You are on page 1of 12

Renewable and Sustainable Energy Reviews 57 (2016) 15561567

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews


journal homepage: www.elsevier.com/locate/rser

Socio-economic determinants of energy consumption: An empirical


survey for Greece
Muhammad Azam a,n, Abdul Qayyum Khan b, Eleni Zafeiriou c, Garyfallos Arabatzis d
a
School of Economics, Finance & Banking, College of Business, Universiti Utara Malaysia, Malaysia
b
Department of Management Sciences, COMSATS Institute of Information Technology, Wah Cantt Campus, Pakistan
c
Department of Agricultural Development, Democritus University of Thrace, Greece
d
Department of Forestry and Management, Environment and Natural Resources, Democritus University of Thrace, Greece

art ic l e i nf o a b s t r a c t

Article history: The present study identies and evaluates the macroeconomic factors that determine energy consumption in
Received 4 April 2015 Greece. This study is based on annual time series data covering the period ranging from 19752013. The
Received in revised form Augmented Dickey Fuller (ADF) test and KPSS test are employed in order to test stationarity of the data,
25 October 2015
conrming that all the variables are I(1). Within the next step of the methodology we employed a VECM model
Accepted 17 December 2015
Available online 13 January 2016
in order to detect causality among the variables through the Granger causality test. Similarly, the impulse
response analysis provides that urbanization, infrastructure, trade, income, population growth, and foreign
Keywords: direct investment are the main determinants of energy consumption in Greece during the period surveyed. The
Energy ndings suggest that in order to fulll the energy demand in the country, the decision-makers need to for-
FDI
mulate energy policy in favor of enlarging the supply of energy and especially renewable energy to foster the
Income
process of economic growth and development. Especially, in the times of crisis the particular determinants are
Exports
Urbanization a necessity to initiate economic growth.
Greece & 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1557
1.1. Overview of energy in the Greece . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1557
2. Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1558
3. Data and methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1559
3.1. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1559
3.2. Model specication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1559
3.3. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1560
3.3.1. Descriptive analysis and correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1560
3.3.2. Unit root tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1561
3.3.3. Johansen cointegration test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1562
3.3.4. Estimation of error vector correction models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1562
3.3.5. Granger causality tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1564
3.3.6. Variance decomposition analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1564
3.3.7. Impulse response analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1564
4. Concluding remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1564
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1567

n
Corresponding author.
E-mail addresses: drazam75@yahoo.com (M. Azam), qayyum72@ciitwah.edu.pk (A.Q. Khan), ezafeir@agro.duth.gr (E. Zafeiriou), garamp@fmenr.duth.gr (G. Arabatzis).

http://dx.doi.org/10.1016/j.rser.2015.12.082
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1557

1. Introduction In a similar vein, the study of Tampakis et al. [46] reveals that it
is necessary to exploit more renewable energy sources for socio-
Improved socioeconomic development is desirable for a better economic development and environmental purposes. Therefore,
standard of living and social welfare. The main objective of every renewable energy in the form of wind energy provides several
state is to maintain and sustain higher level of economic devel- advantages for many countries including Greece. After assessing
opment in order to encourage a healthy economy; that is to the different forms of energy, the study nds that most
increase production and fulll society needs without any envir- environment-friendly energy forms are hydroelectric, solar, and
onmental damages. Presumably, energy is an interesting issue of wind energy. The use of renewable energy sources helps to miti-
research since it is the most important input other than capital and gate carbon dioxide (CO2) emissions, increase energy supply, lower
labor. Energy plays a key role as to all other consumptions and the related costs and contribute to domestic development by
production processes. Energy consumption is vital for human generating new employment prospects in Greece. In a recent
being survival and it is one of the basic indicators of eco empirical study, Dagoumasa and Kitsios [15] evaluated the effect of
nomic development and growth. It is generally believed by the the economic crisis on energy poverty in the context of Greece.
researchers that energy plays an indispensable role in the process The study surveyed the relationship between per capita electricity
of economic and social development and enhanced quality of life consumption and economic growth in comparison to other Eur-
both in developed and developing economies [11,12,17,20, opean countries. According to their ndings the standard of living
29,36,38]. It has also been observed that an increase in per capita in Greece has been improved largely compared to other European
energy consumption has a positive impact on the emergence and countries. Although, the economic crisis has a signicant impact
development of civilization [7,31]. Indeed, energy use is vital for on the electricity consumption and on the ability of users to pay
the attainment of more sustainable development and growth, their electricity bills; however, the inability of people to pay the
while irrational consumption is the main reason of global envir- electricity bills on time, leads to liquidity problems for the Public
onmental deterioration, including extensive utilization of renew- Power Corporation (PPC), allowing the menace of transforming an
able resources and pollution caused by fossil [43]. Therefore, the energy poverty problem to an energy security problem for the
main focus of this study is to identify and evaluate factors deter- country.
mining energy consumption in Greece. According to the aforementioned studies energy is validated as
Energy is considered to be the driving force in every economy an essential input for every type of economic activity expanding
and used in different forms, including chemical, electrical, nuclear, social welfare. Despite the evident contribution of energy to eco-
thermal, sound, and motion etc. Energy can be stocked, trans- nomic development, little attention has been paid to the energy
formed and augmented depending upon its application purpose. consumption and its determinants for Greece. Numerous of factors
Similarly, there are different sources of energy such as fossil (i.e., may well affect energy consumption, but this study focuses only
petroleum, coal, natural gas, oil etc.), renewable (biomass, geo- on the effects of some selected variables, namely population
thermal, hydro, solar, wind and marine etc.,) and ssile (thorium growth rate, host country exports, urbanization, and infrastructure
and uranium etc.) [18]. In a study by Mirza et al. [36] it has been on energy consumption in Greece, a European country that gets
revealed that biomass as a source of energy has been used for through a time of recession. In addition, according to our knowl-
thousands of years by the human-being. In Pakistan, traditional edge, there is no comprehensive empirical study available on the
fuels e.g., dung, rewood, and crop residues provide a major share investigation of the energy consumption function in Greece.
in fullling the daily energy needs of rural and low-income urban Therefore, the main objective of this study is to evaluate empiri-
households. The study suggests that although the government of cally various factors explaining energy consumption in the context
Pakistan is nancing multiple projects pertinent to biomass energy of Greece. The broad contribution of this study is twofold: rst-
development, yet extensive efforts are needed for utilizing the full this study will not only shed light on the signicance of energy
potential and taking maximum benets out of this vital resource of and present previous studies in order to make evident that this is
renewable energy in the country. the rst empirical work on the investigation of the energy con-
The energy crises in the 1970s along with the substantial rise in sumption function for Greece, therefore, it will contribute to lit-
energy prices, specically oil prices, have had an unfavorable erature related to the determinants of energy consumption. Sec-
impact on the economic growth particularly in developing coun- ond- the outcomes will enable Greece to formulate policies
tries. However, many of the advanced countries succeeded to have allowing high cost investment (especially on renewable energy),
gradually condensed energy requirements. Subsequently, the contributing to rene of energy security. Moreover, this survey
energy crisis drew the attention for the execution of energy pre- may well provide tools of scal and monetary policy in order the
servation and management processes. Therefore the factors aforementioned factors through the impact on energy consump-
explaining energy consumption and also the linkage between tion to lead to economic development. Explicitly, the empirical
energy consumption and economic growth employing different ndings are expected to guide the decision-makers to chalk out
econometric techniques and using time series and panel data for appropriate, effective and environment-friendly policies to
empirical investigation have been used a subject of extensive enhance energy, which will consequently contribute largely in the
survey [22,23]. The studies of Koutroumanidis et al. [30] and socio-economic development of Greece.
Arabatzis and Malesios [6] have also endorsed that all over history,
energy resources have attained a prominent signicance for the 1.1. Overview of energy in the Greece
economic development and social wellbeing across the world.
Where, the oil crisis of the 1973 along with many environmental The scal crisis of Greece in 2009 turned quickly into an out-
protection challenges, has led several countries to nd new and standing debt crisis, which eventually turned into a complete
alternative energy sources. In particular, two main renewable recession. It has been observed that as a result, the social cost of
energy sources are biomass and fuel wood which can make a the Greek crisis has been recorded as unreasonably high. Conse-
noteworthy contribution, as potential substitutes for oil. Arabatzis quently, the national income in Greece has decreased by roughly a
et al. [7], shed some light on the signicance of renewable energy quarter. The gap in living standard relative to the rest of Western
sources which play a critical role in increasing the energy demand Europe has been increased [34]. The Greece country report [21],
globally. Due to the economic crisis in Greece, fuel wood con- the National Energy Efciency Action Plans (NEEAPs) and experts'
sumption increased extensively, particularly in the rural areas. survey reveal that Greece has neither striving nor an innovative
1558 M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567

