Available online at www.sciencedirect.

com

ScienceDirect
Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

12th International Strategic Management Conference, ISMC 2016, 21-23 July 2016, Podgorica,
Montenegro

Energy efficiency evaluation of provinces in Turkey using

data envelopment analysis
Beyzanur Cayir Ervurala, Bilal Ervuralb, Selim Zaimc, a*
a,b,c
Istanbul Technical University, Industrial Engineering Department
Macka, 34356, Istanbul, Turkey

Abstract

Renewable energy is an indispensable alternative energy source due to the environmental benefits and price advantages over
conventional energy sources. From global to local level in the world, all policy makers and strategic authorities are aware of the
superiorities of renewable energy. Hence, there are enormous studies, action/master plans and researches about significance of
renewable energy have been conducted to meet growing energy demand with green energy sources in a long term. Renewable
energy potential of Turkey provides an advantage due to the climatic conditions. In this study, a nonparametric method and
powerful benchmarking tool, Data Envelopment Analysis (DEA) has been utilized in order to analyze the renewable energy
efficiencies of the each provinces (81 cities) in Turkey. According to the obtained results, investment decisions should be applied
particularly, in inefficient regions to provide overall technical efficiency for future.
©
© 2016
2016TheTheAuthors.
Authors.Published
Publishedbyby
Elsevier Ltd.Ltd.
Elsevier This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of ISMC 2016.
Peer-review under responsibility of the organizing committee of ISMC 2016.
Keywords: Data envelopment analysis; Efficiency analysis; Renewable energy; Strategic investment

1. Introduction

Turkey is one of the fastest growing countries in terms of population, urbanization and industrialization level
according to IEA (International Energy Agency). All these indicators which are important signs of the economic

* Corresponding author. Tel.: +90-212-293-1300 (2759); Fax:+90-212-240-7260.

E-mail address: cayirb@itu.edu.tr

1877-0428 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of ISMC 2016.
doi:10.1016/j.sbspro.2016.11.009
140 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

expansion of countries that leads to increasing energy demand. Under this viewpoint, Turkey has been forced to carry
out energy actions and strategic plans to respond rapid energy demand with domestic sources in a long term. In global
energy market, Turkey’s strategic position and geographic location make it an important energy player. Dependence
on foreign energy sources is a major challenge for governments to sustain energy security and uninterrupted energy
flow with a low cost. The over-dependence on oil and gas causes long term economic problems because energy costs
constitute the largest share of imports (70 %). For this reason, investments on renewable energy resources should be
increased substantially for economic independence and sustainability of energy. The climatic conditions of Turkey
provide several advantages to obtain energy production from renewable energy sources when compared with other
countries. The Turkish government has increased the share of renewable sources 30 percent to meet increasing energy
demand by 2023 according to energy strategies, plans and regulations. To achieve this, energy investments have been
encouraged with incentives and feed-in tariffs. Energy efficiency investments have gained much significance from the
world over the past decades. In the literature, Data envelopment analysis (DEA) is a commonly used benchmarking
tool in order to measure productive efficiency of decision making units (DMU) to show their performances in many
managerial problems.
Data envelopment analysis (DEA) is a nonparametric method which first introduced by Farrell (1957) then
developed by Charnes, Cooper and Rhodes in 1978. DEA is an easy applicable method, since it does not require any
assumptions of the related function and it can handle multiple input and output variables of various units. Due to the
several advantages, it has an extensive application area to analyze efficiency level of the organizations (companies).
In the literature, lots of studies related to DEA have been broadly applied in the energy sector in order to evaluate the
efficiency analysis of renewable energy power generation companies, and efficiency analysis of alternative renewable
energy source technologies. Barros (2008) analyzed the total productivity of the hydroelectric plants using DEA in
order to rank the plants according to their performances. Jha and Shrestha (2006) measured the performance of
hydropower plants during 2001-2004 using DEA in Nepal. They considered the installed capacity, total operations
and maintenance expenditure and the number of employees as inputs and three outputs which are energy generated by
the plant, winter peaking capacity, and summer peaking capacity. Chien and Hu (2007) utilized DEA to compare
renewable energy technology at the country level. They analyzed 45 countries according to technical efficiency and
they have obtained that OECD countries have higher share on renewable energy sources than non-OECD countries.
San Cristobal (2011) used DEA in order to evaluate renewable energy technologies and they have taken as inputs
investment ratio, implement period, and O & M costs and they have considered power, operating hours, useful life,
and the tons of C2O avoided, as outputs. Kim et al. (2015) evaluated the investment efficiency of renewable energy
technologies in Korea with DEA tool and they concluded that wind power is the most efficient renewable energy for
policy makers. Woo et al. (2015) used DEA to measure environmental efficiency of renewable energy with Malmquist
index between 2004 and 2011 years. Menegaki (2013) has taken into consideration energy inefficiencies of European
countries with DEA model conducting panel data set.
According to the literature surveys, there are a few studies about analyzing efficiency of Turkish energy sector by
utilizing DEA. Bagdadioglu et al. (1996) investigated Turkish electricity distribution sector to analyze performances
of the publicly operated organizations utilizing DEA tool. Another study of Bagdadioglu et al. (2007) which analyzed
efficient level of the merger of the companies using DEA model for the period 1999–2003. Ertürk and Aşık (2011)
has evaluated the natural gas distribution companies by using a non-parametric method, Data Envelopment Analysis
(DEA). Sözen et al. (2012) used DEA for efficiency analyses of hydro-power plants by implementing two various
models which consist of multiple inputs and outputs to assess their relative performances. According to analyses,
Gökçekaya hydro-power plant has the highest performance in both models.
In the literature, the most widely used inputs are the number of employees, transformer capacity and network length
(Azadeh, Ghaderi, Omrani, & Eivazy, 2009; Yadav, Padhy, & Gupta, 2010). The most commonly used outputs are
units of energy delivered, number of customers and size of service area in all efficiency analysis studies concerning
distribution companies (Lins, Sollero, Calôba, & da Silva, 2007; Pombo & Taborda, 2006). Different from other
studies in the literature, we have considered the renewable energy potentials of the regions as input and the gross
energy generation from renewable sources is taken as an output variable, except for other variables, which are network
length (input), total installed power of renewable energy (input), transformer capacity(input), number of consumer
(output), respectively. The main objective is to measure of renewable energy efficiency of the regions according to
renewable energy potentials.
 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148 141

