The selected countries are members of the Organization for Economic Cooperation and Development (OECD), which strives to reduce CO2 emissions with responsibility for common CO2 reduction objectives



IURPWR07

7KHDYHUDJH&R(PLVVLRQV

 







 

 

   



 

 



Fig. 2. Average fuel-energy related CO2 emissions in 12 European countries from 2007 to 2010.





114

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

Table 4 Ranking of countries in terms of CO2 reductions during 2007e2010 with respect to three technical factors. Country

Energy intensity

Fuel mix

Emission coefficient

Austria Belgium Denmark Finland France Germany Italy Luxembourg Netherlands Spain Sweden United Kingdom

6(6) 10(11) 4(4) 3(3) 8(7) 11(10) 2(5) 5(2) 9(9) 1(1) 7(8) 12(12)

5(5) 11(12) 4(1) 6(4) 3(6) 12(10) 8(8) 7(7) 9(11) 1(3) 2(2) 10(9)

9(9) 5(5) 6(4) 8(8) 4(6) 12(12) 1(2) 7(7) 2(1) 11(10) 10(11) 3(3)

* Ranking is placed inside parentheses, based on the CO2 reduction divided by the level of total CO2 emissions in 2014.

committed to by the EU in international agreements including the Kyoto Protocol. Most CO2 emissions are derived from fuel combustion, primarily by fossil fuels (Zou and Ang, 2008). Some CO2 emissions are generated from biomass and biofuel in renewable energy. Accordingly, we calculated fuel energy-related CO2 emissions from the energy consumption of each fuel type, by following the 2006 Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventory (IPCC, 2006). Fuel energy consumption data were collected from the Eurostat database (2016), and were combined with emissions factors from the IPCC for computing CO2 emissions. The data sources for LMDI included total energy consumption, GDP per capita, and national population (Table 3). Reduction of CO2 emissions can be achieved in several ways, including through improvement of energy efficiency, use of renewable or nuclear energy, and CCS. In particular, efficiencyimproving technologies relating to fossil fuels, renewable energy, and storage technology can have a great effect on CO2 emissions mitigation. The International Patent Classification(IPC) codes for eight kinds of GET patents related to fossil fuels, renewable energy, and storage technology were used in the same way as in earlier studies (Lanzi et al., 2011; Noailly and Shestalova, 2013; Xiang et al., 2016). Based on these IPC codes, GET patents were utilized for measuring technical efficiency in the first stage. In the first stage of DEA, technical efficiency was measured in terms of the amount by which GET reduced CO2 emissions. Since CO2 emissions, the output, were calculated directly from fuel consumption, we confined our analysis to technologies related to carbon-containing energies among GET or those that reduce CO2 emissions by improving the efficiency in the process of emitting CO2 from energy consumption of various fuel types. For example, in renewable energy, only efficiency improving technologies that emit CO2 from energy consumption, including biomass energy and waste-to-energy, were included for efficiency measurements. In the second stage of DEA, because all technologies associated with CO2 reduction must be considered for measuring environmental consciousness, we added several technology fields not included in the first stage, including solar power, wind power, nuclear energy, and CCS technology, which are detailed in Appendix A. Since the countries selected for analysis are all European, we utilized patent applications made to the European patent office (EPO) from 2005 to 2008 and from 2011 to 2013 by downloading them through the patent data base WIPSON. The IPC codes from which we collected patent data are described in Appendix A.

4. Results We calculated the energy-related CO2 emissions for the 12 selected European countries from 2007 to 2010 (Fig. 2). Germany, the country with the highest greenhouse gas (GHG) emissions in Europe, has the highest level of CO2 emissions, followed by France, England, Italy, and Spain. We found CO2 emissions varied according to the energy industrial structure, technology level, and scale of energy consumption of each country, though all are OECD countries. Luxembourg emitted 11 MT of CO2, which was the lowest of the 12 countries. Table 4 shows the ranking of CO2 reductions for three technical factors in countries during 2007e2010. When considering the level of total CO2 emissions for each country, Spain, Denmark, and the Netherlands reduced CO2 emissions the most for each technical factor. Despite different levels of CO2 emissions for each country, the technical efficiency of GET could be compared through DEA by focusing on changes in CO2 reductions. Table 5 shows the results of the first stage of DEA. By applying the BCC model, we also measured SE, which can be calculated by dividing the efficiency score from the CCR into that of the BCC. Nine countries have efficient technologies and obtained the highest efficiency score of 1, while three countries did not (France, Germany, and the United Kingdom). Countries with relatively lower technical efficiency scores tended to have many more GET-related patent applications than did other countries. High technical efficiency means that the GET efficiently reduced CO2 emissions. Although the 12 countries had different levels of CO2 emissions and patent applications, we identified many countries that reduced CO2 emissions at an efficient scale. Denmark, Spain, and Italy were recommended as benchmark countries for Germany in the first stage of DEA. Given the level of patent application, it was necessary to reduce the CO2 emissions from the emission coefficient, fuel mix, and energy intensity effects to attain high technical efficiency. Benchmark countries for France were Denmark, the Netherlands, Spain, and Italy. France could also increase technical efficiency by increasing CO2 reduction with respect to the three factors. Finally, benchmarking countries for the United Kingdom were Denmark and the Netherlands. The UK could become a high technical efficiency country by reducing CO2 emissions from the effects of the three factors. When considering the SE of each country, seven countries (Austria, Denmark, Finland, Spain, Sweden, Luxembourg, and Belgium) had efficiency scores of 1 in both technology and scale by the BCC model, which means that they efficiently reduced CO2 emissions at the most productive scale. Germany and France have

