Measuring the efficiency of renewable energy investment in Korea using data envelopment analysis

Measuring the efficiency of the investment for renewable energy
in Korea using data envelopment analysis
Kyung-Taek Kim, Deok Joo Lee n
, Sung-Joon Park, Yanshuai Zhang, Azamat Sultanov
Department of Industrial & Management Systems Engineering, Kyung Hee University, Yongin-Si 446-701, Gyeonggi-Do, Republic of Korea
a r t i c l e i n f o
Article history:
Received 9 July 2014
Received in revised form
23 October 2014
Accepted 8 March 2015
Available online 30 March 2015
Keywords:
Efficiency
Renewable energy
Investment
Data envelopment analysis
a b s t r a c t
New and renewable energy (NRE) has been paid much attention as a core alternative energy that can
respond to the depletion of fossil fuel, the global movement to address climate change, and recent high oil
prices because it is more environment-friendly and sustainable than fossil fuel. As the scale of investment in
NRE has increased, an intriguing issue of the efficiency of the investment has been raised since strategic
selection and focused investment allows policy goals to be achieved with limited resources and budget.
Particularly, since there are various kinds of renewable energy sources, the efficiency of each NRE technology
must be examined to find suitable technologies for the environments of each target country and to
eventually realize efficient investments in NRE. The purpose of this paper is to evaluate the investment
efficiency of three NRE technologies – wind power, photovoltaic, and fuel cells – with the DEA (data
envelopment analysis) method considering the two policy objectives of public investment, technological
development and wider dissemination of NRE in Korea. The results indicate that wind power is the most
efficient renewable energy in Korea from the perspective of government investment.
& 2015 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694
2. Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
3. Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
3.1. DEA model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
3.2. System model of investment in NRE of Korea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
4. Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
4.1. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 697
4.2. Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
4.2.1. Results of efficiency analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
4.2.2. Efficiency improvement projection analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
1. Introduction
New and renewable energy (NRE) is a core alternative energy
that can respond to the depletion of fossil fuel, the global move-
ment of climate change, and recent high oil prices because it is
more environment-friendly and sustainable than fossil fuel.
Despite economic depression after financial crisis in 2008, the
global scale of investment in NRE reached an all-time high of $279
billion in 2011 then slightly decreased in 2012 to $244 billion,
which is nonetheless the second highest NRE investment to date.
Regionally, the largest investor country in the world is China,
having invested $66.6 billion in 2012; the US followed with
$36 billion, while the EU as a whole invested $79.9 billion [1].
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/rser
Renewable and Sustainable Energy Reviews
http://dx.doi.org/10.1016/j.rser.2015.03.034
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
n
Corresponding author. Tel.: þ82 31 201 2911; fax: þ82 31 202 8854.
E-mail address: ldj@khu.ac.kr (D.J. Lee).
Renewable and Sustainable Energy Reviews 47 (2015) 694–702

