Technology gap and regional energy efficiency in China's textile industry: A non-parametric meta-frontier approach

Journal of Cleaner Production 137 (2016) 21e28

Contents lists available at ScienceDirect

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

Technology gap and regional energy efficiency in China's textile industry: A non-parametric meta-frontier approach Boqiang Lin a, b, *, Hongli Zhao a a

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

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 March 2016 Received in revised form 10 June 2016 Accepted 10 July 2016 Available online 11 July 2016

Based on the theory of total factor production, this paper analyzes energy efficiency in China's textile industry at the regional levels using non-parametric meta-frontier approach and a provincial panel data during the period, 2000e2012. We further analyze the regional differences in energy utilization technology gap using the technology gap ratio. Irrespective of the frontier (meta or group), the empirical result depicts a tremendous energy saving potential in China's textile industry. Relative to meta-frontier, the average energy efficiency of China's textile industry is 0.673 during the sample period; hence, the energy saving potential is 32.7% if output remains unchanged. Relative to group frontier, the average efficiency of China's textile industry is 0.797, which may overestimate the true level of energy utilization. From the regional perspective, significant differences exist in energy technology within the textile industry. During the sample period, the energy utilization technology gap ratio (TGR) of the Textile Industry in eastern China remains above 0.95 and it's steadily improving, approaching the optimum for the whole textile industry. Moreover, the textile industries in central and western China have improvement potentials of 19.6% and 27.4%, respectively. Finally, based on the results from the regional energy efficiency analysis, future policy priorities are suggested. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Energy efficiency Non-parametric meta-frontier Technology gap DEA Alternative elasticity of energy

1. Introduction Energy is a powerful driver of social and economic development. With China's rapid development, the dependence of the economy on energy has obviously increased. Moreover, the issue of shortage of resources and energy has increasingly become a serious problem (Lin and Tian, 2016). To solve the energy source gap, we will depend more on the international energy market, which has not only affected China's energy security, but also has made the volatility of the international energy market a serious challenge to national stability (Lin and Ouyang, 2014). China's heavy consumption of non-renewable fossil fuels such as coal and petroleum is a direct and major cause of national environmental deterioration (Xu and Lin, 2015). Therefore, to ease the conflict between economic growth and the environment, energy conservation and

* Corresponding author. Collaborative Innovation Center for Energy Economics and Energy Policy, China Institute for Studies in Energy Policy, Xiamen University, Xiamen, Fujian, 361005, PR China. E-mail addresses: [email protected], [email protected] (B. Lin). http://dx.doi.org/10.1016/j.jclepro.2016.07.055 0959-6526/© 2016 Elsevier Ltd. All rights reserved.

improvement in energy efficiency have become the inevitable choices at present and in the future (Lin and Du, 2013). The textile industry, well known as a traditional pillar industry in China, is an important sector of the national economy and international trade. Its role aids market growth, promote employment, and increase the income of farmers; thus accelerating the process of urbanization and promoting social harmony. It is worth to note that China is the largest textile and garment producer and exporter in the world; sustainable growth of textile exports is therefore essential to ensure stable growth of China's foreign exchange reserves, balance of international payments, and stability of exchange rate. Textile exports (in proportion to the total global textile export) increased from 10.3% in 2000 to 35.2% in 2012 (CEIC China Database, 2015). It can be said that the textile industry has developed rapidly since China's reform and opening up. In 2012, enterprises at national scale in industry had a total industrial output value of 4.7612 billion Yuan (approximately USD 764 million), up by 27.46% from a year earlier, and accounted for 6.58% of that of all designed size enterprises nationwide (CEIC China Database, 2015).