Table 5 Technical efficiency measured in the first stage of data envelopment analysis (DEA). Decision Making Unit (DMU)

CCR (%) (Total Technical Efficiency)

BCC (%) (Pure Technical Efficiency)

Scale Efficiency (SE)

Austria Belgium Denmark Finland Luxembourg Spain Sweden Netherlands Italy United Kingdom France Germany

1 1 1 1 1 1 1 0.58 0.27 0.23 0.15 0.04

1 1 1 1 1 1 1 1 1 0.81 0.88 0.67

1 1 1 1 1 1 1 0.58 0.27 0.28 0.17 0.06

* The BCC score was taken to be the technical efficiency in this study.

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120 Table 6 Voluntary environmental consciousness (VEC) in the second stage of DEA. DMU

VEC

RANK

Denmark France Germany Italy Netherlands Spain Sweden United Kingdom Finland Austria Belgium Luxembourg

1 1 1 1 1 1 1 0.51 0.28 0.20 0.15 0.14

1 1 1 1 1 1 1 8 9 10 11 12

1

considerably higher energy-related CO2 emissions than other countries. France has a relatively low score (0.88) in technical efficiency and a very low SE score. Germany had the lowest score in both technical efficiency and SE out of the 12 countries studied. In the second stage of DEA, we quantified VEC based on levels of CO2 emissions and technology development as measured by the number of relevant patent applications. Because responsibility for resolving climate change in terms of international cooperation requires consistent reduction in CO2 emissions, we identified countries with high VEC. As shown in Table 6, seven countries (Denmark, France, Germany, the Netherlands, Spain, Sweden, and Italy) had the highest VEC scores. On the contrary, except for the UK with an intermediate VEC of 0.51, the other four countries had low VEC scores. Luxembourg, Finland, Austria, and Belgium had VEC scores of 0.14, 0.28, 0.2, and 0.15, respectively. Fig. 3 shows the relative location of the 12 countries classified by technical efficiency and VEC. All countries show technical efficiency of more than 0.5 in general, but vary greatly in terms of VEC. Five countries with the highest technical efficiency (Denmark, the Netherlands, Spain, Sweden, and Italy) also have the highest VEC, belonging to case A as described in Table 1. These countries reduced

VEC (In the latter Stage)

Germany

115

CO2 emissions efficiently in the early stages and continuously endeavored for further mitigation by subsequently applying many patents. These countries provide an outstanding reference group in terms of technical efficiency and VEC. Although Germany and France in case B had relatively low technical efficiency scores, we identified that they consistently strived to reduce CO2 emissions through their high VEC score. Since they predominately applied many GET-related patents in four fields and had high CO2 emissions, they had high VEC. The United Kingdom, in case C, has a unique situation. It showed relatively low scores in both technical efficiency and VEC. Although it had many GET-related patent applications, its GET did not efficiently reduce CO2 emissions. In addition, given its level of CO2 emissions, it did not develop an extended scope of GET, including nuclear energy and CCS, in contrast to other countries with high VEC scores. Four countries qualified for case D, including Luxembourg and Belgium, with the highest technical efficiencies computed in the first stage of DEA and relatively low VEC scores, are located in the bottom right of Fig. 3. These countries reduced much of their CO2 emissions with GET, but thereafter made less effort toward further reduction via GET-related patents. 5. Discussion We discovered that following enforcement of the Kyoto Protocol in 2005, most of the 12 European countries efficiently reduced CO2 emissions with GET except for Germany, France, and the United Kingdom. In terms of VEC, seven countries showed their endeavor to reduce CO2 emissions, while the remaining five had relatively low VEC. The relatively low technical efficiency of Germany and France appears to be related to the large number of GET-related patents as compared to their CO2 reduction. However, their high VEC indicated strong willingness to address climate change. They invested most heavily in environmental technologies, paying a substantial amount of attention to environmental issues such as climate change, and enacting exemplary environmental policies over several decades. Germany, where coal-fired power plants account

France

Denmark Netherlands Spain Sweden Italy

United Kingdom Finland Austria Belgium

0

Luxembourg

0.5

1

Technical Efficiency (In the early stage) * The scale range is different for each axis. Fig. 3. Mapping of relative locations of the 12 countries in terms of technical efficiency and voluntary environmental consciousness (VEC).

