Advertisement

Energy Efficiency

, Volume 11, Issue 8, pp 1941–1953 | Cite as

Energy efficiency measurement of Chinese Yangtze River Delta’s cities transportation: a DEA window analysis approach

  • Xiaohong Chen
  • Yuyan Gao
  • Qingxian AnEmail author
  • Zongrun Wang
  • Luka Neralić
Original Article

Abstract

In this paper, the transportation energy efficiency of Yangtze River Delta’s 15 cities is studied in the period from 2009 to 2013. To measure transportation’s dynamic performance, the window analysis of data envelopment analysis (DEA) is used to estimate the efficiency in cross-sectional and time-varying data. Capital inputs and labor inputs are selected as the two non-energy inputs, energy consumption as the energy input, passenger volume and freight volume are selected as the two desirable outputs, and carbon dioxide is chosen as the undesirable output. The empirical study shows that Shanghai had the highest transportation energy efficiency, followed by Zhejiang province, and the average efficiency of Jiangsu province was the worst. A regression analysis on the efficiency values and their three influencing factors was carried out. The results show that per capita gross domestic product and the per capita area of paved roads in a city negatively influenced the efficiency values, while the number of public transportation vehicles per 10,000 persons positively influenced the transportation efficiency values. Lastly, some policy recommendations, conclusions, and suggestions for further research are given.

Keywords

Window analysis Yangtze River Delta Data envelopment analysis Regression analysis Transportation energy efficiency 

Introduction

China’s “Reform and Opening Up” policy has brought great economic development, which has made the country the world’s second largest economy following the U.S.A. However, the rapid economic growth of China has brought high energy consumption and high environmental pollution. With a large population and relatively low energy resources, China’s per capita energy consumption is only 40% of the world average, which is more than 20 years behind developed countries. China’s energy use is much higher than that of developed countries and the world average. It is approximately three times that of the U.S.A. and 7.2 times that of Japan. Improving energy efficiency has become an urgent issue for the Chinese government and future economic development.

Among all industries, the transportation industry is one of the highest energy-consuming and environmental-polluting sectors. In 2012, Chinese transportation consumed 302 million tons of standard coal, which is 7.2% more than the last 10 years’ consumption rate according to the data from the National Bureau of Statistics of China (NBSC). Due to the huge increase of coal burning, more gases, including carbon dioxide, are generated. Since 2010, the worldwide transportation industry has become the second biggest factor of greenhouse gas production, which makes up 22% of global carbon dioxide emissions, as reported by the International Energy Agency. To solve the problem, China’s 12th Five-Year Plan for 2011–2015 proposed establishing a “green, low-carbon development concept.” Its main feature was “low resource consumption, high utilization efficiency, (and) less environmental pollution.”

The seriousness of the problem makes people pay more attention to the energy efficiency of transportation. An increasing number of scholars have realized and examined the importance of transportation energy efficiency. One of the popular methods is data envelopment analysis (DEA), which is a widely used mathematical programming method to evaluate the relative efficiency of a set of comparable decision-making units (Charnes et al. 1978; Blum 2015; An et al. 2017a, 2017b; Wu et al. 2017). Since DEA can address the multiple-input and multiple-output systems, it has recently been widely extended and applied many areas, including transportation (Azadi et al. 2014; An et al. 2018; Wu et al. 2013; Zaim and Gazel 2017). Cui and Li (2014) proposed a new model named three-stage virtual frontier DEA model for assessing the energy efficiency of transportation. To examine the rationality of the model, data from 30 Chinese provinces between 2003 and 2012 were collected. From the results, it turns out that transportation energy efficiency can be greatly influenced by transportation structure and management measures. Bi et al. (2014) built a non-radial DEA model that consisted of multidirectional efficiency analysis (MEA). This analysis used data from 2006 to 2010 and received unexpected results for the measurement of regional energy and efficiency of the Chinese transportation industry. Overall, 30 regions and 3 major areas of Chinese efficiency patterns have been investigated. Song et al. (2015) analyzed the influence between the railway transportation and the environmental impact of transportation in China. They applied a non-radial DEA model under managerial disposability and a panel beta regression with fixed effects to analyze 30 Chinese provinces’ environmental efficiencies and to model the influence of environmental efficiency in railway transportation, respectively. Azadi et al. (2014) confirmed explicit, accomplishable, and practical targets in order to achieve green supply chain management’s requirements of a company and its transportation service providers (TSPs). They also find targets for two-stage network structures by applying two approaches. To analyze the relationship between perceived airport quality and airport profit margins, Merkert and Assaf (2015) implemented two-stage DEA models. These models helped to analyze a single efficiency score that includes the potentially conflicting indicators of perceived service quality and profitability for airports. Chang et al. (2016) analyzed Chinese environmental efficiency in the transportation industry by applying a non-radial DEA model with the slack-based measure (SBM). After the analyses, it shows that eco-efficient transportation industries exist in a few Chinese provinces and a majority of the provinces are unable to meet 50% of the ideal environmental efficiency levels. Park et al. (2016) also applied a slack-based data envelopment analysis approach to measure the environmental efficiency of U.S. transportation sectors. Chu et al. (2016) built a SBM-DEA model with a parallel computing design for the efficient measurement of transportation systems that can significantly reduce the calculation time. Chen et al. (2017) applied DEA for analyzing the rural transportation management in Quinte West. Liu and Wu (2017) developed a framework based on an SBM-DEA model to measure energy and environmental efficiency of transportation sectors. The proposed model can effectively identify the potential amount of reductions in both energy consumption and CO2 emissions for those sectors.