22

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

However, the textile industry also causes serious pollution (Alkaya and Demirer, 2014). Energy consumption of the industry accounts for approximately 4.3% of total energy consumed by the manufacturing sector,1 and it follows that it is one of China's largest pollution emitters. The rapid development of China's textile industry has resulted in continuous increases in energy consumption and carbon dioxide emissions (see Fig. 1). Simultaneously, China faces severe challenges related to energy resources and environmental constraints; these challenges include inefficient resource use, high energy consumption and serious pollution. Under the background of increasingly scarce energy resources and the need for sustainable development, China's textile industry faces considerable energy constraints. Effective allocation of resources and maximization of energy efficiency have become critical to the development of the industry. In current situation of serious energy and environmental constraints, improving energy efficiency is considered to be the most realistic, most effective, and the lowest cost way of energy saving and emission reduction. Scientific measurement of the energy efficiency of China's textile industry provides a basis for comparison of regional differences in energy technology level, and so can guide the formulation of regional energy saving and emissions reduction policies. This paper therefore evaluates the energy efficiency of China's textile industry in different provinces. Based on technology gap ratio (TGR), the regional energy technology gap of the textile industry is analyzed quantitatively, in order to establish differences in energy saving targets in the textile industry, and provide a basis for decision-making. The rest of the paper is organized as follows: section 2 introduces the methods and data in detail. Section 3 undertakes empirical analysis. Section 4 presents the related discussion. Section 5 gives conclusions and policy suggestions. 2. Methods and data 2.1. Literatures review In recent years, research on energy efficiency measurement methods has made important progress. Methods for measuring energy efficiency can be classified into two types: single factor and total factor. Single factor measurement methods mostly use energy intensity as a metric of energy efficiency, and this is also the case of early studies on energy efficiency. Although this measure is simple, it takes energy as the only factor of production, ignoring other factors. It also does not consider alternative elasticity of energy and other production factors. To make up for this defect, total factor energy efficiency is used. Foreign scholars began studying energy efficiency earlier than Chinese scholars, and there are numerous international studies on energy efficiency. Hu and Kao (2007) developed an energy saving ratio based on energy efficiency, and then calculated this ratio (including per capita calculations) for the 17 APEC economies during 1991e2000. The results showed that China had the highest energy saving rate among the 17 economies compared, implying
ndezthat China faced a serious problem of energy waste. F. Herna Sancho et al. (2011) used the DEA method to analyze energy efficiency and the impact factors of industrial waste water treatment in Spain. The results showed that only 10% of industrial enterprises used energy efficiently. Azadeh et al. (2007) integrated the PCA and DEA methods, and evaluated the energy efficiency of the major OECD countries in energy intensive industries. The results showed that energy saving potential for fossil fuels was far greater than that of electricity. Similarly, Olanrewaju et al. (2012) first proposed the

1

The data is from the annual report of the Chinese textile industry for 2013.

Fig. 1. Energy consumption and CO2 emissions of China's textile industry during 1990e2012. Source: Chinese Energy Statistical Yearbook

integration of the IDA-ANN-DEA method to study energy consumption in 15 industrial sectors in Canada. Toshiyuki and Mika (2012) applied the DEA-DA method to model the scoring and ranking of efficiency of Japanese electric power industry and enterprises, and found almost no changes for Japanese power companies during 2005e2009. In recent years, research on China's energy efficiency has made fruitful achievements. Domestic and foreign scholars have studied China's energy efficiency using different data and models. Hu and Wang (2006) first adopted the DEA model of constant scale return, and studied total factor energy efficiency in China from 1995 to 2002. Based on the factorization method and the structure index method, Shi (1999) argued that since the reform and opening up, Chinese energy efficiency had improved significantly. This study thus concluded that the nation's opening up, industrial structure and economic system were the important determinants of energy efficiency. Yang and Shi (2008), Lin and Long (2015), and Li and Cheng (2008) also studied the total energy efficiency of various regions of China using the DEA method. Shi (2006) studied the regional energy efficiency of China, comparing energy efficiency among different provinces and evaluating the energy saving potential for different regions. This study found that energy efficiency in the southeast coast was higher, while China's inland areas had abundant coal resources and relatively lower energy efficiency. Strictly speaking, the precondition for the comparison of technical efficiency must be that all production units share a similar technical level. Otherwise, we cannot find the real reason for the loss of production unit efficiency because of a lack of uniform standards. Imbalances in regional economies and social development in China have also caused differences in energy utilization technology. Therefore, the technology frontier varies among the regions (eastern, central and western) in China. Energy technology level is clearly higher in the eastern region than in the central and western regions. However, existing researches on energy efficiency are based on the assumption that the three regions share the same technology level and hence evaluate the efficiency of the decisionmaking units with different technology frontier at the same frontier. Examples of such studies include Hu and Wang (2006), Wei and Shen (2007), Qu (2009), Shi (2006) and so on. To overcome the problem that decision making units with different technology fronts are at the same frontier, Battese and Rao (2002) took the technical differences between the groups into consideration. Based on stochastic frontier analysis (SFA), the metafrontier production function research framework was proposed. Assuming that not all production units shared the same technical level, they analyzed and compared the technical efficiency of production units under group frontier and meta-frontier. O'Donnell (2007) further used data envelopment analysis (DEA) to construct

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

nonparametric statistical measures of the leading edge, and solved the problem of the SFA method only being able to deal with the single output problem. In domestic studies, Wang et al. (2011) first adopted the nonparametric common frontier approach to analyze the energy efficiency gap between eastern and western China. Cui et al. (2013) analyzed the reasonable allocation of land resources in three regions by measuring non-agricultural land utilization in cities in eastern, central and western China. This paper takes the total factor production theory as its foundation, and then uses the non-parameter common front method to analyze and compare regional differences of energy technology and the total energy efficiency of the textile industry. Under the framework of a common frontier, we can not only get the unified standards to compare regional total factor energy efficiency, but can also use the index of TGR to study the regional energy use gap between mainstream and cutting-edge technology, in order to offer effective suggestions for the improvement of energy use and energy technology in the textile industry.

