MODELLING
RENEWABLE
ENERGY
MANAGEMENT
INEFFICIENCY IN EUROPE WITH PANEL DATA STOCHASTIC
FRONTIER MODELS
Angeliki N. Menegaki, Department of Languages, Literature and Culture of Black
Sea Countries, Democritus University of Thrace, 69100, Komotini
Abstract
The paper employs three stochastic frontier inefficiency configurations to categorize
European countries according to their inefficiency in renewable energy management.
The results come from an empirical application of a panel with 32 European countries
over a 14 year old period using a translog type production function. In particular the
paper focuses on results from the Alvarez et al. (2006) fixed management model and
compares results with the conventional stochastic frontier model and the random
coefficients model with inputs such as renewable energy, fossil fuel energy,
employment, capital and carbon emissions. The results suggest that renewable energy
deployment does not significantly affect growth in Europe, wherein management
inefficiency becomes eloquent with the aid of the management adapted frontier
model.
Keywords: Europe; fixed management model; inefficiency; renewable energy;
stochastic frontier;
1. Introduction
Renewable energy sources (RES) are currently unevenly and insufficiently exploited
in the European Union, EU (Menegaki, 2011, Menegaki, 2012) with a small
contribution of about 7.8-8% (Eurostat, 2012) to the overall gross inland energy
production. In spite of the various European directives for the promotion of RES