To meet the increasing energy demand, the most efficient way is to provide more investment in renewable energy
resources/technologies due to abundance of renewable sources, including solar energy, wind power, geothermal
energy, and biomass. In this study, the utilization rate of the renewable energy potentials of Turkey’s regions/province
will be analyzed using DEA model. According to the obtained results, strategically important investment decisions in
the regions will be evaluated under energy efficiency concept. Thus, energy resource planning activities can be
implemented realistically to satisfy all provinces/regions’ conditions regarding efficiency frontier analysis.
This study is organized in four main steps. Section 2, explains data envelopment analysis model. Section 3, provides
the analysis results of energy efficiencies of the regions, finally some concluding remarks and future research
directions are given at the last section.

2. DEA Model

DEA is a prominent method for evaluating the productivity of organizations with multiple incomparable inputs and
outputs. DEA was first proposed by Charnes et al. (1978) based on the seminal work of Farrell (1957) on the
measurement of productive efficiency. The objective function in that model was to maximize the ratio of weighted
outputs to weighted inputs for a particular organizational unit (or in DEA terminology, decision making units, DMUs).
DEA is a linear programming based technique for measuring the relative efficiency of DMUs which has gained
considerable interest in recent years due to its advantages over traditional methods (Demirbag, Tatoglu, & Glaister,
2009). DEA provides a peer group comparison using a frontier in order to determine efficient and inefficient units.
DEA can incorporate multiple inputs and multiple outputs which can be expressed in different units of measurement.
Another advantage of DEA that gained more interest from researchers and analysts is its ability to identify the
potential improvement for inefficient units. DEA compares the unit with a convex combination of units located on the
frontier and enables the analyst to indicate the sources and the level of inefficiency for each of its inputs and outputs
(Sevkli, Koh, Zaim, Demirbag, & Tatoglu, 2007).
DEA has two common models which are CCR (Charnes et al., 1978) and BCC (Banker, Charnes, & Cooper, 1984)
models. CCR models consider the overall efficiency and assume constant returns to scale. BCC models provides more
details about the model which thinks pure technical efficiency and assume variable returns to scale. A standard
formulation of the CCR model can be expressed as:

σ೘
ೕసభ ௕ೕ೚ ௬ೕ೚
‫ܼݔܽܯ‬௢ ൌ σ೙
(1)
೔సభ ௔೔೚ ௫೔೚

subject to

σ೘
ೕసభ ௕ೕ೓ ௬ೕ೓
σ೙
൑ ͳ݂‫ ݄ݎ݋‬ൌ ͳǡʹǡ ǥ ǡ ݇ (2)
೔సభ ௔೔೓ ௫೔೓

ܽ௜௛ ǡ ܾ௝௛ ൒ Ͳ݂‫ ݅ݎ݋‬ൌ ͳǡʹǡ ǥ ǡ ݊Ǣ ݆ ൌ ͳǡʹǡ ǥ ǡ ݉Ǣ ݄ ൌ ͳǡʹǡ ǥ ǡ ݇ (3)

where
Zo efficiency score of oth organization
xih observed value of input i for the organization h
yih observed value of output i for the organization h
aih weight attached to input i of organization h
bjh weight attached to output j of organization h
k number of organizations
m number of output variables
n number of input variables