23

facing meta-frontier, and the distance function based on metafrontier is

n o D*t ðxt ; yt Þ ¼ sup d > 0 : ðxt =d; yt Þ2Pt* ðxt Þ " ¼

d

# n o 1 inf d > 0 : ðdxt ; yt Þ2Pt* ðxt Þ d

(5)

D*t ðxt ; yt Þ is the ratio of actual production unit level to metafrontier input levels. Therefore, the technical efficiency based on common frontier can be expressed as:

h i1 0  D*t ðxt ; yt Þ ¼ TEt* ðxt ; yt Þ  1

(6)

Since the meta-frontier is the envelope curve which is not lower than that of the group, the technical efficiency of the production unit under meta-frontier and group frontier can be expressed as follows:

2.2. Methodology

D*t ðxt ; yt Þ  Dkt ðxt ; yt Þ0TEt* ðxt ; yt Þ  TEtk ðxt ; yt Þ

We assume that production technology is the knowledge and ability that transforms input into output, set xt ¼ ðx1 ; x2 ; …; xM Þ2RM as the production input, and set þ yt ¼ ðy1 ; y2 ; …; yN Þ2RN þ as the production output. In the common frontier method, all production units are first classified by the relevant standards, and then K groups are obtained. The production unit of each group can be put into the same technical collective T k ,

O'Donnell et al. (2007) pointed out that the meta-frontier technical efficiency can be decomposed into technical efficiency within the group and the TGR. TGR is the ratio of the technical efficiency of the production units under meta-frontier and group frontiers, which reflects the gap between the group frontier and the common frontier technology. The bigger the TGR, the closer the group frontier technology is to the meta-frontier. If TGR equals 1, no gap exists between the group frontier technology and metafrontier technology.

Ttk ¼ fðxt ; yt Þ : xt can produce yt g

(1) 0  TGRkt ðxt ; yt Þ ¼

Production possibility set PðxÞ can be defined as:

o n Ptk ðxÞ ¼ yt : ðxt ; yt Þ2Ttk ;

(2)

Ptk ðxÞ

where, the upper bound of is the group frontier. At this point, the production frontier that the production unit faces is the group frontier production, while the distance function based on input minimization is

n o Dkt ðxt ; yt Þ ¼ sup d > 0 : ðxt =d; yt Þ2Ptk ðxt Þ " ¼

d

n

inf d > 0 : d

o

ðdxt ; yt Þ2Ptk ðxt Þ

Dkt ðxt ; yt Þ TE*t ðxt ; yt Þ ¼ 1 D*t ðxt ; yt Þ TEkt ðxt ; yt Þ

TE*t ðxt ; yt Þ ¼ TEkt ðxt ; yt Þ  TGRkt ðxt ; yt Þ

#1 (3)

(4)

We assume that there is a k sub technology collection T k in the whole, k ¼ 1,2, …,K, which operates under a common technical collection T * .A common set of techniques is union set of each technical set: Tt* ¼ fTt1 ∪Tt2 ∪/∪Ttk g, and Tt* ¼ fðxt ; yt Þ : xt can produce yt g, at which point the production may be Pt* ðxÞ ¼ fyt : ðxt ; yt Þ2Tt* g, where the upper bound of Pt* ðxÞ is the meta-frontier. Referring to Battese and Rao (2002), metafrontier is different from group frontier, implying that the technical gap between the groups can be surpassed, and all the production units have the same technical potential to pursuit investment minimization. At this point, the production unit is

(8)

(9)

The meta-frontier, the group frontier and the TGR can be illustrated by Fig. 2. Fig. 2 shows three groups of samples; i.e., group 1, 2 and 3, respectively. The other hypothesis is that the meta-frontier is a convex function. Based on the meta-frontier and group frontier, the

Referring to Farrell (1957), the distance function is the ratio of the actual production levels to the frontier production levels, thus it can be used to measure the technical efficiency of the production unit. That is, energy efficiency or total energy efficiency,

h i1 0  Dkt ðxt ; yt Þ ¼ TEtk ðxt ; yt Þ  1

(7)

Fig. 2. A graphical illustration of the meta-frontier, group frontier and TGR.

24

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

technical efficiency and TGR of production unit A can be respectively expressed as:

TE* ðAÞ ¼

0B 0C 0B=0D 0B ; TE2 ðAÞ ¼ ; TGR2 ðAÞ ¼ ¼ 0D 0D 0C=0D 0C

1 ½Dt ðx 8t ; yt Þ ¼ TEðxt ; yt Þ ¼ minq;l q < yit þ Yt l  0; s:t: qxit  Xt l  0; : l  0; i ¼ 1; 2; /I

(10)

(11)

