Progress in Nuclear Energy 88 (2016) 364e374

Contents lists available at ScienceDirect

Progress in Nuclear Energy

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

Nuclear energy, renewable energy, CO2 emissions, and economic

growth for nine developed countries: Evidence from panel Granger
causality tests
Kais Saidi*, Mounir Ben Mbarek
Department of Economics, Faculty of Economics and Management, University of Sfax, Street of Airport, LP 1088, Sfax 3018, Tunisia

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 26 March 2015
Received in revised form
2 October 2015
Accepted 16 January 2016
Available online 10 February 2016

The aim of this paper is to investigate the causal relationship between nuclear energy consumption, CO2
emissions, renewable energy and real GDP per capita using dynamic panel for nine developed countries
over the period 1990e2013. Capital and labor are included as additional variables. Results shown that
there is a unidirectional causality running from renewable energy consumption to real GDP per capita for
the whole panel at short run; this implies that policies for reducing energy consumption may not retard
economic growth and income. However, there is no links between nuclear energy consumption and real
GDP per capita, but a unidirectional causality from nuclear energy consumption to labor. Moreover, a
bidirectional causality between labor and capital, and between CO2 emissions and capital are found. In
addition, there is a unidirectional causal relationship from labor to CO2 emissions, while among other
variables no causal relationship is found.
In the long run, there exists also a bidirectional causality between renewable energy consumption and
real GDP per capita, which complain that renewable energy is a crucial component for economic growth.
In addition, results revealed a unidirectional causality from GDP to CO2 emissions.
2016 Elsevier Ltd. All rights reserved.

Keywords:
Nuclear energy consumption
CO2 emissions
Renewable energy consumption
Economic growth
Panel co-integration
Causality

1. Introduction
The nexus between nuclear energy consumption, CO2 emissions
and economic growth have been recently discussed in the economics literature. The overall results showed that there is a strong
relationship between nuclear energy use and economic growth,
between CO2 emissions and economic growth, and between
renewable energy consumption and economic growth. Nevertheless, there is a great concern about environmental challenge, since
the energy consumption can causes economic growth but it is the
key reason of environmental degradation. Yet, as current electricity
production relies heavily on fossil fuels, it is expected that the
expansion of production technologies based on nuclear energy and
renewable energy would signicantly reduce future emissions of
greenhouse gases emissions (Hoffert et al., 2002; Service, 2005;
Rohatgi et al., 2002).
Nuclear energy is considered as an alternative source to deal

* Corresponding author.
E-mail addresses: saidikais.fsegs2014@gmail.com
benmbarek@yahoo.fr (M. Ben Mbarek).
http://dx.doi.org/10.1016/j.pnucene.2016.01.018
0149-1970/ 2016 Elsevier Ltd. All rights reserved.

(K.

Saidi),

mounir_

with high oil prices and to reduce dependence of foreign countries

for energy requirement in some countries. Nuclear power stations
are capital-intensive, and nuclear energy costs are less vulnerable
to changes in fuel prices generation than coal or gas red.
Furthermore, nuclear energy is an important source in the development of energy and environmental strategies in long-term. Nuclear energy has meet global energy needs in some areas of the
world where growth in energy demand is fast, oil and gas proved
reserves are likely to be exhausted within a few generations,
alternative resources are scarce, security of energy supply is a priority and reducing air pollution and emissions of greenhouse gases
is essential (Fiore, 2006; Toth and Rogner, 2006). Responding to
these difculties, nuclear energy plays a key role in long-term
development and environmental strategies. Nuclear energy meets
to global energy needs worldwide with a growing energy demand.
However, nuclear power growth are facing ongoing controversy,
namely, operational safety, radioactive waste disposal, and the risk
of proliferation of nuclear materials, as well as public perception
and acceptance nuclear energy (Toth and Rogner, 2006).
In the context of their strategy to increase energy security, many
countries have built nuclear power plants, not only to reduce

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

dependence on oil imports and to extend secure energy supplies,

but also to minimize price volatility associated with oil imports
(Toth and Rogner, 2006 . Vaillancourt et al., 2008). In addition, the
global interest in nuclear energy has become even more pressing
since the Kyoto agreement, which requires signatories to have
signicantly reduced their CO2 emissions and their effects on
environment (Becker and Posner, 2005). Moreover, it is commonly
recognized that free-carbon energy sources are the main solution
to face global warming and to insure global energy security (Elliot,
2007; Ferguson, 2007).
The rest of this paper is organized as follows. The second section
provides a literature review on the causal relationship between
economic growth, nuclear energy consumption, CO2 emissions and
renewable energy consumption. Section 3 discusses the results and
concluding remarks are given in Section 4.
2. Literature review
The causal relationship among nuclear energy consumption,
CO2 emissions and economic growth is the issue of various academic researches over the last few decades. Many papers have been
appeared the last few years covering many geographic locations,
using different econometric tools and including a range of control
variables. Several studies have focused on a specic country while
others have relied on a group of countries within a panel data
framework. Economic growth measured in terms of gross domestic
product (real or in per capita) or growth rate of GDP, using different
econometric methodologies, countries and time period and have
found conicting results.
First group of researchers focused on the relationship between
nuclear energy consumption and economic growth. The overall
ndings showed that there is a strong relationship between nuclear
energy consumption and economic growth. Their results are synthesized into four testable hypotheses: feedback, growth, conservation and neutrality hypotheses. For an overview of previous
studies, see Table 1. The growth hypothesis was investigated by Yoo
and Jung (2005). They employ annual data covering the period
1977e2002 using error correction term. The results showed that
there is a unidirectional causality running from nuclear energy
consumption to economic growth in Korea without any feedback
effect. The same study examined by Yoo and Jung (2005) to
investigate the causality between nuclear energy consumption and
economic growth in Korea by applying modems time-series techniques. The unidirectional causality running from economic growth
to nuclear energy consumption without any feedback effects in
France and Pakistan, and from nuclear energy to economic growth
in Korea which supported the conservation hypotheses. Finally, the
neutrality hypothesis was conrmed by Menyah and Wolde Rufael
(2010) who studied the causal relationship between nuclear energy
consumption and economic growth for nine industrialized countries. Their results showed the existence of neutrality hypothesis
for the USA and France.
The second strand of researcher's emphases the causal relationship between nuclear energy consumption and CO2 emissions.
In this strand, Wolde-Rufael et al. (2010) studied the causal relationship between nuclear energy, CO2 emissions and economic
growth for a group of 19 developed countries and developing
countries for the period 1984e2007 using the error correction
model (ECM). The results from the panel Granger causality tests
suggest that in the short-run nuclear energy consumption plays an
important role in reducing of CO2 emissions, but long-run estimation indicated that there is a statistically signicant negative association between nuclear energy consumption and emissions. A
recent study set by the nuclear energy agency support that is a key
element in reducing GHG produced by fossil fuel sector.

