The integrated efficiency of inputs–outputs and energy – CO2 emissions performance of China's agricultural sector

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

The integrated efficiency of inputs–outputs and energy – CO2 emissions performance of China's agricultural sector ⁎

Rilong Feia, Boqiang Linb, a

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

A R T I C L E I N F O

A BS T RAC T

Keywords: Agricultural sector Integrated efficiency Energy–CO2 performance Energy intensity CO2 emission reduction

This research aims at exploring the integrated efficiency of inputs–outputs and unified performance in energy consumption and CO2 emissions for the Chinese agricultural sector, and also examining the reduction potential of energy intensity and CO2 emission intensity in the sector. For this purpose, we adopt a non-radial directional distance function in this study. It incorporates the inefficiency of all input factors and desirable and undesirable outputs to estimate the integrated (operational and environmental) efficiency and energy–CO2 performance of China's agricultural sector. An empirical research of 30 provinces in China is conducted by using the approach. The main practical conclusion follows: First, most of China's provinces and regions did not perform efficiently in input factors as well as the integrated efficiency of inputs-outputs and unified performance in energy consumption and CO2 emissions in the agricultural sector. The average score of integrated efficiency and energy–CO2 performance is 0.447 and 0.425 respectively within the sample period. Provinces in East China mostly performed better than those in Central and West China. Second, with slack and advanced production technology considered, there is vast energy-saving and CO2 emission reduction in the Chinese agricultural sector. In theory, the reduction potential of energy intensity and CO2 emission intensity can reach 59.6177% and 56.4948% respectively of the actual level. The central and western regions show great reduction potential of energy intensity and CO2 emission intensity compared with the eastern region. Based on these findings, some policy suggestions for improving the integrated efficiency of inputs-outputs and unified performance in energy consumption and CO2 emissions are provided for China's agricultural sector.

1. Introduction China's economy has progressed greatly since the reform and opening-up policy started in 1978. According to the Nation Bureau of Statistic of China (NBSC), China's real gross domestic product has grown about a hundred times in 2012 more than the level in 1978. This fast-growing rate makes China the world's biggest economy according to International Monetary Fund [21]. However, such astonishing economic expansion is accompanied with large energy consumption. In 2013, China's energy consumption is 4.17 billion tons of standard coal, which is much greater than 571 million tons in 1978 [40]. The rapid industrialization and rigid demand for energy use make China the largest energy consumer in the word [4]. Over this time span, China's agriculture has also made tremendous strides in boosting economic development and improving the living standards of the people. It feeds more than 20% of the world's population using less

than 10% of global arable land. Moreover, agricultural mechanization in China is currently rapidly and widely popularized, which leads to a significant increase in energy use and its related-CO2 emission. During the period 2001–2012, agricultural output (at 1978 prices) increased from 571.07 billion yuan RMB in 1985 to 980.46 billion yuan RMB in 2012. This is mainly thanks to the implementation of the household contract responsibility system and the promotion of agricultural mechanization. The total mechanical power soared from 551.72 million kilowatt in 2001 to 1025.5 million kilowatt in 2012. It drove energy consumption increase from 43.45 to 74.44 million tons of standard coal equivalent; and as a result CO2 emissions increased from 116.26 to 185.66 million tons during this period, as shown in Fig. 1. As a result, overuse of resources and environmental pollution has hindered the sustainable development of China's agricultural sector. It is common sense that enhancing energy and environmental efficiency is a crucial and scientific way for the world to combat energy

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

http://dx.doi.org/10.1016/j.rser.2016.11.040 Received 19 November 2015; Received in revised form 27 August 2016; Accepted 4 November 2016 1364-0321/ © 2016 Elsevier Ltd. All rights reserved.

Please cite this article as: Fei, R., Renewable and Sustainable Energy Reviews (2016), http://dx.doi.org/10.1016/j.rser.2016.11.040

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

mental efficiency) and energy–environmental performance of China's agricultural sector. Meanwhile, it also examines the reduction potential of energy intensity and CO2 emission intensity during the studied period. Different from Zhou et al. [49], however, the paper considers energy factors as well as non-energy factors (capital and labor) because it focuses on exploring agricultural unified performance. To measure the unified efficiency, we propose a total-factor NDDF (TNDDF) that incorporates inefficiencies for all the input factors and desirable and undesirable outputs. Unlike Zhang et al. [45], when measuring energy– environmental performance, we still follow the same (TNDDF) without fixing non-energy inputs, because we want to measure the potential reduction of energy intensity and CO2 emission intensity under the condition that all inputs and outputs are in their optimal states. According to what we have learnt, this paper is the first to empirically explore the unified efficiency and energy–environmental performance of China's agricultural sector. Using this method, we can estimate the reduced degree of input factors and CO2 emissions in effective output. We take China's agricultural sector as a case study because it is an important and indispensable part of the economy and has contributed to increased energy consumption a lot in recent years. However, the quantitative impact of energy-related pollution treatment on the efficiency of the agricultural sector is still under-explored since agriculture makes smaller contribution to the national economy compared with the industrial and service sector [30]. This gap motivates us to explore the energy and environmental performance of China’ agriculture sector, which can help policy makers in agricultural policy formulation. The rest of the paper is as follows. In Section 2, the research methods and methodologies are illustrated on the whole. In Section 3, the data and variables are presented in details as well as the empirical results. In Section 4, the related discussion is further provided. Section 5 concludes this study and suggests policy implications in the paper.