2.3. Variables and data sources In this paper, the sample used for empirical analysis comprises 2000e2012 panel data on China's textile industry in 29 provinces (some administrative regions, including Tibet and Chongqing, are excluded because of a lack of data). When using a non-parametric meta-frontier, we divide the 29 provinces into different levels according to technology group. Referring to Battese et al. (2004), combining with geographic position and production technology level, the 29 provinces (cities) are divided into three major areas, namely eastern China, central China and western China.2 All value variables are comparable in 2000, as is the logarithm of price. The data mainly come from the China Statistical Yearbook, Energy Statistical Yearbook and Statistical Yearbook of Individual Provinces, and China Industry Economy Statistical Yearbook. The main variables of this paper are as follows: The output of the textile industry (denoted as Y). Considering the input elements containing the middle consumable-energy, total textile industry output value is taken as the output indicator. To eliminate the impact of price factors, we use the metallurgy producer price index to deflate raw data, and convert it to values for the year 2000 to represent constant actual industrial production. Capital stock of textile industry (denoted as K). There exist considerable researches on the estimation of capital stock in China, especially industrial capital stock. Zhang and Choi (2013), Xu (2010), Huang and Ren (2002), He (2011), and others have estimated industrial capital stock. Based on previous researches, we use the perpetual inventory method to measure the capital stock of China's textile industry. The basic formula for productive capital stock can be expressed as:

Kt ¼ Kt1 ð1  dt Þ þ It

(12)

where, Kt represents capital in the textile industry; Kt1 represents capital stock in the base year; dt represents the rate of depreciation; and It represents the total investment. Capital stock in the base year (denoted as Kt1 ). Referring to Zhang and Zhang (2003), we use net value of fixed assets of the textile industry in 1990 as capital stock in the base year, which in turn supplies a basis for calculating capital stock in subsequent years. Corresponding data is sourced from the China Statistical Yearbook. Investment amount (denoted as It ). Zhang and Zhang (2003) used the original value of fixed assets to estimate fixed capital formation in the industrial sector. An investment data series is formed through first order difference of the original value of fixed

2 The eastern China including Beijing, tianjin, hebei, liaoning, Shanghai, jiangsu, fujian, shandong, zhejiang, guangdong, Hainan; The central China including shanxi, jilin, heilongjiang, anhui, jiangxi, henan, hunan, hubei; The Western China including Inner Mongolia, guangxi, sichuan, guizhou, yunnan, shaanxi, gansu, qinghai, ningxia, xinjiang.

assets. This method is also used for estimation in the textile industry. Deflating the investment data series by the price index of investment in fixed assets, we get the investment series at 2005 constant prices. Rate of depreciation (denoted as d). The depreciation rate is computed using the method applied by He (2011). The rate of depreciation ¼ (accumulated depreciation in Period T  accumulated depreciation in Period T1)/the original value of fixed assets in period T1. Energy input in the textile industry (denoted as E). The data on textile industry energy consumption come from the statistical yearbook of provinces and cities. The individual data are obtained by linear interpolation. Non-scalar energy conversion standard coal reference coefficients are taken from the Energy Statistical Yearbook 2012. Labor input in the textile industry (denoted as L). Strictly speaking, labor input should consider both the size and quality of labor force, and should be measured by labor time at standard labor intensity. Under a market economy, labor remuneration can reflect the time and quality of labor investment. However, because of the particularity of the economic system, income distribution and regional differences of wage levels, labor remuneration cannot accurately respond to labor investment. Comparatively speaking, the number of labor force can reflect the objective size of labor input. This paper thus uses the average annual number of employees in the textile industry as a measure of labor input. Labor input data come from Statistical Yearbook of each province. 3. Empirical results Given regional heterogeneity, the energy efficiency index of an industry in a specific region cannot be directly compared with that of the same industry in another region (Xu and Lin, 2016). To solve this problem we consider the energy efficiency of the textile industry in various provinces with technological heterogeneity under the common frontier. Using the DEA method to solve the distance function, this paper takes all the input and output data during the sample period as the reference technology set. Total energy efficiency of different provincial textile industries during 2000e2012 is calculated using Matlab7.6 software at the group frontier and meta-frontier. The results are shown in Table 1 and Fig. 3. Table 1 shows that under the meta and group frontier, average energy efficiencies of Chinese textile industry in the sample period are 0.673 and 0.797, respectively. This shows that current economic output can be maintained if national textile industry energy investment is cut by 32.7% and 20.3%, respectively. China's textile industry has low energy efficiency, and to reach the potential optimal level of the industry, there is 32.7% of the improvement space; meaning there exists huge potential for energy saving and emissions reduction. The average energy efficiency under the metafrontier in the eastern region is 0.806, far larger than 0.542 in the central region and 0.675 in the western region. This shows that the textile industry in eastern China is more efficient in terms of energy investment than it is in central or western China. As shown in Fig. 3, energy efficiency score of the textile industry under the meta-frontier is lower than that of the group frontier. This shows a significant technological gap between the meta-frontier and group frontier, mainly because the metafrontier that reflects the best green production technology nationwide is a potential efficient improvement frontier. While the group frontier is based on the fact that the technology gap among eastern, central, and western China cannot be transcended in the short term, and that the provinces in all regions have already optimized the production technology frontier. This optimal