365

Sadorsky (2009a) estimated an empirical model of renewable

energy consumption for the G7 countries. Panel cointegration estimations showed that in the long term, increases in real GDP per
capita and CO2 per capita are found to be major drivers behind per
capita renewable energy consumption. These results are robust
across two different panel cointegration estimators. Oil price increases have a smaller although negative impact on renewable
energy consumption. In this way, Yoo and Ku (2009) investigated
the causal relationship between nuclear energy consumption and
economic growth using a data from six countries among 20 countries that have used nuclear energy for more than 20 years until
2005. To this end, time-series techniques including the unit roots
tests, co-integration, and Granger causality are employed to
Argentina, France, Germany, Korea, Pakistan, and Switzerland. The
main conclusion is that the causal relationship between nuclear
energy consumption and economic growth is not uniform across
countries. In the case of Switzerland, there exists a bidirectional
causality between nuclear energy consumption and economic
growth. This means that an increase in nuclear energy consumption
directly affects economic growth and that economic growth also
can stimulates further nuclear energy consumption.
In another study, Apergis and Payne (2010) examined the relationship between nuclear energy consumption and economic
growth for sixteen countries within a multivariate panel framework over the period 1980e2005. Pedroni's (1999, 2004) used
heterogeneous panel cointegration test, he noted that there is a
long-run equilibrium relationship between real GDP, nuclear energy consumption, real gross xed capital formation, and the labor
force with the respective coefcients positive and statistically signicant. The results of the panel vector error correction model
showed a bidirectional causality between nuclear energy consumption and economic growth in the short-run while unidirectional causality from nuclear energy consumption to economic
growth in the long-run. Thus, the results provide support for the
feedback hypothesis associated with the relationship between
nuclear energy consumption and economic growth. Similarly,
Apergis et al. (2010) examined the causal relationship between CO2
emissions, nuclear energy consumption, renewable energy consumption, and economic growth for a group of 19 developed and
developing countries for the period 1984e2007 using a panel error
correction model. The long-run estimation indicated that there is a
statistically signicant negative association between nuclear energy consumption and emissions, but a statistically signicant
positive relationship between emissions and renewable energy
consumption. The results from the panel Granger causality tests
suggest that in the short-run nuclear energy consumption plays an
important role in reducing CO2 emissions whereas renewable energy consumption does not contribute to reductions in emissions.
This may be due to the lack of adequate storage technology to
overcome intermittent supply problems as a result electricity
producers which have to rely on emission generating energy
sources to meet peak load demand.
Apergis and Payne (2010b) examined the causal relationship
between renewable energy consumption and economic growth for
13 countries within Eurasia over the period 1992e2007 within a
multivariate panel data framework. The heterogeneous panel
cointegration test reveals a long-run equilibrium relationship exists
between real GDP, renewable energy consumption, real gross xed
capital formation, and labor force. The results from the error
correction models indicate a bidirectional causality between
renewable energy consumption and economic growth in both short
and long-run. Thus, the empirical ndings lend support for the
feedback hypothesis of the interdependent relationship between
renewable energy consumption and economic growth. Menyah and
Rufael (2010) explored the causal relationship between carbon

366

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

Table 1
Summary of empirical studies on the links between nuclear energy consumption, CO2 emissions, renewable energy consumption, and economic growth.
N. Author(s)

Countries

Period

Single e countries studies

1. Yoo and Jung (2005)

Korea

NEC / GDP (feedback)

1977e2002 Unit-roots;
cointegration;
Grangercausality
1960e2007 Granger
NEC/CO2
causality
NEC4RE
ARDL TY
REC has a negative impact on GDP GDP/REC (renewable energy
approach
consumption)

2.

Menyah and Rufael (2010) US

3.
4.

Tugcu (2013)
Ocal and Aslan (2013)

Turkish
Turkey

Multi-country studies
5. Sadorsky (2009a)

G7 countries

6.

Yoo and Ku (2009)

Six countries

7.

Apergis and Payne (2010) Sixteen countries

1980e2005

8.

Apergis et al. (2010)

19 developed and developing countries

1984e2007

9.

Lee and Chiu (2011)

Developed countries

1971e2006

10 Latin American developing countries

1980e2009

10. Bildirici (2013)

11. Apergis and Payne (2010b) 13 countries within Eurasia

12. Apergis and Danuletiu

80 countries
(2014)
13. Ben Jabli and Ben Youssef Five North African countries
(2013)

14. Ben Jebli et al. (2014)

24 subsaharan Africa countries

15. Chang et al. (2014)

Six developed countries

16. Naser (2015)

4 industrialized countries (US, Canada,

Japan, France)

1986e2005

1992e2007

Methodology Findings

Panel
cointegration
Unit-roots;
Cointegration;
Granger
causality
Pedroni's
(1999, 2004)
Panel error
correction
Panel
cointegration
Panel causality
ARDL
approach ECM
Panel
cointegration
test
ECM
Causality test

NEC4GDP (Switzerland) GDP4NEC (feedback in France and

Pakistan)

NEC4GDP (short-run) NEC/GDP (long-run)

CO24NEC
GDP/NEC (long-run) NECsGDP (short-run) (is no causality)

BE4GDP
REC4GDP

RE/GDP

*short-run
GDP/CO2
CRCsGDP
WCsGDP
CRCsCO2
*long-run
CO2/CRC
WC/GDP
CO24GDP (short-run)
1980e2010 EKC; Panel
cointegration; CO2/REC
Granger
GDP4REC
causality
1971e2011 Granger
GDP/NEC (a gross G6 countries)
causality
NEC4GDP (UK)
1965e2010 TY approach
NEC/GDP (Japan)
GDP/NEC (France)
GDPsNEC (US, Canada)

1971e2008 Panel
cointegration
FMOLS DOLS

Notes: NEC: indicates nuclear energy consumption; GDP: indicates real GDP; CO2: indicates carbon dioxide emissions; BE: Biomass Energy, CRE: combustible renewable
consumption, WC: waste consumption, and REC: indicates renewable energy consumption. TY approach refers to TodaeYamamoto approach to Granger causality, EKC refers
environmental Kuznets curve, ECM refers error correction models, and ARDL refers to the auto regressive distributed lag procedure.