This indicates that most advanced countries increase their invest-
ments in NRE based on the long-term perspectives of not only
energy securities, but also sustainable economic development. In
particular, for countries who have been highly dependent upon
energy imports, such as Korea, it is imperative to promote the
development and the utilization of NRE in order to strengthen
energy self-reliance.
Many countries, including Korea, have presented their own
national strategic plans for the development and dissemination of
NRE technology. The Korean government has been establishing
national plans for NRE with a 10-year planning term, and the third
plan is now in progress. The first such basic plan (1997–2006)
focused on the technological development of NRE, and the second
basic plan for NRE technology development, utilization, and
dissemination (2003–2012) covered an efficient dissemination
strategy as well as technology development. In the third plan
(2009–2030), the Korean government established the target pene-
tration rate of NRE among primary energy sources as 11% by 2030.
The third plan emphasizes policy integration between technolo-
gical development and dissemination of NRE in order to resolve
the problems that arose in the execution of prior plans such as
budget deficiency and the overemphasis on dissemination-related
policy. Public investment in NRE of Korea was 854.6 billion KRW1
in 2010, 7.5 times larger than in 2003, and NRE-related invest-
ments are expected to increase continuously.
As the scale of investment is increasing, an intriguing issue of the
efficient allocation of the public funds in NRE has been raised [2]
since strategic selection and focused investment allows the policy
goals of NRE to be achieved with limited resources and budget. In
the case of NRE, since there are various kinds of energy sources such
as wind, solar, biomass, fuel cells, etc., the efficiency of each NRE
technology must be examined to choose suitable technology for the
particular environments of each target country and to eventually
realize efficient investment. The efficiency analysis should be
performed in both systematic and scientific manners in order to
produce meaningful results which provide policy makers with
critical decision-making information regarding efficient resource
allocation for NRE. However, there is no previous study in which a
national NRE investment system is systematically defined and the
investment efficiency analyzed using a quantitative methodology
like multiple criteria decision analysis (MCDA) or data envelopment
analysis (DEA).
In general, NRE policy is composed of technology development
and market dissemination; therefore, investments are also divided
into those two areas. Hence, whether or not a country’s NRE policy
is successful should be judged on the basis of both aspects. This
implies that, when the efficiency of NRE-related investment is
evaluated, the NRE technology development and their dissemina-
tion in energy market should be considered together in a single
investment system model [3]. However, the studies conducted
thus far on the efficiency of NRE-related investments dealt with
the technology development and market dissemination as two
separate entities [4–10].
The purpose of this paper is to evaluate (with the DEA method)
the investment efficiency of three NRE technologies – wind power,
photovoltaic, and fuel cells – considering the two policy objectives
of public investment, technological development and wider dis-
semination of NRE in Korea. We present a system model of public
investments in NRE that includes relevant elements of both
technology development and dissemination policies. Based on this
system model, this paper analyzes the efficiencies of public
investments for NRE technologies by using DEA to measure the
efficiencies of each NRE technology using the available empirical
data of Korea. The remainder of paper is organized as follows:
Section 2 provides a brief review of relevant literature. Section 3
explains the investment system model on which the subsequent
efficiency analysis is based, and introduces the DEA model used in
this paper. Section 4 presents the data and the results of the
analysis. Finally, Section 5 concludes this study with a discussion
of policy implications.
2. Literature review
DEA is a non-parametric methodology for quantitatively analyz-
ing efficiency, particularly the relative efficiencies of a set of
comparable entities, often called decision-making units (DMUs). In
general, each DMU performs the same function by transforming
multiple inputs into multiple outputs, which are characterized by a
system model. DEA has been extensively applied to empirical studies
of efficiency analysis due to its advantages; it does not require any
prior assumptions about the underlying functional relationships
between inputs and outputs, and it is applicable to systems with
various kinds of input or output variables of different units. Since the
late 1980s, DEA has been accepted as a major frontier technique for
benchmarking energy sectors in many countries [11], and a consider-
able set of literature has been produced.
Beginning in the middle of 2000, studies have applied DEA to
the efficiency analysis of NRE areas. Those studies can be categor-
ized into two themes: efficiency analysis of NRE power generation
companies (or facilities), and efficiency analysis of alternative NRE
sources or technologies. Among the former category, Barros [12]
estimated the changes in total productivity for the period 2001–
2004, breaking this down into technically efficient change and
technological change by applying DEA to one of the renewable
energy sources in Portugal, the hydroelectric energy generating
plants of Energias de Portugal (EDP; an electricity company).
Barros [12] concluded that the hydroelectric plants exhibited
average improvement in technical efficiency and technological
change. The increase in technological change was higher than the
increase in technical efficiency. In addition, his results indicated
that EDP should adopt an internal benchmarking procedure to
upgrade the efficiency of the ineffective units. Jha and Shrestha
[13] evaluated the performance of hydropower plants owned by
Nepal Electricity Authority (NEA) during the financial year
2001–2004 period using DEA with three inputs (the installed
capacity of the plant, total operations and maintenance (O&M]
expenditure, and the number of employees) and three outputs
(energy generated by the plant, winter peaking capacity,
and summer peaking capacity). Madlener et al. [14] performed
an assessment of 41 agricultural biogas plants located in Austria to
determine their relative performance in terms of economic,
environmental, and social criteria and corresponding indicators.
DEA and MCDA techniques were used to measure the performance
of biogas plants with two inputs (labor and organic dry
substance) and three outputs (electricity, heat, and greenhouse
gas emissions).
The previously mentioned papers analyzed the efficiency at a
plant or facility level, but Chien and Hu [15] applied DEA to
compare the efficiencies of NRE technology at the country level.
They analyzed the effects of renewable energy on the macroeco-
nomic technical efficiency of 45 countries during the 2001–2002
period using DEA with three inputs (labor, capital stock, and
energy consumption) and a single output (real GDP). Furthermore,
dividing 45 DMUs into OECD member-countries and non-member
countries, Chien and Hu [15] concluded that OECD countries have
higher technical efficiency in renewable energy than non-OECD
countries. Furthermore they found that OECD countries have a
higher share of geothermal, solar, tide, and wind fuels within their
1
Monetary unit of Korea. The exchange rate for $1 is about 1100 KRW in 2014.
K.-T. Kim et al. / Renewable and Sustainable Energy Reviews 47 (2015) 694–702 695