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

25

Table 1 Total factor energy efficiency of the textile industry under the meta-frontier and regional frontier. Province

The average value of meta-frontier

The average value of regional frontier

Beijing Tianjin Hebei Liaoning Shanghai Jiangsu Zhejiang Fujian Shandong Guangdong Hainan Shannxi Jilin Heilongjiang Anhui Jiangxi Henan Hunan Hubei Inner Mongolia Guangxi Sichuan Guizhou Yunnan Shannxi Gansu Qinghai Ningxia Xinjiang

0.709 0.744 0.603 0.789 0.857 0.829 0.754 0.896 0.708 0.926 0.784 0.423 0.612 0.68 0.896 0.651 0.598 0.723 0.659 0.546 0.612 0.528 0.465 0.968 0.444 0.398 0.387 0.405 0.432

0.709 0.744 0.607 0.941 0.857 0.829 0.754 0.969 0.708 0.926 0.784 0.542 0.748 0.79 0.969 0.92 0.826 0.918 0.78 0.736 0.956 0.821 0.627 0.969 0.704 0.598 0.559 0.556 0.646

Eastern China Central China Western China

0.806 0.675 0.542

0.821 0.832 0.741

Nationwide

0.673

0.797

Fig. 3. Total energy efficiency of the provincial textile industries under the meta-frontier and the group frontier. Note: 1e11 are eastern region provinces; 2e19 are central region provinces; and 20e29 are western region provinces.

combination reflects the best energy saving and emission reduction technologies in the relevant regions, and also reflects the advanced technology of the practical efficiency under existing technology conditions. The difference of energy efficiency score in eastern China under both frontiers is obviously less than that of central and western China. Taking Jiangsu in eastern china as an example, relative to group frontier, it has an energy efficiency index of 82.9%, which means that it can improve its energy efficiency by 17.1% through the use of best technology in eastern china. Relative to metafrontier, energy efficiency index of Jiangsu remains 82.9%, while potential improvement in energy efficiency remains 17.1% under nationwide implementation of the best reduction technology. This shows that Jiangsu is simultaneously on the group frontier and the meta-frontier. Table 1 shows that energy efficiencies of most provinces in eastern China do not differ significantly between both

kinds of frontier. This means that the group frontiers of the eastern region mostly coincide with the meta-frontier. The textile industry in the eastern region is the benchmark for the national textile industry in terms of energy saving, and thus represents the nationwide optimum energy-saving and emission reduction technology. According to Table 1 and Fig. 3, the energy efficiency score under the group frontier is generally overestimated. This overestimation is especially notable in central China, followed by western China. The energy efficiency score of central China exceeds that of eastern China (82.1%). However, when the reference technology set is changed to meta-frontier, the results are reversed. Energy efficiency score of 80.6% in eastern China significantly exceeds that of central China (67.5%). The obvious overestimation of energy efficiency under the group frontier is due to the gap between central and eastern regions in terms of geographical location, policy

26

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

support, economic development and openness. Under the group frontier, central China performs best because the textile industry is labor intensive. In addition, great textile provinces are concentrated in the central China, where industry tends to be more labor intensive, and faces a low technology frontier with labor intensiveness. Under existing technological conditions, there is little room for the provinces within a region to be closer to the frontier of the group relying on improving management and allocation efficiency. When the energy efficiency score of central China is evaluated under the meta-frontier, there is a higher decrease in energy efficiency score under the metafrontier, mainly because the meta-frontier is mostly constructed by the eastern provinces with advanced equipment and technology. This suggests that regional technology difference is probably the main reason for the overall low energy efficiency score of the national textile industry; it also shows that eliminating the regional technology gap may realize energy saving in the textile industry. From regional perspective, under the meta-frontier, eastern China has the highest energy efficiency score, followed by central, and finally, western China. Moreover, under the group frontier, energy efficiency is highest in central China, followed by eastern and western. Taking Shaanxi in western China as an example, average energy efficiency score reaches 70.4%, and energy saving space is 29.6% with the energy technology level of western China. If the results for Shanxi are extrapolated nationwide with reference to a meta-frontier, the average energy efficiency score is only 44.4%, while energy saving potential can be improved to be 55.6%, far exceeding 29.6% under group frontier. Similar conclusions can also be obtained from the comparison of other provinces. To verify total factor energy efficiency differences for the two leading edge regional textile industries, this paper further applies the Mann Whitney U test (a non-parametric statistical method). Test results are shown in Table 2. Note: “***” Indicate significance at the 1% level. “**” Indicate significance at the 5% level. According to Table 2, the total energy efficiency of the two frontiers differs significantly below the 5% level in nationwide and the central China, while the total factor energy efficiency of the two kinds of frontier in the west China is significant at 1% level. Possible reasons for this result are as follows: under the group frontier, it is assumed that each region has its own technological frontier, and its efficiency measure can only reflect the energy use level of its own technology; while under the meta-frontier, the reference technology is the optimal level of national energy utilization. A large gap persists in technology and management experience in the textile industry between the central and western regions as well as the eastern coastal areas. This means that energy utilization level under the group frontier is significantly overestimated, verifying the previous analysis. However, the energy efficiency score of the eastern China under both frontiers is not significantly different at 1% level, and the P value reaches 0.673. The main reason for this is that eastern China has the most advanced technology, and provinces with powerful textile industry are concentrated in eastern China. At the same time, China's textile industry in the eastern region, which is considered as the national