dioxide (CO2) emissions, renewable and nuclear energy consumption and real GDP for the US for the period 1960e2007 using a
modied version of the Granger causality test, they found a unidirectional causality running from nuclear energy consumption to
CO2 emissions without feedback but no causality running from
renewable energy to CO2 emissions. The econometric evidence
seems to suggest that nuclear energy consumption can help to
mitigate CO2 emissions, but so far, renewable energy consumption
has not reached a level where it can make a signicant contribution
to emissions reduction.
Lee and Chiu (2011) examined the short-run dynamics and longrun equilibrium relationships among nuclear energy consumption,
oil prices, oil consumption, and economic growth for developed
countries covering the period 1971e2006. The panel cointegration
results show that in the long run, oil prices have a positive impact
on nuclear energy consumption, suggesting the existence of the

substitution relationship between nuclear energy and oil. Furthermore, the panel causality results nd evidence of unidirectional
causality running from oil prices and economic growth to nuclear
energy consumption in the long run, while there is no causality
between nuclear energy consumption and economic growth in the
short run.
Tiwari (2011) analyzed the dynamics links between renewable
energy consumption, economic growth, and CO2 emissions in India,
using structural VAR approach. The ndings of unit root tests
showed that all variables are non-stationary at their level form and
stationary in rst difference form and cointegration analysis,
analyzed through Johansen and Juselius (1990), shows that there is
no evidence of cointegration among the test variables. The innovations analysis of study reveals that there is a positive shock on
the consumption of renewable energy source increases GDP and
decreases CO2 emissions and a positive shock on GDP have a very

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

high positive impact on the CO2 emissions.

Ben Jebli and Ben Youssef (2013) examined the causal relationship between economic growth, combustible renewable and
waste consumption, and CO2 emissions for a balanced panel of ve
North Africa countries during the period 1971e2008. The panel
cointegration test results indicate that there is a unidirectional
causality running from real GDP per capita to per capita CO2
emissions at the short-run. However, there is an evidence of no
causality between combustible renewable and waste consumption
and real GDP and between combustible renewable and waste
consumption and CO2 emissions. In the long-run, they afrmed that
there is evidence of a unidirectional causality running from CO2
emissions and combustible renewable and waste consumption to
real GDP. The results from panel FMOLS and DOLS estimations
showed that emissions is the most signicant variable in explaining
economic growth in the region which is followed by the consumption of combustible renewable and waste. In the long-run,
increases in combustible renewable and waste consumption and
emissions lead to increase economic growth. Ben Mbarek et al.
(2015) analyzed the dynamics relationships between renewable
energy, nuclear energy and economic growth in France. The ndings suggest that there is unidirectional relationship between the
economic growth and the nuclear energy.
Bildirici (2013) used the Autoregressive Distributed Lag bounds
testing (ARDL) approach of cointegration and error correction
models to analyze the causality between biomass energy consumption and economic growth in the selected 10 Latin American
developing countries. It covers annual data from 1980 to 2009. The
cointegration test results show that there is a cointegration between the biomass energy consumption and the economic growth
in Latin American developing countries. There exists a bidirectional causality between biomass energy and economic
growth, while for others only biomass energy Granger causes
economic growth.
Pao and Fu (2013) used annual data on GDP and four types of
energy consumption, namely non-hydroelectric renewable energy
consumption, total renewable energy consumption, non-renewable
energy consumption and the total primary energy consumption,
while Pao and Fu (2013) consider, in addition to the above variables,
total renewable energy consumption and hydroelectric, new
renewable and nuclear energy consumption at disaggregated level.
The two studies are based on a production function framework,
controlling for real gross xed capital formation and labor force.
Mixed results are derived regarding the direction of causality between the variables. However, the authors insist on the role of
renewable energy with its different components in promoting the
Brazil's economic development.
Tugcu (2013) investigated the long- and the short-run relationships between disaggregate energy consumption (i.e., alternative and nuclear, fossil and renewable) and total factor
productivity growth in the Turkish economy. This result highlighted that disaggregates energy consumption is cointegrated to
total factor productivity growth and there exists bi-directional
causal relationships among the variables in consideration, while
Leitao (2014) also investigated the existence of the correlation between economic growth, carbon dioxide emissions, renewable
energy and globalization. His result document showed that there is
a strong and positive link between renewable energy and economic
growth.
Ocal and Aslan (2013) examined the renewable energy consumptioneeconomic growth causality nexus in Turkey. Studies in
the literature can be grouped as country-specic and multi-country
studies. The results of these studies are inconsistent, and there is no
agreement on the existence or the direction of causality between
renewable energy consumption and economic growth. The results

367

of this country-specic study supported the conservation hypothesis. The results of empirical tests from ARDL approach showed that
renewable energy consumption has a negative impact on economic
growth, and the ones of TodaeYamamoto causality tests showed
that there is a unidirectional causality running from economic
growth to renewable energy consumption.
Apergis and Danuletiu (2014) examined for the rst time the
relationship between renewable energy and economic growth for
80 countries under the Canning and Pedroni (2008) long-run
causality test, which indicates that there is long-run positive causality running from renewable energy to real GDP for the total
sample as well as across regions. The empirical ndings provide
strong evidence that the interdependence between renewable
energy consumption and economic growth indicates that renewable energy is important for economic growth and likewise economic growth encourages the use of more renewable energy
source. The presence of causality provides an avenue to continue
the use of government policies that enhance the development of
the renewable energy sector. Ben Mbarek et al. (2014) analyzed the
causality between the economic growth, the energy and the environment, measured by CO2 emissions in Tunisia. The empirical
results conrm the presence of unidirectional relationship between
GDP and CO2 emissions in the short term.
Ben Jebli et al. (2014) based on the Environmental Kuznets Curve
(EKC) hypothesis; this paper used panel cointegration techniques to
investigate the short and the long-run relationship between CO2
emissions, economic growth, renewable energy consumption and
trade openness for a panel of 24 Sub-Saharan Africa countries over
the period 1980e2010. The validity of the EKC hypothesis has not
been supported for these countries. Short-run Granger causality
results reveal that there is a bidirectional causality between emissions and economic growth; bidirectional causality between
emissions and real exports; unidirectional causality from real imports to emissions; and unidirectional causality runs from trade
(exports or imports) to renewable energy consumption. There is an
indirect short-run causality running from emissions to renewable
energy and an indirect short-run causality from GDP to renewable
energy. In the long-run, the error correction term is statistically
signicant for emissions, renewable energy consumption and trade
openness. The long-run estimation suggest that real GDP per capita
and real imports per capita both have a negative and statistically
signicant impact on per capita CO2 emissions. The impact of the
square of real GDP per capita and real exports per capita are both
positive and statistically signicant on per capita CO2 emissions.
Chang et al. (2014) analyzed the causal link between nuclear
energy consumption and economic growth for six developed
countries over the period from 1971 to 2011. Granger causality
procedure based on Meta-analysis in heterogeneous mixed panels
is used to allow for cross-sectional dependency and heterogeneity
across countries. The empirical ndings for the overall panel support the presence of unidirectional causality running from economic growth to nuclear energy consumption across the G-6
countries. However, in the case of UK they found a bidirectional
causality running from nuclear energy consumption to economic
growth; while the results for Germany conrm the growth hypothesis and for the rest of the countries the neutrality hypothesis.
Naser (2015) examined the causal relationship between nuclear
energy consumption and economic growth for four industrialized
countries; the US, Canada, Japan, and France, between 1965 and
2010. In a multivariate framework that accounts for other key determinants such that of oil demand and price, a modied version of
the Granger causality test developed by Toda and Yamamoto (1995)
is applied. Results show that there is one-way causality from nuclear energy consumption to economic growth in Japan denoting
that an energy conservation policy that aims to minimize nuclear

368

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

energy consumption may adversely affect economic growth.