However, transportation energy efficiency in these mentioned papers did not consider a dynamic evaluation. They cannot provide us with more information regarding the efficiency changes over time. In this paper, the window analysis DEA model is used to measure efficiency. DEA window analysis, introduced in Charnes et al. (1985), is a kind of dynamic evaluation method to handle cross-sectional and time-varying data. The data from 29 administrative regions of China from 2000 to 2008 were selected as the sample to calculate the energy and environmental efficiency by applying a modified DEA model (Wang et al. 2013). Moreover, the efficiency in cross-sectional and time-varying data was measured by the DEA window analysis technique. Based on some existing DEA models, Bian and Yang (2010) evaluated the aggregated efficiency of resources and the environment. Both DMUs’ energy efficiencies and environment efficiencies can be appraised by the above models at the same time. Yang and Chang (2009) presented a two-stage range-adjusted model to evaluate the inefficiency and congestion in mixed energy consumption. DEA window analysis was combined with this model to investigate the case over time. Řepková (2015) used the DEA window analysis approach to analyze the data of Czech commercial banks. She examined the efficiency of the Czech banking sector during the period 2003–2012 and used the DEA window analysis approach based on an input-oriented model to measure banking efficiency. In Hunjet et al. (2015), the dynamic relative efficiency of 12 selected Croatian towns was studied using DEA. With the number of employed workers and employed assets as the inputs and income as the output, window analysis was considered for the period from 2004 to 2009 for an input-oriented (output-oriented) DEA model with constant (variable) returns to scale. Computational results on the dynamic relative efficiency of the towns are presented and analyzed. In the study by Žaja et al. (2017), the impact of industry restructuring on the efficiency of 21 Croatian state-owned electricity distribution centers was studied. An input-oriented BCC model for the first stage of DEA with operating costs as the input and total electricity sales, number of customers, and network length as outputs was specified. Using pooled data for the period from 2005 to 2013, the relative efficiency of the electricity distribution centers was estimated each year. In the second stage, the efficiency scores were regressed on contextual variables to estimate their impact on efficiency scores and document approximately 2.8% annual productivity gains over the years following the regulatory changes.

However, there are some gaps in the papers mentioned above. First, a few papers use the window analysis of DEA in the transportation industry. Therefore, these papers in general did not consider a dynamic evaluation. They cannot provide us with more information about the efficiency changes. Second, many domestic researchers considered the “province” as the decision-making unit to study its transportation energy efficiency. They did not take the “city” as a decision-making unit. This paper views “city” as a decision-making unit and applies the window analysis DEA model to measure the efficiency in cross-sectional and time-varying data. This is an analysis that the previous researchers have not performed. This paper intends to evaluate transportation energy efficiency in 15 of the Yangtze River Delta’s cities’ transportation systems from 2009 to 2013. The 15 cities in the Yangtze River Delta region are considered as the 15 different decision-making units.

The rest of this paper is organized as follows. The “Methodology” section introduces the window analysis DEA model. The “Transportation energy efficiency evaluation of the Yangtze River Delta region” section presents the data and the analysis of the transportation energy efficiency of Yangtze River Delta’s 15 cities from 2009 to 2013. Regression analysis and policy recommendations are shown in the “Regression analysis and policy recommendations” section. The conclusions are given in the “Conclusions” section.

Methodology

In this section, we present a DEA model to evaluate the transportation energy efficiency of the Yangtze River Delta’s 15 cities. In the DEA analysis, each DMU corresponds to a city. In addition, in order to consider a dynamic evaluation, the DEA window analysis model is used to calculate transportation energy efficiency.

DEA models for energy efficiency evaluation

Data envelopment analysis, which was initiated by Charnes et al., 1978, is a nonparametric approach to evaluate the relative efficiency of a set of comparable decision-making units (DMUs). The purpose of this paper is to develop a framework for measuring transportation energy efficiency in 15 of the Yangtze River Delta’s cities’ transportation systems. The paper presents a framework based on the variable returns to scale to evaluate the efficiency of the transportation energy of Yangtze River Delta’s 15 cities.

Suppose that there are n DMUs denoted by DMUj(j = 1, …, n), and each of them represents a city in the Yangtze River Delta. Each of them consumes m different non-energy inputs xij(i = 1, …, m) and p different energy inputs ekj(k = 1, …, p) to produce s different desirable outputs yrj(r = 1, …, s) along with d different undesirable outputs bgj(g = 1, …, d). Separately, the corresponding production possibility set, assuming variable returns to scale (VRS), can be characterized as
$$ T=\left\{\left({x}_i,{e}_k,{y}_r,{b}_g\right)\left|\begin{array}{l}\sum \limits_{j=1}^n{\lambda}_j{x}_{ij}+{s}_i^{x-}={x}_i,\sum \limits_{j=1}^n{\lambda}_j{e}_{kj}+{s}_k^{e+}={e}_k,\\ {}\sum \limits_{j=1}^n{\lambda}_j{y}_{rj}-{s}_r^{y+}={y}_r,\sum \limits_{j=1}^n{\lambda}_j{b}_{gj}={b}_g,\kern0.3em \sum \limits_{j=1}^n{\lambda}_j=1\kern0.1em \end{array}\right.\right\},\kern0.5em \mathrm{for}\kern0.5em {\displaystyle \begin{array}{l}i=1,2,..,m,k=1,2,\dots, p,\kern0.5em \\ {}r=1,2,\dots, s,g=1,2,\dots, d,\end{array}} $$
(1)
where \( \kern0.1em {\lambda}_j,{s}_i^{x-},{s}_k^{e+},{s}_r^{y+}\ge 0 \) for all j, i, l, r, g.