Tel.fax. +30 2531072886
Email address: [email protected]
1
being the 2001/77/EC on electricity production from RES (European Commission,
2001), the 2002/91/EC on energy performance of buildings (European Commission,
2002), the 2003/30/EC on the promotion of biofuels and other bioliquids (European
Commission, 2003) and the 2009/28/EC on the promotion of the use of energy from
RES (European Commission, 2009), which demanded that RES in final gross energy
consumption in Europe doubled from 6 to 12% and achieving 22% electricity
production from RES by 2010, also a reduction in primary energy use by 20% and a
reduction of greenhouse gases by 20% below the 1990 levels, still no-long run
relationship between RES consumption and growth has been confirmed providing
evidence for the neutrality hypothesis (Menegaki, 2012), namely evidence that no
dependence of the two magnitudes exists or that renewable energy does not play a
significant role in European growth. Moreover, the share of imported fuels remains
high and is estimated to reach 70% of total energy consumption by 2020 (Loyning,
2010). EU Directive 2009/28/EC repeals EU Directives 2001/77/EC and 2003/30/EC
and paves the way for an effective management of RES across member states in order
to pursue their RES 2020 goals.
The overall 2020 potential for RES in the EU corresponds to a share of 28.5%
of the overall current gross final energy demand. Member States possessing large
RES potentials are France, Germany, Italy, Poland, Spain, Sweden and the UK, while
Austria, Finland, Portugal and Sweden had already by 2005, fulfilled their 2020
planned potential. Sweden is the leader with 44.4% of gross final consumption while
Malta is a laggard with a minor 0.2% (Loyning, 2010). A breakdown of the RES
potential in Europe reveals that the highest target has been set for the heat sector
wherein the largest progress has been made (Ecofys, 2011).
2
The vast majority of member states are confident they will reach their 2020
RES goals and 60% of them expect to exceed them while Italy and Luxemburg plan to
resort to co-operation mechanisms to achieve their goals (EREC, 2011) and smooth
national discrepancies. Furthermore, the relevant industry forecasts a share in final
energy consumption of about four percentage units above the share forecasted in their
national renewable energy action plans (EREC, 2011) despite the financial crisis. By
2020 wind energy will represent 14.1% of the electricity consumption, hydropower
10.5%, biomass 6.5%, photovoltaics 2.35%, solar power 0.5%, geothermal energy
0.3% and ocean energy 0.15% (Eurostat, 2012). The share of RES in heating and
cooling will increase from 10.2% in 2005 to 21.3% in 2020. Renewable energy in
transport would amount to 12.2% by 2020 (EREC, 2011). There also hopeful
estimates that by 2050 renewable electricity will provide 100% of the European
power demand (Zervos et al., 2010).
RES supporting instruments
There are various categories of support instruments: feed-in tariff, premium, quota
obligation, investment grants, tender schemes, tax exemptions and fiscal incentives.
The number of the employed instruments by each country is presented in Figure 2
with data taken from Ecofys (2011). Feed-in tariffs guarantee the price and a secure
demand thereby creating certainty to investors. There is always the possibility of a
large profit for RES producers when the cost of renewable energy has been
overestimated, but this can be alleviated when the cost is regularly updated. Boomsma
et al. (2012) provide evidence of investment behavior under RES support schemes.
They find that feed-in-tariffs encourage earlier investment while as the investment
proceeds, certificate trading creates incentives for larger projects. As regards quota
obligations, they entail the imposition of minimum shares of renewable electricity on
3
suppliers or producers that increase with time. They are often combined with tradable
green certificates. Tender schemes, provided usually for large off-shore projects, draw
attention to the competitive element incorporated in renewable energy projects. Tax
incentives appear either in investments or the production of RES. Fiscal incentives
also include soft and low interest loans with longer repayment periods or tax
exemptions from CO2 or energy taxes. They are attractive because of their direct
message transmitted to final energy consumers about the added value of RES
(OPTRES, 2007). For a detailed overview of the supporting instruments applied in
each country, interested readers should consult Ecofys (2011).
4,5
4
3,5
3
2,5
2
1,5
1
0,5
Au
str
Be ia
lgi
u
Bu m
lga
ria
Cz
C
ec ypr
hR us
ep
u
Geblic
rm
a
De ny
nm
ar
Es k
ton
ia
Sp
ain
Fin
lan
d
Fra
nc
Gr e
ee
Hu ce
ng
ary
Ire
lan
d
It
Lit aly
hu
Lu ania
xe
mb
urg
La
tvi
a
Ne Malt
the a
rla
nd
s
Po
lan
Po d
rtu
Ro gal
ma
n
Sw ia
ed
Slo en
ve
n
Slo ia
va
kia
UK
0
Figure 1: Number of RES supporting instruments in European countries (data taken
from Ecofys, 2011).
As far as the structure of this paper is concerned, after the introduction, the rest of the
paper is organized as follows: Part 2 explains the constituents of an effective and
efficient management of renewable energy resources by countries. Part 3 presents the
fixed management stochastic frontier approach for inefficiency measurement. Part 4
deals with data description and results and last, part 5 is the conclusion.
4
2.
Effective and efficient management of renewable energy resources by
countries
The management that countries need to perform, focuses on increasing RES
penetration (at least according to 2009/28/EC Directive demands) in the market while
at the same time minimising public costs. Management must be applied as such so as
to keep RES industry consumers and producers interested. One of the biggest
challenges is to bring the above three groups into a joint agreement and reconcile their
different interests. Table 1 addresses all the RES management defaults noted today
and proposes solutions consistent with the 2009/28/EC Directive.
For the industry agents, a guarantee of a continuous demand of their applied
technology is crucial to keep them in the game. Investors, depending on their risk
attitudes, are interested in as much high a producer surplus as possible. Consumers are
interested in low prices. Today and until 2010, most countries have performed
Business-as-Usual (BAU) policies and they worked towards a harmonization of
European objectives. Assuming that energy consumption grows with time and that
CO2 prices pass directly to energy prices, increased RES deployment will decrease
CO2 prices. It is crucial to know the potential domestic and realistic supply of energy
from each technology: biogas, biomass, biowaste, on-shore/off-shore wind,
small/large scale hydro, solar, thermal, photovoltaic, tidal, wave and geothermal
energy.
Costs are adapted by endogenously technology-specific learning rates and
estimate the long-run marginal social cost. Policy makers must decide whether they
will remain at BAU solutions leading to the most cost-effective technology which
however excludes expensive in the short-run novel technologies. Existing and new
plants should be distinguished and handled differently by support mechanisms.
Support should be stable but limited upto a certain time and might as well become
5
integrated with other policies such as climate change or agricultural policy.
Technological learning must be reflected in the feed-in-tariff sytem by taking into
account and correct any overcompensation produced by uniform tariffs, namely
introducing tariff degression based on the technological progress made. Local
electricity stations must be lawfully bound at buying green energy at priority. Tariff
levels can vary depending on local conditions, the size or fuel type of the plant. The
idea of setting a plant as a reference point increases transparency, so that not only the
most favourable conditions in a country are exploited and the risk of over-subsidizing
very efficient plants is lowered. Premium tariffs paid on top of the electricity market
prices are a move towards a more market based support instrument. It leads to a more
efficient allocation of the grid costs and gives producers an incentive to feed
electricity into the grid at times of peak demand. Penalty payments for noncompliance can ensure quota fulfillment. Abolishment of subsidies for fossil and
nuclear energy will re-distridute investment interest on RES.
Markets need to be deregulated at a higher degree, so that consumers will not
bear the burden of RES development while producers will go free. Transmission
System Operators (TSOs) must be provided the possibility of acting independently so
that this leads to the development of the transmission infrastructure capable to
integrate RES. The intermittent nature of wind or solar energy and the seasonality of
biomass and hydropower are important to take into account and incorporate in the
demand. According to Zervos et al. (2010), until EU has a fully liberalised electricity
market, RES must have priority access to the grid. In particular Jamasb and Politt
(2005) suggest some interesting form of electricity liberalization which consists of a
combination of competitive energy and retail markets with regulated transmission and
distribution activities.
6
Countries which develop RES, also increase employment (2 million full time
jobs will be created by 2010; Lins, 2008, projecting at 4.4 million jobs by 2030 and