energy efciency policy. In addition, a reasonable interpretation 2. Literature review


for energy save is the economic recession, a fact conrmed for the
reference year (i.e., 2010). According to Katsivelis [26], the energy An abundance of empirical studies referring on the energy
consumption of four European Union (EU) countries (i.e., consumption in the context of developed and developing coun-
Germany19.1%, France 15.3%, Italy10%, UK 12.1%) corre- tries, though, these are negligible in the case of Greece. Pre-
sponds to the 56.5% of the total EU energy consumption. Though, dominantly, the aforementioned studies surveyed the linkages
Greece consumes a very small amount of energy which is almost between energy consumption and economic growth. For instance,
1.6% of the total EU consumed energy of 56.5%. The country's the study of Hondroyiannis and Papapetrou [23] conrms the
energy import dependency reaches 65.3% and is ranked 8th out of existence of a long-run relationship between energy consumption,
27 EU countries in terms of energy import dependency. Similarly, price developments (as an index of economic efciency) and real
Greece consumes almost 19 million tons of oil. This corresponds to gross domestic (GDP) variables for Greece during 19601996 by
a turnover of nearly 9 billion Euros, equivalent to 5% of the GDP employing the Vector Error-Correction model (VEC). The study
and evidently necessitates the recruitment of 50,000 employees, suggests a few policies that intend to improve economic efciency,
which is 1% of the total labor force. Moreover, the data statistics and may encourage energy conservation without hindering eco-
demonstrate that energy consumption fell down from 21.9 million nomic growth. In a similar study, at aggregated levels of energy
tons in 2007 to 15.1 million tons in 2016 [5]. The total gross energy consumption the empirical ndings of Tsani [47] conrmed the
demand was covered largely by oil and petroleum products, solid existence of a unidirectional causal relationship from energy
fuels, natural gas and renewable energy with 53.2%, 27.8%, 11.4%, consumption to economic growth in Greece for the time period
and 7.6 % respectively in 2010. Likewise, the electric power 19602006 and with the application of the Toda and Yamamoto
accounted for 27% of the country's total energy demand, where, test (1995). While, in case of disaggregated levels the results
fossil fuel (i.e., oil and gas) reaches almost 69%. A brief comparative indicate that there is a bidirectional causal connection between
analysis of energy consumption and production of Greek economy industrial and residential energy consumption to economic
with those of China, EU and World is reported in Table 1. growth but results are not for the transport energy consumption.
In another study, Kottari [28] focuses on the signicance of According to the study of Koutroumanidis et al. [30] the auto-
renewable energy source which plays a vital role in Greece's regressive integrated moving average (ARIMA) models and arti-
energy production prole. The study reveals that the existent cial neural networks (ANN) hybrid model provide the best fore-
investment framework needs a prominent expansion in energy casts. The policy makers may well use these results for a better
from wind, solar, geothermal, and biomass/biofuels energy pro- planning of the production and fuel wood market in Greece.
duction, which are likely to help progressively as a transport fuel. The study of Acaravci and Ozturk [1] investigates causal rela-
Greece is also expected to generate 40% of its total electrical power tionships between CO2 emissions per capita, energy consumption
from renewable energy source by 2020. The study further added per capita, real GDP per capita and the square of real GDP per
that due to the development of solar power plant in Greece, within capita by using autoregressive distributed lag (ARDL) bounds
the context of the debt crisis, solar power plant development and testing approach of cointegration for nineteen EU countries. The
particularly the HELIOS project need to attract more foreign bounds F-statistic for cointegration test reveals the existence of a
investments, which in the result would help to create job oppor- long-run relationship between a set of variables for Greece along
tunities and will largely stimulate economic growth of Greece. with some other EU countries. Belke et al. [10] empirically esti-
The rest of the study is structured as follows. Section 2 briey mate the long-run connection between energy consumption,
reviews existing empirical studies on the energy consumption, energy prices and real GDP for 25 OECD countries including
where more emphasis has been given on empirical studies in the Greece during 19812007. The causality tests conrmed the exis-
context of Greece. Section 3 refers to the data and methodological tence of a bi-directional causal linkage between energy con-
issues. Section 4 interprets the empirical results and the conclu- sumption and economic growth, while, the results indicate that
sions of the study. energy consumption is price-inelastic during the period under the

Table 1
Comparative analysis of energy consumption and production.
Source: [13].