In set (1), \( {s}_i^{x-},{s}_k^{e+},{s}_r^{y+} \) are slack variables for non-energy (capital input and labor input), energy (energy input), desirable outputs (passenger volume and freight volume) and undesirable outputs (CO2 emissions). In the process of production, each DMU likes to produce desirable outputs as much as possible, to not produce undesirable outputs and to consume as few resource inputs as possible.

The paper presents a new model on the assumption of variable returns to scale to evaluate the transportation energy efficiency as
$$ {\displaystyle \begin{array}{l}{E}_1=\min \theta \\ {}s.t.\kern0.3em \sum \limits_{j=1}^n{\lambda}_j{x}_{ij}+{s}_i^{x-}={x}_{ij_0},\kern0.7em i=1,\dots, m\\ {}\kern1.5em \sum \limits_{j=1}^n{\lambda}_j{e}_{kj}+{s}_k^{e+}=\theta {e}_{kj_0},\kern0.7em k=1,\dots, p\\ {}\kern1.5em \sum \limits_{j=1}^n{\lambda}_j{y}_{rj}-{s}_r^{y+}={y}_{rj_0},\kern0.8000001em r=1,\dots, s\\ {}\kern1.5em \sum \limits_{j=1}^n{\lambda}_j{b}_{gj}=\theta {b}_{gj_0},\kern0.7em g=1,\dots, d\\ {}\kern1.5em \sum \limits_{j=1}^n{\lambda}_j=1\\ {}\kern1.5em {\lambda}_j,{s}_i^{x-},{s}_k^{e+},{s}_r^{y+}\ge 0,\kern0.9000001em \mathrm{for}\kern0.2em \mathrm{all}\kern0.2em j,i,l,r.\end{array}} $$
(2)
In model (2), θ represents constant returns to scale (CRS) technical efficiency, which is between 0 and 1 for the efficiency value of a city. If the variable is larger, the condition of the respective city acts better in energy saving and environmental protection. When θ = 1 and \( {s}_i^{x-},{s}_k^{e+},{s}_r^{y+}=0 \), the DMU is efficient. It means that the city is considered to be environmentally inefficient. It has potential to reduce energy utilization and environmental pollutant emissions. When θ = 1, and some of the slacks are not 0, the DMU is weakly efficient for DEA. When θ < 1, the DMU is inefficient. It means that the city’s environmental efficiency is low, and it has the potential to reduce energy utilization and environmental pollution emissions. Model (2) proposed that the environmental efficiency measure is a kind of radial efficiency, which may have the weak ability to distinguish transportation energy efficiencies. According to Bian and Yang (2010), model (3) will be extended to a non-radial measure as
$$ {\displaystyle \begin{array}{l}{E}_2=\min \frac{1}{2}\left(\frac{1}{p}\sum \limits_{k=1}^p{\theta}_k^e+\frac{1}{d}\sum \limits_{g=1}^d{\theta}_g^b\right)\\ {}s.t.\kern1.2em \sum \limits_{j=1}^n{\lambda}_j{x}_{ij}+{s}_i^{x-}={x}_{ij_0},\kern0.7em i=1,\dots, m\\ {}\kern2.6em \sum \limits_{j=1}^n{\lambda}_j{e}_{kj}+{s}_k^{e+}={\theta}_k^e{e}_{kj_0},\kern0.7em k=1,\dots, p\\ {}\kern2.6em \sum \limits_{j=1}^n{\lambda}_j{y}_{rj}-{s}_r^{y+}={y}_{rj_0},\kern0.7em r=1,\dots, s\\ {}\kern2.6em \sum \limits_{j=1}^n{\lambda}_j{b}_{gj}={\theta}_g^b{b}_{gj_0},\kern0.7em g=1,\dots, d\\ {}\kern2.6em \sum \limits_{j=1}^n{\lambda}_j=1\\ {}\kern1.5em {\lambda}_j,{s}_i^{x-},{s}_k^{e+},{s}_r^{y+}\ge 0,\kern0.9000001em \mathrm{for}\kern0.2em \mathrm{all}\kern0.2em j,i,l,r.\end{array}} $$
(3)

Model (3) adjusted the energy consumption and pollutant emissions by using different proportions. Energy consumption accounted for \( {\theta}_k^e \) and CO2 emissions accounted for \( {\theta}_g^b \). It showed that the energy consumption and pollutant emissions can be reduced in different proportions in order to allow the evaluation of the city to reach its best environmental efficiency boundary. In model (3), this paper sets the weights to 1/2. In general, decision-makers can assign different proportions. Model (3) has a higher discriminatory ability than model (2), so model (3) is used to evaluate the transportation energy efficiency of the 15 cities in the Yangtze River Delta.

DEA window analysis for Yangtze River Delta’s cities transportation systems

In this part, this paper will evaluate the transportation energy efficiency in 15 of Yangtze River Delta’s cities’ transportation systems from 2009 to 2013, which is considered a dynamic evaluation. It provides more information about efficiency changes. Therefore, applying DEA window analysis in the transportation industry is meaningful and practical to explore transportation energy efficiency.

DEA window analysis, proposed in Charnes et al. (1985), is a dynamic evaluation model that can measure dynamic effects. The basic principle of DEA window analysis is that the same decision-making units that are in different time periods are regarded as different decision units through methods such as the moving average to choose different reference sets to evaluate the relative efficiency. DEA window analysis can be compared to DMU for the horizontal efficiency at different times. It can also be compared to different DMU longitudinal efficiencies at the same time and can compare the relative efficiencies from different DMU periods, which makes the analysis of panel data more flexible.