6.1 million jobs by 2050; Zervos et al., 2010) and industrial development and at the
same time decrease local pollution (728 million tons/year of CO2 emission reduction
in 2020, representing 17.3% of the total GHG emisions in 1990; Lins, 2008). Crossborder power trading through the building of grid infrastructure can be enhanced. The
creation of a single power market based on RES also imposes the need for a super and
smart grid to interconnect all RES installations both in a centralised and decentralised
way. New power capacity equal to 42% of current EU capacity must be built by 2020
(Zervos et al., 2010).
Co-operation among countries is imposed for a variety of reasons one of
which being the interaction among the different support schemes which for example
can have an impact on the price of CO2 permit trading system or the price of power in
these regions and across borders. Each country must take into accout its political,
technological and market risks in whose context a RES plan must be built or adapted
for the country. All countries together need to work for an EU-wide carbon tax
reflecting all costs incurred when using conventional energy sources.
Good management entails the establishment of a one-stop shop which assigns
only to one authority the permit and support related procedures of RES projects, while
at the same time will reduce lead times and lack of the co-ordination among
authorities.
Moreover, apt RES management must concentrate urgently on grid matters
giving priority to a pan-European grid system. Sufficient and reinforced capacity has
to be built. Transparent and swift procedures for grid connection must be established.
7
Campaigns must lower public opposition and disseminate the hidden benefits
from RES. Emphasis must be placed to sustainable development through job creation.
Participation of local public, the authorities and other stakeholders in large projects in
their area through co-operation programmes is crucial and will not jeopardise the
implementation of the projects. A stable and transparent environment will uphold
other financial barriers such as the lack of trust from banks because of subsidies'
unpredictability due to the vague administrative framework.
Table 1: Management defaults and solutions for renewable energy sources.
RES Management default
Solution
1. Planning delays and restrictions often caused by - Establish preclusion effect (i.e. objections
reaction to objection and issuance.
can be raised only withing a given time).
2. Lack of co-ordination between authorities
- One-stop shop (through a central agency).
because of their large number being involved and
the unclear administrative framework (e.g. broad
margins of descretion, lack of transparency,
corruption etc).
3. Lengthy procedures in obtaining authorization
- One-stop shop.
e.g. 3-6 years in Greece, France etc.
- Provide a fast lane for RES projects.
- Set obligatory limits for authorization
provision; if not met, this will lead to tacit
approval.
4. Costs of obtaining permission (e.g. 30% of the
One-stop shop.
overall costs for small PV projects).
5. Insufficient or hostile spatial planning affecting
 Include stakeholders financially, e.g.
large projects caused by deliberate actions,
ensure that municipalities directly benefit
social opposition and environmental effects.
from RES projects in their area.
 Involve local population and encourage
their participation in the project.
 Teach them to make the trade-off between
the additional local impact and the avoided
import of fossil fuels.
6. No exemption of small-scale systems from
- Adapt legislation as such.
authorization.
7. Number of permits required is going up to 40
One-stop shop.
(wind energy in Greece and Cyprus).
8. Grid connection and access problems (both
- Extension of existing electricity networks
within the country and within neighbouring
and their development into smart
countries), because they are not ensured by law, networks.
there is resistance by TSOs and DSOs
 Enhance cross-border trade in electricity
(Distribution system operator) and the fact that
between member states by increasing
the existene of old infrastructure demands huge
interconnection capacity.
financial support.
 Establish a priority grid access for
electricity by RES.
 Deregulate energy markets more.
 Introduce an efficient sanction scheme for
TSOs and DSOs.
8
RES Management default
Solution
9. Limited information and awareness on RES
 Prepare the public to accept the impact on
benefits, on support measures, pilot projects and landscape from large RES projects.
insufficient funding for information campaigns.
 Dissemination of best practice projects.
10. Barriers for build environment (40% of final
 Exemplary role of public buildings.
energy demand in Europe is consumed in
 Establish renewable energy obligations for
buildings), because of lax energy performance,
new buildings.
requirements by national laws, conservativeness  Make clear that areas under monument
of construction industry, lack of financial motives, protection will be useless derelicts if they
problems in tenancy and property laws.
do not have modern provisions of energy
and water.
11. Lack of certification schemes for installers due - Mandate member states to appoint such a
to lack of certification bodies, lack of training and body both at a national and european
guidelines.
level.
12. Lack of market transparency and market prices - Establish a monitoring body both at
not including the external costs of energy.
national and international level.
- Promote deregulation.
13. Old fashioned legislation tailored to use fossils Adapt legislation.
instead of RES.
14. Non-intentional barriers such as complete lack Adapt legislation.
of administration procedures because there is no
demand for them.
15. Heterogeneous application of laws:
Adapt legislation and provide one-stop shop
contradictory fragmentation of political
at a cental agency.
competences (region, provinces and
municipalities).
16. Delay or lack of promotion in biogas because of - Introduction of subsidies.
lack of infrastructure, equipment and incentives - Develop legislation with regard to access
and due to the precence of bureaucracy.
and grid codes etc.
17. Delays in establishing district heating or cooling. - Disseminate examples from Scandinavian
countries and provide the know-how.
- Create incentives.
Note: The table has been summarized from ECORYS (2010).
3. The fixed management stochastic frontier approach for ineffieciency
measurement
The concept of technical efficiency has been widely used in a variety of
benchmarking applications; in the banking sector (Behr, 2010), dairy firms (Yélou et
al., 2010), manufacturing firms (Bhaumik et al., 2011), sawmill industry (Helvoigt et
al., 2009), container ports (Cullinane, 2006), oil and gas industry (Managi et al. 2006),
fisheries (Tingley et al., 2006), pig finishing farms (Van Meensel, 2010), wind farms
(Iglesias, 2010) to quote only a few of the most recent. However, none, to the best of
the author’s knowledge has applied the frontier analysis and more specifically the
9
fixed management stochastic frontier (Alvarez et. al., 2006) or the management
adapted efficiency frontier from now on, method country-wide or particularly in the
European countries growth framework with an emphasis on renewable energy as an
input. Typically studies across countries perform cointegration and causality analyses
between GDP, renewable energy and other factors (Menegaki, 2011; Apergis and
Payne, 2011) and they do not perform any benchmarking, except for cases where a
country is evidenced to contribute to heterogeneity and therefore it can be included as
a dummy variable for a more precise reflection of the long-run relationship among the
variables.
The main advantage of the stochastic frontier approach over fixed effects
panel regression models (Bhaumik et al., 2011) is that it can provide evidence not
only on whether or not the average country has GDP efficiency hysteresis in our
example, but also it estimates a measure of the degree of the hysteresis for each
country and ideally for each time period (however, time was not significant in the
current application), and also the marginal impact of country characteristics on this
measure. While incorporating heterogeneity in the conventional and random frontier
models did not lead to the generation of significant models and therefore could not
disentangle the contribution of specific inputs or countries to heterogeneity,
the
management adapted efficiency frontier model was significant and could provide
some equivalent explanation for the disparate progress European countries have made
on renewable energy resources penetration in their economies. To start with,
observing the graph of RES and GDP (Figure 2), it becomes evident that there are
factors other than wealth or the size of inputs an economy is using in its production
process that affect the penetration of RES in it, because we observe countries with a
high GDP and a very low attainment in RES and vice versa.
10
80
70
60
50
RE S
40
30
20
10
0
-10
200
100
0
300
GDP
Figure 2: Renewable energy and GDP in European countries.
Following the stylised specification in literature we will describe a production frontier
uU
model representing the maximum GDP attainable, namely y   x  v  u with
.
Thıs is Aigner, Lovell and Schmidt’s (1977) normal-half model. Also note that