Energy Greece China European Union World

a b c
Electricity production 54.98 billion kWh 5.65 trillion kWh 3.26 trillion kWh 22.29a trillion kWh
Electricity consumption 52.02d billion kWh 5.523b trillion kWh 2.798d trillion kWh 20.98a trillion kWh
Natural gas proved reserves 991.1b million cu m 3.3b trillion cu m 1.573b billion cu m 206.4b trillion cu m
Natural gas production 5c million cu m 947c million cu m 146.8c billion cu m 3.488a trillion cu m
Electricity installed generating capacity22.3d million kW 1.505b billion kW NA 5.752a billion kW
Electricity from fossil fuels 70.4%d of total installed 67.3%b of total installed 50.5%a of total installed 67.2%a of total installed
capacity capacity capacity capacity
Electricity from hydroelectric plants 11.4%d of total installed 22.2%b of total installed 16.3%a of total installed 18.2%a of total installed
capacity capacity capacity capacity
Electricity from other renewable sources 15.1%d of total installed 9%b of total installed 19.6%a of total installed 7.5%a of total installed
capacity capacity capacity capacity
Natural gas consumption 3.6c billion cu m 181.8b billion cu m 438.1c billion cu m 3.408a trillion cu m
Rened petroleum products production 498,000c bbl/day 9.648c million bbl/day 11.62c million bbl/day 84.52e million bbl/day
Rened petroleum products consumption 284,000c bbl/day 10.76c million bbl/day 12.77c million bbl/day 90.05c million bbl/day
CO2 emissions from energy consumption 78.8c million Mt 10c billion Mt 3.914c billion Mt 33.95d billion Mt

a
2011 estimated.
b
2014 estimated.
c
2013 estimated.
d
2012 estimated.
e
2010 estimated.
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1559

study. Arabatzis and Malesios [6] collected primary data from 385 and high income countries. The study nds that the inverted U-
households or individuals, which took place in 2009 (20082009 shaped association exists between per capita income and energy
heating season) in Grevena prefecture, which is situated in consumption intensity. Sbia et al. [45] nd that incoming FDI,
Northern Greece. The empirical investigation of the study is based trade and CO2 emissions cutback energy demand, while economic
on several household fuelwood consumption models includes growth and clean energy have constructive and signicant inu-
GLM, Tobit and Heckman regression-based techniques. The nd- ence on energy consumption in UAE during 1975Q12011Q4.
ings of the study exhibit that household sociological and eco- Similarly, several previous studies empirically examine the rela-
nomical characteristics as well as overall environmental issues are tionship between urbanization and energy consumption, where
appropriate to explicate variances towards fuelwood consumption these studies have found a signicantly positive relationship
for space cooking and heating Grevena prefecture part of Northern between urbanization and energy consumption [25,33,37,39,41].
Greece. In a similar vein, the empirical study of Salima and Shaei [44]
The empirical ndings of Fuinhas and Marques [19] show the depicts that both population and urbanization variables have
presence of bidirectional causality between energy consumption positive impacts on non-renewable energy consumption, while,
and economic growth in both the long-run and short-run, for population density has a negative effect on non-renewable energy
Portugal, Italy, Greece, Spain and Turkey during 19652009 and consumption in OECD countries over the period from 19802011.
used ARDL bounds test approach. Georgantopoulos [22] carried Azam et al. [9] observe that inward FDI, economic growth, trade
out an empirical study in order to explore for the presence and openness and human development index have signicantly posi-
direction of causality between electricity consumption and real tive effects on energy use in three ASEAN countries (Indonesia,
GDP for Greece during 19802010. The empirical results conrm Malaysia and Thailand) for the time period 19802012.
the existence of cointegrating relationship between the variables Having presented the aforementioned empirical studies the
under review, where the VAR/VEC approach indicates that uni- novelty of the present study stands on the fact of a survey of
directional causal linkage exists which runs from capital formation energy consumption function for Greece intending to examine
and electricity consumption to real GDP in the short-run sug- socio-economic factors that inuence energy consumption in
gesting that the Greece's economy is largely energy dependent. Greece.
The results further reveal that rising consumption of electricity
and positive economic growth rates are expected in the future.
Dergiades et al. [16] explore the linear and non-linear causal 3. Data and methodology
relationship between total energy consumption and economic
growth for Greece during 19602008. The nding suggests sig- 3.1. Data
nicant unidirectional both linear and non-linear causal relation-
ship running from total useful energy consumption to economic The present paper employs annual time series data ranging
growth, implying that the impact of real GDP on energy con- from 1975 to 2013 for the achievement of the objective. The data
sumption was found statistically insignicant. were obtained from the World Development Indicator (2014), the
Similarly, Dritsaki and Dritsaki [17] note that numerous prior World Bank database (http://data.worldbank.org/news/release-of-
studies have made an effort to examine the direction of causality
world-development-indicators-2014). To be more specic, the
between energy consumption, economic growth and CO2 emis-
data employed include the following variables; per capita energy
sions mainly in developing economies. However, their study
consumption in kg of oil equivalent (EC), urbanization growth rate
focused only on three countries of Southern Europe namely
(Ugr), total population growth rate (Tpop), infrastructure (Infr),
Greece, Spain, and Portugal during 19602009 and used the panel
Gross Domestic Product growth rate (GDPg), trade as a percentage
unit root tests, panel cointegration methods and panel causality
of GDP Trd and foreign direct investment of Greece (FDI). The
test to investigate the relationship between the energy con-
majority of the aforementioned variables are represented by
sumption, economic growth and CO2 emission. The empirical
proxies that will be subtle described in the next section.
results show that there is a short-run bi-directional causal relation
between the observed variables. While, in the long-run, a uni-
3.2. Model specication
directional causality running from CO2 emissions to energy con-
sumption, and economic growth is validated, as well as a bidir-
In this study, we adopt a single multivariate framework,
ectional causality between energy consumption and economic
methodology where energy consumption is the response variable
growth in Greece, Spain, and Portugal. Bilgen [11] expounds that
[2,4,8,14,22,24,32,3]1. All the variables employed are transformed
there is a close association between energy consumption and
into semi log-linear form and thus a multiple regression model is
economic growth. In a study Ucan et al. [48] investigate the link-
age between renewable and non-renewable energy consumption used in order to examine the impacts of various determinants of
and economic growth for a panel of fteen EU countries during energy consumption, namely urbanization, Gross domestic pro-
19902011. The ndings reveal that there is a long-run equilibrium duct growth rate (income), Total population growth rate, a proxy
liaison between real GDP, renewable and non-renewable energy for infrastructure (utilization of landline telephone per 1000
consumption, greenhouse gas emissions and research and devel- people), trade as a percentage of GDP and foreign direct invest-
opment. Moreover, the Granger-causality results validate the uni- ment on energy consumption. The evolution of the aforemen-
directional causality between non-renewable energy use and tioned variables is depicted in Fig. 1.
real GDP. In terms of mathematics the proposed energy function for
Some erstwhile studies conducted in a few countries excluding Greece is provided by the following formula:
Greece, empirically investigated the various factors that impact EC t f Ugr t; Tpop t; Inf r t; Trd t; GDP g t; FDI t 1
energy consumption; for example, Mielnik and Goldemberg [35]
nd that a decrease in energy intensity is allied with an expansion EC indicates per capita energy consumption in kg of oil
in incoming FDI for 20 developing economies during 19871998. equivalent, Ugr denotes urbanization growth rate, Tpop is total
The ndings of Pu-yang et al. [42] exhibit that FDI inows can
stimulate energy efciency, decline energy consumption intensity 1
See these studies for omitted variables mis-specication problems of
while using data over the period of 19852008 for 74 low, middle the model.
1560 M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567

Fig. 1. The evolution of the variables employed in the study to be here.

population growth rate, Infr is infrastructure, GDPg is Gross combination with the assistance of all the other aforementioned
Domestic Product growth rate, Trd is trade as a percentage of GDP are econometric tools employed in our study. Though, before the
and FDI denote foreign direct investment of Greece. The sign of the implementation of the aforementioned econometric tools the next
coefcients are necessary in order to quantify the impact of the section presents the descriptive statistics of the variables as well as
aforementioned variables on the energy consumption (Fig. 2). the correlation among the variables employed.