In our study, there are 15 cities and a time period of 5 years (from 2009 to 2013), which means n = 15 and T = 5. Following Zhang et al. (2011) and Halkos and Tzeremes (2009), reliable transportation energy efficiency results are obtained using a narrow window with a width of three (w = 3). Therefore, the first window contains the first 3 years of 2009, 2010, and 2011. Then, the window moves on a 1-year period by dropping the original year and adding a new year. The next 3 years of 2010, 2011, and 2012 form the second window. This process continues until the last window, which contains the last 3 years of 2011, 2012, and 2013, and is constructed. Finally, each city gets three windows, and the number of decision-making units (Yangtze River Delta in China) becomes 45 (n × w = 15 × 3). The radial and non-radial energy and environmental efficiency (E2 and E3) of the Yangtze River Delta’s 15 cities in each window can be obtained using DEA window analysis. Each year has three transportation energy efficiency values for each region.

In addition, 2009 and 2013 have only one value, 2010 and 2012 have two values, and 2011 has three values. Then, this paper calculates the average results of transportation energy efficiency of each city in the same year in order to get a new efficiency result for the 15 cities.

Transportation energy efficiency evaluation of the Yangtze River Delta region

Description of the areas of Yangtze River Delta region

The Yangtze River Delta city group is located in the east of China at the intersection of the “Gold Coast” and the Yangtze River “golden waterway.” Therefore, internal and external economic links are very convenient. It is China’s largest economic zone, and China’s comprehensive strength is in this strongest economic center. It is an important international gateway for the Asia Pacific region and the world’s major advanced manufacturing base. As a result, China took the lead among the world-class urban agglomeration areas. In the past 30 years, the Yangtze River Delta has been China’s fastest growing region. This region was formed with Shanghai as the leader of the Jiangsu, Zhejiang Economic Zone. Therefore, this region is China’s fastest growing regional economy, the largest share of China’s total economy, and it contains the most promising economic sectors. Therefore, we study the Yangtze River Delta’s transportation energy efficiency.

Yangtze River Delta’s 15 cities are defined as the DMUs. The data on labor input, passenger volume, and freight volume is obtained from the “China City Statistical Yearbook.” The data on capital inputs are obtained from the statistical yearbooks of each city. For example, the data on Nanjing’s capital input are obtained from the “Nanjing Statistical Yearbook.”

In this paper, we evaluate the transportation energy efficiency of the Yangtze River Delta region from 2009 to 2013. Due to the lack of data, we choose 15 cities to evaluate and divided the 15 cities into 3 categories, as shown in Table 1. The data from 2013 are taken as an example and are shown in Table 9.
Table 1

Three bigger areas

Area

City

Shanghai

Shanghai

Jiangsu

Nanjing, Wuxi, Xuzhou, Changzhou, Suzhou, Nantong, Lianyungang, Yangzhou, Zhenjiang

Zhejiang

Hangzhou, Ningbo, Wenzhou, Huzhou, Shaoxing

As shown in Table 1, there are three areas. The map of the Yangtze River Delta region is shown in Fig. 1. The first area is Shanghai. The second area is Jiangsu province, which consists of nine cities (Nanjing, Wuxi, Xuzhou, Changzhou, Suzhou, Nantong, Lianyungang, Yangzhou, and Zhenjiang). The third area is Zhejiang province, which consists of five cities (Hangzhou, Ningbo, Wenzhou, Huzhou, and Shaoxing).
Fig. 1

The Yangtze River Delta region map

This paper evaluated the transportation energy efficiency of the Yangtze River Delta region. We select three input indicators and three output indicators. Capital inputs (transportation investment in fixed assets) and labor inputs (the number of staff in the transportation sector) are selected as the two non-energy inputs, energy consumption (transportation energy consumption volume) as the energy input, passenger volume and freight volume as the two desirable outputs, and carbon dioxide as the undesirable output. The abovementioned input and output indicators are summarized in Table 2.
Table 2

Variables of inputs and outputs

Indicators

Variable

Units

Non-energy inputs

Capital

10 thousand yuan

Labor

10 thousand people

Energy input

Energy

10 thousand tons

Desirable outputs

Passenger volume

10 thousand people

Freight volume

10 thousand tons

Undesirable output

CO2

Tons

DEA window analysis

The efficiency values (E3) of Yangtze River Delta’s 15 cities from the year 2009 to 2013 are measured through model (3). Then, the DEA window analysis is applied to obtain the transportation energy efficiency. Table 3 lists the transportation energy efficiency in the Yangtze River Delta region.
Table 3

The transportation energy efficiency of the Yangtze River Delta region

Region

City

Year

    

Statistical measure

2009

2010

2011

2012

2013

Mean

Max

Min

SD

Shanghai

Shanghai

0.911

0.887

1.000

      
 

0.854

0.963

1.000

     
  

0.963

1.000

0.955

0.948

1.000

0.854

0.050

Jiangsu

Nanjing

0.636

0.800

1.000

      
 

0.771

1.000

1.000

     
  

1.000

1.000

0.402

0.846

1.000

0.402

0.202

Wuxi

0.240

0.243

0.270

      
 

0.248

0.274

0.347

     
  

0.289

0.363

0.203

0.275

0.363

0.203

0.049

Xuzhou

0.384

0.871

0.789

      
 

0.838

0.749

0.586

     
  

0.753

0.594

1.000

0.729

1.000

0.384

0.172

Changzhou

0.351

1.000

0.400

      
 

1.000

0.398

0.418

     
  

0.420

0.431

0.360

0.531

1.000

0.351

0.252

Suzhou

0.590

0.660

1.000

      
 

0.650

1.000

1.000

     
  

1.000

1.000

0.430

0.814

1.000

0.430

0.216

Nantong

0.485

0.694

1.000

      
 

0.660

0.854

1.000

     
  