Fig. 1. Agricultural mechanical power, energy consumption and CO2 emissions in China, 2001–2012.

and CO2 challenge. Hence, accurate evaluation of energy input and environmental efficiency is of great importance. In many previous studies, they have adopted data envelopment analysis (DEA) technique to explore energy and CO2 performance from a perspective of production efficiency ([22,23,32,37]). In the case of agriculture, there are also a lot of similar studies ([2,17,26,27,33,35,36,39,46]). In comparison to the traditional DEA models, DDF which proposed afterwards measures production efficiency by increasing desirable outputs (e.g., agricultural output) and reducing undesirable outputs (e.g., CO2 emissions) at the same time. This is regarded as a radial efficiency measure with few shortcomings that it may overestimate efficiency when there exist some slacks [15] and cannot distinguish between environmental and operational performance [38]. Several papers have developed the conventional DDF into the non-radial directional distance function (NDDF) by incorporating slacks into efficiency measurement [1,13]. In recent years, Zhou et al. [49], Zhang et al. [44,45] and Lin and Du [28] used this method to measure performance in energy and CO2 emission at the industrial and regional levels. We use Fig. 2 to illustrate the differences between NDDF and the traditional DEA method. We consider the EDFC area as the production sets which describe the environmental production technology T. Suppose point M is the decision point and g is the policy directional vector. If we use the DDF method to estimate the efficiency of point M, point F is the benchmark to realize efficiency maximization. However, when we use the NDDF method to estimate the efficiency of point K, the benchmark is point D. It can be seen that, at point D, it can decrease more undesirable output while keep the desirable output constant. Thus, the distance DF is the non-zero slack of the undesirable output. In the DDF method, it does not take the non-zero slack into consideratio n, and as a result, it may underestimate the potentiality of the inefficiency. The paper employed the NDDF based on Zhou et al. [49] and Zhang et al. [46] to explore the unified efficiency (operational and environ-

D

F

2. Methodology and method 2.1. DEA modeling with undesirable outputs As mentioned above, the DEA method is usually thought to be a popular tool in measuring energy and environmental performance [37,44,6]. There have been several approaches dealing with the undesirable output based on the DEA. The first category takes the undesirable outputs as input factors, and the representative literatures include Haynes et al. [19], Lee et al. [24], and Hailu and Veeman [18]. This method is easily grasped and operated by investigators and satisfies the requests that undesirable outputs are smaller. However, it is not consistent with the actual situation. The second one is conducting data transformation to undesirable outputs. Specifically, the undesirable outputs that are the-smaller-the-better are transferred to new variables treated as desirable outputs that are the-larger-thebetter. Then the traditional DEA method can be employed to explore the efficiency of energy or the environment [20]. The third category distinguishes the weak and strong disposability between the undesirable and desirable outputs. Based on the joint-production framework, researchers and scholars have developed several models for assessing energy or environmental performances [8,14,43,44,48]. Among these measurements, the DDF method raised by Chung et al. has been largely used in empirical application, which allows for increase in desirable output and decrease in undesirable output and the input factors at the same time [12,3,31,34,10]. However, the DDF may underestimate the efficiency loss of the assessed DMU due to its radial efficiency measure [10]. Fortunately, Zhou et al. [49] developed a nonradial directional distance function (NDDF) method that allows for disproportional adjustments of input factors, desirable and undesirable outputs. Hence, the non-radial directional distance function (herein after referred to as NDDF) has a higher discriminating ability than the directional distance function DDF. Thereafter, this method is further

C

E

B M

g

A O Fig. 2. The illustration of non-rational direction distance functions.

2

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

→ ⎯ D (K, L, E, Y, C;g) = maxψK ηK + ψLηL + ψEηE + ψY ηY + ψCηC

developed for estimating technology heterogeneity [48], investigating the dynamic change in CO2 emission performance [42] and the integrated energy–carbon performance [28,44]. Due to the comparative advantages of the NDDF method, we apply it in this paper. Generally, in the frontier modeling, each DMU explores input factors X = (X1, X2,……,XN )to produce agricultral outputs Y = (Y, 1 Y2,……,YM ) and CO2 emissionsC = (C1, C 2,……,CR). With a collection of DMUs of N assessed areas, then the production technology set Ω can be formulated as follows:

Ω = {(K, L, E, Y, C):(K, L, E)can produce(Y, C)} N

∑n =1 γn*Ln ≤ L,

N

∑n =1 γn*Cn = C ,

∑n =1 γn*Kn ≤ K , ∑n =1 γn*Yn ≥ Y , n = 1,

2,

N

S. T. ∑

γK n =1 n n

(3)

≤ K − ηK gK ,

N

∑n =1 γnLn ≤ L − ηL gL , N

∑n =1 γnEn ≤ E − ηE gE , N

∑n =1 γnYn ≥ Y − ηE gY ,

(1)

N

N

∑n =1 γnCn = C − ηE gC ,

∑n =1 γn*En ≤ E ,

n = 1,

N

2,

……, N

ηK , ηL , ηE , ηY , ηC ≥0

……, N

→ ⎯ Under this theoretical frame, if D (K, L, E, Y, C;g)=0 , then the evaluated DUM is located on the best-practice frontier in the explicit g direction.

Methodologically, the production technology set Ω generally possesses the following properties: (1) if C = 0 and (K, L, E, Y, C) ∈ Ω , then Y = 0 . It indicates that the desirable output cannot be produced without the undesirable output. This property is termed as null-jointness. (2) if (K, L, E, Y, C) ∈ Ω , and 0 ≤ ρ ≤ 1, then (K, L, E, ρY, ρC) ∈ Ω . (3) if (K, L, E, Y, C) ∈ Ω , and Y 1 < Y , then (K, L, E,Y 1,C) ∈ Ω . This means that the input factor and desirable output are endowed with the assumption of weak disposability. The excessive input factors and desirable output can be disposed without any cost.

2.2. The integrated efficiency of inputs-outputs indicator (AUEI) In the previous literature, the efficiency indicators are always with regards to one factor such as energy or CO2. However, such efficiency indicator cannot reflect the unified efficiency of the whole production system which contains so many inputs and outputs. Agricultural sustainable economic development needs coordination between input factors and the achievement of maximum agricultural output and minimum CO2 emissions. Hence, we need to seek a unified efficiency index to measure this comprehensiveness which includes all input and output. Considering there are three input factors (capital, labor and energy), one desirable output (agricultural output) and one undesirable output (CO2), so the normalized weight vector is set as (1/9, 1/9, 1/9, 1/3, 1/3) and the directional vectors as is set as (-K, -L, -E, Y, -C) based on Zhou et al. [49], Barros et al. [1], Lin and Du [28]. Sueyoshi and Goto [38] describe the integrated efficiency indicator as the average of all individual efficiencies. Zhou et al. [49] describe the energy efficiency indicator as the ratio of actual energy input to potential optimal energy input and the CO2 performance indicator as the ratio of optimal CO2 emissions to actual CO2 emissions. Based on these papers, the integrated efficiency of inputs-outputs index for agricultural sector (AUEI) is an average efficiency of each factor. Suppose that η=(ηK*, ηL*, ηE*, ηY*, ηC*)T indicates the optimal solution to Eq. (4). Then the AUEI can be described as follows:

Besides, the production technology setΩ is always assumed to be convex, closed and bounded [11]. Thus, such a nonparametric production technology set Ω is built. Then the NDDF can be adopted to calculate the technical efficiency in our study. As discussed above, the conventional DDF is a radial efficiency. It may overestimate the efficiency of the assessed DMU when there exists nonzero slacks [15,16]. Another shortcoming of the directional distance function is that the radial DDF can merely deliver the same rate of inefficiency because it cannot separate the environmental and operational performance apart [38]. Therefore, it is difficult to provide an integrated unified efficiency measure. In contrast, the NDDF is hence advocated to overcome this limitation [5,7,49,6,47,26]. As a result, this paper selects the non-radial DDF method and defines it as:

→ ⎯ D (K, L, E, Y, C;g) = sup {ΨT η :(K, L, E, Y, C)+ diag(η)*g∈Ω }

N

(2)

1

AEUI= 4

T

where ψ=(ψK , ψL , ψE , ψY , ψL) is a normalized vector denoting the weights assigned to each variables; η=(ηK , ηL , ηE , ηY , ηL )T is a vector of scaling factors which measures the departure of real production activity from the optimal state. In other words it measures maximum scaling possibility of the reduction inputs and undesirable output and expansion of the desirable output; g=(gK , gL , gE , gY , gL )T is a directional vector determining the directions in which each variable is scaled, and for measuring the inefficiency of all variables, and diag (β) represents a diagonal matrix with β. Besides, the directional vector g and the weight vector Ψ can be set in different ways for the different policy goals. Unlike Zhou et al. [49] and Zhang et al. [45], this paper not only pays attention to energy efficiency, but also considers non-energy efficiency because it intends to evaluate the integrated efficiency performance of the whole agricultural production system. Thus, when we define the total factor non-radial directional distance function (herein after referred to as TNDDF), the inefficiency of all input factors, desirable and undesirable outputs are all incorporated into the objective function and constraints. Thus, the TNDDF based on production technology can be formulated as the following linear programming:

[(1−ηK*) + (1−ηL*) + (1−ηE*)+(1−ηC*)] 1+ηY*

1 1− 4 *(ηK*

=

+ ηL* + ηE* + ηC*) 1+ηY*

(4)

By defining it this way, we can examine the overall efficiency of agricultural sector which includes all factor indices. Then, for estimating the pure environmental performance of the agricultural sector at the optimal state, we define an energy–CO2 performance indicator based on Zhou et al. [49]. Unlike Zhang et al. [46], we define this indicator with the same non-radial directional distance function as the integrated efficiency of inputs-outputs index. This is because we want to measure the energy–CO2 performance under the condition that all inputs and outputs are in their optimal states, as shown in Eq. (5). 1

ECPI= 2

[(1−ηE*)+(1−ηC*)] 1+ηY*

1

=

1− 2 *(ηE* + ηC*) 1+ηY*

(5)

Obviously, the AUEI and the ECPI both range from 0 to 1. The higher the AUEI (ECPI), the better the integrated efficiency and energy–CO2 performance in the agricultural sector. If the AUEI 3

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

Table 1 The integrated efficiency of inputs–outputs index of China's agricultural sector, 2001–2012.

beijing tianjin hebei Liaoning shanghai jiangsu zhejiang fujian shandong guangdong hainan shanxi Mongolia jilin heilj anhui jiangxi henan hubei hunan guangxi chongqing sichuan guizhou yunnan shaanxi gansu qinghai ningxia xinjiang avg

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

avg

0.765 0.258 0.485 0.535 0.265 0.601 0.316 0.729 0.526 0.499 0.659 0.103 0.278 0.497 0.334 0.387 0.574 0.465 0.396 0.363 0.874 0.186 0.533 0.112 0.283 0.417 0.261 0.299 0.263 0.182 0.415

0.881 0.223 0.484 0.748 0.278 0.601 0.294 0.682 0.561 0.526 0.659 0.115 0.280 0.511 0.381 0.410 0.590 0.469 0.387 0.365 0.912 0.183 0.569 0.110 0.298 0.449 0.275 0.300 0.274 0.601 0.447

0.941 0.282 0.508 0.768 0.292 0.602 0.305 0.697 0.567 0.561 0.674 0.127 0.260 0.826 0.369 0.382 0.591 0.471 0.424 0.375 1.000 0.188 0.552 0.107 0.309 0.431 0.279 0.313 0.287 0.203 0.456

1.000 0.282 0.538 0.776 0.317 0.629 0.326 0.706 0.565 0.564 0.653 0.166 0.263 0.530 0.356 0.413 0.646 0.491 0.494 0.368 0.869 0.200 0.502 0.112 0.234 0.435 0.305 0.326 0.301 0.205 0.452

1.000 0.269 0.521 0.463 0.295 0.661 0.328 0.491 0.528 0.454 0.662 0.178 0.375 0.462 0.409 0.435 0.529 0.480 0.455 0.286 0.913 0.188 0.534 0.117 0.252 0.436 0.293 0.359 0.307 0.200 0.429

1.000 0.272 0.555 0.459 0.366 0.668 0.334 0.337 0.334 0.566 0.640 0.175 0.273 0.460 0.407 0.450 0.562 0.496 0.372 0.431 0.901 0.165 0.528 0.122 0.262 0.422 0.286 0.369 0.311 0.202 0.424

0.929 0.270 0.559 0.457 0.360 0.559 0.335 0.362 0.339 0.587 0.736 0.211 0.272 0.430 0.398 0.466 0.601 0.524 0.384 0.442 0.849 0.175 0.538 0.138 0.270 0.449 0.301 0.402 0.315 0.211 0.429

0.919 0.278 0.551 0.483 0.418 0.582 0.356 0.446 0.571 0.598 0.644 0.177 0.250 0.855 0.476 0.468 0.650 0.552 0.401 0.448 1.000 0.176 0.511 0.225 0.291 0.472 0.313 0.344 0.312 0.210 0.466

0.985 0.275 0.549 0.492 0.420 0.572 0.369 0.378 0.578 0.624 0.485 0.182 0.227 0.895 0.436 0.483 0.679 0.578 0.335 0.454 1.000 0.212 0.522 0.225 0.297 0.491 0.321 0.351 0.319 0.211 0.465