3.3. Methodology
3.3.1. Descriptive analysis and correlation
A brief summary of the descriptive statistics is presented in
The present study involves the specication of the relationship
Table 2. According to our ndings the most important refers to the
between energy consumption and different socio-economic indi-
actual value deviation from the mean value for the ve variables
cators in order to evaluate the particular model. Different unit root
tests were employed in order to determine the rank of integration employed in or study. To be more specic, the value of standard
for the variables employed that will be tested also for cointegra- deviation for the variables of energy consumption and urbaniza-
tion. The estimation of Vector Error Correction Model is the step tion growth is equal to 0.222 and 0.622 respectively. Furthermore
post to the determination of the number of the cointegration the same statistics for gross domestic product growth rate total
vectors, while the variance decomposition and impulse response population growth rate, trade as a percentage of GDP and foreign
analysis are the nal steps of the present methodology employed. direct investment is equal to 3.556, 12.61, 0.652, 6.577 and 0.392
Additionally, we estimated the causal relationships among the respectively. The rest of the results derived can be observed in
variables in order to become able to realize an effective strategy in Table 2.
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1561

Fig. 2. Variance decomposition graphs.

Table 2 while population growth has a signicant positive relationship


Descriptive statistics analysis. with trade. Infrastructure has a signicant positive relationship
Variables Minimum Maximum Mean Standard deviation
with total population and trade. FDI has positive but insignicant
linkage with urban growth and income.
LnEC 7.17 7.91 7.6309 0.22290
Ugr  0.15 2.43 0.8499 0.62201 3.3.2. Unit root tests
GDPg  7.11 7.25 1.5658 3.55691
Before proceeding with any empirical estimation of the model,
Infr 18.65 57.56 41.3440 12.61148
Tpop 9.05 11.19 10.4148 0.65387 it is indispensable to examine the time series data. The data ana-
Trd 40.29 65.33 51.7863 6.57745 lysis presupposes the stationarity the time series employed. Dis-
FDI 0.04 2.07 0.8903 0.39183
tribution of data is analyzed through descriptive statistics and
correlations among the variables are analyzed by using a Pearson
Table 3 illustrates the Pearson correlation result for correlation correlation matrix. Augmented DickeyFuller (ADF) test examines
among the seven variables used in the study. It is evident from the hypothesis that the variable in question has a unit root while
Table 3 that energy consumption has statistically signicant rela- KPSS test examines the null hypothesis of stationarity. Two tests
tionship with urbanization, total population growth, infrastructure are employed for stationarity since ADF is characterized by limited
and trade in Greece. Energy consumption has also positive, but a power. If the series has a unit root, then we employ the data is
statistically insignicant connection with gross domestic product, appropriate to make it stationary, in parliamentary procedure to
and negative and insignicant relationship with foreign direct obviate the problem of spurious regression arising from non-
investment. It is observed that urbanization has a signicant stationarity in the time series. Thus, results of unit root tests are
negative relationship with total population growth, infrastructure reported in Table 4. All the seven variables included in the study
and trade, and positive signicance relationship with income, are non-stationary at a level when the intercept is included only,
1562 M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567

Table 3
Pearson correlation analysis.

Variables LnEC Ugr GDPg Infr Tpop Trd FDI

LnEC Pearson correlation 1  0.726nn  0.135 0.969nn 0.971nn 0.627nn  0.092


Sig. (2-tailed) 0.000 0.414 0.000 0.000 0.000 0.580
Ugr Pearson correlation 1 0.501nn  0.740nn  0.793nn  0.679nn 0.160
Sig. (2-tailed) 0.001 0.000 0.000 0.000 0.239
GDPg Pearson correlation 1  0.149  0.285  0.276 0.128
Sig. (2-tailed) 0.365 0.079 0.089 0.438
Infr Pearson correlation 1 0.967nn 0.557nn  0.213
Sig. (2-tailed) 0.000 0.000 0.192
Tpop Pearson correlation 1 0.676nn  0.152
Sig. (2-tailed) 0.000 0.396
Trd Pearson correlation 1 0.011
Sig. (2-tailed) 0.948
FDI Pearson correlation 1
Sig. (2-tailed)

nn
Show that correlation is signicant at the 0.05 level (2-tailed).

Table 5
Table 4 Results of KPSS test.
Results of ADF unit root test.
Variable KPSS test P-value Result
Variables Include intercept only Include intercept and trend Result
ECT 1.8632(1) p-value 0.01 Non-stationary
Test statisticsa Critical value Test statisticsa Critical 0.4851(1) p-value 0.06502 Stationary I(1)
value Urb KPSS level1.1739, (1) p-value 0.01 Non-stationary
KPSS level0.0961, (1) p-value 0.1 Stationary I(1)
EC  2.0004[1]b  3.6171  0.313983  4.2165 I(1)* GDPg KPSS level0.2974, (1) p-value 0.1 Stationary
 4.7699c[0]  3.6171  5.309279  4.2242 I(1)** Infr KPSS level1.2974, (1) p-value 0.01 Non-stationary
Ugr  2.7582[1]  3.6171  1.609194  4.2165 I(1)* KPSS level0.2007, (1), p-value 0.1 Stationary I(1)
 4.9192[0]  3.6171  4.977426  4.2242 I(1)** Tpop KPSS level1.3138 (1), p-value 0.01 Non-stationary
GDPg  2.0298[1]  3.6171  2.132661  4.2165 I(1)* KPSS level0.1746, (1), p-value 0.1 Stationary I(1)
 5.6962[0]  3.6228  5.682838  4.2242 I(1)** Trd KPSS level1.4376, (1), p-value 0.01 Non-stationary
Infr  1.9159[1]  3.6171  0.104618  4.2165 I(1)* KPSS level0.0747, (1), p-value 0.1 Stationary I(1)
 4.0311[0]  3.6171  4.905604  4.2242 I(1)** FDI KPSS level1.8222, (1), p-value 0.01 Non-stationary
Tpop  0.9583[1]  3.6171 1.836872  4.2165 I(2)* KPSS level0.0234, (1), p-value 0.1 Stationary I(1)
 0.4500[0]  3.6171  1.234295  4.2242 I(2)**
 4.9.92[0]  3.6228  4.878411  4.2324 
Trd  2.4666[1]  3.6171  3.541533  4.2165 I(1)*
 5.6645[1]  3.6171  5.540086  4.2242 I(1)**
Table 6
FDI  2.8382[2]  3.6228  4.234700  4.2242 I(1)
Johansen cointegration test results.
 4.0197[0]  3.6289   I(0)