renewable energy supply, but non-OECD countries have a higher
overall share of renewable energy within their total energy supply
than OECD countries.
Among the second category which studies efficiency of alternative
NRE sources, Halkos and Tzeremes [16] applied a bootstrapped DEA
model to evaluate the financial performance of the 78 firms operat-
ing in the Greek renewable energy industry and concluded that firms
operating in the wind power energy sector had higher financial
efficiency than firms operating in the hydropower energy sector. Lins
et al. [17] used the DEA method to incorporate socio-environmental
factors as well as purely economic variables in assessing the
performance of alternative renewable energy resources in the
Brazilian power sector. They compared the DEA-based efficiencies
of 11 alternative energy sources for generating electricity in Brazil
using two inputs (investment cost and O&M costs) and three outputs
(the number of jobs generated, potential electricity provided, and
greenhouse gas emissions). Lins et al. [17] showed that technologies
using solid wastes to generate energy should be assigned higher
priority than the other options analyzed, including those fueled by
natural gas and other renewable energy resources. A similar research
paper by San Cristóbal [18] may be the first attempt to define DMUs
as technology types and analyze the efficiencies at the technology
level; the former three papers dealt with efficiencies of country or
company level. San Cristóbal [18] evaluated the efficiency of 13
renewable technologies related to wind power, hydroelectric, solar,
biomass and biofuel through multiple criteria DEA model with three
inputs (investment ratio, implement period, and O&M costs) and four
outputs (power, operating hours, useful life, and the tons of CO2
avoided).
Among studies analyzing the Korean energy sector, few studies
used the DEA method to analyze the efficiency of electric power
companies ([19–23]), and only one study conducted an efficiency
analysis of NRE [24]. Analyzing the investment efficiency of
national R&D projects using various methodologies, Kim [24]
applied the DEA in analyzing the efficiency of national R&D
projects for NRE using a single input (accumulated investment
costs) and two outputs (number of papers and number of patent
applications). However, this was not an efficiency analysis of the
NRE technology itself.
Table 1 summarizes the relevant literature which used the DEA
approach in analyzing the efficiency of NRE.
3. Model
3.1. DEA model
Since the seminal paper of Charnes et al. [25], much attention has
been paid to DEA by operations research (OR) and management
science (MS) researchers, economists, and efficiency experts from
Table 1
Summary of literatures.
Authors Energy Variable Year Ref.
Input Output
Barros Hydroelectric Number of workers 2008 [12]
Capital Production in MWh
Operating costs Capital utilization
Investment
Jha and Shrestha Hydropower Installed capacity of the plant Energy generated by the plant 2008 [13]
Total operating and maintenance expenditure Winter peaking capacity
Number of employees Summer peaking capacity
Madlener et al. Biogas Labor Electricity 2009 [14]
Organic dry substance Heat
Greenhouse gas emissions
Chien and Hu Renewable energy Labor employment Real GDP 2007 [15]
Capital stock
Energy consumption
Halkos and Tzeremes Renewable energy Debt/equity ratio Return on equity 2012 [16]
Current ratio Return on asset
Assets Turnover ratio Gross profit margin
Operating profit margin
Lins et al. Alternative Energy Greenhouse gases emission 2012 [17]
Potential job creation Operating and Maintenance cost
Potential distributed generation Investment cost
San Crisὀbal Renewable energy Investment ratio Power generation 2011 [18]
Implement period Operating hours
Operating and maintenance costs Useful life
Tons of CO2 avoided
Lee Electric industry Labor Electricity 1999 [19]
Capital
Fuel
Kim and Jo Electric industry Revenue 2000 [20]
Number of employees Profit
Installed capacity Electricity
Lee Electric industry Cost of sales Revenue 2006 [21]
Selling and administrative expenses
Tangible fixed asset
Number of employees
Ko et al. Electric industry Installed capacity Revenue 2008 [22]
Employees Electricity
Ku and Kim Regional efficiency of energy Labor GDP 2011 [23]
Capital
Kim Renewable energy Accumulated investment costs The number of papers 2010 [24]
The number of patents
K.-T. Kim et al. / Renewable and Sustainable Energy Reviews 47 (2015) 694–702
696