U ~ N 0,  u2
 and

v ~ N 0,  u2
. As
aforementioned, we employ the Alvarez et al.
(2006) fixed management model, in a production frontier for European economies in
order to push forward the hidden role of management in those economies and the size
effect it bears on the contribution of renewable energies in European growth. Each
country is supposed to encompass a time invariant factor labeled management so that
the production frontier appears as:
yit  f xit ,1 , xit , 2 ,..., xit , K , mi 
EQ1
The functional form employed for the estimation is a stochastic frontier model of a
translog type with five inputs is depicted in Equation 2 (Greene, 2007):
11
K
log yit  ai    k ,i ln xit ,k  k 1 m1  km ln xit ,k ln xit ,m  vit  uit
K
K
k 1
EQ2
1 
ai  a   a wi   aa  wi2 
 2  and  k ,i   k  k wi
where,
wi ~ N 0,1, vit ~ N 0, 2 , uit  N 0,  u2




The second order terms in the translog allow for cross-relationships while the squares
for non-linearities. The translog function allows elasticities to vary by the magnitude
of the explanatory variable. However, regardless of the above mathematical niceties
of this function, the reason it was deployed in the current empirical application, was
its better fit to the data. Therefore we have neglected issues such as monotonicity and
concavity. We have dropped elements that were deteriorating the model fit since our
aim is not to use the frontier for predictions but to provide a snapshot of the
management inefficiency across countries.
It is understood that
it vit uit and this will be conditioned on m . The
i
random parameters model relies on Mundlak approach (Greene, 2007) to incorporate
the group means of the variables in the model as in Equation 3. Also with fi being the
structural random variable that drives the random parameters, we write Equation 3 as:
K
wi   k log xi ,k  fi
k 1
EQ3
To measure ui, we will use the Jondrow et al. (1982) estimator to estimate it. This is
provided by Equation 4
 









z

E
u


z
,
v

u
,
z


2


1

1


z


which, if conditioned on m, can be re-written as in Equation 5






 











/
m
/
/
m
it
i
it
i


E
u
/
,
m


it
it
i


2




1



/
m
/
it
i


EQ4
EQ5
12
Since we are interested in the inefficiency of each country, we calculate it as
suggested in Greene (2007), and described in Equation 6
y
Eff