a
N. hypothesis A. hypothesis Maximum likelihood ratio test
Figures in square brackets besides each statistic represent optimum lags
selected using the minimum AIC value.
b
Statistics Critical value
Figures in square brackets besides each statistics represent optimum lags,
selected using the minimum AIC value.
c
r 0 r 1 188.4800nn 133.57
Figures in sub second row of main row represent value of variable at rst
r r1 r 2 124.3028nn 103.18
difference.
*
r r2 r 3 88.75936nn 76.07
Result when only intercept is included only.
**
r r3 r 4 57.92080nn 54.46
Results when both intercept and trend is included.
r r4 r 5 30.23787 35.65
r r5 r 6 12.87976 20.04
but when the trend is included in the model then FDI becomes r r6 r 7 2.605592 6.65
stationary at level. All the variables become stationary in both the nn
Show that signicant at 5% level.
cases (when the intercept and is only included when both inter-
cept and trend is included) by taking the rst difference except
null hypothesis are rejected as Maximum Likelihood Ratio statis-
total population which becomes stationary after taking second
tics are greater than the critical value at 1 percent level.
difference. This result indicates that all the time series employed
Thus according to the results we conrm the existence of at
are integrated of rank 1. least one cointegration relationship of the socio economic vari-
In addition, we implemented another stationarity test KPSS ables with energy consumption, at 5% level. Therefore, we may say
that is considered a more reliable test through which we test that, for the most part, the socioeconomic factors of Greece have
stationarity of the variables employed. The results of the test are signicant long-term impact on energy consumption.
provided in the next Table 5.
3.3.4. Estimation of error vector correction models
3.3.3. Johansen cointegration test Given the fact that there is not a sole cointegration relationship
The next step in our analysis involves the implementation of that connects the variables under review, and the non stationarity
the Johansen cointegration test. The results of the implementation of the variables does not allow us to employ an estimation rela-
of the Johansen co-integration test are provided in Table 6. tionship with the assistance of OLS we estimated the VECM
According to our ndings it is conrmed the existence of four co- models based on the existence of four cointegrating vectors.
integrating relationships among the variables given that rst four Explicitly, for those cases and based on Johansen technique we
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1563

Table 7
VECM model estimates.

Error correction: D(ECT) D(FDI) D(GROWTH) D(INFR) D(TPOP) D(TRADE) D(URB)

Error correction term  1.33167  0.00292  0.03102 0.020563  0.00084 0.022555  0.00016
(0.23072) (0.00217) (0.01026) (0.01094) (0.00057) (0.01690) (0.00076)
[  5.77171] [  1.34890] [  3.02396] [1.87951] [  1.47302] [1.33436] [  0.20220]
D(ECT(  1)) 0.705387 0.003406 0.031967  0.00291 0.001383  0.0242 0.001296
(0.21735) (0.00204) (0.00966) (0.01031) (0.00053) (0.01592) (0.00072)
[3.24542] [1.66870] [3.30849] [  0.28253] [2.58722] [  1.51965] [1.79934]
D(ECT(  2)) 0.687746 0.001677 0.008100  0.00115 0.001849  0.01255 0.001950
(0.19857) (0.00186) (0.00883) (0.00942) (0.00049) (0.01455) (0.00066)
[3.46345] [0.89907] [0.91758] [  0.12233] [3.78710] [  0.86268] [2.96348]
D(FDI(  1))  131.642  1.11607  2.64965 6.042642  0.03144 1.937031 0.226503
(65.2672) (0.61291) (2.90142) (3.09485) (0.16051) (4.78169) (0.21623)
[  2.01697] [  1.82093] [  0.91323] [1.95248] [  0.19585] [0.40509] [1.04751]
D(FDI(  2))  24.3384  0.37277  1.28353 1.472435 0.027692 1.827840 0.143661
(41.8256) (0.39278) (1.85933) (1.98329) (0.10286) (3.06428) (0.13857)
[  0.58190] [  0.94906] [  0.69032] [0.74242] [0.26922] [0.59650] [1.03676]
D(GROWTH(  1))  4.39886  0.06307  0.44254  0.36843  0.002 0.595571  0.0006
(5.73468) (0.05385) (0.25493) (0.27193) (0.01410) (0.42014) (0.01900)
[  0.76706] [  1.17122] [  1.73593] [  1.35487] [  0.14184] [1.41755] [  0.03133]
D(GROWTH(  2))  11.2537  0.06616  0.46861  0.26859  0.03129 0.589217  0.0288
(5.20648) (0.04889) (0.23145) (0.24688) (0.01280) (0.38144) (0.01725)
[  2.16148] [  1.35313] [  2.02465] [  1.08793] [  2.44340] [1.54470] [  1.66977]
D(INFR(  1))  44.7212  0.20016  1.12534 0.524844  0.046 0.436628  0.0513
(8.15704) (0.07660) (0.36262) (0.38679) (0.02006) (0.59761) (0.02702)
[  5.48252] [  2.61299] [  3.10339] [1.35692] [  2.29306] [0.73062] [  1.89812]
D(INFR(  2))  20.2129  0.07462  0.09601  0.08389  0.01497 0.892065  0.00864
(6.91802) (0.06497) (0.30754) (0.32804) (0.01701) (0.50684) (0.02292)
[  2.92178] [  1.14861] [  0.31218] [  0.25573] [  0.87974] [1.76007] [  0.37697]
D(TPOP(  1)) 285.6954 1.877768  10.4153  11.9947 0.760572 2.611003  0.01867
(166.456) (1.56316) (7.39970) (7.89302) (0.40937) (12.1951) (0.55147)
[1.71634] [1.20127] [  1.40753] [  1.51966] [1.85793] [0.21410] [  0.03386]
D(TPOP(  2))  210.689  1.08063 13.19998  13.9418  0.54455 11.38480  0.77721
(256.872) (2.41224) (11.4191) (12.1804) (0.63173) (18.8193) (0.85102)
[  0.82021] [  0.44798] [1.15596] [  1.14461] [  0.86200] [0.60495] [  0.91328]
D(TRADE(  1))  16.2499  0.02344  0.50545 0.028309  0.00204 0.362969  0.0092
(4.26567) (0.04006) (0.18963) (0.20227) (0.01049) (0.31252) (0.01413)
[  3.80946] [  0.58521] [  2.66547] [0.13996] [  0.19435] [1.16144] [  0.65103]
D(TRADE(  2))  14.4729  0.04358 0.044878  0.06887  0.02182 0.428344  0.02325
(5.46132) (0.05129) (0.24278) (0.25897) (0.01343) (0.40011) (0.01809)
[  2.65007] [  0.84976] [0.18485] [  0.26595] [  1.62468] [1.07056] [  1.28479]
D(URB(  1))  33.2912  0.45799 13.77385 5.287924  0.34201  2.38001 0.144015
(139.231) (1.30749) (6.18942) (6.60206) (0.34241) (10.2005) (0.46127)
[  0.23911] [  0.35028] [2.22538] [0.80095] [  0.99883] [  0.23332] [0.31221]
D(URB(  2)) 250.8195 1.211967  7.63114 9.550220 0.527987  13.3062 0.629655
(195.322) (1.83423) (8.68290) (9.26178) (0.48035) (14.3099) (0.64710)
[1.28414] [0.66075] [  0.87887] [1.03114] [1.09916] [  0.92986] [0.97304]
C 55.00254 0.122499  0.08964 0.105214  0.05804 0.156746  0.07811
(15.4808) (0.14538) (0.68819) (0.73407) (0.03807) (1.13417) (0.05129)
[3.55295] [0.84263] [  0.13026] [0.14333] [  1.52444] [0.13820] [  1.52291]
R-squared 0.881940 0.824960 0.861968 0.744417 0.773856 0.755670 0.745355
Adj. R-squared 0.704849 0.562401 0.654921 0.361042 0.434639 0.389174 0.363388
Sum sq. resids 18432.48 1.625511 36.42616 41.44502 0.111483 98.93621 0.202314
S.E. equation 39.19235 0.368048 1.742272 1.858427 0.096386 2.871356 0.129844
F-statistic 4.980169 3.141994 4.163140 1.941748 2.281303 2.061879 1.951357
Log likelihood  142.999 1.709465  46.4872  48.488 43.24492  61.9747 34.00770
Akaike AIC 10.45157 1.115518 4.224984 4.354064  1.56419 5.224172  0.96824
Schwarz SC 11.33046 1.994414 5.103879 5.232959  0.68529 6.103067  0.08934
Mean dependent 18.52321  0.02832  0.24284 0.895577  0.04071 0.306636  0.05241
S.D. dependent 72.14056 0.556373 2.965900 2.324928 0.128189 3.673908 0.162736