various application fields including energy and environment [11].
DEA is a well-established technique to measure and compare the
efficiencies of DMUs with multiple inputs and outputs by applying
mathematical programming.
Suppose that there are n DMUs and they produce s types of
outputs by using m types of inputs. Let yrj be the amount of r‐th
output produced by j‐th DMU, xij be the amount of i‐th input
utilized by j-th DMU, ur be the weight of r‐th output, and vi be the
weight of i‐th input. Based on the efficiency concept in economics,
the efficiency of a DMU, called DMU0 ð0 ¼ 1; 2; …; nÞ, can be
calculated by the ratio of total amounts of output to total amounts
of input. When there are multiple inputs and outputs, the total
amounts can be interpreted as the weighted sum of the amounts of
outputs or inputs. Charnes et al. [25] introduced an optimization
model in which each DMU seeks the optimal weight value of each
input and output in order to maximize its own efficiency defined as
the ratio of weighted sum of outputs to inputs while keeping the
efficiencies of all other DMUs not more than certain value, for
example one. This model, called the CCR model, can be expressed in
mathematical form as follows:
Max
fur ;vig
θo ¼
X
s
r ¼ 1
uryro
X
m
i ¼ 1
vixio
s:t:
X
s
r ¼ 1
uryrj
X
m
i ¼ 1
vixij
r1; j ¼ 1; ⋯; n
ur; vi Z0 f or all r and i ð1Þ
The above formulation can be transformed equivalently into a
more tractable linear programming (LP) model as follows [26]:
Max
X
s
r ¼ 1
uryro
s:t:
X
m
i ¼ 1
vixio ¼ 1
X
s
r ¼ 1
uryrj
X
m
i ¼ 1
vixij r0 ðj ¼ 1; ⋯; nÞ
ur; vi Z0 f or all r and i ð2Þ
Using the above LP model, the efficiency scores of DMU1 to
DMUn can be derived by solving n such models. The optimal value
of the objective function of (2) measures the technical efficiency.
Technical inefficiency results from the difference between actually
used amount of inputs and minimum inputs to produce certain
amount of outputs. Note that the CCR model presumes constant
returns to scale.
3.2. System model of investment in NRE of Korea
To analyze NRE investment efficiency in Korea, a system model
regarding the investment in NRE should first be systematically
established. Accordingly, after we investigated the NRE-related
policies in Korea, it was found that there are two main directions
of NRE policy: supporting NRE dissemination and supporting R&D
investment in NRE technology development. The supporting dis-
semination policy for NRE, centered on users, is meant to increase
the diffusion of NRE by providing motivations (through various
kinds of subsidy) to use NRE voluntarily. This policy has the
intention to eventually create a critical mass of demand for NRE
and thus attract private investments to set up a viable NRE
industry in Korea. Second, technology development policy nur-
tures potential NRE providers (via supporting R&D investment) to
grow a market leader who can commercialize self-developed NRE
products with sufficient technological capabilities. Considering
such aspects, the present paper proposes a system model, com-
posed of inputs, process, and outputs, of NRE investment in Korea
as shown by Fig. 1.
The inputs are assumed to be made up of two elements:
investments for NRE dissemination and investments for NRE
technology development. In this study, the subsidy for NRE usage
promotion which has been provided by Korean government is
selected as the proxy measure of investments for NRE dissemina-
tion and the public R&D expenditure for NRE technology as the
proxy measure of technology development investment. Such
investments would be allocated to various types of NRE technol-
ogy. In this study, fuel cells, photovoltaic, and wind power, which
have been selected as three core NRE technologies by the Korean
government [27], are considered. These NRE technologies consti-
tute the elements of the process in the investment system model
and correspond to DMUs in the DEA analysis. The outputs of the
investment system consist of three elements: the advancement of
NRE technology, increase in diffusion of NRE, and improvement of
economic viability of NRE in Korea. As for the representative
variables for each output element, we choose the number of
patents for the advancement of NRE technology, the total volume
of power generation for the increase in diffusion of NRE, and the
unit cost of power generation for the improvement of economic
viability.
4. Analysis
4.1. Data
The values of data used in the DEA analysis are illustrated
in Table 2. The subjects of DEA analysis are photovoltaic, wind
power, and fuel cells which are regarded as the three major
renewable energy sources in Korea. The period of analysis is set
as the five years from 2007 to 2011. We consider an individual
DMU of DEA analysis as each energy source in each year. There-
fore, although the energy source is the same, the data values used
in DEA should be different if the year is different. For example, as
wind power in 2007 (WP07) and wind power in 2008 (WP08) are
treated as different individual DMUs, so their efficiencies would be
analyzed independently and compared each other.
The data of two input variables were obtained from the ‘2012
New & Renewable Energy White Paper,’ published by the Ministry
of Knowledge Economy of the Korean government and the Korea
Energy Management Corporation [28]. We collected the data of
investment in NRE technology development based on the govern-
ment funded R&D expenditure for each renewable energy source.
The data of investment in NRE dissemination summed the
expenses of subsidy for NRE usage promotion policies by Korean
government, such as the 1 million green homes, regional deploy-
ment subsidy, and loans and tax incentive programs.
Among the output variables, the data of total volume of power
generation were obtained from the ‘New and Renewable Energy
Fig. 1. System of renewable energy investment.