Exp
(

u
)
Optimal
_
y
EQ6
4. Data description and results
Annual data ranging from 1997 to 2010 for 31 European countries were obtained from
Eurostat database. The multivariate frontier framework encompasses real GDP per
capita in PPP terms (Variable: GDP), percentage of RES in gross inland energy
production (variable: RES), final energy consumption in 1000 toe (variable: CON),
greenhouse emissions on CO2 equivalents with base year 1990 (variable: GRE),
employment rate (variable: EMP) calculated as the number of people with ages
between 15 and 64 years old divided by the corresponding population size and capital
being the real gross fixed capital formation (variable: CAPI). Summary statistics are
listed for each of the variables across countries in Table 4 in the Appendix.
Table 2. Estimated frontier models
Variable
Stochastic Frontier
Constant
-83.768
(10.980)*
-0.788
(0.618)
3.488
(0.589)*
6.744
(2.211)*
24.872
(2.716)*
-0.157
(0.067)*
-34.972
(12.200)*
0.301
(0.564)
1.713
(0.775)*
2.923
(1.817)*
10.440
(3.273)*
-0.010
(3.023)
0.077
(0.013)*
0.416
0.041
(0.009)*
0.000
Model 1
Renewable Energy (RES)
Energy Consumption
(CON)
Carbon Emissions (GRE)
Employment (EMPLO)
Capital (CAPI)
Random Coefficients Model
Means of Random Parameters Management  Inputs
Model 2
Model 3
Management
RES  RES
RES  CON
0.172
(0.923)
0.019
(0.018)
0.028
(0.007)*
-0.181
(0.046)*
0.183
(0.194)
0.053
(0.135)
0.092
(0.019)*
13
RES  GRE
RES  EMPLO
RES  CAPI
CON  GRE
CON  EMPLO
CON  CAPI
GRE  EMPLO
GRE  CAPI
EMPLO  CAPI
CAPI  CAPI
(0.012)*
0.289
(0.507)*
-0.299
(0.179)**
-0.034
(0.048)
0.356
(0.037)*
-1.205
(0.142)*
-0.050
(0.034)
-2.599
(0.512)*
0.149
(0.098)
0.620
(0.205)*
-0.530
(0.103)*
(0.011)
-0.213
(0.050)*
0.157
(0.100)
0.003
(0.072)
0.136
(0.034)*
-0.491
(0.168)*
0.024
(0.1086)
-1.025
(0.417)*
0.072
(0.220)
-0.112
(0.856)
0.000
(0.000)
Variable means in unobserved management
-0.712
(0.058)*
CON
-0.441
(0.179)
GRE
0.800
(0.358)
EMPLO
1.492
(0.729)
CAPI
-1.840
(0.222)
λ
0.766
2.039
(0.128)*
(0.991)
σ
0.266 (0.000)*
0.111
(0.014)*
LogL
15.839
393.052
Notes: Standard errors in parentheses, * and **indicate 5% and 10% significance respectively
RES
We have estimated a stochastic production frontier with five inputs (time effects were
not significant). First order coefficients are typically interpreted as output elasticities
evaluated at the mean of the sample. However, as aforementioned we will omit all
these technical niceties (Christensen et al. 1973), because the purpose of the paper is
not to compute elasticities or make predictions which would presuppose special
handlings of the data, e.g. be corrected by their geometric mean before estimation
(See and Coell, 2011). The paper aims to highlight the differences across the three
inefficiency types and make evident the management hysteresis in RES.
14
In model 1, all inputs are significant except for RES. Capital has a negative
sign. Next, in model 2, we estimate the random coefficients model with 1000 Halton
draws in each replication. RES remains insignificant. Also capital is non significant in
this version of frontier model. In model 3, the coefficients of management (βkm) are
significant only for energy consumption and carbon emissions but not for renewable
energy and capital. The separate coefficient of management (βm) is significant. The
positive sign of management can be interpreted as evidence that management has a
positive sign with an increasing effect on production (growth). We have applied the
Wald test with one restriction on whether all betas are commonly zero to find out that
the estimated function does not collapse to a Cobb-Douglas frontier since the χ2 test
was not significant (p=0.94). Noteworthy is the fact that while the constant term was
high and significant in the conventional and random parameters models, this term is
not significant in the management adapted frontier model (Model 3), while the
independent term accounting for management becomes significant. Last, RES is the
only significant variable in the means of unobservable management.
After applying Equation 6, mean inefficiency has been estimated to be 0.98
(st.dev.=0.0007), min=0.9807 and max=0.9882. Table 3 lists the descriptive statistics
for the inefficiency estimates produced from the conventional stochastic frontier
model, the random coefficients stochastic frontier model and the management adapted
frontier model (Models 1, 2 and 3 respectively). Apparently in the latter case, is both
the mean and the standard deviation smaller, providing evidence that if management
is not taken into account, then inefficiency is biased, producing thus an inaccurate
estimation of the overall amount and its constituents.
15
Table 3. Descriptive statistics for the three inefficiency models.
Conventional Stochastic frontier model
Random coefficiens stochastic frontier model
Management adapted frontier model
Mean
0.121
0.084
-1.010
St.deviation
0.054
0.068
1.299
Min
0.000
0.000
-3.570
Max
0.392
0.443
4.192
4
3
2
Managementadaptedinefficiency
1
0
0
10
20
30
40
50
60
70
80
-1
-2
-3
Figure 3. Management adapted inefficiency and RES
It is interesting to observe Figure 3 with management adapted inefficiency (MAI),
because observations have the largest range and it resembles a hyperbolic shape. The
other two inefficiency measures, conventional stochastic frontier (CSF) and the
random stochastic frontier (RSF) have small ranges and resemble almost straight
lines. The hyperbolic shape of the graph means that as RES increase, inefficiency