Converting this to standard linear form, value of outputs / value of inputs ≤ 1 implies value of outputs - value of
inputs ≤ 0
142 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

‫ܼݔܽܯ‬௢ ൌ σ௠
௝ୀଵ ܾ௝௢ ‫ݕ‬௝௢ (4)

subject to

σ௡௜ୀଵ ܽ௜௢ ‫ݔ‬௜௢ ൌ ͳ

σ௠ ௡
௝ୀଵ ܾ௝௛ ‫ݕ‬௝௛ െ σ௜ୀଵ ܽ௜௛ ‫ݔ‬௜௛ ൑ Ͳ݂‫ ݄ݎ݋‬ൌ ͳǡʹǡ ǥ ǡ ݇ (5)

ܽ௜௛ ǡ ܾ௝௛ ൒ Ͳ݂‫ ݅ݎ݋‬ൌ ͳǡʹǡ ǥ ǡ ݊Ǣ ݆ ൌ ͳǡʹǡ ǥ ǡ ݄݉ܽ݊݀ ൌ ͳǡʹǡ ǥ ǡ ݇ (6)

3. Methodology

Turkey has 780,576 km2 total area and divided into 81 provinces. Turkey is also subdivided into 7 regions, which
is given in Figure 1, and 21 sub regions for geographic, demographic and economic purposes.

Fig. 1. Regions of Turkey

This paper utilizes a data set of 81 regions (provinces) renewable energy potentials, network length, total installed
power of renewable energy, transformer capacity, gross energy generation from renewable energy and total number
of consumers knowledge which are utilized from the electricity service system. All these data have been collected
from energy sector databases, private electricity companies and Turkish Electricity Distribution Corporation
(TEDAS), then the data analyzed in terms of their overall technical, pure and scale efficiency. The descriptive
statistics of the variables adopted in the analysis are given in Table 1. The correlation between input and output
variables is statistically verified which is given in Table 2. Short definitions of the input and output variables used in
the analysis are as follows:

X1 Total Renewable Energy Potential-All the renewable energy resources have been analyzed according to climatic
conditions of the provinces and the abundance of sources in the regions. The intensity of wind, solar, hydro, geo
thermal and biomass energy sources in the regions are identified and the total renewable energy potential has been
determined as a rate by using multi criteria decision making tools. Wind speed, sunshine duration, river regimes,
percentages of the biomass waste and thermal power of locations are considered for the evaluation of overall potential
rate while the renewable energy potentials are assessed.
X2 Network Length- Distribution line measured in (km) per province.
X3 Total Installed Power of Renewable Energy- Total installed power means the theoretical capacity of the
generators if working at maximum output (MW).
X4 Transformer Capacity- The maximum capacity of transformers which is loaded electricity in order to connect
to the distribution system (MVA).
Y1 Gross Energy Generation from Renewable Sources- The energy production obtained from the renewable energy
power plants throughout the year (MW).
Y2 Number of Consumer- Measured as the total number customer of the delivered electricity service.
 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148 143

Table 1. Descriptive statistic of variables of the regions

Gross Energy
Total Installed Transformer Generation from
Total Renewable Network Power of Renew. Capacity Renewable Number of
Energy Potential Length(km) Energy(MW) (MVA) Sources(MW) Consumer
Max 0.0721 67642 3128.25 21546 3128000 542435
Min 0.0004 1783 0 103 0 813
Average 0.01234 13381.950 390.3830 1660.777 280950.991 25154.283
SD 0.01206 11134.211 558.9335 2791.645 471181.996 67713.269

The number of DMU’s should be at least twice the number of inputs and outputs (Sarkis, 2007). In this study the
number of DMU is eighty one, which is higher than the number of input and output variables.

Table 2. Correlation coefficient between input and output variables

Energy Generation
from Renewable Number of
Sources (MW) (Y1) Consumer(Y2)

Total Renewable Energy

0.82015 0.66064
Potential (X1)
Network Length(km) (X2) 0.6126 0.7876
Total Installed Power of
0.9478 0.6612
Renew. Energy(MW) (X3)
Transf. Capacity (MVA) (X4) 0.6898 0.8520

All of correlation coefficients between input and output items exhibit positive correlation, as shown in Table 2.
Thus, the input and output items are well complies with the prerequisite condition of the DEA model.