116

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

for a large proportion (about 44% in 2014) of electricity production, has continued to invest significant amounts in fossil fuel energy efficiency and CCS technologies, with by far the largest number of energy-related patent applications (Kraeusel and Most, 2012; Sabrina and Julian, 2015). However, it is now actively promoting the development of a low-carbon energy conversion system with the Energiewende policy (2011), which focuses more on a reduction in the use of coal-fired power plants and on renewable energy technology (Ehret and Bonhoff, 2015). France, which continuously expanded its nuclear electricity production to 77% in 2014 (IAEA, 2015), also enforced the Grenelle environmental policy in 2009, which emphasizes the significance of renewable energy, and the need to continually strive for development of renewable energy (European Commission, 2009). In addition, France introduced energy conversion method for the green growth in 2014, promoting a structural conversion of the energy industry, mainly with respect to renewable energy (Matthes et al., 2015). Considering this, its high VEC is expected to be maintained. Under the UK Low Carbon Transition Plan (2009) by The Department of Energy & Climate Change (DECC), the UK Office for Renewable Energy Development has actively made efforts to increase the rate of electricity production from renewable energy for CO2 reduction (Foxon, 2013). In particular, it focused on developing CCS technology and applying it to energy infrastructure with governmental support, and had many renewable and CCS-related patent applications (IEA, 2009). However, reduction in CO2 emissions from 2007 to 2010 was mainly derived from reduced production related to GDP per capita. The magnitude of effect from the three technological factors on CO2 reduction was small due to the relatively low technical response to increased CO2 emissions. Countries in case A, including Sweden, Spain, and the Netherlands, were the most outstanding reference group in terms of technical efficiency and VEC, which implies that these countries met the international agreement to address climate change well, and have had a significant role in reducing CO2 emission. Sweden, the top renewable energy user in 2013, had remarkable growth in the renewable energy field by implementing a tradable green certificate scheme that promoted an increase in renewable energy production, such as from bio-fuel or wind (Oikonomou and Mun€derholm, 2009). The high VEC for daca, 2008; Pettersson and So Sweden shows that efforts to reduce CO2 emissions in Sweden have been successful. In Spain, as industrial activities sharply declined during the economic recession in 2009, CO2 emissions also declined along with the shrinking industrial activity. This reduction made a major contribution to Spain being selected for the reference group (Alves and Moutinho, 2013). With its strong will for making large CO2 reductions, the Netherlands has invested much in the renewable energy field by enforcing renewable energy-supporting policies, such as the Sustainable Energy Production Plus policy (2008) or the Energy Research Subsidy (2009) (De Bruijne et al., 2016; Hrusc et al., 2011). It attained an advanced level in renewable energy technologies, including wind and solar energy production, which was boosted by strong governmental support. Conversely, countries in case D, such as Finland, Austria, Luxembourg, and Belgium, made less effort to consistently develop GET, despite having efficient GET for CO2 reduction. With small total CO2 emissions in these countries, the number of patents related to GET was also relatively low. In these countries, R&D investment in GET needs to be increased, as do efforts to apply for more GET-related patents as a way to address climate change and develop VEC.