0.860

1.000

0.389

0.771

1.000

0.389

0.216

Lianyungang

1.000

0.975

1.000

      
 

1.000

1.000

1.000

     
  

1.000

1.000

0.967

0.994

1.000

0.967

0.012

Yangzhou

1.000

0.772

0.644

      
 

0.809

0.721

0.686

     
  

0.789

0.751

0.604

0.753

1.000

0.604

0.108

Zhenjiang

0.815

0.653

0.530

      
 

0.676

0.590

0.575

     
  

0.688

0.652

0.625

0.645

0.688

0.625

0.077

Zhejiang

Hangzhou

0.658

0.848

1.000

      
 

0.848

1.000

0.866

     
  

1.000

0.987

1.000

0.912

1.000

0.658

0.111

Ningbo

0.980

1.000

0.933

      
 

1.000

0.923

0.985

     
  

0.939

1.000

1.000

0.973

1.000

0.923

0.031

Wenzhou

1.000

1.000

1.000

      
 

1.000

1.000

0.981

     
  

1.000

1.000

0.910

0.988

1.000

0.910

0.028

Huzhou

1.000

1.000

1.000

      
 

1.000

1.000

1.000

     
  

1.000

1.000

0.878

0.986

1.000

0.878

0.038

Shaoxing

0.576

0.621

0.521

      
 

0.680

0.569

0.564

     
  

0.664

0.653

0.690

0.616

0.690

0.521

0.057

All cities

Average

0.708

0.802

0.811

0.816

0.694

0.785

   
In Table 3, the average efficiencies of all cities in the different years from 2009 to 2013 are calculated, and the maximums, minimums, and the standard deviations are presented. Through the observations in Table 3, we find the following:
  1. (i)

    Not all cities in the Yangtze River Delta region have efficient transportation sectors. This means that average efficiency values are all below 1.

     
  2. (ii)

    Among all cities, the one that has the highest transportation energy efficiency is Lianyungang, with an efficiency score of 0.994. Wenzhou is ranked the second with an efficiency score of 0.994. Wuxi has the lowest efficiency score at 0.275. According to the data we gathered, the passenger volume and freight volume of Lianyungang are almost the same to Wuxi, whose input amount and CO2 emissions are twice those of Lianyungang, demonstrating that transportation energy efficiency of Lianyungang is higher than Wuxi. Both Lianyungang and Wuxi are port cities and Lianyungang is one of the ten biggest port cities. It connects 50 shipping lines and through which ships can travel to major worldwide ports. The Wenzhou port is also one of the 25 main Chinese ocean ports, and it plays an important role in the Chinese integrated transportation network. However, Wuxi has no ports. Thereby, compared with Lianyungang and Wenzhou, the transportation efficiency of Wuxi is low.

     
  3. (iii)

    The average value of transportation energy efficiency of Shanghai is 0.948, while that of Jiangsu province is 0.678. Among the nine cities of Jiangsu province, there are six cities (Nanjing, Xuzhou, Suzhou, Nantong, Lianyungang, and Yangzhou) whose efficiency scores are above 0.7 and 3 cities between 0.25 and 0.7. The average transportation energy efficiency scores of Zhejiang is 0.889. Among the five cities in Zhejiang province, there are four cities (Hangzhou, Ningbo, Wenzhou, and Suzhou) whose scores are all above 0.9, and only that of Shaoxing is 0.615. Overall, Jiangsu province has the lowest transportation energy efficiency, which may be caused by local government policies.

     
DEA window analysis is used to obtain efficiency values. According to the results of Table 3, the average annual efficiency values of the 15 cities are calculated. The average efficiency values of the 15 cities in different years are shown in Table 4.
Table 4

The average annual transportation energy efficiency of the Yangtze River Delta region

Region

City

2009

2010

2011

2012

2013

Shanghai

Shanghai

0.911

0.870

0.975

1

0.955

Jiangsu

Nanjing

0.636

0.786

1

1

0.402

Wuxi

0.240

0.245

0.278

0.355

0.203

Xuzhou

0.384

0.855

0.763

0.590

1

Changzhou

0.351

1

0.406

0.424

0.360

Suzhou

0.590

0.655

1

1

0.430

Nantong

0.485

0.677

0.905

1

0.389

Lianyungang

1

0.988

1

1

0.967

Yangzhou

1

0.791

0.718

0.718

0.604

Zhenjiang

0.815

0.665

0.603

0.613

0.625

Zhejiang

Hangzhou

0.658

0.848

1

0.927

1

Ningbo

0.980

1

0.9314

0.993

1

Wenzhou

1

1

1

0.990

0.9104

Huzhou

1

1

1

1

0.8780

Shaoxing

0.576

0.651

0.585

0.609

0.690

In Table 4, we find that the efficiency values of Huzhou from 2009 to 2012 are 1. However, in 2013, it is below 0.9. As seen from the collected data, the labor in Huzhou in 2013 is 19,800 people and from 2009 to 2012 is all below 10,000 people. Other indicators have not changed substantially. The labor input is excessive in 2013, resulting in a waste of resources, which reduces its efficiency value. Transportation energy efficiency values in Wuxi are all basically below 0.3. The efficiency values of Lianyungang and Wenzhou are more than 0.9 per year. This is because Wuxi does not have any sea ports, while Wenzhou and Lianyungang are important sea port cities. Therefore, Wuxi’s passenger volume and freight volume are relatively low. Consequently, its transportation energy efficiency is low. Thus, it can be concluded that transport structure has important impact on transportation energy efficiency.