Oppositely, increasing real GDP causes additional nuclear energy
consumption in France. In the US and Canada, there is evidence that
support the neutrality hypothesis. Looking at the other investigated
channels, the level of real oil prices seems to have a vital role in
deriving the demand for nuclear power in three out of four countries. There is also a causal linkage between oil and nuclear energy
consumption in the US, Japan, and France, suggesting that the uncertainty surrounding the global oil market plays a key role in
determining the demand for nuclear energy. This means that the
policies in these countries should endeavor to overcome the
constrain on nuclear energy consumption to face any un-expected
hikes in oil prices, which may adversely affect economic growth in
such oil importing countries.
Saidi and Hammami (2015) examined the impact of energy
consumption and the CO2 emissions on economic growth using
simultaneous-equation models with panel data for 58 countries
over the period 1990e2012. The empirical results show that energy
consumption has a positive impact on economic growth. This implies that energy consumption played an important role in the increase of economic growth in the investigated economies but with
the consequence of high pollution. Since energy is an important
ingredient for economic growth, strong energy policies are
required to realize economic growth. On the other hand, the CO2
emissions have a negative impact on economic growth.
3. Methodological approach and data
3.1. Methodological approach
On the basis of modern econometric techniques, we study the
causal relationship between NEC, CO2 emissions, RE, and GDP. The
test procedure involves the following steps. First, we determine the
order of integration of the series using unit root tests in panel.
Whether the variables contain a unit root, the second step is used
panel cointegration tests to examine the existence of a long-term
relationship between the series and to estimate the long-term
equations Fully Modied OLS (FMOLS). Finally, if a long-term
relationship between the variables is found, we study the size
and direction of the causal link between the series following a
dynamic approach to the panel estimation.
3.1.1. Panel unit root tests
In the analysis of the common components of nuclear energy
consumption, renewable energy, GDP per capita and CO2 emissions,
standard time series unit root tests can be applied. The traditional
unit root test Augmented DickeyeFuller (ADF) has a problem of low
power of rejection of the null hypothesis of stationary of the series,
especially for the short period of data. Thus the recent literature
shows that the panel stationarity tests are more powerful than
those based on individual time series Al-Iriani (2006). Among the
recently developed tests, we, the unit root test LLC, Levin et al.
(2002); IPS, Im et al. (2003), and Beitung (2000). Narayan and
Smyth (2009) used the test Breitung (2000) as this test is generally more robust than any test of the rst generation of unit root
tests in panel.
In this paper, Hadri, IPS and LLC tests are used to test unit roots
in the panel data. Apergis and Payne (2009) use the IPS test that is
also more general and more popular, based on the ADF test principle. The unit root test panel Im et al. (2003) assumes a heterogeneous unit root (the alternative hypothesis allows the presence
of a subset of N1 size of individuals with the variable of interest
following a stationary process. Im unit root tests, Pesaran and Shin
(2003) therefore take into account the heterogeneity (Hurlin and
Mignon, 2005), While Breitung (2000) and Levin et al. (2002)

assumes a homogenous unit autoregressive root (under the alternative hypothesis, the autoregressive coefcient is the same for all
individuals to test LLC). The underlying autoregressive model can
be expressed as follows:
The IPS test is based on the following model1:

Dyit ai bi yi;t1

Pi
X

rij Dyi;tj it ; i 1; ; N;

j1

(1)

t 1; ; T
where yit is the series for country, i in the panel over period t, pi is
the number of lags selected for the ADF regression and it are
independently and normally distributed random variables for all i
and t with zero means and nite heterogeneous variances, s2i.
IPS tests the null hypothesis of the unit root for each individual
(country) in the panel, that is, H0: bi 0, against the alternative H1:
bi < 0, i 1 .,N1; bi 0, i N11, .,N, which allows for some of
the individual series to be integrated. The proposed Ztbar (p, r)
statistic converges in distribution to a standard normal variate
sequentially, as T/ followed by N.
The LLC unit root test is also based on model (1) but it considers
the coefcients of the autoregressive term as homogeneous across
all individuals, that is, bi b vi. LLC tests the null hypothesis that
each individual in the panel has integrated time series, that is, H0:
bi b 0 vi, H1: bi b < 0 vi. Therefore, under the alternative, all
single series are stationary. The resulting statistic, t*, asymptotically
follows a standard normal distribution. Hadri (2000) proposed a
residual-based Lagrange Multiplier test for the null of level or trend
stationarity allowing the case of heterogeneous disturbance terms
across i.

3.1.2. Panel cointegration test

Several tests are developed through panels; Tests of no cointegration panel data proposed by Pedroni (1995, 1997, 1999, 2004),
Kao (1999) and Bai and Ng (2001) are similar residual tests
testing proposed by Engle and Granger (1987) as part of the time
series. Larsson et al. (2001) and Groen and Kleibergen (2003), have
in turn inspired by the work of Johansen (1991, 1995) to propose
tests based on the likelihood ratio in a system where a priori the
number of relationships cointegration is not known.
Pedroni (1999) proposed a methodology to test the panel data
cointegration that can be considered as an extension of the traditional Engle and Granger (1987) two-step residual-based method.
Pedroni tests take into account the heterogeneity through parameters that may differ between individuals. Thus, under the alternative hypothesis, there is a cointegration relationship for each
individual, and the parameters of the cointegrating relationship are
not necessarily the same for each individual panel (Hurlin and
Mignon, 2005). We use Pedroni's (1999) method to test for cointegration in a heterogeneous panel data and consider the following
cointegrating regression:

GDPit ai di t b1i NCEit b2i CO2it b3i RECit b4i Kit

b5i Lit it ;
i 1; ; N; t 1; ; T

(2)

where ai is the country-specic intercept and dit is a deterministic

time trend specic to individual countries in the panel. The slope
coefcients b1i, b2i, b3i, b4i and b5i can vary from one individual to
another allowing the cointegrating vectors to be heterogeneous a

The model can be generalized by introducing linear time trends.