selected the cointegrating vector characterized by greater eigen- rst lag of FDI. Thus, for the case of Greece we conrm as statistically
value and the results derived are reported in Table 7. signicant, the impact of all macroeconomic variables on energy
Having estimated the VECM and according to our ndings when consumption. Evidently, this result has many policy implications for
growth and energy consumption are the dependent variables then the case of Greece. The result derived is an expected one given that
the error correction term is statistically signicant a result that Greece is an oil-importing country and therefore all the economic
validates the existence of a long run relationship among the vari- activity is strongly related to the energy consumption. Furthermore
ables of the model. Furthermore regarding the equation for which for the case of growth and the short term dynamics we may con-
the energy consumption is the dependent variable it is impressive clude that energy consumption, infrastructure and urbanization
and the results taken concerning also the short term dynamics. To be seem to have a statistically signicant effect on growth.
more specic, for that model all the variables are statistically sig- Another worth mentioning result concerning the short term
nicant not only the rst but also the second lag with exception the dynamics involves the impact of energy consumption and
1564 M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567

infrastructure and growth to total population. Moreover, as a last independent variables, and conduct the Johansen cointegration
but certainly not least result refers to the high values of R squares test among these variables over a period of 10 years. First of all the
while no problems of heteroscedasticity or serial correlation are volatility in energy consumption is mainly attributed to the vola-
detected. tility of growth as expected with opposite signs though. The
volatility in infrastructure is closely related to volatility in FDI and
3.3.5. Granger causality tests urbanization that stand for a reasonable explanation. In addition
For the case of VECM we employed a causality test (Granger the population growth is closely related to growth with opposite
causality) in order to conrm the impact of each variable on the signs. Growth and infrastructure may well interpret the volatility
other (uni-directional or bi-directional). The results are presented in trade. The volatility in the variable of urbanization may well be
in the next Table 8. According to the results a bidirectional rela- interpreted by population growth while the impact of all the other
tionship is conrmed among energy consumption and growth as variables is limited.
expected. Infrastructure and trade are related with a uni-
directional relationship with energy consumption. 3.3.7. Impulse response analysis
All the aforementioned results are compatible with stylized The next step in our analysis involves impulse response ana-
lysis. To be more specic, we employed the generalized impulse
facts of other developed countries, while we may have a more
responses as described by Pesaran and Shin [40] and Koop et al.
complete picture of the situation with the assistance of other
[27]. The particular tool provides a description of the dynamics of
econometric tools like impulse response and variance decom-
the reaction of all the variables to a standard deviation shock to
position analysis.
the residual of one of them. In Fig. 3 we can see the graphs of
impulse response analysis for a ten year time period.
3.3.6. Variance decomposition analysis
According to the rst graph it is evident that a shock of one
In order to compare the contribution extents of Greece's var-
standard deviation to energy consumption leads to changes of dif-
ious socio-economic indicators to the change in the consumption
ferent signs to the other variables. That is for growth within the rst
of energy, the variance decomposition approach is adopted over two time period it increases with high speed, while then the slope
the sample period. First, we take the energy consumption as the of the curve though increasing becomes less sharp. The three time
dependent variable, while economic indicators together as period is followed by a 5 period time that is decreasing and then it
stabilizes. The same pattern can be observed for foreign direct
Table 8 investment while infrastructure follows the opposite pattern. The
Results of Granger causality tests.
variable representing the total population growth presents a slight
Null hypothesis: Obs F-statistic Prob. volatility due to the change in the energy consumption. Negative is
the change in trade within the ten time period and stable after the
FDI does not Granger Cause ECT 33 0.71524 0.4978 fth period. In the second graph of Fig. 3 we can see that the
ECT does not Granger Cause FDI 2.32770 0.1161
response of all the variables to one standard deviation of FDI is
GROWTH does not Granger Cause ECT 36 4.10306 0.0262
ECT does not Granger Cause GROWTH 2.64662 0.0869 slight with different signs while it disappears after four time period.
INFR does not Granger Cause ECT 36 5.63509 0.0082 The response of urbanization and total population growth on the
ECT does not Granger Cause INFR 1.05003 0.3620 other variables is similar to that of energy consumption to the other
TPOP does not Granger Cause ECT 36 0.32195 0.7271 variables. A shock of one standard deviation to growth leads to
ECT does not Granger Cause TPOP 1.07733 0.3529
TRADE does not Granger Cause ECT 36 0.18648 0.8308
changes the graphs (Graphs 13).
ECT does not Granger Cause TRADE 2.99962 0.0644
URB does not Granger Cause ECT 36 1.28251 0.2916
ECT does not Granger Cause URB 2.20892 0.1268 4. Concluding remarks
GROWTH does not Granger Cause FDI 34 0.48494 0.6206
FDI does not Granger Cause GROWTH 0.77257 0.4711
INFR does not Granger Cause FDI 34 5.61888 0.0087 This study made an attempt to empirically evaluate the impacts
FDI does not Granger Cause INFR 0.45673 0.6378 of some selected explanatory variables namely urbanization
TPOP does not Granger Cause FDI 34 1.65803 0.2081 growth rate, infrastructure, population growth rate, trade, gross
FDI does not Granger Cause TPOP 0.02649 0.9739 domestic product and foreign direct investment on energy con-
TRADE does not Granger Cause FDI 34 2.79305 0.0777
FDI does not Granger Cause TRADE 1.11644 0.3411
sumption in the context of Greece. Stationarity of every variable is
URB does not Granger Cause FDI 34 1.26966 0.2961 surveyed with ADF unit root test, while we also employed KPSS
FDI does not Granger Cause URB 0.06189 0.9401 test. According to our ndings, the majority of the variables are I
INFR does not Granger Cause GROWTH 37 0.90624 0.4142 (1) with exception that of GDP that is found to be stationary. The
GROWTH does not Granger Cause INFR 0.51722 0.6011
Johansen co-integration results indicate that there is at least one
TPOP does not Granger Cause GROWTH 37 0.21668 0.8064
GROWTH does not Granger Cause TPOP 1.29799 0.2871 co-integrating relationship in the long run among the variables.
TRADE does not Granger Cause GROWTH 37 0.70948 0.4995 Further, we estimated the error correction model based on one of
GROWTH does not Granger Cause TRADE 0.41005 0.6671 the four cointegrating vectors (selected with the eigenvalue as a
URB does not Granger Cause GROWTH 37 0.03258 0.9680 criterion) in order to detect the relationships among the variables
GROWTH does not Granger Cause URB 2.99437 0.0643
TPOP does not Granger Cause INFR 37 0.10483 0.9008
in the short run as well. According to the results when energy
INFR does not Granger Cause TPOP 0.07957 0.9237 consumption and growth are the dependent variables the error
TRADE does not Granger Cause INFR 37 0.36044 0.7002 correction terms are statistically signicant implying the existence
INFR does not Granger Cause TRADE 1.85786 0.1724 of a relationship among the variables employed in the long term.
URB does not Granger Cause INFR 37 0.32494 0.7249
Regarding the results in the short term when growth end energy
INFR does not Granger Cause URB 0.01279 0.9873
TRADE does not Granger Cause TPOP 37 2.72637 0.0807 consumption are the dependent variable almost all the macro-
TPOP does not Granger Cause TRADE 2.36426 0.1102 economic variables play an important role in their value. In order
URB does not Granger Cause TPOP 37 1.56943 0.2238 to detect the causal relationships among the variables in pair we
TPOP does not Granger Cause URB 1.62520 0.2127 employed causality tests through which we conrmed a bidirec-
URB does not Granger Cause TRADE 37 0.82053 0.4492
TRADE does not Granger Cause URB 1.49972 0.2385
tional relationship among energy consumption and economic
growth while infrastructure causes energy consumption and
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1565