drops, initially at a larger rate and then at a smaller rate, because RES increase at a
smaller rate, probably due to learning effects. Hence a critical difference between
MAI and the other two measures, CSF and RSF, is that MAI allows various
inefficiency grades based on RES attainment level by each economy. The largest
correlation between RES and the three inefficiency measures is borne by MAI, equal
to -0.52. Countries with positive MAI are Ireland, Luxemburg, Cyprus and Malta also
being some of the RES laggards in Europe. Above the mean MAI but below the
16
horizontal axis are: Bulgaria, Denmark, Slovenia, Belgium, Greece, Netherlands and
the UK. The rest of the countries are below the mean MAI with the following ranking
from first to last: Hungary, Slovakia, Czech Republic, Lithuania, Switzerland,
Portugal, Finland, Germany, Italy, Poland, Estonia, France, Austria, Sweden, Norway,
Spain, Romania and Latvia. Of course it must be pinpointed that this ranking does not
come solely from inefficiency on RES but it is rather the combined inefficiency from
all inputs but with a high correlation with RES.
4
Management adapted
Random
Conventional
3
1
0
Be
lg
Bu ium
C lg
ze a
ch ria
D Rep
en .
m
G ar
er k
m
a
Es ny
to
n
Ir e i a
la
G nd
re
ec
Sp e
a
Fr in
an
ce
Ita
C ly
yp
ru
L s
Li atvi
th a
Lu ua
xe ni
m a
b
H urg
un
ga
ry
N
et Ma
he lta
rla
n
Au ds
st
Po ria
l
Po and
rt
R uga
om l
a
Sl nia
ov
e
Sl nia
ov
a
Fi kia
nl
Sw and
ed
en
U
C K
ro
a
Ic tia
el
a
N nd
Sw or
itz way
er
la
nd
Mean Inefficiency
2
-1
-2
-3
Country
Figure 6. Comparison of mean inefficiency measures across European countries.
Figure 6 includes the graphs of all three types of inefficiency examined in this
empirical application. The CIF does not have large fluctuations and differences among
countries. The RSF has larger fluctuations while the MAI, has the largest fluctuations
and has differences among groups of countries. Conventional and Random
inefficiencies are positively correlated with each other with correlation equal to 0.447.
17
80
70
Mean management adapted inefficiency
60
50
40
30
20
10
ro
U
K
at
i
Ic a
el
an
N d
Sw orw
itz ay
er
la
nd
-10
C
Be
lg
iu
Bu m
C lga
ze
r
ch ia
R
D ep
en .
m
G ark
er
m
an
Es y
to
ni
Ire a
la
n
G d
re
ec
e
Sp
a
Fr in
an
ce
Ita
C ly
yp
ru
s
La
t
Li via
th
Lu uan
x e ia
m
bu
H rg
un
ga
ry
M
N
et
a
he lta
rla
nd
Au s
st
r
Po ia
la
Po nd
rtu
R ga
om l
a
Sl nia
ov
e
Sl nia
ov
ak
Fi i a
nl
a
Sw nd
ed
en
0
RES
Figure 7. Mean management adapted inefficiency, and mean RES across European
countries.
Based on Figure 7, we can arbitrarily distinguish several groups of countries. RES and
MAI opposite co-movements become more evident in one group of countries, where
clearly a kind of elliptic circle is formed by the two graphs. These countries are
Ireland, Greece, Spain, France, Italy and Cyprus. This argument is further
strengthened by the fact that the correlation between RES and MAI is -0.52.
Moreover, this almost perfectly elliptic pattern encompasses most Mediterranean
countries. Other groups observed with distinctive opposite co-movements in the two
magnitudes, are; Group 1: Belgium, Bulgaria, Czech Republic, Denmark, Germany,
Estonia, Ireland, Group 2: Greece, Spain, France, Italy, Cyprus, Group 3: Latvia,
Lithouania and Luxemburg, Group 4: Hungary, Malta, Group 5: Netherlands, Austria,
Poland, Portugal, Romania, Slovenia, Slovakia, Group 6: Finland, Sweden, UK,
Group 7: Croatia, Iceland, Norway and Switzerland. The full ranking of all European
countries is listed in Table 5 in the Appendix.
18
5. Conclusion
European economies need to adopt managerial strategies consistent with the
2009/28/EC Directive in order to address all the relevant obstacles and bolster RES
expansion Europe wide. Efficiency from a managerial point of view is crucial, due to
the need of reaching the deadlines posed by the European Commission and stipulated
by European countries participating in the EU. The deployment of renewable energy
does not participate in the production and growth of European economies in a drastic
way and there is a lot of way to be covered by many countries.
This paper aims to unveil the management hysteresis in the deployment of
renewable energy in Europe while at the same time it highlights the differences of
three inefficiency models that aim to do that. In essence, the stochastic frontier
approach is used for estimating efficiency scores and the consequent benchmarking.
To the best of the author's knowledge, this is the first empirical application of
management adapted frontier models, country wide, and with particular reference to
European countries benchmarking for renewable energy levels in their economies.
Observing the existence of countries with high GDP and low renewable energy
levels, it becomes understandable that renewable energy penetration in European
economies does not go on a pari passo with income.
Renewable energies
management in Europe defaults in many aspects and there is a lot of margin for their
full potential to unfold in the majority of countries. Complying with the demands of
the 2009/28/EC Directive, will be a major step towards the homogenization of energy
policy in Europe, leading to energy supply security, climate change mitigation,
industrial and employment growth. Most of the hysteresis will be covered if one-stop
shop is established which will reduce delays, lack of co-ordination and transparency,
lead-in costs, social resistance or apathy, legislation gaps and other kinds of barriers.