6. Conclusion CO2 emissions derived from energy consumption have caused serious environmental problems such as climate change and global warming. Countries facing these problems have been striving to mitigate CO2 emissions in several sectors. Efforts have included developing GET that can contribute to reducing CO2 emissions generated by energy consumption. However, as environmental issues become more acute, the importance of efforts to increase the performance of GET for CO2 emissions reduction and to promote active international cooperation for solving climate change is crucial. In order to induce such efforts, it is necessary to measure the current status. In this study, we proposed a framework to quantify such measures by using two-stage DEA. This framework was applied to compare the technical efficiency of GET and the VEC for 12 European countries. In the first stage of DEA, which reflected time lag, we calculated the technical efficiency of GET in each country for mitigating CO2. Thereafter, the degree of technological effort by each country to environmental problems in terms of VEC was calculated. Based on these two analyses, we classified European countries into four groups. The five countries (Denmark, Netherlands, Spain, Sweden, and Italy) that were assigned to the reference group consistently made technological efforts to reduce CO2 emissions with their highly efficient GET for CO2 reduction. Although France and Germany showed low technical efficiencies in the first stage, they were very active in developing GET, including nuclear energy and new renewable energy, giving them both a high VEC score. On the other hand, four countries (Finland, Luxembourg, Austria, and Belgium) with high technical efficiency proved to be less environmentally conscious about CO2 emissions reduction than other countries. In the case of the United Kingdom, both technical efficiency and environmental consciousness need to be enhanced by developing GET. Our findings are helpful for managing effective and consistent CO2 emissions reductions. Information related to the technical efficiency of each country, including the reference countries, is conducive to making technology policy and setting the direction of technology development for CO2 emissions reductions. In addition, we quantified national interest and sense of responsibility regarding environmental problems with VEC scores. These could be utilized for urging reluctant countries to have greater environmental consciousness and responsibility in attaining international environmental goals. Moreover, the proposed two-stage DEA framework with time lag can be applied to other countries to identify and compare their technical efficiencies and VEC scores. However, there are some limitations to our research. First, in the process of measuring technical efficiency, though GET as we defined it had a great effect on national CO2 emissions, it did not completely account for all CO2 reductions. Further research is needed to include other technology sectors that are related to the reduction of CO2 emissions so as to more comprehensively and accurately measure the technical efficiency of CO2 reduction technology. If transferred or licensed technologies are considered in addition to the applied patents, GET can also cover a wider range of technologies related to CO2 reduction. Second, we measured the technical efficiency and VEC of eight countries during a specific period. In order to identify change trends over time, dynamic analysis will be necessary. Acknowledgement This work was supported by the National Research Foundation of Korea grant funded by the Korea government (MSIP) (2016R1A2A1A05005270).

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

117

APPENDIX A Appendix A-1. IPC classes for fossil fuel technologies for efficiency improvement in energy generation.

Technology Fossil Fuel Energy

Description

IPC classes

1. Coal (first & second stages) Coal Production of fuel gases by carbureting air or other gases without pyrolysis 2. Combustion (first & second stages) Hot-Gas Hot-gas or combustion-product positive-displacement engine; use of waste heat of combustion engines not otherwise provided for Turbines Gas-turbine plants; air intakes for jet-propulsion plants; controlling fuel supply in air-breathing jetpropulsion plants Ignition [Classes listed below excluding combinations with B60, B68, F24, F27] Engines characterized by fuel-air mixture compression ignition Engines characterized by air compression and subsequent fuel addition; with compression ignition Engines characterized by the fuel-air charge being ignited by compression ignition of an additional fuel Engines characterized by both fuel-air mixture compression and air compression, or characterized by both positive ignition and compression ignition, e.g., in different cylinders Engines characterized by the introduction of liquid fuel into cylinders by use of auxiliary fluid; compression ignition engines using air or gas for blowing fuel into compressed air in the cylinder Methods of operating air-compressing compression-ignition engines involving introduction of small quantities of fuel in the form of a fine mist into the air in the engine intake Burners Combustion apparatus; combustion processes Fluidized bed combustion Chemical or physical processes in general, conducted in the presence of fluids and solid particles; apparatus for such processes; with liquid as a fluidizing medium Chemical or physical processes in general, conducted in the presence of fluids and solid particles; apparatus for such processes; Fluidized-bed furnaces; other furnaces using or treating finely-divided material in dispersion 3. Steam (first & second stages) Steam Steam generation Engines Steam engine plants; steam accumulators; engine plants not otherwise provided for; engines using special working fluids or cycles

C10J

F02G F02C

F02B1/12-14 F02B3/06-10 F02B7 F02B11 F02B13/02-04 F02B49 F23 B01J8/20-22 B01J8/24-30 F27B15 F22 F01K

Source: Lanzi et al. (2011) and Noailly and Shestalova (2013).

Appendix A-2. IPC classes for renewable energy technologies for efficiency improvement in energy generation.