In addition to Zhenjiang, Hangzhou, Xuzhou, and Shaoxing, other cities had transportation efficiency effects that decreased in 2013. By observing the data of these cities, governmental departments in 2013 increased labor inputs so that more people were involved in the transportation industry. However, the increasing labor inputs did not result in an increase of outputs. In other words, these resources have not been effectively utilized. The increase of the labor force does not improve the energy efficiency, but it wastes human resources. Governmental departments should reduce the waste of resources and use resources more rationally.

This paper selects 15 cities for the evaluation. It divides the 15 cities by region into Shanghai, Jiangsu province, and Zhejiang province. The efficiency results for the three different areas are shown in Table 5.
Table 5

The efficiencies for the three different areas from 2009 to 2013

Region

2009

2010

2011

2012

2013

Average

Shanghai

0.911

0.870

0.975

1.000

0.955

0.942

Jiangsu

0.611

0.740

0.741

0.745

0.553

0.678

Zhejiang

0.843

0.900

0.903

0.904

0.896

0.889

In Table 5, 15 cities are classified into three categories, which are Shanghai, Zhejiang province, and Jiangsu province. Shanghai gets the highest transportation energy efficiency mean value at 0.942, which is higher than the overall average efficiency mean value among the 15 cities at 0.785. (The value is obtained from the Mean value in Table 3.) This is higher than the mean efficiency values of the Zhejiang and Jiangsu provinces at 0.889 and 0.678, respectively. Jiangsu province’s average annual transportation capital input is 124.8 billion yuan, its energy input is 267.6 million tons, and its CO2 emissions are 63 thousand tons. By contrast, Zhejiang province’s average annual transportation capital input is 71.7 billion yuan, its energy input is 119 million tons, and its CO2 emissions are 24 thousand tons, which is approximately half that of Jiangsu province. However, their average annual passenger volumes and freight volumes are about the same. These gaps lead to different transportation energy efficiencies.

The result might be associated with some local government rules and policies. Shanghai has the highest transportation energy efficiency, which may be because the government of Shanghai has better transportation policies than those of Zhejiang and Jiangsu, such as restricting car purchases. In addition, management abilities may have important impacts on transportation energy efficiency.

Regression analysis and policy recommendations

Regression analysis on the environmental efficiency of transportation

In the real situation, some factors will affect the efficiency value. To explore the relationship between the influencing factors and the efficiency value, this paper selects three factors: per capita gross domestic product, the number of public transportation vehicles per 10,000 persons, and the per capita area of paved roads in the city. Per capita gross domestic product is defined as a reflection of the impact of the economic level on the energy efficiency of transportation. The number of public transportation vehicles per 10,000 persons is an indicator of the developmental level of urban public transportation and the traffic structure status. The per capita area of paved roads in a city can reflect the rationality of the urban road area.

The data of per capita gross domestic product, the number of public transportation vehicles per ten thousand persons and the per capita area of paved roads in the city are obtained from the “China City Statistical Yearbook.” SPSS software is used to conduct the regression analysis. Before taking the regression analysis, we excluded the multiple collinearity between the three environmental variables. The results of the analysis are shown in Tables 6 and 8. The adjusted R2 value is equal to 0.210 in Table 7. Although the R2 value is slightly small which may be caused by the small size of our samples, it does not affect our research results because it well represents the contribution of the explanatory variables in our model.
Table 6

Correlationsa

Variable

Values

Per capita GDP

Number of public transportation vehicles per 10,000 persons

Per capita area of paved roads in city

Values

1

   

Per capita GDP

− 0.393***

1

  

Number of public transportation vehicles per ten thousand population

0.030**

0.461

1

 

Per capita area of paved roads in city

− 0.319*

0.332

0.062

1

*Significant at 10%; **significant at 5%; ***significant at 1%

aPredictors: (constant) per capita GDP, number of public transportation vehicles per 10,000 persons, and per capita area of paved roads in city

Table 7

Model summaryb

Model

R

R 2

Adjusted R2

Std. error of the estimate

The Durbin-Watson

1

0.492a

0.242

0.210

0.220060

1.829

aPredictors: (constant) per capita GDP, number of public transportation vehicles per 10,000 persons, and per capita area of paved roads in city

bDependent variable: values

According to the literature (Kraft and Tırtıroğlu, 1998; Cullinane and Song, 2006), if the correlation coefficient (the first column in Table 6) of a factor is positive, it is positively correlated with the efficiency value. That is, its increase will make the transportation energy efficiency increase. If its coefficient is negative, its increase will reduce the efficiency value, and its increase will lower traffic energy efficiency.

Through the observations in Tables 6 and 8, we find the following:
  1. (i)

    Per capita gross domestic product. Its coefficient (− 0.393) is negative and significant for the efficiency value at the 1% significance level. This result is inconsistent with our expected results. In general, the increase in per capita gross domestic product will increase transportation energy efficiency. However, blind expansion may lead to excessive investment in production factors, resulting in wasted resources, which will reduce the efficiency.

     
  2. (ii)

    Number of public transportation vehicles per 10,000 persons. Its coefficient (0.030) is positive and significant for the efficiency value at the 5% significance level. A higher number of public transportation vehicles per 10,000 persons can improve transportation energy efficiency.

     
  3. (iii)

    Per capita area of paved roads in city. Its coefficient (− 0.319) is negative and significant for the efficiency value at the 10% significance level. This result is contrary to our expected results. Most likely, this may be because the road design is not standardized and is unreasonable. Therefore, a higher per capita area of paved roads in a city does not improve the transportation energy efficiency.

     
Table 8

Coefficients

Model

Unstandardized coefficients

Standardized coefficients

t

Sig.