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

cross countries.
Pedroni (1999) presented seven different statistics, four of
which are based on pooling along the within-dimension, while the
other three are based on pooling along the between-dimension.
The former, known as panel cointegration statistics, test the null
hypothesis of no cointegration, H0: 4i 1vi, against the alternative
H1: 4i 4 < 1 ci in the residuals from the panel regression (2),
c
it 4i d
i;t1 mit . By contrast, the latter, known as group mean
panel cointegration statistics, test the null hypothesis of no cointegration against the alternative H1: 4i < 1 ci, which allows the
possibility of an additional heterogeneity source across the countries. All the statistics are asymptotically normally distributed.
Pedroni (2000) proved that the panel OLS estimator is biased
when the variables are cointegrated. If the variables are cointegrated, then the FMOLS estimator produces asymptotically unbiased estimates of elasticities and efcient, normally distributed
standard errors at long-run. Moreover, Kao (1999) also proposed
the test of null hypothesis of no cointegration: DickeyeFuller test
type and DickeyeFuller Augmented test type. Unlike tests Pedroni,
Kao considers the particular case where the cointegrating vectors
are assumed to be homogeneous across individuals. In other words,
these tests do not take into account the heterogeneity under the
alternative hypothesis and are also valid for a bivariate system (ie
when only one regressor is present in the cointegrating
relationship).
3.1.3. Panel DOLS and FMOLS estimates
A more powerful test with comparison to single equation
methods, which investigates directly the condition on the cointegrating vector, that is required for strong relation to hold was
suggested by Pedroni (2001, 2004). This method stipulates the null
hypothesis in a more natural form.
Thus, the relationship of whether strong relationship between
renewable energy consumption, nuclear energy consumption, CO2
emissions and economic growth holds consistently for all panel
countries is tested. The regression between these four factors is
reported in equation (6). Our model is based on the regression
between these four factors. The dependent variables are either,
renewable energy consumption, nuclear energy consumption, GDP
per capita, and CO2 emissions.

369

growth, renewable energy consumption, CO2 emissions, nuclear

energy consumption, capital, and labor can be written as follows2:

k
X

DGDPit

b1j DGDPi;tj

j1
k
X

k
X

g1j DNECi;tj

j1

j1

w1j DCO2i;tj

j1

k
X

d1j DRECi;tj

k
X

r1j DKi;tj

j1

q1j DLi;tj

j1

l1 ECTi;t1 Dm1it
(3)

k
X

DNECit

b2j DGDPi;tj

j1
k
X

k
X

k
X

g2j DNECi;tj

j1

d2j DRECi;tj

w2j DCO2i;tj

j1

k
X

j1

k
X

r2j DKi;tj

j1

q2j DLi;tj

j1

l2 ECTi;t1 Dm2it
(4)

DCO2it

k
X

b3j DGDPi;tj

j1

k
X

g3j DNECi;tj

j1

k
X

d3j DRECi;tj

j1

k
X

w3j DCO2i;tj

j1

k
X

r3j DKi;tj

j1

k
X

q3j DLi;tj

j1

l3 ECTi;t1 Dm3it
(5)

DRECit

k
X

b4j DGDPi;tj

j1

3.1.4. Panel causality tests

Granger Causality is a statistical hypothesis test which veries
whether one time series is capable of forecasting another (Granger,
1969). Granger causality becomes a powerful tool to investigate the
causal effect and functional relation from numerous temporal data
which are easy to source today (Luo et al., 2011).
An underlying assumption of Granger causality is that a variable
X Granger causes Y if Y can be better predicted using the histories of
both X and Y than it can use the history of Y alone. Engle and
Granger (1987) explained that if co-integration exists between
two variables in the long run, and then there must be either unidirectional or bi-directional Granger Causality between these two
variables. As mentioned earlier, if they have one unit root and are
cointegrated, then the bivariate VECM is specied and estimated.
The Granger causality test is then conducted in the context of the
VECM. If the two series have one unit root and are not cointegrated,
then the bivariate VAR is specied and estimated.
The existence of cointegration between series conrms that
there ought to be at least, one causal relationship, but it fails to give
its direction. Hence, we follow the famous procedure from Engle
and Granger (1987) to examine the short-run as well as the longrun causal dynamics between the competing variables. Following
Engle and Granger (1987), a vector error correction model (VECM)
which is used for testing the Granger causality among economic

k
X

k
X

g4j DNECi;tj

k
X

j1
k
X

d4j DRECi;tj

j1

w4j DCO2i;tj

j1
k
X

r4j DKi;tj

j1

k
X

q4j DLi;tj

j1

l4 ECTi;t1 Dm4it
(6)

DKit

k
X

b5j DGDPi;tj

j1

k
X

g5j DNECi;tj

j1

k
X

d5j DRECi;tj

j1

k
X
j1

k
X

w5j DCO2i;tj

j1

r5j DKi;tj

k
X

q5j DLi;tj

j1

l5 ECTi;t1 Dm5it
(7)

2
It should be noted that only equations where the null hypothesis of no cointegration is rejected will be estimated within the Granger causality framework.
Hence, no error correction model will be estimated for the equation where CO2
variable is set as dependent variable since no cointegrating relationship was found.