Fig. 3. Impulse response graphs.

Histogram Normal P-P Plot of Regression Standardized Residual


Dependent Variable: LnEC Dependent Variable: LnEC
10
1.0
8
Expected Cum Prob
Frequency

0.8
6

4 0.6

2
0.4
Mean =5.99E-14
Std. Dev. =0.918
0 N =39
-2 -1 0 1 2 3
0.2
Regression Standardized Residual

Graph 1. Histogram of residual. 0.0


0.0 0.2 0.4 0.6 0.8 1.0
energy consumption has an impact on the volume of trade. In
Observed Cum Prob
addition and a next step in our analysis involved a variance
decomposition analysis in order to detect whether the volatility of Graph 2. PP plot of standardized residual.
each variable is interpreted by the others. Based on our ndings
the volatility of energy consumption that is our main concern is the impulse response analysis. Actually, that is the most important
interpreted mainly by growth and infrastructure in the long term. one for the objective of the present survey. According to our
As a last but certainly not least econometric tool implemented, is ndings, a shock of one standard deviation to energy consumption
1566 M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567

Box-2:Ugr
Box-1:LnEC
2.50
8.00

2.00
7.80
1.50
7.60
1.00

7.40
0.50

7.20 0.00

7.00 -0.50

LnEC Ugr

Box-3: GDPg Box-4: Infr


10.00 60.00

50.00
5.00

40.00
0.00
30.00

-5.00
20.00

-10.00 10.00

GDPg Infr

Box-5: Tpop
11.50
Box-6: Trd
70.00

11.00 65.00

10.50 60.00

55.00
10.00
50.00
9.50
45.00

9.00 40.00

Tpop Trd

Box-7 FDI
2.50

2.00

1.50

1.00

0.50

0.00

FDI

Graph 3. Outliers of variables. Note: 1. Pachauri (2007); Mirza et al. (2008); Koroneos et al., (2011);Georgantopoulos (2012);Bilgen (2009, 2014) and Dritsak&Dritsaki (2014)
2. Kum et al. (2012) and Arabatzis et al. (2012). 3. Rennings et al. (2013) 4. Ferreira (2013) 5. Matsaganis (2013) 6. See also Annual Energy Outlook (AEO2014) 7. See these
studies for omitted variables mis-specication problems of the model (Cole & Elliott, 2003; Ang, 2009; Halicioglu, 2009; Jalil & Mahmud, 2009; Apergis& Payne, 2009, 2010;
Lean & Smyth, 2010; Arouri et al., 2012)
M. Azam et al. / Renewable and Sustainable Energy Reviews 57 (2016) 15561567 1567