Table 2 The results of Mann-Whitney U test for total factor energy efficiency differences under the two frontiers in the textile industry. Index

Eastern China

Central China

Western China

Nationwide

U test value Prob > U

0.274 0.673

2.153 0.019**

2.695 0.006***

2.787 0.013**

Note: “***” Indicate significance at the 1% level. “**” Indicate significance at the 5% level.

textile industry's benchmark for energy saving and emission reduction, represents the highest level of energy technology in the textile industry. Since both reference technologies are basically the same, measurement results for energy efficiency do not show obvious differences. 4. Discussion Using the meta-frontier approach to study regional total energy efficiency of the textile industry, the most important evaluation index is the technical gap ratio (TGR). This indicator reflects a measurement of how close the group frontier to the meta-frontier in different regions. Based on the measurement of energy efficiency in the textile industry, we can find that the regional differences in energy technology in the textile industry are analyzed using the index of TGR. Table 3 and Fig. 4 show the differences and changes of TGR of the textile industry in the three major regions during the period, 2000e2012. From Table 3, we can find that according to the nonparametric Kruskal-Wallis test, the null hypothesis is significantly rejected at the level of 1%. It shows that TGRs are different among different regions. The average TGR in western, central, and eastern regions reach 0.961, 0.804 and 0.726, respectively. Among them, given constant economic output, eastern China can achieve 96.1% of potential energy use (the highest in China's textile industry). As can be seen from Fig. 4, during the sample period, the energy use TGR in the eastern area remains stands above 0.95; showing steady improvement, and having almost a little difference with potential energy utilization level nationwide. The main reasons are that the eastern region enjoys advantages of geographic location, higher openness, and stronger ability to attract foreign investment. Consequently, there are advanced textile techniques and management methods. Therefore, the energy use of China's textile industry in the eastern region basically represents the highest level of the national textile industry. In contrast, the TGR of the textile industry in the central and western regions are lower than that of the textile industry in the eastern region, that is, 0.814 and 0.734, respectively. Compared with the common frontier, the improvement space for energy technology in these two regions is 18.6% and 26.6%, respectively. Fig. 5 shows the time trends of energy technology gaps in the three regions. As we can see from the figure, energy technology gap of the textile industry between central and west China has been shrinking since 2000, while the energy technology gap curve between eastern and western China had an inflection point in 2009, with a continuous narrowing. The energy technology gap between eastern and central regions of China is likely to expand. The reasons for this may be due to the fact that the provinces in central and western China are landlocked, and thus lack the superior national conditions enjoyed by eastern region. The interior provinces also lack technological exchanges and cooperation with the eastern provinces, resulting in the suffering weakness of textile technology. On the other hand, with the rise of production factors price in the coastal areas, especially the rise of labor force

Table 3 The descriptive statistics of TGRs for regional groups in China's textile industry. Region

Minimum value

Maximum value

Average value

Standard deviation

Eastern China Central China Western China

0.937 0.802 0.716

1 0.814 0.749

0.961 0.804 0.726

0.012 0.006 0.08

Kruskal-Wallis test, Chi-squared ¼ 18.786, p-value ¼ 0.05**. Note: “***”, “**”, “*” represent 1%, 5%, 10% of the significance level.

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

27

Fig. 4. The TGRs in different regional groups in China's textile industry.

Fig. 5. Energy technology gap in the three regions of China's textile industry.

price, the traditional advantages of the textile industry in the coastal areas are gradually declining, and thus the textile industry is moving to the western region. In recent years, with more transfer from the eastern region to the central and western regions, many provinces (in the central and western regions) have reduced the entry barriers for projects to attract more foreign investment. This may cause disorder in the development of the textile industry and have a negative impact on the development of energy technology.

5. Conclusions and policy implications

the meta-frontier. The results of energy efficiency measured under the group frontier may significantly be overestimated. (3) During the sample period, the values of TGR in eastern China have remained above 0.95 with a steady increase. Moreover, slight difference exists between regional and national potential energy use levels. (4) Since 2000, energy technology gap of the textile industry between central and western China has been shrinking. The curve of the energy technology gap between the eastern and western regions of China had an inflection point in 2009 and experienced a shrink, while that between the eastern and central regions grew.