We used equation (4-6) to derive the efficiency index for each of the firms and the reference sets in which all
DMUs’ efficiency equals to one (Charnes et al., 1978). A total of 11 provinces appeared in the reference set as efficient
units, as shown in Table 3. Additionally, the total count of occurrence is obtained as 212 in DEA analysis.

Table 3. The count of occurrence of efficient provinces in the reference sets

Count of
Province
occurrence
BATMAN 38
ELAZIG 1
GAZIANTEP 30
GUMUSHANE 34
IGDIR 46
IZMIR 16
KAHRAMANMARAS 16
KIRIKKALE 5
KUTAHYA 11
SANLIURFA 4
YALOVA 11
Total 212
144 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

Norman and Stoker (1991) classified the DMUs into four categories according to the efficiency levels:
(1) The robustly efficient units. The DMU is efficient and also it is appeared on many reference sets as their
benchmarks.
(2) The marginal efficient units. The DMU is efficient but it is not other DMU’s benchmark in that the DMU may
has different characteristics from others in the same category.
(3) The marginal inefficient units. These will have an efficiency rating, between 0.9 and 1, and if they adjust their
inputs or outputs, they could soon raise their score towards 1.0
(4) The distinctly inefficient units. With an efficiency score of less than 0.9, these units would have difficulty in
making themselves efficient in the short term.
Based on the above criteria, we then regroup the total sample of 81 provinces into four categories as noted above.
Table 4 shows the number and percentage of cities for the different investment origins.
As it is apparent from Table 4, 13.5 percent of provinces belong to the robustly efficient unit, while only 3.7 percent
of the DMUs fit in a marginal inefficient units. Also nearly 82.71 percent of the provinces are classified into the
distinctly inefficient unit. These results imply that most of the provinces are inefficient when we compare to the
efficient units in Turkey.

Table 4. Number and percentage of provinces for each efficiency category

All Provinces
Number %
Robustly efficient units 11 13.5
Marginal efficient units 0 0
Marginal inefficient units 3 3.7
Distinctly inefficient units 67 82.71
Total 81 100

IRS means that increase in input will result in a greater than proportionate increase in output, whereas DRS is the
case where the result is less than the proportionate increase in output. CRS is exhibited where the result is the
proportionate increase in output (Lin, Madu, Kuei, & Lu, 2004). Table 5 shows that there are 31 DMUs in the condition
of increasing return scale (IRS), and there are 14 DMUs in the condition of decreasing return scale (DRS), and there
are 36 DMUs in the condition of constant return scale (CRS).

Table 5. Categories of scale returns for each group of provinces

Provinces
Number Percent
IRS 31 38.27
DRS 14 17.28
CRS 36 44.44
Total 81 100

According to Table 6,
i. The Central Anatolia Region has the highest percentage (6.17 per cent) in the DRS category,
ii. The percentage of regions classified into CRS, The Black Sea Region is greater than others, and
iii. The percentage of firms in the IRS category, The Black Sea Region and The Eastern Anatolia Region
(9.88 per cent) are similar and greater than remained regions.

The implication drawn from item (i) means that The Central Anatolia Region expend more input to obtain the same
level of output compared to the other regions. Our results from item (ii) indicate that The Black Sea Region tend to
have higher efficiency than other regions. Item (iii) also indicates that The Black Sea Region and The Eastern Anatolia
 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148 145

Region tend to perceive the need to create more fit by enhancing their outputs, renewable energy generation and
number of customers, as compared to other regions.