0.950 0.265 0.534 0.523 0.428 0.561 0.307 0.379 0.581 0.640 0.429 0.188 0.185 0.865 0.468 0.493 0.699 0.593 0.351 0.452 0.941 0.198 0.536 0.253 0.303 0.509 0.316 0.353 0.324 0.222 0.461

0.976 0.252 0.430 0.526 0.421 0.508 0.310 0.375 0.590 0.639 0.422 0.191 0.196 0.889 0.441 0.505 0.729 0.575 0.337 0.296 0.925 0.202 0.561 0.273 0.334 0.517 0.323 0.356 0.339 0.222 0.455

1.000 0.242 0.391 0.571 0.379 0.487 0.310 0.380 0.659 0.642 0.423 0.189 0.178 1.000 0.366 0.529 0.792 0.582 0.341 0.283 1.000 0.187 0.556 0.291 0.354 0.543 0.334 0.385 0.351 0.217 0.465

0.946 0.264 0.509 0.567 0.353 0.586 0.324 0.497 0.533 0.575 0.590 0.167 0.253 0.685 0.403 0.452 0.637 0.523 0.390 0.380 0.932 0.188 0.537 0.174 0.291 0.464 0.300 0.346 0.309 0.241 0.447

(ECPI) is equal to 1, then the observation reflects the best integrated efficiency (energy–CO2 performance) located on the technology frontier. 2.3. The potential reduction of energy intensity and CO2 emission intensity

(6)

PRC=ηC**Cit

(7)

PRC=ηY**Yit

(8)

(9)

PCI = {Cit − ηC**Ckt}/{Ykt + ηY**Ykt}

(10)

PCIR=Cit /Ykt −{Cit − ηE**Cit}/{Ykt + ηY**Ykt

(12)

Although diverse research techniques have been involved in previous research, some universally applicable and insightful conclusions can be summarized by analysis and comparison: (i) China is still at a stage of lower energy and environmental efficiency across industries and at the aggregate level. (ii) There exists distinct performance disparity among different regions in China. In this part, we use the agricultural input-output data during 2001–2012 to conduct our empirical research. The agricultural input data include labor, energy and capital. Labor and energy data are obtained from China Energy Statistical Yearbook, and data on capital is extracted from [41]. We supplement the data on capital for 2007–2012 based on his method. The output data include agricultural GDP and CO2 emissions. The data on agricultural GDP are obtained from China Statistical Yearbook and are converted into 1978 price level. The data on CO2 emissions are obtained by the product of energy inputs and their corresponding emission coefficient (IPCC, 2007). 3.1. Total factor integrated efficiency of inputs-outputs and energy– CO2 performance

Furthermore, the potential energy and CO2 emissions intensity can be reduced as described by Eqs. (9) and (10)

PEI = {Eit − ηE**Eit}/{Ykt + ηY**Ykt}

(11)

3. Empirical results

Because of simple calculation and intelligibility, energy and CO2 intensity are always taken as the basis during international comparison of energy or CO2 efficiency and the task allocation of energy saving and CO2 reduction [9,25]. However, the efficiency measurement described in our paper is based on relative efficiency under the framework of total factors, which is substantially different from those based on singlefactor. As mentioned above, the NDDF allows for crediting increment in input factors and desirable output while simultaneously crediting reduction in CO2 emissions. This property is in conformity with the requirement of sustainable economic development. Based on Eqs. (1)–(3), and its optimal solution η=(ηK*, ηL*, ηE*, ηY*, ηC*)T , the potential reduction of energy and CO2 emissions and growth of agricultural output relative to the bestpractice frontier are calculated by Eqs. (6)–(8)

PRE=ηE**Eit

PEIR=Eit / Ykt −{Eit − ηE**Eit}/{Ykt + ηY**Ykt}

The optimal solution (ηK*, ηL*, ηE*, ηY*, ηC*)T of the 30 provinces from 2001 to 2012 are first calculated by solving Eq. (3), and then the results of the unified efficiency index (AUEI) and the energy–environmental performance index (ECPI) of China's agricultural sector are obtained and displayed in Tables 1 and 2. These scores could be referred to as static performance indices and provide important information on the rankings of the provinces in terms of their integrated efficiency of

Correspondingly, the potential energy and CO2 intensity reduction can be calculated as: 4

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

Table 2 Energy–environmental performance index of China's agricultural sector, 2001–2012.

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

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

avg

0.731 0.254 0.314 0.767 0.294 0.322 0.298 0.569 0.360 0.249 0.495 0.120 0.275 0.677 0.317 0.303 0.448 0.364 0.407 0.386 0.808 0.182 0.483 0.121 0.282 0.376 0.285 0.330 0.258 0.196 0.376

0.849 0.256 0.300 0.558 0.310 0.366 0.274 0.535 0.437 0.256 0.513 0.132 0.278 0.678 0.377 0.331 0.463 0.370 0.402 0.379 0.865 0.181 0.505 0.121 0.282 0.433 0.297 0.282 0.271 0.546 0.395

0.897 0.366 0.332 0.624 0.319 0.418 0.289 0.614 0.417 0.298 0.557 0.135 0.273 0.728 0.366 0.304 0.451 0.408 0.453 0.375 1.000 0.195 0.426 0.120 0.309 0.387 0.301 0.300 0.284 0.212 0.405

1.000 0.362 0.359 0.661 0.332 0.507 0.315 0.666 0.398 0.307 0.526 0.164 0.262 0.709 0.490 0.326 0.511 0.399 0.322 0.369 0.739 0.225 0.355 0.125 0.224 0.360 0.317 0.309 0.301 0.217 0.405

1.000 0.330 0.339 0.624 0.289 0.551 0.319 0.213 0.347 0.455 0.561 0.176 0.263 0.560 0.587 0.370 0.232 0.353 0.428 0.301 0.826 0.207 0.388 0.131 0.240 0.337 0.307 0.355 0.313 0.209 0.387

1.000 0.333 0.421 0.621 0.392 0.588 0.331 0.332 0.334 0.316 0.527 0.176 0.293 0.564 0.581 0.375 0.268 0.358 0.404 0.153 0.801 0.181 0.356 0.136 0.252 0.289 0.314 0.358 0.323 0.214 0.386