Propagation Statistics,’ published by the Korea Energy Manage-
ment Corporation [29] and the number of patents related to each
NRE source were collected from the National Technical Informa-
tion Service (NITIS) database [30]. In the case of the unit cost of
power generation, there was a subtle problem in gathering
relevant data. It is well known that economic infeasibility some-
times inhibits the use of NRE for generating electric power. In
order to promote the use of NRE even under this disadvantageous
condition, the Korean government has compensated for the gaps
between the power generation costs using NRE and fossil fuel.
Thus, it is reasonable to assume that the unit cost of power
generation of NRE should be equal to the actual charging cost
plus the subsidy of the fore-mentioned compensation. In the case
of wind power, there was little subsidy in 2010 and no subsidy in
2011, indicating that the technological level of wind power is
already high enough to be cost-competitive with fossil fuels. If we
apply the assumption to wind power with no subsidy data, the
unit cost could increase over time even if there would be
technological advancements. To escape from this fallacious case,
we collected the data of the unit cost of power generation
respectively to best align with each NRE source. The unit costs of
photovoltaic and fuel cells were calculated according to the above
assumption using raw data from the Electric Power Statistics
Information System (EPSIS) database [31] and Feed in Tariff (FIT)
statistics [28]. The unit price of wind power generation, was
estimated with ‘the Standard Price of Power Source’ [28]. It should
be noted that the inverse value of the unit cost was used in DEA
because a lower unit cost of power generation represents stronger
technological competitiveness.
Descriptive statistics of the data are presented in Table 3. The
differences between minimum and maximum values of inputs and
outputs are very large. In detail, for the input variables, the
investments in NRE dissemination ranges from a minimum of
2.3 billion KRW to 493.0 billion KRW, and the investments in NRE
technology development from a minimum of 16.9 billion KRW to a
maximum of 84.4 billion KRW. For the output variables, the total
volume of power generation is 8.5 GWh at the minimum and
917.2 GWh at the maximum. The number of patents ranges from a
minimum of 2 to a maximum of 104, and the unit cost of power
generation ranges from a minimum of 101.0 KRW/kWh to a max-
imum of 692.5 KRW/kWh. Those great differences between mini-
mum and maximum values of data led us to infer that the efficiencies
of DMUs might show a significant difference according to both NRE
source and year.
4.2. Analysis
4.2.1. Results of efficiency analysis
The DEA efficiency score was evaluated on the basis of the
CCR model [25], and the output-oriented model was applied to
maximize the output level at the given input level. The analysis
was conducted using DEA-Solver Pro 5.0 of SAITECH Inc.
The efficiency scores of each DMU evaluated by using the CCR
model for 2007–2011 are shown in Table 4 and Fig. 2. An efficiency
score of 1.0 indicates an efficient DMU, and a score less than
1.0 refers to a relatively inefficient DMU. Note that the reference
set, the last column of Table 4, is the set of efficient DMUs from
which an inefficient DMU’s inefficiency is determined. Wind
Table 2
Input and output data.
NRE sources Year DMU Inputs Outputs
Investment for NRE technology
(Mil. KRW)
Investment for NRE dissemination
(Mil. KRW)
Power generation (MWh) Patent (unit) Unit cost (KRW/kWh)
Photovoltaic 2007 PV07 17,100 197,421 71,279 6 692.54
2008 PV08 56,700 376,043 284,315 3 685.43
2009 PV09 70,600 423,649 566,191 7 675.89
2010 PV10 84,400 449,659 772,801 13 642.83
2011 PV11 76,800 493,015 917,198 43 609.54
Wind power 2007 WP07 16,900 16,914 375,641 6 107.29
2008 WP08 17,500 39,829 436,034 3 107.29
2009 WP09 37,500 29,455 685,353 2 105.14
2010 WP10 40,500 21,217 816,950 7 103.04
2011 WP11 41,600 18,487 862,884 19 100.98
Fuel cell 2007 FC07 31,100 2,256 8,522 104 282.51
2008 FC08 63,900 6,056 20,310 50 282.42
2009 FC09 55,000 22,103 89,270 35 280.84
2010 FC10 51,400 50,927 196,960 51 273.60
2011 FC11 42,000 64,135 294,621 69 265.16
Table 3
Descriptive statistics of data.
Inputs Outputs
Investment for
NRE technology
(Mil. KRW)
Investment for NRE
dissemination (Mil.
KRW)
Power
generation
(MWh)
Patent
(unit)
Unit
cost
(KRW/
kWh)
Max 84,400 493,015 917,198 104 692.5
Min 16,900 2,256 8,522 2 101.0
Avg. 46,867 147,411 426,555 28 347.6
S.D. 21,408 187,071 323,230 30 241.5
Table 4
Results of CCR model.
DMU Score Rank Reference set
PV07 0.2626 13 WP08, FC07
PV08 0.2067 15 WP08, FC07
PV09 0.3349 12 WP08, FC07
PV10 0.3944 11 WP08, FC07
PV11 0.6207 8 WP08, FC07
WP07 1.0000 1 WP07
WP08 1.0000 1 WP08
WP09 0.8441 6 WP07, WP11
WP10 0.9626 5 WP07, WP11
WP11 1.0000 1 WP11
FC07 1.0000 1 FC07
FC08 0.4458 9 WP07, FC07
FC09 0.2617 14 WP07, FC07
FC10 0.4396 10 WP08, FC07
FC11 0.7534 7 WP08, FC07
698