370

DLit

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

k
X

b6j DGDPi;tj

j1

k
X

g6j DNECi;tj

j1

k
X

d6j DRECi;tj

j1

k
X
j1

k
X

w6j DCO2i;tj

j1

r6j DKi;tj

k
X

q6j DLi;tj

j1

l6 ECTi;t1 Dm6it
(8)
where, D is the rst difference operator; bj, gj, yj, dj, qj, rj (j 1, 2, 3,
4, 5, 6) are parameters to estimated; mjit (j 1, 2, 3, 4, 5, 6), ECT is the
error correction term derived from the corresponding long-run
equilibrium relationship obtain from the residuals of the estimate
cointegrating equation (2). The coefcients lj (j 1, 2, 3, 4, 5, 6) of
the ECTs represent the deviation of the dependent variables from
the long-run equilibrium.
The error correction model allows testing for the existence of
Granger causality in three possible ways (Sebri and Abid, 2012).
First, the short-run Granger causality is investigated by testing the
signicance of the sum of lagged differences of explanatory variables by using the partial F-statistic. Second, the long-run causality
is checked by examining the coefcients of the ECTt1 based on tstatistics. Particularly, a long-run Granger causality exists if this
coefcient is negative and statistically signicant.
3.2. Data specications
This study based on annual data for nine developed countries;
Canada, France, Japan, Netherlands, Spain, Sweden, Switzerland;
Kingdom and the United States. The variables in this study include
renewable energy consumption (REC) is measured by combustible
renewable and waste % of total energy dened in thousands of
metric tons, gross xed capital formation in billion of constant 2005
US $, total labor force, and per capita real GDP (GDP) (constant 2005
US $) which is used as the proxy of economic growth are downloaded from the World Bank's Development Indicators (WDI, 2014).
CO2 emissions (CO2) (metric tons per capita), nuclear energy consumption (NEC) is expressed in terms of Terawatt-hours (TWh) are
obtained from the British Petroleum Statistical Review of World
Energy (BP, 2013). The data is for the period from 1990 to 2012.
Table 2 below shows the mean value, standard deviation and the
different variables coefcient of variation for the different countries
and also to the panel. The highest means of real GDP (50259.33) in
the Switzerland, and nuclear consumption (761.5517), CO2 emissions (6053.214) are in the US, while the highest mean of the variable related to the renewable consumption (16.58566) is in
Sweden. The lowest means of real GDP (23182.88) in the Spain, REC
(1.309450) in the Japan, NEC (3.851009) in the Netherlands, and
CO2 emissions (43.96015) in the Sweden. Then, UK is the highest
volatility country in real GDP (0.1478), followed by Japan (0.2659)
in NEC, Spain (0.1732) in CO2 emissions, and UK (0.6966) in REC.
4. Empirical results
4.1. Panel unit root results
The presence of a unit root in the series is tested using the Levin,
Lin & Chu (LLC), Im, Pesaran and Shin (IPS), Maddala & Wu and
Hadri tests. The panel based ADF test proposed by LLC (2002) assumes homogeneity in the dynamics of the autoregressive coefcients for all panel units. Alternatively, Maddala and Wu (1999)
employ nonparametric methods in conducting panel unit root tests
with the Fisher-ADF and Fisher-PP tests which has the advantage of

allowing for as much heterogeneity across units as possible. Under

the Levin et al. (2002), Fisher- ADF, and Fisher-PP tests, the null
hypothesis is the presence of a unit root and the alternative hypothesis is no unit root. In addition, our unit root testing analysis
makes use of the Im et al. (2003) test (i.e. IPS test) that permits to
solve Levin and Lin's serial correlation problem by assuming heterogeneity between units in a dynamic panel framework. The test
shows that under the null hypothesis of non stationary, the statistic
follows the standard normal distribution asymptotically.
The results of the unit root tests for the series of GDP, NEC, K, L,
CO2, and REC variables are shown in Table 3. All tests reject nullhypothesis of non stationary when variables are used at rst difference. This implies that series of variables GDP per capita, NEC, K,
L, CO2, and REC are integrated of order one, and I (1) process. These
results are substance with notation that most of macroeconomics
variables are no stationary at level, but become stationary after rst
differencing (Nelson and Plosser (1982)). Consequently, as pooled
data is stationary at rst difference, the series follow stochastic
trends and therefore can be cointegrated as well.

4.2. Panel cointegration test results

Based on the results of the unit root test panel, we proceed to
the cointegration test panel. The concept of co-integration can be
dened as a systematic long-term co-movement between two or
more economic variables (Yoo, 2006).
Pedroni tests are reported in Table 4. Results show existence of
cointegration between variables at 1% signicant level as for all six
models these reject the null of no cointegration. Therefore, it is
concluded that the variables are cointegrated and a long run relationship exist for group as a whole and the members of the panel.
From the Kao residual cointegration result reported in Table 5,
strong evidence is found to reject the null hypothesis of no cointegration at 5% level of signicance. Therefore, it is concluded that
there exist a strong evidence of long-run cointegration relationship
between the variables for the multicounty panel. These results are
consistent with Lee (2005), Sadorsky (2009).

4.3. DOLS and FMOLS results

FMOLS and DOLS results are reported in Table 6. The results of
individual panel FMOLS, DOLS are shown in this table indicate
whether renewable energy consumption, nuclear energy consumption, CO2 emissions stimulate economic growth in developed
countries.
In this table, FMOLS test shows a positive relationship running
from GDP to renewable energy consumption, and CO2 per capita, at
1% signicance. This means that, a 1% increase in GDP per capita,
leads to 9.64% increase of CO2 per capita, 1798% increase of
renewable energy consumption, and 8.47% increase of nuclear
consumption. Furthermore, a 1% increase in nuclear energy decreases renewable energy consumption by 4%. This means that
policies based on income.
Besides, a 1% increase in renewable energy consumption increases is associated to a 0.0003% increase in GDP per capita, reduces CO2 emissions by 0.03%, and reduces nuclear consumption by
0.007%. Also, a1% increase in CO2 emissions reduces renewable
energy consumption by 39%, increases GDP per capita by 0.02% and
reduces nuclear consumption by 0.19%. According to panel DOLS
results, 1% increase in GDP per capita is associated to an increase by
772% of CO2 emissions. Also, renewable energy consumption reduces CO2 emissions and decreases nuclear energy consumption, at
1% signicance.

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

371

Table 2
Descriptive statistics of the used variables.
Descriptive statistics
Canada

France

Japan

Netherlands

Spain

Sweden

Switzerland

UK

US

Panel

Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV
Means
Std.dev.
CV

GDP

NEC

CO2

REC

32668.80
3849.498
0.1178
31690.68
2562.913
0.0808
34350.82
1754.073
0.0510
36385.95
4432.491
0.1218
23182.88
2844.001
0.1226
36683.19
5369.711
0.1463
50259.33
3329.752
0.0662
33543.04
4959.634
0.1478
39940.47
4581.697
0.1147
35411.69
7841.875
0.2214

86.29795
10.23077
0.1185
405.1959
40.85801
0.1008
262.8216
69.88590
0.2659
3.851009
0.388396
0.1008
58.35405
3.319915
0.0568
67.35973
6.211656
0.0922
25.68829
1.592136
0.0619
80.09604
12.93223
0.1614
761.5517
81.93649
0.1075
194.5796
238.1445
1.2238

20.99813
5.817138
0.2770
36.60626
5.757350
0.1572
105.0950
8.808695
0.0838
11.22878
1.821793
0.1622
25.29031
6.447458
0.2549
5.895616
1.350404
0.2290
7.896452
0.872233
0.1104
32.22175
6.641073
0.2061
235.1378
51.65336
0.2196
53.37446
72.55963
1.3594

16.72656
1.607215
0.0960
27.76774
1.472464
0.0530
66.59329
1.013125
0.0152
8.107720
0.720119
0.0888
19.29858
2.889086
0.1497
4.683048
0.198123
0.0423
4.124727
0.252783
0.0612
30.12290
1.327758
0.0440
145.9957
10.21855
0.0699
35.93558
43.15227
1.2008