Beta

Std. error

Beta

 

Constant

0.977

0.106

 

9.217

0.000

Per capita GDP

0.000

0.000

− 0.445

− 3.590

0.001

Per capita area of paved roads in city

− 0.09

0.005

− 0.186

− 1.689

0.096

Number of public transportation vehicles per 10,000 persons

0.016

0.007

0.246

2.103

0.039

Dependent variable: value

Policy recommendations

The transportation energy efficiency of the Yangtze River Delta region is high compared to other areas of China, but it is still low compared to developed countries. China has been dedicated to improving energy efficiency and has been successful. The practice in China shows that effective policies play an important role in forcing resource-using units to improve transportation energy efficiency.

Through this paper’s analysis, we can conclude that transportation structures and management measures have important impacts on transportation energy efficiency. The increasing investment leads to higher prices and lower consumption. The increase of passengers and freight falls behind the increase in transportation infrastructure inputs, so the transportation energy efficiency decreases. Transportation energy efficiency should be highly valued by government departments. Government departments should take effective measures to reduce the waste of resources and the emissions of harmful gases in the transportation sector. Government departments can also establish incentives in the region(s) whose transportation energy efficiency is high.

To explore the relationship between the influencing factors and the efficiency value, this paper selects three influencing factors (per capita gross domestic product, number of public transportation vehicles per 10,000 persons, and per capita area of paved roads in city). Considering the three factor’s influences on traffic efficiency, the government should take corresponding measures:
  1. (i)

    The government should not blindly pursue the rapid growth of per capita GDP. This can lead to excessive investments in production factors, resulting in wasted resources, which reduces efficiency.

     
  2. (ii)

    The government should properly regulate the design of the road and should not blindly build roads. This will improve traffic efficiency.

     
  3. (iii)

    Developing environmentally friendly city busses is an effective way to control air pollution in cities. Government departments should understand the needs of public transportation vehicles and improve their development.

     
  4. (iv)

    As much as possible, the government can drive urban development with low traffic efficiency and reduce the gap between cities. The efficiency of cities with low transportation efficiencies in the Yangtze River Delta region is only 0.203 during the 5 years from 2009 to 2013, and the gap is larger than that of cities with efficiencies of 1.

     

Rational use of resources, reduced traffic congestion, reduced environmental pollution, and improved social justice are our development goals. With the joint efforts of the government and the traffic departments, we can improve the efficiency of energy use and create a green, safe, comfortable, low energy consumption and low pollution environment.

Conclusions

With the rapid development of China’s economy, the transportation industry has become one of the highest energy-consuming and environmental-polluting sectors. Great public attention should be paid to the energy efficiency problem of the transportation sector. In this paper, we use the data envelopment analysis method to evaluate the transportation energy efficiency of the Yangtze River Delta region.

Within a joint production framework of undesirable outputs (CO2), energy inputs (transportation energy consumption volume), and non-energy inputs (labor and capital inputs), this study employs a data envelopment analysis based model to evaluate the transportation energy efficiency of 15 cities and three areas of the Yangtze River Delta. In addition, this study applies a DEA window analysis technique to measure the efficiency in cross-sectional and time-varying data from 2009 to 2013. The empirical results show that Shanghai has higher transportation energy efficiency than Zhejiang province, and the efficiency of Jiangsu province is the worst. The efficiencies of all three regions have similar variation trends, and in general, the transportation energy efficiency of the Yangtze River Delta increased from 2009 to 2012. The transportation energy efficiencies of Zhejiang province have a more balanced performance than those of Jiangsu province. It can be concluded that the transportation structure and management measures have important impacts on transportation energy efficiency.

This study has some limitations. First, the efficiency was compared only among the Yangtze River Delta region. The efficiency scores would likely have been changed if the data included other Chinese cities. Furthermore, owed to the lack of data, this paper evaluates the transportation energy efficiency only of these 15 cities in the Yangtze River Delta. The other nine cities of Yangtze River Delta region have not been evaluated. To better study the traffic system, the data of more areas should be collected in the future. All these remain avenues for future research. In future works, the transportation energy efficiency of all Chinese cities could be studied.