0.902 0.326 0.434 0.631 0.396 0.815 0.332 0.359 0.341 0.361 0.712 0.223 0.292 0.548 0.565 0.374 0.371 0.376 0.414 0.151 0.699 0.199 0.365 0.147 0.262 0.302 0.320 0.389 0.331 0.226 0.405

0.914 0.343 0.422 0.694 0.499 0.847 0.354 0.459 0.455 0.384 0.577 0.180 0.265 0.870 0.710 0.341 0.470 0.409 0.430 0.138 1.000 0.204 0.313 0.211 0.286 0.296 0.323 0.262 0.325 0.223 0.440

0.971 0.342 0.426 0.727 0.499 0.815 0.368 0.532 0.473 0.444 0.751 0.184 0.242 0.905 0.632 0.344 0.559 0.472 0.483 0.139 1.000 0.206 0.339 0.211 0.295 0.307 0.318 0.261 0.337 0.225 0.460

0.965 0.325 0.405 0.794 0.477 0.785 0.404 0.532 0.484 0.492 0.626 0.195 0.220 0.880 0.693 0.334 0.624 0.511 0.506 0.138 0.921 0.188 0.365 0.234 0.306 0.353 0.317 0.409 0.344 0.239 0.469

0.966 0.304 0.464 0.806 0.464 0.759 0.405 0.522 0.508 0.510 0.600 0.201 0.205 0.925 0.641 0.333 0.699 0.480 0.474 0.288 0.927 0.193 0.391 0.255 0.343 0.380 0.325 0.416 0.366 0.237 0.480

1.000 0.291 0.422 0.817 0.441 0.704 0.403 0.531 0.650 0.533 0.595 0.191 0.207 1.000 0.487 0.354 0.846 0.497 0.478 0.279 1.000 0.182 0.366 0.272 0.366 0.442 0.344 0.262 0.383 0.230 0.486

0.933 0.319 0.386 0.694 0.393 0.623 0.341 0.489 0.434 0.384 0.587 0.173 0.256 0.754 0.537 0.341 0.495 0.416 0.433 0.258 0.882 0.195 0.388 0.174 0.287 0.355 0.314 0.328 0.320 0.248 0.425

inputs-outputs and energy–CO2 efficiency. It can be seen from Table 1 that only a few values of the AUEI are equal to unity, revealing that the agriculture sector in most provinces in China did not perform efficiently. The average ECPI in China's agricultural sector within the sample period was only 0.447. This means that China is still at a low stage of developing an inputs-efficient and environmental-friendly agricultural economy. Besides, it reveals that China's agricultural sector performed better during the study period as a whole. The (ECPI) for all provinces ranges from 0.12 to 1 (average=0.425). This implies that, on average, the 30 provinces together can achieve a 57.5% increase in their energy–CO2 efficiency relative to the production technology frontier. The average AUEI score is higher than the average EEPI, indicating that the agricultural sector shows better performance in AUEI than EEPI. For individual provinces, Beijing and Guangxi show the higher AUEI and EEPI values than other provinces. Provinces such as Chongqing and Guizhou show the lowest AUEI and EEPI values. This reflects the provincial imbalance in the agricultural sector in terms of CO2 performance. From Fig. 3, we can see that the score of the AUEI in the eastern region was the highest but has seen a downward trend since 2004, which decreased from 0.578 in 2004 to 0.4986 in 2012. In contrast, the score in the central region increased and began to catch up with the eastern region since 2008. As for the western region, the AUEI score was far behind that of the eastern and central regions and sustained steady growth. In terms of the ECPI described in Fig. 4, all the three regions show upward trends of average ECPI scores in China’ agricultural sector. From the figures, we can see that the eastern region takes a lead in terms of energy–CO2 efficiency. Most ECPI scores of provinces in the eastern region stay at a high level (their average value is 0.5074) during 2001–2012, which indicates that the frontier of the eastern region is quite close to the meta-frontier. ECPI scores of provinces in the western region generally fall behind provinces in the eastern and central regions, and the performance gap between these regions does

0.7000 0.6000

values

0.5000 0.4000 0.3000 0.2000 0.1000 0.0000 eatern

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

0.5126 0.5396 0.5636 0.5780 0.5158 0.5030 0.4994 0.5314 0.5206 0.5087 0.4954 0.4986

central 0.4270 0.4420 0.4824 0.4597 0.4523 0.4527 0.4577 0.5277 0.5271 0.5234 0.5083 0.5261 western 0.2819 0.3399 0.2965 0.2911 0.2985 0.2964 0.3111 0.3169 0.3278 0.3349 0.3474 0.3574

Fig. 3. The average scores of AUEI in the three regions: 2001–2012.

0.7000 0.6000

values

0.5000 0.4000 0.3000 0.2000 0.1000 0.0000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

eatern 0.4229 0.4230 0.4664 0.4940 0.4571 0.4722 0.5099 0.5407 0.5770 0.5717 0.5736 0.5807 central 0.4106 0.4274 0.4492 0.4291 0.4096 0.3974 0.4012 0.4813 0.4961 0.5022 0.5173 0.5340 western 0.2795 0.3241 0.2814 0.2704 0.2764 0.2692 0.2822 0.2713 0.2776 0.3061 0.3228 0.3161

Fig. 4. The average scores of ECPI in the three regions: 2001–2012.

5

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

2012

2001 0.1 0.08

2002

0.06

2011

2003

0.04

AEI

0.02 2010

Fig. 5. The provincial potential reduction of energy consumption in agricultural sector: 2001–2012 (unit: Mtc).

2004

0

2009

PEI

2005

2008

2006 2007

Fig. 7. Comparison of actual and potential energy intensity of China's agricultural sector, 2001–2012. (unit: Mtc/billion yuan RMB).

2012 Fig. 6. The provincial potential reduction of agricultural CO2 emissions: 2001–2012 (unit: Mtc).

2001 0.25 0.2

2002

0.15

2011

2003

0.1

not show a significant shrink in the period.