576.1852
54.22720
0.0941
419.6405
17.50921
0.0417
1313.748
74.56048
0.0567
246.6822
15.27556
0.0619
316.8897
54.88654
0.1732
43.96015
1.469520
0.0334
585.4967
30.47943
0.0520
6053.214
348.8216
0.0576
6053.214
348.8216
0.0576
1734.337
2343.680
1.3513

4.415509
0.300081
0.0697
4.822941
0.556129
0.1153
1.309450
0.371485
0.2836
2.714856
1.103654
0.4065
4.005739
0.766883
0.1914
16.58566
3.376451
0.2035
7.410849
0.927631
0.1251
1.408893
0.981560
0.6966
3.439257
0.362692
0.1054
5.123684
4.611326
0.9000

Notes: Std. Dev indicates standard deviation, GDP indicates real GDP, CO2: indicates carbon dioxide emissions, K indicates real capital, L indicates labor force, REC indicates
renewable energy consumption, and NEC indicates nuclear energy consumption.

Table 3
Panel unit root results for GDP, NEC, K, L, CO2 and REC.
Null: Unit root
Variables

Levin, Lin & Chu

(LLC)

Null: No unit root

Im, Pesaran and Shin (IPS) W- MW-ADF Fisher Chistat

square

MW-PP Fisher Chisquare

Hadri Z-stat

Heteroscedastic consistent Zstat

1.78432 (0.9628)

7.39130 (0.9864)

7.03912 (0.9898)

9.45476 (0.0000)* 9.42924 (0.0000)*

1.15069 (0.1249)

23.6223 (0.1678)

43.1117 (0.0008)*

7.10144 (0.0000)* 4.27638 (0.0000)*

2.11324 (0.9827)
0.3680 (1.0000)
0.97416 (0.1650)

7.81673 (0.9812)
10.1394 (0.9273)
22.4299 (0.2135)

5.74100 (0.9971)
11.4520 (0.8742)
19.5366 (0.3595)

7.95768 (0.0000)* 8.35272 (0.0000)*

9.97191 (0.0000)* 8.47075 (0.0000)*
4.53158 (0.0000)* 5.52846 (0.0000)*

7.70843 (1.0000)

4.73581 (0.9992)

3.38756 (0.9999)

8.97224 (0.0000)* 6.76888 (0.0000)*

4.32745* (0.0000)
5.78249* (0.0000)
6.05070* (0.0000)
4.55537* (0.0000)*
7.73895 (0.0000)*
7.83128* (0.0000)

51.1211* (0.0001)
69.1802* (0.0000)
69.0148* (0.0000)
53.4524 (0.0000)*
94.8209 (0.0000)*
102.770* (0.0000)

73.5811
244.112
67.0927
50.7568
165.037
125.973

1.14263
7.67160
0.49248
5.50327
7.48915
6.19933

Level
GDP 1.73502 (0.0414)
**
NEC 1.77068 (0.0383)
**
K
0.36008 (0.3594)
L
0.33714 (0.3680)
CO2 2.05220 (0.0201)
**
REC 6.34449 (1.0000)
First difference
GDP 4.47849* (0.0000)
NEC 4.42595* (0.0000)
K
7.20727* (0.0000)
L
5.63847* (0.0000)
CO2 8.08176* (0.0000)
REC 9.53977* (0.0000)

(0.0000)*
(0.0000)*
(0.0000)*
(0.0001)*
(0.0000)*
(0.0000)*

(0.1266)
(0.0000)*
(0.3112)
(0.0000)*
(0.0000)*
(0.0000)*

1.36920
3.11331
0.55755
4.85872
3.23620
4.90703

(0.0855)***
(0.0009)*
(0.2886)
(0.0000)*
(0.0006)*
(0.0000)*

Note: *, **, and *** represents signicance at the 1%, 5%, and 10% levels, respectively, of signicance (bold entries). The null hypothesis is that the variable follows a unit root
process, except for the Hadri Z-stat and the Heteroscedastic Consistent Z-stat. p-values are given in parentheses. Probabilities for the Fisher-type tests are computed using an
asymptotic Chi-square distribution. All other tests assume asymptotic normality.

4.4. Panel causality tests results

The short-run and long-run panel Granger causality results from
estimating panel based error correction model set out in (3), (4),
(5), (6), (7) and (8) are reported in Table 7. The rst column shows
the effect of GDP on other variables, the second column shows the
capital effect and the third presents the impact of labor on other
variables. The fourth investigates the effect of renewable energy
consumption on other variables. The fth and sixth column shows

the effect of renewable energy consumption and CO2 emissions on

other variables. The last column shows the long-term relationship
between the variables.
The rst column shows that GDP has a positive and signicant
effect on renewable energy consumption at the 1% level in the short
term and the last column shows that there is a unidirectional
causality from economic growth to renewable energy consumption
in long term. But the effect of renewable energy consumption on
economic growth is not signicant. In column 2, short-term, capital

372

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

Table 4
Pedroni and Kao cointegration tests.
Methods

Series: GDP, K, L, NEC, REC and CO2

Pedroni (2004)

Within dimesion (panel statistics)

Test
Statistics
Panel v-Statistic
1.239992
Panel rho-Statistic
1.181497
Panel PP-Statistic
2.526626
Panel ADF-Statistic
2.485001
(weighted statistic)
Panel v-Statistic
3.044672
Panel rho-Statistic
0.334349
Panel PP-Statistic
4.954896
Panel ADF-Statistic
4.673903

Prob
0.8925
0.8813
(0.0058)*
(0.0065)*

Between dimension (individuals statistics)

Test
Statistics
Group r-statistic
1.182066
Group pp-statistic
5.433082
Group ADF-statistic
4.799413

Prob
0.8814
(0.0000)*
(0.0000)*

0.9988
0.6309
(0.0000)*
(0.0000)*

Notes: *, ** indicate the rejection of the null hypothesis at 1%, 5%, where the null hypothesis is that the variables are not cointegrated.