Table 6. Categories of scale returns for each region of Turkey

Central Eastern Mediterrane Southeastern Total

Marmara Black Sea Aegean
Anatolia Anatolia an Anatolia Percent

IRS 4.93 9.88 2.47 6.17 9.88 1.23 3.7 38.27

DRS 2.46 1.23 2.47 6.17 0 2.47 2.47 17.28
CRS 7.41 9.88 4.94 3.7 6.17 6.17 6.17 44.44

We have applied output oriented model, which aims to maximize outputs while the inputs proportions remain
unchanged to realize DEA efficiency analysis using DEA Solver software. CCR and BCC models are utilized to get
overall technical and pure technical efficiencies. According to Table 7, if the efficiency value equals 1, the DMU is
efficient; if it is less than 1, the evaluated region is inefficient. When we look at the overall efficiency scores in CCR
model, 11 cities are efficient (100 %). Besides that, pure efficiency scores are obtained from BCC model which gives
information about 16 cities are efficient (100 %). These cities uses their inputs efficiently. Scale efficiency score is
obtained with using the following formulation, CCR/BCC. 11 regions have the scale efficiency. It means these
locations have optimal scale size.
In more details, we can express these findings: There are 11 cities with the overall technical efficiency value of 1
among the 81 provinces, which means all their pure technical efficiency values and scale efficiency values are 1,
indicating that the resource utilization of such regions, whether in technique or scale, reaches the fittest. These point
that these DMU’s are in the stage of constant returns to scale (CRTS), an optimal status for the combination of input
factors and production scale and there is no need for any improvement (Lee, 2009). Also, 31 provinces are in the stage
of increasing returns to scale (IRTS) among 81 regions. If they can expand their production scale, they should be able
to improve the overall operational efficiency. By the way, 14 provinces are in the stage of decreasing returns to scale
(DRTS) among 81 regions. They should decrease their inputs and production scale in order to improve their overall
operational efficiency. In brief, 36 regions (i.e. 44.44% of the total provinces) are in the stage of CRTS; 31 regions
(i.e. 38.27% of the total provinces) are in the stage of IRTS; 14 of them (i.e. 17.28 % of the total provinces) are in the
stage of DRTS.
According to ranks of the regions, Yalova, Gaziantep, Elazig, Sanliurfa, Kutahya, Kirikkale, K.maras, Izmir, Igdir,
Gumushane, and Batman are efficient cities in terms of the utilizing renewable energy potential. There is no need for
any improvement since the proportionate increase in input is exactly equal to the increase in output. After that Giresun,
Karabuk, Sinop, Diyarbakir, Siirt and other provinces are coming as top ranks in terms of utilizing domestic renewable
energy sources.
Agri, Amasya, Ardahan, Artvin, Bartın, Bayburt, Bilecik, Bingol, Bitlis, Cankiri, Duzce, Erzincan, Eskisehir,
Hakkari, Isparta, Karabuk, Karaman, Kars, Kırklareli, Kırsehir, Kilis, Mugla, Mus, Siirt, Ordu, Tekirdag, Trabzon,
Tunceli, Usak, Yozgat, Zonguldak are in the stage of increasing returns to scale. If they expand their scales, they can
improve the overall efficiency. In this respect, some investment studies on the network length, transformer capacity,
and installed power of renewable energy should be increased in order to catch the efficiency in the related regions.
Afyonkarahisar, Aksaray, Ankara, Antalya, Denizli, Diyarbakir, Giresun, Istanbul, Konya, Mardin, Nevsehir,
Nigde, Osmaniye, Sakarya are in the stage of decreasing returns to scale which occurs when the proportion of output
is less than the desired increased input during the production process. At this point, number of consumer and energy
generation from renewable energy outputs should be increased to improve efficiency frontier.
146 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