leads to changes of different signs to the other variables. But the [19] Fuinhas JA, Marques AC. Energy consumption and economic growth nexus in
most important nding is the response of energy consumption to Portugal, Italy, Greece, Spain and Turkey: an ARDL bounds test approach
(19652009). Energy Econ 2012;34:5117.
the volatility of all other variables employed in our study with [20] Georgantopoulos A. Electricity consumption and economic growth: analysis
different time of reaction, a result with an abundance of policy and forecasts using VAR/VEC approach for Greece with capital formation. Int J
implications. Energy Econ Policy 2012;2(4):26378.
[21] Greece Country Report. Energy efciency in Europe: assessment of energy
The empirical ndings of this study appear to be useful for efciency action plans and policies in EU Member States. Supported by
onward policy consideration. The ndings suggest that policy Intelligent Energy Europe. www.energyefciencywatch.org; 2011.
makers need to create investment-friendly environment in order [22] Halicioglu F. An econometric study of CO2 emissions, energy consumption,
income and foreign trade in Turkey. Energy Policy 2009;37:115664.
to enhance more FDI inows into the country. Moreover, gradual [23] Hondroyiannis G, Lolos S, Papapetrou E. Energy consumption and economic
population transfers from rural areas to urban area (urbanization) growth: assessing the evidence from Greece. Energy Econ 2002;24:31936.
and infrastructure development is also of paramount importance [24] Jalil A, Mahmud SF. Environment Kuznets curve for CO2 emissions: a coin-
tegration analysis for China. Energy Policy 2009;37:516772.
as all these factors expand energy demand. More importantly, in
[25] Jones D. Urbanization and energy use in economic development. Energy J
order to fulll the energy demand in the country, the policy option 1989;10(4):2944.
is that energy policy should to be adjusted in favor of enlarging [26] Katsivelis PS. Energy in Greece. Basic information and gures. Power imbal-
supply of energy to give a boost to economic growth and devel- ance-alternative for the energy sector in Greece and its European and Global
context. Public conference organized by Rosa Luxemburg Stiftung Brussels and
opment, a process that in days of crisis is a necessity. In addition, Athens; 1012 Oct. 2013.
an empirical survey that would have as a subject the inter rela- [27] Koop G, Pesaran MH, Potter SM. Impulse response analysis in nonlinear
tionships among the variables having as a dependent the renew- multivariate models. J Econom 1996;74(1):11947.
[28] Kottari M. Energy brains-energy fact sheet, Greece. Retrieved from http://www.
able energy through which the country may well become energy energybrains.org/docs/EFS/EnergyBrains_EFS_Greece_MK_2014.pdf; 2014.
independent in order to regain its economic independency and to [29] Koroneos CJ, Nanaki EA, Xydis GA. Exergy analysis of the energy use in Greece.
get in the path of growth after a long time of recession. Energy Policy 2011;39:247581.
[30] Koutroumanidis T, Ioannou K, Arabatzis G. Predicting fuelwood prices in
Greece with the use of ARIMA models, articial neural networks and a hybrid
ARIMA-ANN model. Energy Policy 2009;37(9):362734.
References [31] Kum H, Ocal O, Aslan A. The relationship among natural gas energy con-
sumption, capital and economic growth: bootstrap-corrected causality tests
from G-7 countries. Renew Sustain Energy Rev 2012;16(5):23615.
[1] Acaravci A, Ozturk I. On the relationship between energy consumption, CO2 [32] Lean HH, Smyth R. CO2 emissions, electricity consumption and output in
emissions and economic growth in Europe. Energy 2010;35(12):541220. ASEAN. Appl Energy 2010;87(6):185864.
[2] Apergis N, Payne JE. Energy consumption and economic growth: evidence from [33] Madlener R, Sunak Y. Impacts of urbanization on urban structures and energy
the Commonwealth of Independent States. Energy Econ 2009;31(5):6417. demand: what can we learn for urban energy planning and urbanization
[3] Apergis N, Payne JE. Energy consumption and growth in South America: management? Sustain Cities Soc 2011;1(1):4553.
evidence from a panel error correction model. Energy Econ 2010;32:14216. [34] Matsaganis M. The Greek Crisis: Social Impact and Policy Responses. Retrieved
[4] Ang JB. CO2 emissions, research and technology transfer in China. Ecol Econ from http://library.fes.de/pdf-les/id/10314.pdf; 2013.
2009;68(10):265865. [35] Mielnik O, Goldemberg J. Foreign direct investment and decoupling between
[5] Annual Energy Outlook (AEO2014). Annual energy outlook prepared by the U. energy and gross domestic production in developing countries. Energy Policy
S. Energy Information Administration; 2014. 2002;30:879.
[6] Arabatzis G, Malesios C. An econometric analysis of residential consumption of [36] Mirza UK, Ahmad N, Majeed T. An overview of biomass energy utilization in
fuelwood in a mountainous prefecture of Northern Greece. Energy Policy Pakistan. Renew Sustain Energy Rev 2008;12(7):198896.
2011;39(12):808897. [37] Mishra V, Smyth R, Sharma S. The energy-GDP nexus: evidence from a panel
[7] Arabatzis G, Kitikidou K, Tampakis S, Soutsas K. The fuelwood consumption in of Pacic Island countries. Resour Energy Econ 2009;31(3):21020.
a rural area of Greece. Renew Sustain Energy Rev 2012;16(9):648996. [38] Pachauri S. An energy analysis of household consumption: changing pattern of
[8] Arouri MH, Youssef AB, Mhenni H, Rault C. Energy consumption, economic direct and indirect use in India. Dordrecht, NLD: Springer; 2007.
growth and CO2 emissions in Middle East and North African countries. Energy [39] Parikh J, Shukla V. Urbanization, energy use and greenhouse effects in eco-
Policy 2012;45:3429. nomic development. Global Environ Change 1995;5(2):87103.
[9] Azam M, Qayyum K, Zaman K, Mehboob. Factor determining energy con- [40] Pesaran HH, Shin Y. Generalized impulse response analysis in linear multi-
sumption: evidence from Indonesia, Malaysia and Thailand. Renew Sustain variate models. Econ Lett 1998;58(1):1729.
Energy Rev 2015;42:112331. [41] Poumanyvong P, Kaneko S. Does urbanization lead to less energy use and
[10] Belke A, Dreger C, de Haan F. Energy consumption and economic growth lower CO2 emissions? A cross-country analysis Ecol Econ 2010;70(2):43444.
new insights into the cointegration relationship. Ruhr economic papers no. [42] Pu-yang S, Li-chao W, Si-yang C. On the nonlinear relationship between FDI
190; 2010. and energy consumption intensity in an open economy: analysis based on the
[11] Bilgen S. Structure and environmental impact of global energy consumption. framework of environmental Kuznets curve. J Financ Econ 2011;37(8):7990.
Renew Sustain Energy Rev 2014;38:890902. [43] Rennings K, Brohmann B, Nentwich J, Schleich L, Traber T, Wustenhagen R.
[12] Bilgen S. Calculation and interpretation of the standard chemical exergies of Sustainable energy consumption in residential buildings. Berlin, Germany:
elements using the chemical reference species. ActaPhys-Chim Sin Springer-Verlag; 2013.
2009;25:16459. [44] Salima RA, Shaei S. Urbanization and renewable and non-renewable energy
[13] Central Intelligence Agency (CIA). The World Bank Factbook. Retrieved from consumption in OECD countries: an empirical analysis. Econ Model
https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html; 2015. 2014;38:58191.
[14] Cole MA, Elliott RJR. Determining the trade-environment composition effect: [45] Sbia R, Shahbaz M, Hamdi H. A contribution of foreign direct investment, clean
the role of capital, labour and environmental regulations. J Environ Econ energy, trade openness, carbon emissions and economic growth to energy
Manag 2003;46(3):36383. demand in UAE. Econ Model 2014;36:1917.
[15] Dagoumasa A, Kitsios F. Assessing the impact of the economic crisis on energy [46] Tampakis S, Tsantopoulos G, Arabatzis G, Rerras I. Citizens' views on various
poverty in Greece. Sustain Cities Soc 2014;13:26778. forms of energy and their contribution to the environment. Renew Sustain
[16] Dergiades T, Martinopoulos G, Tsouldis L. Energy consumption and economic Energy Rev 2013;20:47382.
growth: parametric and non-parametric causality testing for the case of [47] Tsani SZ. Energy consumption and economic growth: a causality analysis for
Greece. Energy Policy 2013;36:68697. Greece. Energy Econ 2010;32:58290.
[17] Dritsak C, Dritsaki M. Causal relationship between energy consumption, eco- [48] Ucan O, Aricioglu E, Yucel F. Energy consumption and economic growth
nomic growth and CO2 emissions: a dynamic panel data approach. Int J Energy nexus: evidence from developed countries in Europe. Int J Energy Econ Policy
Econ Policy 2014;4(2):12536. 2014;4(3):4119.
[18] Ferreira G. Alternative energies: updates on progress. Berlin, Germany:
Springer-Verlag; 2013.

You might also like