Renewable Energy 1. Carbon-free renewable energy (second stage) Wind powera Wind motors Solar energya Devices for producing mechanical power from solar energy Use of solar heat, e.g., solar heat collectors Drying solid materials or objects by processes involving the application of heat by radiation, e.g., from the sun Devices consisting of a plurality of semiconductor components sensitive to infra-red radiation or light that are specially adapted for the conversion of this radiant energy into electrical energy Semiconductor devices sensitive to infra-red radiation, light, electromagnetic radiation of shorter wavelengths, or corpuscular radiation, specially adapted as devices for the conversion of this radiant energy into electrical energy, including as a panel or array of photoelectric cells, e.g., solar cells Generators in which light radiation is directly converted into electrical energy Geothermal energya Devices for producing mechanical power from geothermal energy Production or use of heat, not derived from combustion; using geothermal heat a Marine (Ocean) energy Tide or wave power plants Submerged units incorporating electric generators or motors characterized by using wave or tide energy Ocean thermal energy conversion Hydro powera Water-power plants: layout, construction or equipment, methods of, or apparatus for, but not tide or wave power plants Machines or engines for liquids of reaction type; water wheels; power stations or aggregates of water-storage type; machine or engine aggregates in dams or the like; controlling machines or engines for liquids; and NOT submerged units incorporating electric generators or motors characterized by using wave or tide energy 2. Carbon-containing renewable energy (first & second stages) Biomass energy Solid fuels based on materials of non-mineral origin, animal or vegetable substances Engines or plants operating on gaseous fuels from solid fuel, e.g., wood Waste-to-energy

F03D F03G6 F24J2 F26B3/28 H01L27/142

H01L31/042-058

H02N6 F03G4 F24J3/08 E02B9/08 F03B13/10-26 F03G7/05 E02B9; and not E02B9/08 [F03B3,F03B7, F03B13/06-08 or F03B15] and not F03B13/10-26

C10L5/42-44 F02B43/08 C10L5/46-48 (continued on next page)

118

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

(continued ) Solid fuels based on materials of non-material origin: sewage, town or house refuse, industrial residues or waste materials Incineration of waste and recuperation of heat Incinerators or other apparatus consuming waste, field organic waste Liquid carbonaceous fuels; gaseous fuels; solid fuels; and dumping solid waste; destroying solid waste or transforming solid waste into something useful or harmless; incineration of waste; incinerator Plants for converting heat or fluid energy into mechanical energyduse of waste heat; profiting from waste heat of combustion engines; machines, plant, or systems, using particular sources of energydusing waste heat, and incineration of waste; incinerator constructions; incinerators or other apparatus specially adapted for consuming specific waste or low grade fuels

F23G5/46 F23G7/10 [C10L1, C10L3, C10L5] and [B09B1, B09B3, F23G5, F23G7] [F01K27, F02G5, F25B27/02] and [F23G5, F23G7]

Source: Lanzi et al. (2011) and Noailly and Shestalova (2013). a Technologies that were added in the second stage of DEA.

Appendix A-3. IPC classes for storage technologies in energy generation.

Storage Technology

1. Accumulators in power generation (first & second stages) Storage Lead-acid accumulators Alkaline accumulators Gastight accumulators Other types of accumulators not provided for elsewhere 2. CO2 capture and storage (second stage) a CCS Separation of gases or vapors; recovering vapors of volatile solvents from gases; chemical or biological purification of waste gases, e.g., engine exhaust gases, smoke, fumes, flue gases or aerosols Carbon; compounds thereof Solid sorbent compositions or filter aid compositions; sorbents for chromatography; processes for preparing, regenerating or reactivating thereof Methods or apparatus for obtaining oil, gas, water, soluble or meltable materials or a slurry of minerals from wells Processes or apparatus for separating the constituents of gaseous mixtures involving the use of liquefaction or solidification Chemical, physical, or physicochemical processes in general Semi-permeable membranes for separation processes or apparatus characterized by the material; manufacturing processes specially adapted thereof Arrangements of devices for treating smoke or fumes Separation Processes or apparatus for liquefying or solidifying gases or gaseous mixtures Chemical or physical processes in general, conducted in the presence of fluids and solid particles; apparatus for such processes Degasification of liquids Semi-permeable membranes for separation processes or apparatus characterized by their form, structure or properties; manufacturing processes specially adapted thereof

H01M10/06-18 H01M10/24-32 H01M10/34 H01M10/36-40 B01D-053 C01B-031 B01J-020 E21B-043 F25J-003 B01J-019 B01D-071 F23J-015 B01D-000 F25J-001 B01J-008 B01D-019 B01D-069

Source: Lanzi et al. (2011), Noailly and Shestalova (2013), and Xiang et al. (2016). a Technologies that were added in the second stage of DEA.

Appendix A-4. IPC classes for nuclear power technologies in energy generation.

Nuclear Power Technology

1. Nuclear power generation (second stage) Nuclear engineeringa Fusion reactors; thermos nuclear fusion reactors and low-temperature nuclear fusion reactors Nuclear reactors; reactors, reactor elements, control (monitoring and testing), emergency protection, manufacture, adaptation of reactors for experimentation or irradiation Nuclear power plant; details and control of nuclear power plant, arrangements of reactor and engine in which reactor-produced heat is converted into mechanical energy, arrangements for direct production of electric energy from fusion or fission reactions, arrangements to provide heat for purposes other than conversion into power

Source: http://www.wipo.int/classifications/ipc/en/est/ a Technologies that were added in the second stage of DEA.