power in 2007, wind power in 2008, wind power in 2011, and fuel
cells in 2007 are the most efficient among fifteen DMUs. It is
intriguing that overall, the most efficient NRE source might be
wind power in Korea because there are three DMUs related to
wind power among four efficient DMUs (Fig. 2). However, more
rigorous analyses are needed in order to confidently conclude
which NRE source is the most efficient from the perspective of
investment efficiency, thereby the further analyses are subse-
quently performed in this paper.
We calculated each NRE source’s average efficiency score across
five years and illustrated them in Table 5 with some other
statistics. Wind power is given the highest average efficiency score
of 0.9613, while that of fuel cells is 0.5801 and photovoltaics is
0.3639. In terms of simple average value over a five-year period,
wind power is verified as the most efficient NRE source from the
perspective of the efficiency of public investment. In addition,
photovoltaics energy shows the lowest average efficiency score
during the same period.
In order to more rigorously clarify whether the differences in
efficiencies among NRE sources are significant, a statistical
approach with analysis of variance (ANOVA) with Tukey’s honest
significant difference (HSD) post-hoc test was utilized. Table 6
shows the results of the one-way ANOVA test and Table 7 shows
the results of the Tukey’s HSD post-hoc test.
Significant differences are observed when comparing the average
efficiency scores of three NRE sources under a significance level of
0.01. Based on the observation that there exist significant differences
in efficiencies between three kinds of NRE sources, we perform the
Turkey’s HSD post-hoc test in order to rank the statistically signifi-
cant order of efficiency between NRE sources. As a result, it is verified
that there exist significant differences in the efficiency of investment
not only between wind power and photovoltaics, but also between
wind power and fuel cells, whereas there is no evidence that fuel
cells and photovoltaics have a significant gap in efficiency.
The results of DEA and supplementary statistical analysis assert
that wind power is the most efficient NRE source in Korea from the
perspective of public investment efficiency. It is worthwhile to
compare this result to cases in other countries. However, it is difficult
to collect the same input and output data set from other countries,
making performing an international comparative analysis on the
efficient NRE source using DEA impractical here. Instead, in this
paper, we briefly review the present status of NRE applications of
countries with climatic conditions similar to Korea.
In order to select the countries with climate analogous to Korea,
we compared the annual average temperature and wind speed of
world cities during 2001–2013 using the ‘National Climate Data
Service System’ of the Korea Meteorological Administration (KMA);
as a result we selected Spain, Italy, France, and Japan as countries
Fig. 2. Results of CCR model.
Table 5
Descriptive statistics of the efficiency scores for NRE sources.
NRE Average Standard deviation Min Max
Photovoltaic 0.3639 0.1602 0.2067 0.6207
Wind power 0.9613 0.0675 0.8441 1.0000
Fuel cell 0.5801 0.2939 0.2617 1.0000
Table 6
Results of one-way ANOVA.
Sums of squares Degrees of freedom Mean square F-value
Efficiency score 0.915 2 0.458 11.772n
Within groups 0.466 12 0.039
Total 1.382 14
Significance level: n
po0.01.
Table 7
Results of one-way ANOVA with Tukey’s HSD post-hoc test.
Photovoltaic Wind power Fuel cell
Photovoltaic – 0.5975nn
0.2162
Wind power - 0.3812n
Fuel cell -
Significance level: nn
po0.01, n
po0.05.
Table 8
Climatic conditions of comparative countries.
City Country Temperature Wind speed
Average S.D Average S.D
Madrid Spain 14.7 0.50 5.9 0.45
Rome Italy 16.0 0.42 6.9 0.19
Tokyo Japan 16.6 0.37 5.8 0.12
Paris France 11.8 0.47 6.6 0.09
Seoul South Korea 12.7 0.40 5.4 0.05
Jeju South Korea 16.0 0.32 6.5 0.16

with climate analogous to Korea. Table 8 illustrates the annual
average temperature and wind speed of each country.
Among the four countries, Spain and Italy are relatively active
in adopting NRE. In Spain, the proportion of wind power in the
total amount of generation continuously increased from 2.26% in
2000 to 17.6% in 2012 [32]. In 2013, the total capacity of wind
power facilities amounts to 22.96 GW, the fourth largest wind
power capacity in the world [33]. From the fact that this capacity
accounts for 58.7% of the total amount of generation by NRE, wind
power could be considered a dominant NRE source in Spain. On
the other hand, photovoltaic capacity is proportionately less than
wind power: 5.6 GW of the total NRE capacity (3.29% of the total
amount of NRE power generation) [32]. San Cristóbal [18] also
presented the result that wind power is the most efficient
technology among various NRE sources in Spain. Considering both
papers adopted the same methodology of DEA, and that Korea and
Spain have similar climatic conditions, it is noteworthy that both
identify wind power as the most efficient NRE. The penetration
rate of NRE in the total power generation of Italy is also as high as
29.77% in 2011. In the total amount of power generated from NRE,
the proportion of wind power goes up from 1.10% in 2000 to
13.56% in 2012, which implies that the significance of wind power
is increasing in Italy [32].
Compared to Spain and Italy, France and Japan show low
utilization of NRE in power generation. Because France is more
than 80% dependent on nuclear power in energy, utilization of
NRE is still at a low level. However, the government of France set
up a long-range energy plan called the ‘National Renewable
Energy Development Plan’ with an NRE target penetration rate
of 27% by 2020, and the government is putting more than fifty
relevant policies in action now [32]. The proportion of wind power
in the total amount of power generated from NRE increases from
0.11% in 2000 to 18.0% in 2012 and that of photovoltaics increases
also from 0.01% in 2000 to 4.82% in 2012, although these are less
than tidal power or biomass in France [32]. Japan has tried to
expand the coverage of NRE utilization in order to realize nuclear-
free society since the recent nuclear power plant explosion in
Fukushima. In so doing, Japan constructed a new 6.9-GW capacity
photovoltaic facility in 2013, which results in a total photovoltaic
capacity of 13.6 GW, the fourth largest of this kind in the world
[34]. However, in Japan, the utilization of wind power is still at a
low level.
Table 9
Results of efficiency improvement projection analysis.
DMU Variable Original Improved Variations Variation rate (%)
PV07 Investment in technology 17,100 17,100 0.0 0.00
Investment in dissemination 197,421 25,089 172,332 87.29
Power generation 71,279 271,396 200,117 280.75
Patents
Unit cost 6 23 17 280.75
692.5 154.3 538.2 77.71
Patents 3 15 12 383.80
Unit cost 685.4 33.8 651.6 95.07
Patents 7 21 14 198.62
Unit cost 675.9 27.4 648.4 95.94
Patents 13 33 20 153.53
Unit cost 642.8 23.5 619.3 96.34
Patents 43 69 26 61.11
Unit cost 609.5 29.9 579.7 95.10
FC08 Investment in technology 63,900 63,900 0.0 0.00
Investment in dissemination 6,056 6,056 0 0.00
Patents 50 209 159 318.22
Unit cost 282.4 125.9 156.5 55.42
Patents 35 134 99 282.14
Unit cost 280.8 73.5 207.3 73.83
Power Generation 196,960 448,040 251,080 127.48
Patents 51 116 65 127.48
Unit cost 273.6 75.6 198.0 72.36
Patents 69 92 23 32.73
Unit cost 265.2 89.0 176.1 66.42
700