Table 7. Output oriented DEA model results

Overall Pure Overall Pure
Scale Returns to Scale Returns to
DMU Efficiency Efficiency DMU Efficiency Efficiency
Efficiency Rank Scale Efficiency Rank Scale
(CCR) (BCC) (CCR) (BCC)
ADANA 0.16582 0.23014 0.72051 67 Constant K.MARAS 1.00000 1.00000 1.00000 1 Constant
ADIYAMAN 0.55058 0.55232 0.99685 36 Constant KARABUK 0.93289 1.00000 0.93289 13 Increasing
AFYON 0.23291 0.26089 0.89276 64 Decreasing KARAMAN 0.77539 0.78249 0.99093 20 Increasing
AGRI 0.44801 0.99988 0.44806 42 Increasing KARS 0.63211 0.66266 0.95390 29 Increasing
AKSARAY 0.08540 0.11083 0.77057 79 Decreasing KASTAMON 0.16462 0.17180 0.95822 68 Constant
AMASYA 0.57485 0.58183 0.98801 33 Increasing KAYSERI 0.29639 0.33561 0.88312 58 Constant
ANKARA 0.39218 0.49509 0.79214 49 Decreasing KILIS 0.83904 1.00000 0.83904 17 Increasing
ANTALYA 0.63347 0.75754 0.83623 28 Decreasing KIRIKKALE 1.00000 1.00000 1.00000 1 Constant
ARDAHAN 0.15288 0.99990 0.15289 71 Increasing KIRKLARELI 0.03630 0.04110 0.88321 81 Increasing
ARTVIN 0.12654 0.12721 0.99473 76 Increasing KIRSEHIR 0.75808 0.76639 0.98916 21 Increasing
AYDIN 0.49006 0.53765 0.91149 40 Constant KOCAELI 0.56399 0.56858 0.99193 35 Constant
BALIKESIR 0.16352 0.23382 0.69936 69 Constant KONYA 0.13961 0.40093 0.34822 72 Decreasing
BARTIN 0.15395 0.15794 0.97475 70 Increasing KUTAHYA 1.00000 1.00000 1.00000 1 Constant
BATMAN 1.00000 1.00000 1.00000 1 Constant MALATYA 0.29746 0.30449 0.97693 57 Constant
BAYBURT 0.42085 1.00000 0.42085 46 Increasing MANISA 0.06450 0.06570 0.98174 80 Constant
BILECIK 0.08713 0.11070 0.78712 78 Increasing MARDIN 0.50750 0.71074 0.71405 39 Decreasing
BINGOL 0.83260 0.97121 0.85728 18 Increasing MERSIN 0.45358 0.52967 0.85634 41 Constant
BITLIS 0.23825 0.28493 0.83617 63 Increasing MUGLA 0.60071 0.60190 0.99803 31 Increasing
BOLU 0.38051 0.38085 0.99911 51 Constant MUS 0.69009 0.73376 0.94048 25 Increasing
BURDUR 0.38701 0.38735 0.99912 50 Constant NEVSEHIR 0.40439 0.41195 0.98166 47 Decreasing
BURSA 0.20741 0.24481 0.84724 65 Constant NIGDE 0.13496 0.14047 0.96079 74 Decreasing
CANAKKALE 0.13588 0.18541 0.73284 73 Constant ORDU 0.70765 0.71560 0.98889 24 Increasing
CANKIRI 0.25925 0.28800 0.90016 60 Increasing OSMANIYE 0.63490 0.64549 0.98360 27 Decreasing
CORUM 0.42159 0.42966 0.98123 45 Constant RIZE 0.68287 0.68577 0.99576 26 Constant
DENIZLI 0.54317 0.59132 0.91857 38 Decreasing SAKARYA 0.32553 0.39674 0.82052 55 Decreasing
DIYARBAKIR 0.85614 1.00000 0.85614 15 Decreasing SAMSUN 0.37242 0.48047 0.77512 53 Constant
DUZCE 0.44402 0.48785 0.91015 43 Increasing SANLIURFA 1.00000 1.00000 1.00000 1 Constant
EDIRNE 0.11699 0.11941 0.97977 77 Constant SIIRT 0.84979 0.88677 0.95829 16 Increasing
ELAZIG 1.00000 1.00000 1.00000 1 Constant SINOP 0.90109 0.90493 0.99575 14 Constant
ERZINCAN 0.36912 0.37051 0.99624 54 Increasing SIRNAK 0.72267 0.79964 0.90375 22 Constant
ERZURUM 0.62072 0.69822 0.88901 30 Constant SIVAS 0.59131 0.61554 0.96064 32 Constant
ESKISEHIR 0.42360 0.42709 0.99183 44 Increasing TEKIRDAG 0.13186 0.13187 0.99992 75 Increasing
GAZIANTEP 1.00000 1.00000 1.00000 1 Constant TOKAT 0.57351 0.58395 0.98212 34 Constant
GIRESUN 0.93482 0.98674 0.94739 12 Decreasing TRABZON 0.79912 0.80118 0.99744 19 Increasing
GUMUSHAN 1.00000 1.00000 1.00000 1 Constant TUNCELI 0.71098 1.00000 0.71098 23 Increasing
HAKKARI 0.37410 0.47444 0.78851 52 Increasing USAK 0.30977 0.31915 0.97062 56 Increasing
HATAY 0.39923 0.40148 0.99441 48 Constant VAN 0.25779 0.27868 0.92504 61 Constant
IGDIR 1.00000 1.00000 1.00000 1 Constant YALOVA 1.00000 1.00000 1.00000 1 Constant
ISPARTA 0.25647 0.25888 0.99069 62 Increasing YOZGAT 0.26433 0.27131 0.97428 59 Increasing
ISTANBUL 0.19086 0.31570 0.60456 66 Decreasing ZONGULDA 0.54519 0.59644 0.91407 37 Increasing
IZMIR 1.00000 1.00000 1.00000 1 Constant
 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148 147

Table 8. The average of input excesses and output deficits

Average Improvement
Input Factors Rank
Potential
Total Renewable Energy Potential 0.00074 4
Network Length 1082.61616 1
Total Installed Power of Renew. Energy 12.20963 3
Transformer Capacity 75.53578 2
Overall 1170.3623 1
Average 292.5906 1
Average Improvement
Output Factors Rank
Potential
Energy Generation from Renewable
2787.5071 1
Sources
Number of Consumer 285.6490 2
Overall 3073.1560 1
Average 1536.5780 1