Notes

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

References

  1. An, Q., Chen, H., Xiong, B., Wu, J., & Liang, L. (2017a). Target intermediate products setting in a two-stage system with fairness concern. Omega, 73, 49–59.CrossRefGoogle Scholar
  2. An, Q., Wen, Y., Xiong, B., Yang, M., & Chen, X. (2017b). Allocation of carbon dioxide emission permits with the minimum cost for Chinese provinces in big data environment. Journal of Cleaner Production, 142, 886–893.CrossRefGoogle Scholar
  3. An, Q., Meng, F., Xiong B. (2018). Interval cross efficiency for fully ranking decision making units using DEA/AHP approach. Ann. Oper. Res.  https://doi.org/10.1007/s10479-018-2766-6.
  4. Azadi, M., Shabani, A., Khodakarami, M., & Saen, R. F. (2014). Planning in feasible region by two-stage target-setting DEA methods: an application in green supply chain management of public transportation service providers. Transport. Res. E- Log, 70, 324–338.CrossRefGoogle Scholar
  5. Bi, G., Wang, P., Yang, F., & Liang, L. (2014). Energy and environmental efficiency of China’s transportation sector: a multidirectional analysis approach. Mathematical Problems in Engineering, 2014, 1–12.  https://doi.org/10.1155/2014/539596.CrossRefGoogle Scholar
  6. Bian, Y. W., & Yang, F. (2010). Resource and environment efficiency analysis of provinces in China: a DEA approach based on Shannon’s entropy. Energy Policy, 38(4), 1909–1917.CrossRefGoogle Scholar
  7. Blum, H. (2015). The economic efficiency of energy-consuming equipment: a DEA approach. Energy Efficiency, 8(2), 281–298.CrossRefGoogle Scholar
  8. Chang, Y. T., Park, H., & Zou, B. (2016). Passenger facility charge vs. airport improvement program funds: a dynamic network DEA analysis for U.S. airport financing. Transport. Res. E- Log., 88, 76–93.CrossRefGoogle Scholar
  9. Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), 429–444.MathSciNetCrossRefGoogle Scholar
  10. Charnes, A., Clark, C. T., Cooper, W. W., & Golany, B. (1985). A developmental study of data envelopment analysis in measurement the efficiency of maintenance in the US air forces. Ann. Oper. Res., 2(1), 95–112.CrossRefGoogle Scholar
  11. Chen, C., Achtari, G., Majkut, K., & Sheu, J. B. (2017). Balancing equity and cost in rural transportation management with multi-objective utility analysis and data envelopment analysis: a case of Quinte West. Transport. Res. A- Pol., 95, 148–165.Google Scholar
  12. Chu, J. F., Wu, J., & Song, M. L. (2016). An SBM-DEA model with parallel computing design for environmental efficiency evaluation in the big data context: a transportation system application. Ann. Oper. Res., DOI.  https://doi.org/10.1007/s10479-016-2264-7.
  13. Cui, Q., & Li, Y. (2014). The evaluation of transportation energy efficiency: an application of three-stage virtual frontier DEA. Transport. Res. D -Tre, 29, 1–11.CrossRefGoogle Scholar
  14. Cullinane, K., & Song, D. W. (2006). Estimating the relative efficiency of European container ports: a stochastic frontier analysis. Res. Transport. Econ., 16, 85–115.CrossRefGoogle Scholar
  15. Halkos, G. E., & Tzeremes, N. G. (2009). Exploring the existence of Kuznets curve in countries’ environmental efficiency using DEA window analysis. Ecological Economics, 68, 2168–2176.CrossRefGoogle Scholar
  16. Hunjet, D., Neralić, L., & Wendell, R. E. (2015). Evaluation of the dynamic efficiency of Croatian towns using Data Envelopment Analysis. Cent. Eur. J. Oper. Res., 23(3), 675–686.CrossRefGoogle Scholar
  17. Kraft, E., & Tırtıroğlu, D. (1998). Bank efficiency in Croatia: a stochastic-frontier analysis. Journal of Comparative Economics, 26(2), 282–300.CrossRefGoogle Scholar
  18. Liu, X. H., & Wu, J. (2017). Energy and environmental efficiency analysis of China’s regional transportation sectors: a slack-based DEA approach. Energy Systems, 8(4), 747–759.CrossRefGoogle Scholar
  19. Merkert, R., & Assaf, A. G. (2015). Using DEA models to jointly estimate service quality perception and profitability—evidence from international airports. Transport. Res. A -Pol., 75, 42–50.Google Scholar
  20. Park, Y. S., Lim, S. H., Egilmez, G., & Szmerekovsky, J. (2016). Environmental efficiency assessment of US transport sector: a slack-based data envelopment analysis approach. Transport. Res. D. Tre.,  https://doi.org/10.1016/j.trd.2016.09.009.
  21. Řepková, I. (2015). Banking efficiency determinants in the Czech banking sector. Procedia Economic and Finance, 23, 191–196.CrossRefGoogle Scholar
  22. Song, M., Zhang, G. J., Zeng, W. X., Liu, J. H., & Fang, K. N. (2015). Railway transportation and environmental efficiency in China. Transport. Res. D-Tre., 48, 488–498.CrossRefGoogle Scholar
  23. Wang, K., Yu, S. W., & Zhang, W. (2013). China’s regional energy and environmental efficiency: a DEA window analysis based dynamic evaluation. Mathematical and Computer Modelling, 28(5), 1117–1127.CrossRefGoogle Scholar
  24. Wu, C., Li, Y., & Liu, Q. (2013). A stochastic DEA model considering undesirable outputs with weak disposability. Mathematical and Computer Modelling, 58(5), 980–989.MathSciNetCrossRefGoogle Scholar
  25. Wu, J., Yu, Y., Zhu, Q., An, Q., Liang, L. (2017). Closest target for the orientation-free context-dependent DEA under variable returns to scale. The Journal of the Operational Research Society, doi: https://doi.org/10.1080/01605682.2017.1409865.
  26. Yang, H. H., & Chang, C. Y. (2009). Using DEA window analysis to measure efficiencies of Taiwan’s integrated telecommunication firms. Telecommunications Policy, 33(1), 98–108.CrossRefGoogle Scholar
  27. Zaim, O., & Gazel, T. U. (2017). Overcoming the shortcomings of energy intensity index: a directional technology distance function approach. Energy Efficiency,  https://doi.org/10.1007/s12053-017-9589-1.
  28. Žaja, M. M., Banker, R. D., Fang, S., Hunjet, D., Neralić, L., & Wendell, R. E. (2017). Efficiency gains in Croatia’s electricity distribution centers following industry structure changes. Data. Envelop. Anal. J., 3(1–2), 119–150.CrossRefGoogle Scholar
  29. Zhang, X. P., Cheng, X. M., Yuan, J. H., & Gao, X. J. (2011). Total-factor energy efficiency in developing countries. Energy Policy, 39(2), 644–650.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  1. 1.School of BusinessCentral South UniversityChangshaChina
  2. 2.Key Laboratory of Hunan Province for Mobile Business IntelligenceHunan University of CommerceChangshaChina
  3. 3.Industrial Systems Optimization Laboratory, Charles Delaunay Institute and UMR CNRS 6281University of Technology of TroyesTroyesFrance
  4. 4.Faculty of EconomicsUniversity of ZagrebZagrebCroatia

Personalised recommendations