G21B G21C

G21D

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

References Alves, M.R., Moutinho, V., 2013. Decomposition Analysis for Energy-related CO2 Emissions Intensity over 1996-2009 in Portuguese Industrial Sectors. In Technical Report CEFAGE-UE Working Paper, Portugal, Spain. Ang, B.W., Pandiyan, G., 1997. Decomposition of energy-induced CO2 emissions in manufacturing. Energy Econ. 19 (3), 363e374. Ang, B.W., Zhang, F.Q., Choi, K.H., 1998. Factorizing changes in energy and environmental indicators through decomposition. Energy 23 (6), 489e495. Ang, B.W., Zhang, F.Q., 2000. A survey of index decomposition analysis in energy and environmental studies. Energy 25 (12), 1149e1176. Ang, B.W., Liu, F.L., 2001. A new energy decomposition method: perfect in decomposition and consistent in aggregation. Energy 26 (6), 537e548. Ang, B.W., Liu, F.L., Chew, E.P., 2003. Perfect decomposition techniques in energy and environmental analysis. Energy Policy 31 (14), 1561e1566. Bale
zentis, T., Li, T., Streimikiene, D., Bale
zentis, A., 2016. Is the Lithuanian economy approaching the goals of sustainable energy and climate change mitigation? Evidence from DEA-based environmental performance index. J. Clean. Prod. 116, 23e31. Barua, A., Brockett, P.L., Cooper, W.W., Deng, H., Parker, B.R., Ruefli, T.W., Whinston, A., 2004. DEA evaluations of long-and short-run efficiencies of digital vs. physical product “dot com” companies. Socioecon. Plan. Sci. 38 (4), 233e253. Chang, Y.T., Zhang, N., Danao, D., Zhang, N., 2013. Environmental efficiency analysis of transportation system in China: a non-radial DEA approach. Energy policy 58, 277e283. Charnes, A., Cooper, W.W., Rhodes, E., 1978. Measuring the efficiency of decision making units. Eur. J. Oper. Res. 2 (6), 429e444. Charnes, A., Cooper, W.W., Seiford, L., Stutz, J., 1982. A multiplicative model for efficiency analysis. Socioecon. Plan. Sci. 16 (5), 223e224. Chen, L., Duan, Q., 2016. Decomposition analysis of factors driving CO2 emissions in Chinese provinces based on production-theoretical decomposition analysis. Nat. Hazards 84 (Suppl. 1), 1e11. Corsatea, T.D., 2014. Technological capabilities for innovation activities across Europe: evidence from wind, solar and bioenergy technologies. Renew. Sustain. Energy Rev. 37, 469e479. Deng, C.G., Liu, T., Wu, J., 2007. Efficiency analysis of China’s commercial banks based on DEA: negative output investigation. China-USA Bus. Rev. 6 (2), 50e56. De Bruijne, R., van Erp, F., Leguijt, T., 2016. Get set to bid for the next offshore wind projects in The Netherlands. Renew. Energy Focus 17 (1), 15e16. Ehret, O., Bonhoff, K., 2015. Hydrogen as a fuel and energy storage: success factors for the German Energiewende. Int. J. Hydrogen Energy 40 (15), 5526e5533. European Commission, 2009. Environmental Policy Review, Luxembourg. Accessed 4 January 2017. http://ec.europa.eu/environment/archives/pdf/policy/EPR_ 2009.pdf. European Commission, 2016. Report from the Commission to the European Parliament and the Council, Brussels, Belgium. Accessed 7 March 2017. ec. europa.eu/clima/sites/clima/files/eu_progress_report_2016_en.pdf. Foxon, T.J., 2013. Transition pathways for a UK low carbon electricity future. Energy Policy 52, 10e24. Hadjipaschalis, I., Poullikkas, A., Efthimiou, V., 2009. Overview of current and future energy storage technologies for electric power applications. Renew. Sustain. Energy Rev. 13 (6), 1513e1522. Hall, B.H., Helmers, C., 2010. The Role of Patent Protection in (Clean/green) Technology Transfer (No. w16323). Natl. Bur. of Econ. Res. (NBER), Massachusetts, USA. Han, H.K., Sohn, S.Y., 2011. DEA application to grouping military airbases. Mil. Oper. Res. 16 (2), 31e40. Hrusc, E., Lackovic, P., Oncul, B.O., Wirtz, L.E., 2011. Eur. Renew. Energy Policy Integration. Aalborg University, Aalaborg, Denmark. IAEA, 2015. Nuclear Power Reactors in the World, Reference Data Series No.2. International Atomic Energy Agency, Vienna, Austria. www-pub.iaea.org/MTCD/ Publications/PDF/rds2e35web-85937611.pdf. Accessed 4 January 2017. IEA, 2009. Technology Roadmap: Carbon Capture and Storage. Paris, France. www. iea.org/publications/freepublications/publication/CCSRoadmap2009.pdf. Accessed 4 March 2017. IPCC, 2006. The Intergovernmental Panel on Climate Change, Draft 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Geneva, Switzerland. www.ipcc.ch/meetings/session25/doc4a4b/vol2.pdf. Accessed 4 January 2017. Kao, C., Hwang, S.N., 2008. Efficiency decomposition in two-stage data envelopment analysis: an application to non-life insurance companies in Taiwan. Eur. J. Oper. Res. 185 (1), 418e429. Kim, D.H., Kang, K.Y., Sohn, S.Y., 2016. Spatial pattern analysis of CO2 emission in seoul metropolitan city based on a geographically weighted regression. J. Korean Inst. Indus. Eng. 42 (2), 96e111. Kim, K., Kim, Y., 2012. International comparison of industrial CO2 emission trends and the energy efficiency paradox utilizing production-based decomposition. Energy Econ. 34 (5), 1724e1741. Kim, Y.S., Han, E.J., Sohn, S.Y., 2017. Demand forecasting for heavy-duty diesel engines considering emission regulations. Sustainability 9 (2), 166. Krause, D., 1993. Environmental consciousness an empirical study. Environ. Behav. 25 (1), 126e142. Kraeusel, J., Most, D., 2012. Carbon Capture and Storage on its way to large-scale deployment: social acceptance and willingness to pay in Germany. Energy