We derived the input excesses and the output deficits individually for every inefficient province. The results of
averaging the input excesses for each input factor and the averaging output deficits for output factor are summarized
in Table 8. It is observed that network length and transformer capacity rank very high in input excesses. For example,
Table 8 shows that energy policy makers or analysts tend to put too much effort on managing network length as a way
of upgrading their efficiency. Thus, we can conclude that the provinces in Turkey, are spending more resources and
efforts on managing network length that may influence the efficiency of their renewable energy management practices.
In terms of the overall average of output deficits, energy generation from renewable sources and number of consumer
have significantly high scores and hence, they are the critical reasons for inefficiency.
Additionally, according to improvable spaces of input and output items of DMUs, Kirklareli with the worst overall
technical efficiency (0.0363), for instance, the improvable spaces of this DMU’s input item is (0.0001), and the
improvable spaces of its output item is (4102.5914). That is, to enable such DMU to utilize resources so efficiently as
other DMUs with the overall technical efficiency values of 1 do, its “total renewable energy potential” has to decrease
by 0.0001 percentage, and “energy generation from renewable sources” has to increase by 4102.5914 MW; after these
adjustments, Kirklareli will become a relatively efficient province.

4. Conclusion

Energy is the most important input for countries’ economy, and it considerably helps to shape international relations
and sustainable development strategies all over the world. Governments give top priority to energy policies for their
development goals. Hence, it is important to evaluate the potential capacity of local renewable energies. Establishing
renewable energy plants plays an important role in economy, environmental sensitiveness and social status of citizens.
This study will shift Turkey’s energy policies to renewable and sustainable energy sources which are sensitive to
environment. Increase in local energy resources will aid economic progress of Turkey.
In this study, DEA is applied on each provinces with multiple inputs and outputs located in Turkey in order to
assess their efficiencies. In the analysis, scale efficiency values are obtained by CCR and BCC models. According to
the obtained results, Yalova, Sanliurfa, Kutahya, Kirikkale, K.maras, Izmir, Igdir, Gumushane, Gaziantep, Elazig and
Batman demonstrated the best performance among all provinces in the model. The Black Sea and Marmara regions
are respectively more efficient among the regions. The main reason of the efficiency of these regions is having high
renewable energy potential and utilization of the potential efficiently depend on regional technological progress. In
recent years, renewable energy investments, capacity expansion decisions and system integration activities have been
expanded particularly in these regions. All these factors raises energy efficiencies stage by stage.
Energy policy makers and investors can constitute an action plan to increase renewable energy utilization rate at a
high level according to analysis results. With this study, all authorities in energy sector can take into account current
148 Beyzanur Cayir Ervural et al. / Procedia - Social and Behavioral Sciences 235 (2016) 139 – 148

conditions of the regions/provinces which have renewable energy potential at what level. All these works provide our
country to reduce dependence on foreign energy sources and to maintain sustainable development.
For future researches, this study will provide a scientific basis for creating strategic energy decisions and increasing
the renewable energy investments. For future directions, this study will be analyzed adding new inputs/outputs and
the results will be compared with Super-efficiency DEA.