4.2.2. Efficiency improvement projection analysis
In this section, an efficiency improvement projection analysis
[35], by which we can examine the level of improvement in inputs
and outputs required for any inefficient technology (e.g. photo-
voltaics and fuel cells) to be as efficient as wind power, is
performed. This analysis is useful since it provides targets of
efficiency improvement for currently inefficient investments.
Results of the analysis for photovoltaics and fuel cells are shown
in Table 8.
Overall, the investment in dissemination is excessive for both
inefficient NRE sources in the case of input variables. Moreover, in
the case of output variables, the unit cost of power generation
should be decreased with a simultaneous increase in one of the
other output variables. Focusing on the recent inefficient invest-
ments, let’s look into the results of DMUs of photovoltaics in 2011
and fuel cells in 2011. For the investment of photovoltaics in 2011
to be efficient, the results indicate that the investment in dis-
semination should be reduced to 135.8 billion KRW, 72.4% lower
than the current level, whereas the investment in technology need
not be changed from the present level. On the other hand, two
output variables (amount of power generation and number of
patents) should be simultaneously increased by 61.1%. The output
variable needing the greatest rate of change is the unit cost of
power generated from photovoltaics; which should be decreased
to 29.9 KRW/kWh, 95.1% lower than the current level.
In the case of fuel cells in 2011, it is necessary to reduce the
amount of investment in dissemination by 42.3%, from 64.1 to 37.0
billion KRW. For output variables, power generation must be
increased by 32.7%, from 294.6 to 391.1 GWh, and the number of
patents also must be increased by 32.7%, from 69 to 92. The unit
cost of power generation from fuel cells should be decreased by
66.4%, from 265.2 to 89.0 KRW/kWh.
In conclusion, it is imperative to reduce the level of dissemination
investment in photovoltaics and fuel cells in Korea. Furthermore, in
order to improve efficiency it is necessary to simultaneously increase
the number of patents and power generation, and decrease the unit
cost of power generation. However, it is generally difficult to produce
more outputs with reductions in investment. Therefore, excruciating
efforts are necessary to attain high efficiency of investment in
photovoltaics and fuel cell sectors Table 9.
5. Conclusions
Korea has established a long-term plan for NRE in order to secure
stable energy supplies and promote future growth. Under this long-
term plan for NRE, the Korean government will consistently invest in
technology development and dissemination of NRE in order to
achieve an 11% of NRE penetration rate in energy supply by 2030.
Thus, discussions regarding how to realize the efficient investment
are necessary and the first step to make such discussions effective
would be to measure the efficiency of investments in NRE.
The present paper was an application of DEA to analyze the
efficiency of investment in three major NRE sources in Korea using
empirical data: photovoltaics, wind power, and fuel cells. The
results of DEA indicated that the investments for fuel cells in 2007
and wind power in 2007, 2008 and 2011 turned out to be most
efficient. Further statistical analysis on the efficiency scores of DEA
asserted that wind power is the most efficient NRE technology in
Korea from the perspective of government investment. In addition,
we attempted efficiency improvement projection analysis to
examine the required level of improvement of input and output
factors needed to increase the efficiency of investment in the
photovoltaic and fuel cell technologies. As a result, it was verified
that reducing the level of dissemination investment in photovol-
taics and fuel cells is necessary in Korea.
We expect that, in the future, this work will provide the basis
for creating a policy for making strategic decisions and increasing
the efficiency of investment. This study is limited by the fact that
some important factors in the efficiency of energy technology such
as CO2 emissions have not been considered for some reason.
Therefore, it is necessary to conduct a study taking into account
such important factors and the intangible impacts on the environ-
ment in the field of new and renewable energy.
Acknowledgments
This work was supported by the National Research Foundation
of Korea grant funded by the Korean Government (NRF-2012-
R1A1A2007445).
References
[1] REN21. Renewables 2013 global status report; 2013.
[2] Kim H. Renewable energy and sustainable development. Proc Spring Conf
Korean Soc New Renewable Energy 2007:747–52.
[3] Park SJ, Lee DJ, Kim KT. Forecasting the grid parity of solar photovoltaic energy
using two factor learning curve model. IE Interfaces 2012;25(4):441–9.
[4] Park JH, Kim JH, Heo EY. Peer review of renewable energy dissemination
program. J Korean Soc New Renewable Energy 2010;6(2):40–8.
[5] Ha YJ. An analysis on economic effects of RD investment on energy
technology for technology Innovation. PhD dissertation. Korea Polytechnic
University; 2009.
[6] Lee YS. A study on more efficient performance management and promotion
strategy of commercialization system of renewable energy RD program. PhD
dissertation. Chonbuk National University; 2010.
[7] Lee DG, Moon CG, Heo EY. Outcome analysis on renewable energy dissemina-
tion program through correlation analysis and effectiveness indicator. J Korean
Soc New Renewable Energy 2010;6(2):33–9.
[8] Jin SH, Kim SW. A study on the economic effects of new renewable energy
program by using input–output Table. Korean Resour Econ Rev 2011;20
(2):309–33.
[9] Jin SH, Hwang IC. An analysis on policy performance of new renewable energy
dissemination by region and by energy sources. J Korean Reg Dev Assoc
2011;23(1):15–32.
[10] Kim JB. The basic study on the ripple effect of industrial technological policy
for new renewable energy. J Korean Sol Energy Soc 2012;32(5):18–24.
[11] Zhou P, Ang BW, Poh KL. A survey of data envelopment analysis in energy and
environmental studies. Eur J Oper Res 2008;189(1):1–18.
[12] Barros CP. Efficiency analysis of hydroelectric generating plants: a case study
for Portugal. Energy Econ 2008;30:59–75.
[13] Jha DK, Shrestha R. Measuring efficiency of hydropower plants in Nepal using
data envelopment analysis. IEEE Trans Power Syst 2006;21(4):1502–11.
[14] Madlener R, Antunes GH, Dias LC. Assessing the performance of biogas plants
with multi-criteria and data envelopment analysis. Eur J Oper Res
2009;197:1084–94.
[15] Chien T, Hu JL. Renewable energy and macroeconomic efficiency of OECD and
non-OECD economies. Energy Policy 2007;35(7):3606–15.
[16] Halkos GE, Tzeremes NG. Analyzing the Greek renewable energy sector: a data
envelopment analysis approach. Renewable Sustainable Energy Rev 2012;16
(5):2884–93.
[17] Lins ME, Oliveira LB, Silva ACM, Rosa LP, Pereira Jr AOP. Performance assessment
of alternative energy resources in Brazilian power sector using data envelopment
analysis. Renewable Sustainable Energy Rev 2011;16(1):898–903.
[18] San Cristóbal JR. A multi criteria data envelopment analysis model to evaluate
the efficiency of the renewable energy technologies. Renewable Energy
2011;36(10):2742–6.
[19] Lee HJ. Non-parametric data envelopment analysis of efficiency and produc-
tivity: an application to electric industry. Ph.D Dissertation. Seoul National
University; 1999.
[20] Kim TW, Jo SH. Measuring efficiency of global electricity companies using data
envelopment analysis model. Korean Resour Econ Rev 2000;9(2):349–71.
[21] Lee YS. The research about changes of structure and efficiency of the electric
power industry according to environmental changes of the electric power
industry. Research report. Korea Energy Economics Institute; 2006.
[22] Ko S C, Sim GS, Kim JY. An operating efficiency evaluation regarding the use of
DEA in Korea’s electric power generation industry. J Mil Oper Res Soc Korea
2008;34(1):61–77.
[23] Ku JY, Kim SD. An analysis of energy efficiency in Korean and Japanese
economic regions using data envelopment analysis. J Korean Soc Geosyst Eng
2011;48(1):79–88.
[24] Kim CY, Study on the improvement of investment efficiency budget control
system of government Research and development. Research report. Korea
Institute of Science Technology Evaluation and Planning; 2010.