119

Policy 49, 642e651. Lanzi, E., Verdolini, E., Ha
s
ci
c, I., 2011. Efficiency-improving fossil fuel technologies for electricity generation: data selection and trends. Energy Policy 39 (11), 7000e7014. Lin, B., Ouyang, X., 2014. Analysis of energy-related CO2 (carbon dioxide) emissions and reduction potential in the Chinese non-metallic mineral products industry. Energy 68, 688e697. Lin, S.J., Lu, I.J., Lewis, C., 2006. Identifying key factors and strategies for reducing industrial CO2 emissions from a non-Kyoto protocol member’s (Taiwan) perspective. Energy Policy 34 (13), 1499e1507. Liu, L.C., Ying Fan, G.W., Wei, Y.M., 2007. Using LMDI method to analyze the change of China’s industrial CO2 emissions from final fuel use: an empirical analysis. Energy Policy 35, 5892e5900. Matthes, F.C., Newbery, D., Colombier, M., Mathieu, M., Rüdinger, A., 2015. The Energy Transition in Europe: Initial Lessons from Germany, the UK and France. Centre On Regulation in Europe, Brussels, Belgium. Miao, Z., Geng, Y., Sheng, J., 2016. Efficient allocation of CO2 emissions in China: a zero sum gains data envelopment model. J. Clean. Prod. 112 (5), 4144e4150. Moraga-Gonzalez, J.L., Padron-Fumero, N., 2002. Environmental policy in a green market. Environ. Resour. Econ. 22 (3), 419e447. Noailly, J., Shestalova, V., 2013. Knowledge Spillovers from Renewable Energy Technologies, Lessons from Patent Citations (No. 262). CPB Netherlands Bureau for Economic Policy Analysis, Hague, Netherlands. Oggioni, G., Riccardi, R., Toninelli, R., 2011. Eco-efficiency of the world cement industry: a data envelopment analysis. Energy Policy 39 (5), 2842e2854. Popp, D., 2001. The effect of new technology on energy consumption. Resour. Energy Econ. 23 (3), 215e239. Prodan, I., 2005. Influence of research and development expenditures on number of patent aplications: selected case studies in OECD countries and central Europe, 1981-2001. Appl. Econ. Intern. Dev. 5 (4), 5e22. Quadrelli, R., Peterson, S., 2007. The energyeclimate challenge: recent trends in CO2 emissions from fuel combustion. Energy policy 35 (11), 5938e5952. Rotela Junior, P., Pamplona, E.D.O., Rocha, L.C.S., Valerio, V.E.D.M., Paiva, A.P., 2015. Stochastic portfolio optimization using efficiency evaluation. Manag. Decis. 53 (8), 1698e1713. Sabrina, S., Julian, S., 2015. G7 Coal Phase Out: Germany. E3G, London, UK. https:// www.e3g.org/docs/Germany_G7_coal_analysis_September_2015.pdf. Accessed 4 January 2017. Sarrica, M., Brondi, S., Piccolo, C., Mazzara, B.M., 2016. Environmental consciousness and sustainable energy policies: Italian parliamentary debates in the years 2009e2012. Soc. Nat. Resour 29 (8), 932e947. S anchez, J., Lafuente, R., 2010. Defining and measuring environmental consciousness. Rev. Inter. De. Sociol. 68 (3), 731e755. Seiford, L.M., Zhu, J., 1999. Profitability and marketability of the top 55 US commercial banks. Manag. Sci. 45 (9), 1270e1288. Shi, G.M., Bi, J., Wang, J.N., 2010. Chinese regional industrial energy efficiency evaluation based on a DEA model of fixing non-energy inputs. Energy Policy 38 (10), 6172e6179. Skjærseth, J.B., Bang, G., Schreurs, M.A., 2013. Explaining growing climate policy differences between the European Union and the United States. Glob. Environ. Polit. 