References

Azadeh, A., Ghaderi, S. F., Omrani, H., & Eivazy, H. (2009). An integrated DEA–COLS–SFA algorithm for optimization and policy making of
electricity distribution units. Energy Policy, 37(7), 2605–2618.
Bagdadioglu, N., Price, C. W., & Weyman-Jones, T. (2007). Measuring potential gains from mergers among electricity distribution companies in
Turkey using a non-parametric model. The Energy Journal, 83–110.
Bagdadioglu, N., Waddams Price, C. M., & Weyman-Jones, T. G. (1996). Efficiency and ownership in electricity distribution: A non-parametric
model of the Turkish experience. Energy Economics, 18(1–2), 1–23. http://doi.org/10.1016/0140-9883(95)00042-9
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment
Analysis. Management Science, 30(9), 1078–1092. http://doi.org/10.1287/mnsc.30.9.1078
Barros, C. P. (2008). Efficiency analysis of hydroelectric generating plants: A case study for Portugal. Energy Economics, 30(1), 59–75.
http://doi.org/10.1016/j.eneco.2006.10.008
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research,
2(6), 429–444. http://doi.org/10.1016/0377-2217(78)90138-8
Chien, T., & Hu, J.-L. (2007). Renewable energy and macroeconomic efficiency of OECD and non-OECD economies. Energy Policy, 35(7), 3606–
3615. http://doi.org/10.1016/j.enpol.2006.12.033
Demirbag, M., Tatoglu, E., & Glaister, K. W. (2009). Equity-based entry modes of emerging country multinationals: Lessons from Turkey. Journal
of World Business, 44(4), 445–462. http://doi.org/10.1016/j.jwb.2008.11.009
Ertürk, M., & Türüt-Aşık, S. (2011). Efficiency analysis of Turkish natural gas distribution companies by using data envelopment analysis method.
Energy Policy, 39(3), 1426–1438. http://doi.org/10.1016/j.enpol.2010.12.014
Farrell, M. J. (1957). The Measurement of Productive Efficiency. Journal of the Royal Statistical Society. Series A (General), 120(3), 253–290.
http://doi.org/10.2307/2343100
Jha, D. K., & Shrestha, R. (2006). Measuring Efficiency of Hydropower Plants in Nepal Using Data Envelopment Analysis. IEEE Transactions on
Power Systems, 21(4), 1502–1511. http://doi.org/10.1109/TPWRS.2006.881152
Kim, K.-T., Lee, D. J., Park, S.-J., Zhang, Y., & Sultanov, A. (2015). Measuring the efficiency of the investment for renewable energy in Korea
using data envelopment analysis. Renewable and Sustainable Energy Reviews, 47, 694–702. http://doi.org/10.1016/j.rser.2015.03.034
Lee, C.-C. (2009). Analysis of overall technical efficiency, pure technical efficiency and scale efficiency in the medium-sized audit firms. Expert
Systems with Applications, 36(8), 11156–11171. http://doi.org/10.1016/j.eswa.2009.02.092
Lin, C., Madu, C. N., Kuei, C., & Lu, M. H. (2004). The relative efficiency of quality management practices: A comparison study on AmericanǦ,
JapaneseǦ, and TaiwaneseǦowned firms in Taiwan. International Journal of Quality & Reliability Management, 21(5), 564–577.
http://doi.org/10.1108/02656710410536572
Lins, M. P. E., Sollero, M. K. V., Calôba, G. M., & da Silva, A. C. M. (2007). Integrating the regulatory and utility firm perspectives, when
measuring the efficiency of electricity distribution. European Journal of Operational Research, 181(3), 1413–1424.
http://doi.org/10.1016/j.ejor.2005.10.072
Menegaki, A. N. (2013). Growth and renewable energy in Europe: Benchmarking with data envelopment analysis. Renewable Energy, 60, 363–
369. http://doi.org/10.1016/j.renene.2013.05.042
Norman, M., & Stoker, B. (1991). Data envelopment analysis: the assessment of performance. Chichester ; New York: Wiley.
Pombo, C., & Taborda, R. (2006). Performance and efficiency in Colombia’s power distribution system: Effects of the 1994 reform. Energy
Economics, 28(3), 339–369. http://doi.org/10.1016/j.eneco.2005.08.001
San Cristóbal, J. R. (2011). A multi criteria data envelopment analysis model to evaluate the efficiency of the Renewable Energy technologies.
Renewable Energy, 36(10), 2742–2746. http://doi.org/10.1016/j.renene.2011.03.008
Sarkis, J. (2007). Preparing Your Data for DEA. In J. Zhu & W. D. Cook (Eds.), Modeling Data Irregularities and Structural Complexities in Data
Envelopment Analysis (pp. 305–320). Boston, MA: Springer US. Retrieved from http://link.springer.com/10.1007/978-0-387-71607-7_17
Sevkli, M., Koh, S. C. L., Zaim, S., Demirbag, M., & Tatoglu, E. (2007). An application of data envelopment analytic hierarchy process for supplier
selection: a case study of BEKO in Turkey. International Journal of Production Research, 45(9), 1973–2003.
http://doi.org/10.1080/00207540600957399
Sözen, A., Alp, İ., & Kilinc, C. (2012). Efficiency assessment of the hydro-power plants in Turkey by using Data Envelopment Analysis. Renewable
Energy, 46, 192–202. http://doi.org/10.1016/j.renene.2012.03.021
Woo, C., Chung, Y., Chun, D., Seo, H., & Hong, S. (2015). The static and dynamic environmental efficiency of renewable energy: A Malmquist
index analysis of OECD countries. Renewable and Sustainable Energy Reviews, 47, 367–376. http://doi.org/10.1016/j.rser.2015.03.070
Yadav, V. K., Padhy, N. P., & Gupta, H. O. (2010). A micro level study of an Indian electric utility for efficiency enhancement. Energy, 35(10),
4053–4063. http://doi.org/10.1016/j.energy.2010.06.011