19
The management adapted inefficiency frontier models is superior to the
conventional and random coefficients frontier models in that it clearly distinguishes
several groups of countries based on their management status. While the two
commonplace measures account for small fluctuations in inefficiency thus inhibiting a
decisive categorization of countries, the management adapted model allows to a
categorization of distinctive groups.
REFERENCES
Aigner, D., Lovell, K. AD & Schmidt, P. (1977). Formulation and estimation of
stochastic frontier production function models, Journal of Econometrics, 6, 21-37.
Alvarez, A., Arias, C. & Greene, W. (2006). Fixed management and time invariant
efficiency in a random coefficients model, manuscript, Department of Economics,
University of Oviedo, Oviedo.
Apergis, N. & Payne, J. E. (2012). Renewable and non-renewable energy consumptiongrowth nexus: Evidence from a panel error correction model, Energy Economics,
forthcoming.
Behr, A. (2010). Quantile regression for robust bank efficiency score estimation,
European Journal of Operational Research, 200(2), 568-581.
Bhaumik, S. K., Das, P. K. & Kumbhakar, S. C. (2012). A stochastic frontier approach
to modelling financial constraints in firms: An application to India, Journal of
Banking and Finance, 36(5), 1311-1319.
Boomsma, T.K., Meade, N. & Fleten, S-F. (2012). Renewable energy investments under
different support schemes: A real options approach, European Journal of
Operational Research, forthcoming.
Cullinane, K., Wang, T.-F., Song, D.-W. & Ji, P. (2006). The technical efficiency of
container ports: Comparing data envelopment analysis and stochastic frontier
analysis, Transportation Research Part A: Policy and Practice, 40(4), 354-374.
Ecofys, (2011). Financing Renewable Energy in the European Energy Market, Final
Report,
available
from:[http://ec.europa.eu/energy/renewables/studies/doc/renewables/2011_financing
_renewable.pdf] accessed on: 01/03/2012
20
ECORYS (2010). Assessment of non-cost barriers to renewable energy growth in EU
Member States – AEON DG TREN No. TREN/D1/48 – 2008, available from:
[http://ec.europa.eu/energy/renewables/studies/doc/renewables/2010_non_cost_barri
ers.pdf], accessed on 05.02.2012
European Commission, (2001). Directive 2001/77/EC of the European Parliament and
of the Council on the Promotion of Electricity Produced from Renewable Energy
Sources in the Internal Electricity Market, Official Journal of the European
Communities, L 283/33-40.
European Commission, (2002). Directive 2002/91/EC of the European Parliament and
of the Council on the Energy Performance of Buildings, Official Journal of the
European Communities, L 001/65-71.
European Commission, (2003). Directive 2003/30/EC of the European Parliament and
of the Council on the promotion of the use of biofuels and other renewable fuels for
transport, Official Journal of the European Communities, L 123/42-46.
European Commission, (2009). Directive 2009/28/EC of the European parliament and
of the council of 23 April 2009 on the promotion of the use of energy from
renewable sources and amending and subsequently repealing Directives 2001/77/EC
and 2003/30/EC, Official Journal of the European Union, L 140/16-62.
EREC: European Renewable Energy Council, (2011). Mapping Renewable Energy
Pathways
towards
2020,
EU
Roadmap,
available
from:
[http://www.erec.org/fileadmin/erec_docs/Documents/Publications/EREC-roadmapV4_final.pdf], accessed on 02/03/2012
Eurostat (2012). Statistics database, Available from:
[http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home], accessed on
01/02/2012
Greene, (2007). Limdep, Version 9.0., Econometric modeling Guide, Vol.2
Helvoigt, T. L. & Adams D. M. (2009). A stochastic frontier analysis of technical
progress, efficiency change and productivity growth in the Pacific Northwest
sawmill industry, Forest Policy and Economics, 11(4), 280-287.
Inglesias, G., Castellanos, P. & Seijas, A. (2010). Measurement of productive efficiency
with frontier methods: A case study for wind farms, Energy Economics, 32(5), 11991208.
Jamasb, T. & Politt, M. (2005). Electricity Market Reform in the European Union:
21
Review of Progress toward Liberalization & Integration, The Energy Journal, 26:
11-41.
Jondrow, J., Lovell, K., Materov, I. & Schmidt, P. (1982). On the estimation of technical
efficiency in the stochastic frontier production function model, Journal of
Economerics, 19: 233-238.
Lins, C. (2008). Trends in the global energy scenario-the European perspective, Global
Renewable
Energy
Forum,
Foz
do
Iguacu,
available
from:
[http://www.unido.org/fileadmin/media/documents/pdf/Energy_Environment/rre_br
azilforum_Lins_080519.pdf]
Loyning, L. (2010). Renewable Energy in Europe, Talking Points, United Nations
Intergovernmental
Panel
on
Climate
Change,
available
from:
http://www.rsmi.com/attachments/approved/renewable-energy-ineurope/en/RenewableenergyinEurope.pdf
Managi, S., Opaluch, J. J., Jin, D. & Grigalunas, T. A. (2006). Stochastic frontier
analysis of total factor productivity in the offshore oil and gas industry, Ecological
Economics, 60(1), 204-215.
Menegaki, A. (2011). Growth and renewable energy in Europe: A random effect model
with evidence for neutrality hypothesis, Energy Economics, 33, 257-263.
Menegaki, A. (2012). A social marketing mix for renewable energy in Europe based on
consumer stated preference surveys, Renewable Energy, 39, 30-39.
OPTRES, (2007). Assessment and optimization of renewable energy support schemes in