5.1. Conclusions In this paper, the regional differences in total energy efficiency of the textile industry from 2000 to 2012 are analyzed under the non-parametric meta-frontier approach. The TGR is applied to study the energy technology gap of the textile industry among eastern, central and western regions. The main conclusions are as follows: (1) The average score of energy efficiency in China's textile industry is 0.673 under the meta-frontier and 0.797 under the group frontier. From the national perspective, whether based on the meta-frontier or group frontier, the overall energy efficiency of China's textile industry is still at a low stage, and has large potentials for energy savings. (2) The score of energy efficiency under the meta-frontier does not exceed those of group frontier, which demonstrates a significant technological gap between the group frontier and

5.2. Policy implications Energy efficiency improvement is conducive for energy saving to a certain degree. Base on the analysis, the following policy recommendations are suggested. (1) To realize energy saving and emissions reduction in the textile industry, the government should encourage technological innovation, technological transformation, introduction of advanced technology, and also strengthen the support of technical projects. In terms of the technical aspects, it is necessary to eliminate the production equipments with high energy and water consumption in the printing, dyeing, and chemical fiber industries. The key to realizing such improvements lies in promoting air conditioning and air compressor energy saving, high temperature waste heat and

28

B. Lin, H. Zhao / Journal of Cleaner Production 137 (2016) 21e28

water recycling. Other key issues to help achieve such improvements include the introduction of energy saving technologies and energy efficient recycling technologies, as well as energy saving spinning cooling, and spinning heat medium circulating heating equipment. (2) Inter-regional differences in technological levels will serve as a long-term constraint to the improvement of overall energy efficiency in the textile industry. We should make different polices base on the current and specific situations of textile industry and local energy resources utilization. Firstly, eastern region grasps the optimal technology for energy conservation and emissions reduction; but there is room for improvement by enhancing management capabilities and improving allocation efficiency. These improvements can be made by overcoming management inefficiency factors and also by enhancing allocation efficiency. For example, it will be appropriate to reduce the role of some state-owned enterprises with many administrative levels and low efficiency to improve the production efficiency in the textile industry. Various forms ownerships exist in China's textile industry, but state-owned textile enterprises are the dominant ones.3 Therefore, to appropriately reduce the role of some state-owned enterprises with many administrative levels and low efficiency would be a development direction for the eastern textile industry where the frontier of the region has been very close to the common frontier though some provinces within the region have not yet reached the frontier of the group. Secondly, in central China, the efficiency index performs best under the group frontier, while the performance is worse under the meta-frontier. Central China should therefore focus on the optimal production technology set when making energy-saving and emission reduction policy. It is essential to continuously encourage technology innovation of energy saving and low-carbon to ensure that the group frontier moves to the meta-frontier, gradually eliminating the energy inefficiency caused by the technology gap. On the one hand, the government can improve on the openness of central region, and promote the free flow of elements and the penetration of advanced production technology from both China and abroad. Moreover, the central region must also constantly raise its level of economic development to expand the accumulation of physical capital and human capital, and to enhance its regional scientific and technical level, thus narrowing the technology gap with the advanced regions. Thirdly, for western region, performance of the efficiency index under the group and meta-frontier lies in the middle; they must therefore start with management and technology to improve the performance efficiency. Resources allocation efficiency can be enhanced by improving management levels with existing technology. Considering the geographical position of western region and the fact that it is difficult to spread advanced energy-saving and emission reduction technology from a more advanced eastern China, policy makers must provide special attention and more resources to western China. This includes increasing investment in new energy and renewable energy equipment to promote the penetration and diffusion of green production technology to western China. Acknowledgements The paper is supported by Newhuadu Business School Research Fund, the Grant for Collaborative Innovation Center for Energy

3 These content are from the annual report of the Chinese textile industry for 2013.

Economics and Energy Policy (No: 1260-Z0210011), Xiamen University Flourish Plan Special Funding (No:1260-Y07200).