0.05

3.2. The potential energy and CO2 intensity of China's agricultural sector

2010

Using Eqs. (6)–(12), we can obtain the reduction of energy input and CO2 emissions that are attributed to technological inefficiencies. The aggregate results for each province are presented in Figs. 5 and 6. In terms of the reduction potential of energy, the total amount for all provinces can be calculated as 329.65 million tons during 2001– 2012, accounting for 36.5% of the actual energy consumption. It means that each province, on average, can reduce 10.99 million tons of energy consumption. Specifically, Henan province has the largest energy consumption reduction potential with a total amount of 38.56 million tons, followed by Shandong (34.05 million tons) and Guangdong (30.54 million tons). Additionally, provinces such as Hebei, Jiangsu and Sichuan also show vast reduction potential of energy consumption. From the perspective of regional differences, the eastern region shows the largest amount of energy consumption reduction potential. On balance, the amount of PER in the eastern region is more than that of the sum of the rest regions. With regard to the reduction potential of CO2 emissions, provinces in the western regions generally show small reduction potential. In contrast, the provinces in the eastern and central regions show relatively large amount of reduction potential of CO2 emissions. The total emission reduction potential of all provinces reached 447.56 million tons during 2001–2012, accounting for 25.23% of actual CO2 emissions. Specifically, Hunan province shows the largest amount at 57.96 million tons, followed by Guangdong (43.65 million tons) and Shandong provinces (41.65 million tons). Although both Shandong and Hebei are located in the eastern region and possess a relatively advanced technology, they have great potential for CO2 emission reduction due to their managerial inefficiency. After the advanced production technology under meta- frontier is taken into account, the optimized combination of agricultural output, energy input and CO2 emissions for a DMU under the meta-frontier is

2004

0

2009

ACI PCI

2005

2008

2006 2007

Fig. 8. Comparison of actual and potential CO2 emission intensity of China's agricultural sector, 2001–2012. (unit: Mtc/billion yuan RMB).

((1+ηY*)y), (1− ηE*)e, (1− ηC*)c). Hence, the potential energy and CO2 emission intensity under the meta-frontier can be calculated, as shown in Figs. 7 and 8. These figures show a comparison of actual energy and CO2 emission intensity and potential energy CO2 emission intensity of agricultural sector during the period of 2001–2012. The variation tendency of actual and potential energy intensity is generally consistent, but the actual energy intensity is significantly higher than the potential energy intensity during the study period. In 2007, energy input per billion GDP in the agricultural sector is 0.0807 t, while the optimal value was 0.0315 t. The actual energy intensity changed a little from 0.0774 in 2001 to 0.0772 in 2012, while the potential energy intensity relatively changed more from 0.0284 in 2001 to 0.0339 in 2012. This is in line with the above findings that most provinces in China have an energy-saving potential in the agricultural sector. A detectable gap can be seen that actual and potential energy intensity has been narrowing since 2007. With regard to the reduction potential of CO2 emission intensity, provinces in central and western regions show relatively large amount of PCRTG. Specifically, the actual CO2 emission intensity dropped from 0.207 in 2001 to 0.1925 in 2012, while the potential CO2 emission

6

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

Fig. 9. Boxplots of the AEUI (left) and ECPI (right) for different regions. 1

intensity increased from 0.0789 in 2001 to 0.097 in 2012. As a whole, the actual CO2 emission intensity are gradually falls and the potential CO2 emission intensity gradually rises, which indicates the reduction potential of CO2 emission intensity is narrowing.

0.9 0.8 Axis Title

0.7

4. Discussion

0.6 0.5 0.4 0.3 0.2

The average score of ECPI in the agricultural sector is 0.425, which is a little lower than the national average [28]. Nevertheless, they both have an upward tendency on the whole. For the AUEI, its average score is much lower than the fossil-fuel power plant [45]. This is probably because China's agricultural sector is still at the primary stage of modernization. Production factors of the agricultural sector tend to be less efficient for effective output. The aforementioned empirical results may be due to the fact that provinces in eastern China are more motivated to improve management performance in terms of the AUEI and the ECPI than provinces in central and western China. Fig. 9 shows the boxplots of the AUEI (left) and ECPI (right) for different regions, based on which we can compare the medians and the standard deviations of the AUEI and the ECPI of China's agricultural sector between the different regions. It can be seen that eastern China shows the largest medians for both the UEI and the ECPI, with lower standard deviations for the UEI and the higher standard deviations for the ECPI. This indicates that provinces in this region do not operate under relatively similar technology conditions. For instance, although Tianjin and Zhejiang provinces are both in this region, they show relative smaller AUEI and ECPI than other province in eastern China. Meanwhile, take Hebei province for example, its average AUEI is 0.5097 and ranked highly in the region. However, its average ECPI is only 0.386 and ranked backward in the same region. Central China shows the medians for both the AUEI and the ECPI, with larger standard deviations for the AUEI and the smaller standard deviations for the EEPI. As for the western region, both AUEI and ECPI are the smallest among the three regions. The medians and the standard deviations of the UEI and the EEPI in western China are both small, indicating that province in this region generally operate under relatively similar backward technology conditions. The main reason causing these differences among different regions is that the efficiency of different input factors are always not the same. As shown in Fig. 10 within the study period, capital efficiency is always falling, energy efficiency is rising and labor efficiency produces modestly changes. This is the main reason that contributed to the AEUI and ECPI changes in the agricultural sector of different regions in China. Meanwhile, we examine the difference between the actual and potential energy intensity and CO2 emission intensity in the study period (see Figs. 7 and 8). It indicates that, with slack and advanced production technology considered, there is vast energy-saving and CO2

0.1 0

2001

2002

capital 0.7991 0.7599 0.7559 0.7285 0.7439 0.7072 0.6852 0.6417 0.6139 0.5803 0.5781 0.5915 labour 0.814 0.836 0.8832 0.8658 0.8283 0.8377 0.8227 0.8588 0.811 0.7885 0.8096 0.8058 energy 0.7179 0.7051 0.7208 0.7006 0.6849 0.6861 0.7018 0.674 0.6821 0.6943 0.7317 0.7419

Fig. 10. The efficiency of different input factors in agricultural sector, 2001–2012.