the European electricity market, Intelligent Energy, Karlsruhe, available
from:[http://ec.europa.eu/energy/renewables/studies/doc/renewables/2007_02_optre
s.pdf], accessed on 01/04/2012.
See, K. F., & Coelli, T. (2012). An analysis of factors that influence the technical
efficiency of Malaysian thermal power plants, Energy Economics, 34(3), 677-685.
Tingley, D., Pascoe, S. & Coglan, L. (2005). Factors affecting technical efficiency in
fisheries: stochastic production frontier versus data envelopment analysis
approaches, Fisheries Research, 73(3), 363-376.
Van Meensel, J., Lauwers, L., Van Huylenbroeck, G. & Van Passel, S. (2010).
Comparing frontier methods for economic–environmental trade-off analysis.
European Journal of Operational Research, 207(2), 1027-1040.
22
Yélou, C., Larue, B. & Tran, K. C. (2010). Threshold effects in panel data stochastic
frontier models of dairy production in Canada, Economic Modelling, 27(3), 641647.
Zervos, A., Lins, C. & Muth, J. (2010). Re-thinking 2050. A 100% Renewable Energy
Vision for the European Union, European Renewable Energy Council (EREC),
available from: http://www.erec.org
23
Appendix
Table 4. Descriptive statistics of variables
Country
Belgium
Bulgaria
Czech Rep.
Denmark
Germany
Estonia
Ireland
Greece
Spain
France
Italy
Cyprus
Latvia
Lithuania
Luxemburg
Hungary
Malta
Netherlands
Austria
Poland
Portugal
Romania
Slovenia
Slovakia
Finland
Sweden
UK
Croatia
Iceland
Norway
Switzerland
GDP
Mean
121.25
33.92
74.74
126.95
117.90
55.20
133.59
89.39
100.10
111.71
110.12
91.50
44.68
48.77
250.39
60.52
79.52
131.39
126.96
51.74
77.24
34.49
83.95
59.58
114.97
123.72
118.02
55.56
128.28
165.70
142.05
St.dev.
3.45
6.51
4.68
3.74
2.89
10.32
9.71
4.23
3.59
3.38
6.85
4.67
7.67
8.15
20.70
4.30
2.31
2.24
3.07
5.11
1.51
8.13
4.22
8.99
2.17
2.17
3.92
5.12
8.24
16.19
5.32
RES
Mean
2.35
6.30
4.28
14.25
5.37
13.68
2.55
6.05
7.96
8.36
6.25
2.55
31.77
10.64
1.78
4.12
0.20
2.87
23.76
5.63
18.02
14.30
12.27
5.23
24.90
33.91
1.59
9.52
72.47
50.52
16.88
St.dev.
1.18
3.22
2.80
4.08
3.03
4.86
1.27
1.51
2.89
2.33
1.25
1.08
1.56
3.99
0.56
2.17
0.00
0.67
3.49
1.87
4.16
5.11
3.12
2.88
3.94
8.91
0.66
1.55
2.66
10.11
1.34
CON
Mean
37989.60
9252.71
25105.60
15170.10
221350.00
2747.79
11320.40
19763.90
86743.60
154960.00
127471.00
1758.64
3846.71
4400.21
3930.79
16766.50
413.50
51143.80
25605.40
59474.50
17880.50
24189.30
4700.14
10713.90
25262.0
33960.10
150094.00
5908.21
2191.71
18418.90
20801.60
St.dev.
1680.90
555.35
989.02
343.57
4200.11
215.69
1335.42
1391.73
9997.28
3828.35
6141.97
157.47
348.24
406.99
462.98
849.34
65.47
1206.99
2035.48
3676.60
1064.24
1806.58
274.41
373.42
1103.65
1113.21
5024.63
528.87
195.04
515.87
560.09
GRE
Mean
95.95
52.96
74.47
99.14
80.57
45.15
121.12
118.60
135.98
97.25
105.47
172.13
42.97
44.40
86.90
66.74
139.38
99.84
108.76
69.85
135.27
52.60
97.52
67.17
104.10
93.48
83.74
92.81
118.52
107.55
98.97
St.dev.
5.76
4.59
3.21
7.92
3.30
3.42
5.24
3.90
12.06
2.98
5.10
14.64
2.37
3.19
11.50
4.15
8.70
4.10
6.21
3.45
9.16
3.93
3.62
3.29
7.96
5.63
5.43
9.16
14.24
2.36
1.57
EMPLO
Mean
60.31
55.07
65.55
75.87
66.72
64.01
64.85
58.67
59.02
62.95
55.85
68.35
61.85
61.31
62.98
56.03
54.44
73.75
69.45
55.75
67.60
60.10
64.90
58.82
67.87
72.63
71.06
54.61
82.78
76.52
78.34
St.dev.
1.56
5.24
0.94
1.14
2.47
3.09
3.02
2.26
4.67
1.52
2.47
1.93
3.56
2.25
1.45
1.17
0.64
2.15
1.58
3.21
0.95
2.63
2.20
1.81
1.83
1.26
0.69
1.57
2.03
1.08
0.65
CAPI
Mean
20.63
20.20
27.00
19.85
18.98
29.63
21.74
17.71
27.01
19.00
20.18
19.19
26.35
21.86
19.94
22.29
19.95
20.36
22.46
21.03
24.04
22.83
25.17
26.88
19.72
17.78
16.80
NA
NA
20.42
21.33
St.dev.
0.91
0.36
1.57
1.07
1.65
4.47
4.05
2.38
2.16
1.23
0.76
1.78
4.19
3.25
0.96
1.59
2.70
1.43
1.32
2.20
2.79
4.15
1.90
4.08
0.86
1.11
0.78
NA
NA
2.39
0.74
N.A.=Not available. The sample of countries was dependent upon data availability by Eurostat at
http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/search_database.
Data come from Eurostat database under the codes tec00114, tsdcc110, ten00095, tsien010, tsdec210
Table 5. Countries' ranking based on the three types of inefficiency
Country
Ranking 1
Conventional
Inefficiency
Ranking 2
Random
Inefficiency
Ranking 3
Management
adapted inefficiency
Belgium
30
1.007
4
0.065
22
-0.561
Bulgaria
25
1.690
11
0.097
19
-1.089
Czech Rep.
24
0.152
2
0.054
16
-1.216
Denmark
17
0.137
22
0.146
20
-0.951
Germany
2
0.084
9
0.078
11
-1.680
Estonia
22
0.144
17
0.110
8
-1.802
Ireland
3
0.100
7
0.068
27
0.204
Greece
16
0.136
3
0.062
23
-0.438
Spain
19
0.139
20
0.126
3
-2.029
France
11
0.115
6
0.066
7
-1.820
Italy
14
0.126
25
0.697
10
-1.709
Cyprus
23
0.150
26
0.715
29
1.649
Latvia
7
0.109
19
0.125
1
-2.577
Lithuania
26
0.175
15
0.106
15
-1.357
Luxemburg
29
0.577
18
0.114
28
0.775
Hungary
9
0.110
24
0.629
18
-1.126
Malta
18
0.138
29
0.881
30
3.425
Country
Ranking 1
Conventional
Inefficiency
Ranking 2
Random
Inefficiency
Ranking 3
Management
adapted inefficiency
Netherlands
15
0.128
5
0.065
24
-0.309
Austria
4
0.102
10
0.082
6
-1.886
Poland
21
0.142
21
0.141
9
-1.716
Portugal
27
0.201
28
0.754
13
-1.558
Romania
28
0.203
23
0.192
2
-2.325
Slovenia
6
0.108
14
0.104
21
-0.938
Slovakia
5
0.107
13
0.102
17
-1.153
Finland
12
0.120
27
0.723
12
-1.680
Sweden
13
0.124
16
0.107
5
-1.895
UK
20
0.141
1
0.000
25
-0.190
Croatia
1
0.000
1
0.000
26
0.000
Iceland
1
0.000
1
0.000
26
0.000
Norway
8
0.109
12
0.099
4
-1.908
Switzerland
10
0.113
8
0.070
14
-1.369
Note: Rankings have been performed from smallest to largest and not based on mean inefficiency.