[25] Charnes A, Cooper WW, Rhodes E. Measuring the efﬁciency of decision
making units. Eur J Oper Res 1987;2(6):429–44.
[26] Cooper WW, Seiford LM, Tone K. Data envelopment analysis. Kluwer Aca-
demic Publishers; 2000.
[27] Ministry of Industry Resource. The second basic plan for technology develop-
ment and usage promotion of renewable energy in Korea, Mimeo; 2003.
[28] Ministry of Knowledge Economy, Korea Energy Management Corporation New and
Renewable Energy Center. New and renewable energy white paper 2012; 2013.
[29] Korea Energy Management Corporation, New and Renewable Energy Propa-
gation Statistics 2012; 2013.
[30] National Technical Information Service (NITIS) database (〈www.nitis.go.kr〉).
[31] Electric Power Statistics Information System (EPSIS) database (epsis.kpx.or.kr).
[32] EIA database (〈www.eia.gov〉).
[33] Global Wind Energy Council. Global wind report 2013; 2014.
[34] REN21. Renewables 2014 Global Status Report; 2014.
[35] Suzuki S, Nijkamp P, Rietveld P, Pels E. A distance friction minimization
approach in data envelopment analysis: a comparative study on airport
efﬁciency. Eur J Oper Res 2010;207:1104–15.
702

Measuring the efficiency of renewable energy investment in Korea using data envelopment analysis

Recommended

Recommended

More Related Content

What's hot

What's hot (9)

Similar to Measuring the efficiency of renewable energy investment in Korea using data envelopment analysis

Similar to Measuring the efficiency of renewable energy investment in Korea using data envelopment analysis (20)

Recently uploaded

Recently uploaded (20)

Measuring the efficiency of renewable energy investment in Korea using data envelopment analysis