Fig. 11. Comparison of actual and potential energy intensity of the three regions, 2001– 2012. (unit: Mtc/billion yuan RMB).

emission reduction in China's agricultural sector. In this section, we further provide an in-depth analysis of the particular case of China's agricultural sector in each region. Fig. 11 plots and compares the actual and potential energy intensity in the three regions. The eastern region not only shows the lowest actual energy intensity but also displays the smallest reduction potential of energy intensity. Its potential energy intensity increased 7

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

the inefficiencies for all inputs and outputs are incorporated to measure the integrated efficiency and the energy–environmental performance of the Chinese agricultural sector. We conduct the empirical analysis using a sample of 30 provinces in China for the year 2001–2012. The research results reveal huge differences in both integrated efficiency of inputs-outputs and energy –CO2 performance across provinces as well as regions. The main findings of this paper are as follows. (1) Generally, most provinces and regions did not perform efficiently in energy consumption and CO2 emissions in China's agricultural sector. Provinces in eastern China generally performed better in AUEI and ECPI than the rest provinces. Provinces in western China generally have the lowest unified efficiency and the energy– environmental performance. (2) Most provinces and regions show the potential of energy and CO2 emission reduction. According to our empirical results, the reduction potential of energy intensity and CO2 emission intensity can reach 59.6177% and 56.4948% respectively of the actual level in theory. Meanwhile the vast central and western regions show great reduction potential of energy intensity and CO2 emission intensity compared with the eastern region. Based on these empirical findings, it is very important to improve total-factor efficiency of production factors. The government should further increase investment in R & D in the agricultural sector [29], use scale management innovation, promote rural collective property rights system reform and build a modern agricultural science and technology innovation system in order to improve the integrated efficiency of inputs-outputs and performance of energy use and CO2 emissions in China's agricultural sector. On one hand, specializing division and adjusting measures to local conditions help to improve the regional efficiency. On the other hand, large-scale land operation contributes to capital, technology, management, which can improve the efficiency of resource combination and the spillover effects of scale operation. Now China's energy consumption and CO2 emission of agricultural sector are bound to increase further in the near future due to agricultural mechanization. However, agricultural activities tend to be environmental-friendly under the idea of sustainable development. The relative better methodology employed in this paper can be considered to be a reliable policy assessment tool for such a meaningful estimation. In the long run, technology diffusion across provinces and regions will play an important role in energy saving and emission reduction. Therefore, the Chinese government should spare no effort to push the less developed central and western regions to catch-up and embrace the advanced technology of the well-developed eastern regions.

Fig. 12. Comparison of actual and potential CO2 emission intensity of the three regions, 2001–2012. (unit: Mtc/billion yuan RMB).

from 0.0262 in 2001 to 0.0356 in 2012, and the reduction potential of energy intensity dropped from 0.0424 in 2001 to 0.0364 in 2012. In contrast, the western region not only shows the highest actual energy intensity but also displays the biggest reduction potential of energy intensity. Its potential energy intensity dropped from 0.1035 in 2001 to 0.0901 in 2012, and the reduction potential of energy intensity dropped from 0.0733 in 2001 to 0.0364 in 2012. The central region seems to be in a moderate situation compared with the eastern and western regions. Fig. 12 plots and compares the actual and potential CO2 emission intensity of each region. During the period 2001–2012, the actual CO2 emission intensity drooped from 0.1643 in 2001 to 0.1571 in 2012, and the reduction potential of CO2 emission intensity dropped from 0.0953 in 2001 to 0.0532 in 2012, in the eastern region. However, in the western region, the actual CO2 emission intensity also shows a declining tendency, dropping from 0.0318 in 2001 to 0.2545 in 2012. It also shows the biggest reduction potential of CO2 emission intensity in the western region, which dropped from 0.02319 in 2001 to 0.1669 in 2012. According to the above-mentioned results, it seems that the regions with low integrated efficiency of inputs-outputs index (AUEI) and energy–CO2 performance (ECPI) should be allocated with more quotas of energy and CO2 emission intensity reduction. However, the fact is that in China, the central and western regions generally fall behind in production technology. Nevertheless, it is impossible to eliminate the technology gap among the regions in the short run. It is important that the distribution plan of reducing allowance should base on the comparable group frontiers [10].

Acknowledgements The paper is supported by Xiamen University - Newcastle University Joint Strategic Partnership Fund, the Grant for Collaborative Innovation Center for Energy Economics and Energy Policy (No: 1260-Z0210011), and Xiamen University Flourish Plan Special Funding (No:1260-Y07200), Shandong Provincial Natural Science Foundation, China (No. ZR2016GB10),National Natural Science Foundation of China (No.71603148).

5. Conclusion

References

Many papers have used the DEA approach to explore the energy and environmental performance of the industrial and service sectors, but few have adopted the NDDF method for the agricultural sector. Seeing that the conventional DDF may overestimate real efficiency, this study employed a newly developed indicator – non-radial DDFs proposed by Zhou et al. [49] and Zhang et al. [45] – to measure the integrated efficiency of inputs-outputs and performance in energy use and CO2 emissions of China's agricultural sector. Under the TNDDF,

[1] Barros CP, Managi S, Matousek R. The technical efficiency of the Japanese banks: non-radial directional performance measurement with undesirable output. Omega 2012;40(1):1–8. [2] Bian Y, Hu M, Wang Y, et al. Energy efficiency analysis of the economic system in China during 1986–2012: a parallel slacks-based measure approach. Renew Sustain Energy Rev 2016;55:990–8. [3] Boyd GA, McClelland JD. The impact of environmental constraints on productivity