13 (4), 61e80. Sohn, S.Y., 2006. Random effects logistic regression model for ranking efficiency in data envelopment analysis. J. Oper. Res. Soc. 57 (11), 1289e1299. Sohn, S.Y., Choi, H., 2006. Random effects logistic regression model for data envelopment analysis with correlated decision making units. J. Oper. Res. Soc. 57 (5), 552e560. Sohn, S.Y., Jeon, J., Han, E.J., 2015. A new cost of ownership model for the acquisition of technology complying with environmental regulations. J. Clean. Prod. 100, 269e277. Sohn, S.Y., Kim, Y., 2012. DEA based multi-period evaluation system for research in academia. Expert Syst. Appl. 39 (9), 8274e8278. Sueyoshi, T., Goto, M., 2010. Should the US clean air act include CO2 emission control? Examination by data envelopment analysis. Energy Policy 38 (10), 5902e5911. Sueyoshi, T., Goto, M., 2011. DEA approach for unified efficiency measurement: assessment of Japanese fossil fuel power generation. Energy Econ. 33 (2), 292e303. Sun, J., 1998. Changes in energy consumption and energy intensity: a complete decomposition model. Energy Econ. 20 (1), 85e100. Wang, C., Chen, J., Zou, J., 2005. Decomposition of energy-related CO2 emission in China: 1957e2000. Energy 30 (1), 73e83. Wang, Q., Chiu, Y.H., Chiu, C.R., 2015. Driving factors behind carbon dioxide emissions in China: a modified production-theoretical decomposition analysis. Energy Econ. 51, 252e260. Weina, D., Botang, H., Xin, Z., Mazzanti, M., 2015. How does green technology influence C02 emission in China? An empirical research based on provincial data of China. J. Environ. Biol. 36 (4), 745. WIPSON, 2017. WIPS Corporation. http://www.wipson.com. Accessed 7 March 2017. Wu, F., Fan, L.W., Zhou, P., Zhou, D.Q., 2012. Industrial energy efficiency with CO2 emissions in China: a nonparametric analysis. Energy Policy 49, 164e172. Xiang, Y., Dongii, Z., Dan, H., 2016. Reducing carbon emission-foresight of CCS technology developing, IP management and global technology transfer. In: IAMOT 2016 Conference. Zhou, P., Ang, B.W., 2008a. Linear programming models for measuring economywide energy efficiency performance. Energy Policy 36 (8), 2911e2916.

120

D.S. Kwon et al. / Journal of Cleaner Production 151 (2017) 109e120

Zhou, P., Ang, B.W., 2008b. Decomposition of aggregate CO2 emissions: a production-theoretical approach. Energy Econ. 30 (3), 1054e1067. Zhou, P., Sun, Z.R., Zhou, D.Q., 2014. Optimal path for controlling CO2 emissions in China: a perspective of efficiency analysis. Energy Econ. 45, 99e110. Zhang, N., Zhou, P., Choi, Y., 2013a. Energy efficiency, CO2 emission performance and technology gaps in fossil fuel electricity generation in Korea: a meta-frontier non-radial directional distance function analysis. Energy Policy 56, 653e662.

Zhang, X.P., Zhang, J., Tan, Q.L., 2013b. Decomposing the change of CO2 emissions: a joint production theoretical approach. Energy Policy 58, 329e336.  Szere ka, A., nyi, Z.M., Sze chy, A., Kocsis, T., 2013. Greening due to environmental Zso education? Environmental knowledge, attitudes, consumer behavior and everyday pro-environmental activities of Hungarian high school and university students. J. Clean. Prod. 48, 126e138.