References Alkaya, E., Demirer, G.N., 2014. Sustainable textile production: a case study from a woven fabric manufacturing mill in Turkey. J. Clean. Prod. 65, 595e603. Azadeh, A., Amalnick, M.S., Ghaderi, S.F., Asadzadeh, S.M., 2007. An integrated DEA PCA numerical taxonomy approach for energy efficiency assessment and consumption optimization in energy intensive manufacturing sectors. Energy Policy 35 (7), 3792e3806. Battese, G.E., Rao, D.S.P., 2002. Technology gap, efficiency, and a stochastic metafrontier function. Int. J. Bus. Econ. 1 (2), 87e93. Battese, G.E., Rao, D.S.P., O'Donnell, C.J., 2004. A meta-frontier production function for estimation of technical efficiencies and technology gaps for firms operating under different technologies. J. Prod. Anal. 21, 91e103. CEIC China Database, 2015. Available at: http://ceicdata.securities.com/cdmWeb/. China Energy Statistical Yearbook, 2012. China Statistics Press, Beijing (in Chinese). China Industry Economy Statistical Yearbook, 2001e2013. China Statistics Press, Beijing (in Chinese). China Statistical Yearbook, 2001e2013. China Statistics Press, Beijing (in Chinese). Cui, W., Miao, J.J., Yang, Y., 2013. Urban non-agricultural eco-efficiency and affecting factors based on carbon emission. China Popul. Resour. Environ. 23 (7), 63e69 (in Chinese). Farrell, M.J., 1957. The measurement of productive efficiency. J. R. Stat. Soc. 120 (3), 253e290. He, X.P., 2011. The industrial energy conservation potential and the influencing factors in China. J. Financial Res. 2, 34e46 (in Chinese).
ndez-Sancho, F., Molinos-Senante, M., Sala-Garrido, R., 2011. Energy efficiency Herna in Spanish wastewater treatment plants: a non-radial DEA approach. Sci. Total Environ. 409 (14), 2693e2699. Hu, J.L., Kao, C.H., 2007. Efficient energy-saving targets for APEC economies. Energy Policy 35, 373e382. Hu, J.L., Wang, S.C., 2006. Total-factor energy efficiency of regions in China. Energy Policy 34, 3206e3217. Huang, Y.F., Ren, R.E., 2002. Chinese manufacturing capital stock, perpetual inventory method to estimate. Economics 2, 378e395 (in Chinese). Li, S.X., Cheng, J.H., 2008. Analysis on the energy efficiency about main industrial provinces in China: 1990~2006. J. Quant. Tech. Econ. 10, 34e43 (in Chinese). Lin, B.Q., Du, K.R., 2013. Technology gap and China's regional energy efficiency: a parametric metafrontier approach. Energy Econ. 40, 529e536. Lin, B.Q., Long, H.Y., 2015. A stochastic frontier analysis of energy efficiency of China's chemical industry. J. Clean. Prod. 87, 235e244. Lin, B.Q., Ouyang, X.L., 2014. Electricity demand and conservation potential in the Chinese nonmetallic mineral products industry. Energy Policy 68, 243e253. Lin, B.Q., Tian, P., 2016. The energy rebound effect in China's light industries: a translog cost function approach. J. Clean. Prod. 112, 2793e2801. Olanrewaju, O.A., Jimoh, A.A., Kholopane, P.A., 2012. Integrated IDA-ANN-DEA for assessment and optimization of energy consumption in industrial sectors. Energy 46 (1), 629e635. O'Donnell, C.J., Rao, D.S.P., Battese, G.E., 2007. Metafrontier frameworks for the study of firm level efficiency and technology ratios. Empir. Econ. 34, 231e255. Qu, X.E., 2009. China's interprovincial total factor energy efficiency change decomposes. J. Quant. Tech. Econ. 8, 30e43 (in Chinese). Shi, D., 1999. The influence of industrial structure changes on energy consumption demand. J. Quant. Tech. Econ. 12, 50e52 (in Chinese). Shi, D., 2006. Regional differences in China's energy efficiency and conservation potentials. China ind. Econ. 10, 49e58 (in Chinese). Toshiyuki, S., Mika, G., 2012. Efficiency-based rank assessment for electric power industry: a combined use of data envelopment analysis (DEA) and DEAdiscriminant Analysis (DA). Energy Econ. 34 (3), 634e644. Wang, K.L., Yang, B.Ch., Yang, Li, 2011. The regional disparities of China's total factor energy and energy technology. Sci. Res. Manag. 33 (5), 56e63 (in Chinese). Wei, C., Shen, M.H., 2007. Energy efficiency and energy productivity: a comparison based on the panel data by province. J. Quant. Tech. Econ. 9, 111e120 (in Chinese). Xu, J., 2010. Re-estimating the capital stock of China. Stat. Res. 12, 73e79 (in Chinese). Xu, B., Lin, B.Q., 2015. Factors affecting CO2 emissions in China's transport sector: a dynamic nonparametric additive regression models. J. Clean. Prod. 101, 311e322. Xu, B., Lin, B.Q., 2016. Regional differences of pollution emissions in China: contributing factors and mitigation strategies. J. Clean. Prod. 112, 1454e1463. Yang, H.L., Shi, D., 2008. Energy-efficiency methods and comparing the energy efficiencies of different areas in China. Econ. Theory Bus. Manag. 3, 12e19 (in Chinese). Zhang, N., Choi, Y., 2013. Total-factor carbon emission performance of fossil fuel power plants in China: a metafrontier non-radial malmquist index analysis. Energy Econ. 40, 549e559. Zhang, J., Zhang, Y., 2003. Reestimation of capital stock Chinese K. Econ. Res. 7, 35e42 (in Chinese).