8

Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx

R. Fei, B. Lin

[28] Lin BQ, Du KR. Energy and CO2 emissions performance in China's regional economies: do market-oriented reforms matter?. Energy Policy 2015;78:113–24. [29] Lin BQ, Fei RL. Analyzing inter-factor substitution and technical progress in the Chinese agricultural sector. Europ. J Agron 2015;66:54–61. [30] Liu Z, Qin CX, Zhang YJ. The energy-environment efficiency of road and railway sectors in China: Evidence from the provincial level. Ecol Indic 2016;69:559–70. [31] Oggioni G, Riccardi R, Toninelli R. Eco-efficiency of the world cement industry: a data envelopment analysis. Energy Policy 2011;39:2842–54. [32] Picazo-Tadeo AJ, Gómez-Limón JA, Reig-Martínez E. Assessing farming ecoefficiency: a Data Envelopment Analysis approach. J Environ Manag 2011;92:1154–64. [33] Rahim E, Marzie S. Investigation of CO2 emission reduction and improving energy use efficiency of button mushroom production using data envelopment analysis [15 September]. J Clean Prod 2015;103:112–9. [34] Riccardi R, Oggioni G, Toninelli R. Efficiency analysis of world cement industry in presence of undesirable output: application of data envelopment analysis and directional distance function. Energy Policy 2012;44:140–52. [35] Seyed H, et al. Study on energy use efficiency for paddy crop using data envelopment analysis (DEA) technique [July–August]. Appl Energy 2009;86(7– 8):1320–5. [36] Seyed H, et al. Improving energy use efficiency of canola production using data envelopment analysis (DEA) approach [May]. Energy 2011;36(5):2765–72. [37] Song M, an Q, Zhang W, Wang Z, Wu J. Environmental efficiency evaluation based on data envelopment analysis: a review. Renew Sustain Energy Rev 2012;16:4465–9. [38] Sueyoshi T, Goto M. DEA approach for unified efficiency measurement: assessment of Japanese fossil fuel power generation. Energy Econ 2011;33:292–303. [39] Vlontzos G, Niavis S, Manos BA. DEA approach for estimating the agricultural energy and environmental efficiency of EU countries. Renew Sustain Energy Rev 2014;40:91–6. [40] Wu J, et al. A comprehensive analysis of China's regional energy saving and emission reduction efficiency: from production and treatment perspectives. Energy Policy 2015;84:166–76. [41] Y.R. Wu, China's Capital Stock Series by Region and Sector. University of WesternAustralia 2009 (Discussion Paper, No. 0902). [42] Zhang N, Choi Y. A comparative study of dynamic changes in CO2 emission performance of fossil fuel power plants in China and Korea. Energy Policy 2013;62:324–32. [43] Zhang N, Choi Y. Environmental energy efficiency of China regional economies: a non-oriented slacks-based measure approach. Soc Sci J 2013;50:225–34. [44] Zhang N, Choi Y. A note on the evolution of directional distance function and its development in energy and environmental studies 1997–2013. Renew Sustain Energy Rev 2014;33:50–9. [45] Zhang N, Kong F, Choi Y, Zhou P. The effect of size-control policy on unified energy and carbon efficiency for Chinese fossil fuel power plants. Energy Policy 2014;70:193–200. [46] Zhou DQ, Wu F, Zhou X, et al. Output-specific energy efficiency assessment: a data envelopment analysis approach. Appl Energy 2016;177:117–26. [47] Zhou P, Poh KL, Ang BW. A non-radial DEA approach to measuring environmental performance. Eur J Oper Res 2007;178:1–9. [48] Zhou P, Ang BW, Han J. Total factor carbon emission performance: a Malmquist index analysis. Energy Econ 2010;32:194–201. [49] Zhou P, Ang BW, Wang H. Energy and CO2 emission performance in electricity generation: a non-radial directional distance function approach. Eur J 2012;221:625–35.

improvement in integrated paper plants. J Environ Econ Manag 1999;38:121–42. [4] BP. -Statistical_Review_of_world_energy__workbook; 2014. [5] Chang TP, Hu JL. Total-factor energy productivity growth, technical progress, and efficiency change: an empirical study of China. Appl Energy 2010;87(10):3262–70. [6] Chen S, Golley J. ‘Green’ productivity growth in China's industrial economy. Energy Econ 2014;44:89–98. [7] Choi Y, Zhang N, Zhou P. Efficiency and abatement costs of energy-related CO 2, emissions in China: a slacks-based efficiency measure. Appl Energy 2012;98(5):198–208. [8] Chung YH, Färe R, Grosskopf S. Productivity and undesirable outputs: a directional distance function approach. Microeconomics 1995;51(3):229–40. [9] Cornillie J, Fankhauser S. The energy intensity of transition countries. Energy Econ 2002;26(3):283–95. [10] Du K, Lu H, Yu K. Sources of the potential CO2 emission reduction in China: a nonparametric metafrontier approach. Appl Energy 2014;115(4):491–501. [11] Färe R, Grosskopf S. New Directions: efficiency and Productivity. New York: Springer; 2005. [12] Färe R, Grosskopf S, Pasurka CA. Jr, Environmental production functions and environmental directional distance functions. Energy 2007;32:1055–66. [13] Färe R, Grosskopf S, Pasurka CA. Jr., Toxic releases: an environmental performance index for coal-fired power plants.. Energy Econ 2010;32(1):158–65. [14] Färe R, Grosskopf S, Lovell CK, Pasurka C. Multilateral productivity comparisons when some outputs are undesirable: a nonparametric approach. Rev Econ Stat 1989;71:90–8. [15] Fukuyama H, Weber WL. A directional slacks-based measure of technical efficiency. Socio-Econ Plann Sci 2009;43:274–87. [16] Fukuyama H, Weber WL. A slacks-based inefficiency measure for a two-stage system with bad outputs. Omega 2010;38:389–409. [17] George V, Spyros N, Basil M. A DEA approach for estimating the agricultural energy and environmental efficiency of EU countries [December]. Renew Sustain Energy Rev 2014;40:91–6. [18] Hailu A, Veeman TS. Non-parametric productivity analysis with undesirable outputs: an application to the Canadian pulp and paper industry. Am J Agric Econ 2001;83:605–16, [?]. [19] Haynes KE, Ratick S, Cummings-Saxton J. Pollution prevention frontiers: a data envelopment simulation. Environ Program Eval: a Prim 1998:270–90. [20] Hua Z, Bian Y, Liang L. Eco-efficiency analysis of paper mills along the Huai River: an extended DEA approach. Omega 2007;35:578–87. [21] International Monetary Fund. February 2014. IMF Country Report No. 14/54; 2014. [22] Khoshnevisan B, Rafiee S, Omid M, et al. Reduction of CO 2, emission by improving energy use efficiency ofgreenhouse cucumber production using DEA approach. Energy 2013;55(1):676–82. [23] Khoshnevisan B, Rafiee S, Omid M, et al. Applying data envelopment analysis approach to improve energy efficiency and reduce GHG (greenhouse gas) emission of wheat production. Energy 2013;58(3):588–93. [24] Lee JD, Park JB, Kim TY. Estimation of the shadow prices of pollutants with production/environment inefficiency taken into account: a nonparametric directional distance function approach. J Environ Manag 2002;64:365–75. [25] Liddle B. Revisiting world energy intensity convergence for regional differences. Appl Energy 2010;87:3218–25. [26] Lin B, Tan R. Ecological total-factor energy efficiency of China's energy intensive industries. Ecol Indic 2016;70:480–97. [27] Lin B, Zhao H. Technology gap and regional energy efficiency in China's textile industry: a non-parametric meta-frontier approach. J Clean Prod 2016;137:21–8.

9