Table 5
Kao residual cointegration test result.
Model specication: No deterministic trend
ADF t-statistics

1.995550 (0.0230)**

Table 6
FMOLS and DOLS.
Independent
variables

GDP

NEC

REC

CO2

K
L
CO2
REC
NEC
GDP
K
L
CO2
REC
GDP
K
L
CO2
NEC
REC
GDP
K
L
NEC

Panel group
FMOLS

DOLS

Coefcient

Prob

Coefcient

Prob

46.28113
402.6036
9.640271
1798.030
8.479132
0.000596
0.973370
3.884055
0.032359
4.808538
0.000318
0.005614
0.039122
0.002869
0.007703
39.04659
0.022933
8.783339
26.39156
0.192013

(0.1826)
(0.0085)*
(0.0000)*
(0.0000)*
(0.3366)
(0.6217)
(0.0086)*
(0.0213)**
(0.1784)
(0.0907)***
(0.0000)*
(0.6964)
(0.5534)
(0.0013)*
(0.0327)**
(0.0016)*
(0.0000)*
(0.0000)*
(0.0003)*
(0.6479)

27.28649
772.4263
7.676616
1068.284
19.99613
0.000913
1.073668
3.516923
0.033656
2.010711
0.000270
0.054200
0.005665
0.005048
0.017496
30.67441
0.012477
10.38961
26.70435
0.153385

(0.6743)
(0.0011)*
(0.0757)***
(0.0000)*
(0.1650)
(0.2973)
(0.0166)**
(0.0173)**
(0.2006)
(0.2194)
(0.0001)*
(0.2926)
(0.9706)
(0.0576)***
(0.0545)***
(0.0022)*
(0.0005)*
(0.0000)*
(0.0000)*
(0.7861)

has positive and signicant effect on work and short term CO2
emission and the last column shows the same relationship between
capital and labor and between capital and CO2 emissions and longterm economic growth.
Then, in the third column the work has a positive and

statistically signicant effect at the 1% level on the variables:

capital, nuclear energy consumption, and CO2 emissions. Therefore can be concluded that there is a bidirectional relationship
between capital and labor, a unidirectional relationship from
labor at the nuclear energy consumption, and a unidirectional
relationship from CO2 emissions to labor, but there is no relationship between labor and renewable energy consumption in
the short-run.
Finally, the CO2 emission has a positive and signicant effect on
capital, and capital has a positive effect on CO2 emissions at the 1%
level, this implies that there is a bidirectional relationship between
capital and CO2 emissions.
In long-run, for GDP per capita equation, the estimated coefcient on error correction term is negative and statistically signicant. It shows that short-term adjustments to equilibrium are
driven by adjustment back to long-run equilibrium through error
correction term. For REC equation, the estimated coefcient on
error correction term is negative and statistically signicant indicating that REC is responsive to adjustments back to equilibrium.
This implies that there is signicant bidirectional causation between GDP per capita and REC.

5. Conclusion
This study will rst determine the direction of causality between economic growth and energy consumption in nine
developed countries, namely Canada, France, Japan, Netherlands,
Spain, Sweden, Switzerland; Kingdom and the United States. To
do so after the unit root tests Im et al., and Levin and Lin, cointegration and causality panel on data sets covering the period
1990e2012.
Results of FMOLS and DOLS reveal that GDP has a positive effect
on renewable energy and vice versa. Also, renewable energy reduces CO2 emissions. This means that policies oriented to the
promotion of renewable energy consumption is effective to environment protection.

Table 7
The VECM Granger causality analysis.
Short run

Long run

Dependent variables D(GDP)

D(GDP)
D(K)
D(L)
D(NEC)
D(REC)
CO2

e
0.425055 (0.8085)
2.358487 (0.3075)
1.719782 (0.4232)
10.53086* (0.0052)
1.298267 0.5225

D(K)

D(L)

D(NEC)

2.395582 (0.3019)
e
60.76038* (0.0000)
0.549590 (0.7597)
0.461554 (0.7939)
9.251823* 0.0098

4.132011 (0.1267)
15.99586* (0.0003)
e
13.25463* (0.0013)
0.201789 (0.9040)
9.502559* 0.0086

0.655503
0.382987
3.817273
e
0.203469
2.052287

D(REC)

CO2

(0.7205) 1.515776 (0.4687) 0.648058 (0.7232)

(0.8257) 0.542293 (0.7625) 12.40990* (0.0020)
(0.1483) 0.801898 (0.6697) 0.474039 (0.7890)
1.636908 (0.4411) 2.160265 (0.3396)
(0.9033) e
1.011407 (0.6031)
(0.3584) 0.046075 0.9772 -

Notes: ECT represents the coefcient of the error correction term. * indicates that the parameter estimates are signicant at the 1% level.

ECMt1
0.351683* [-2.75879]
0.002268* [-3.63072]
1.61E-05 [-0.43643]
0.002624 [-0.79132]
0.000217* [-2.65732]
0.050034* [-3.95692]

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

The results show that there is a unidirectional causality from

REC to GDP per capita at the 1% level for the whole panel at short
run. However, there is no causality between NEC and economic
growth. A unidirectional causality from NEC to labor at the 1% level
is found. Moreover, there is existence of bidirectional causality
between labor and capital, and bidirectional causality between CO2
emissions and capital. In addition, there is a unidirectional causal
relationship working with CO2 emissions, while among other variables no statistically signicant causal relationship is found. Results show a short-run unidirectional relationship between GDP
per capita and renewable energy consumption, this implies that
policies for reducing energy consumption may not retard economic
growth and income.
In the long run, there exists also a bidirectional causality between renewable energy consumption and GDP per capita, which
complain that energy consumption is a crucial component in
growth, a complement to capital and labor as input factors of
production and vice versa. In the long-run, results reveal a unidirectional causality between CO2 emissions and GDP per capita.
References
Al-Iriani, Mahmoud A., 2006. Energy-GDP relationship revisited: an example from
GCC countries using panel causality. Energy Policy 34, 3342e3350.
Apergis, N., Payne, J.E., 2009. CO2 emissions, energy usage, and output in Central
America. Energy Policy 37, 3282e3286.
Apergis, N., Danuletiu, D.C., 2014. Renewable energy and economic growth: evidence from the sign of panel long-run causality. Int. J. Energy Econ. Policy 4,
578e587.
Apergis, N., Payne, J.E., 2010. A panel study of nuclear energy consumption and
economic growth. Energy Econ. 32, 545e549.
Apergis, N., Payne, J.E., 2010b. Renewable energy consumption and growth in
Eurasia. Energy Econ. 32, 1392e1397.
Apergis, N., Payne, J.E., Menyah, K., Wolde-Rufael, Y., 2010. On the causal dynamics
between emissions, nuclear energy, renewable energy, and economic growth.
Ecol. Econ. 69, 2255e2260.
Bai, J., Ng, S., 2001. A panic attack on unit roots and cointegration. Econometrica 72,
1127e1177.
Becker, G., Posner, R., 2005. Nuclear Power: Has its Time Come (Again)? http://
www.becker-posner-log.com/archives/2005/05/nuclear_power_h.html.
Ben Jebli, M., Youssef, Ben, Ozturk, I., 2014. The Role of Renewable Energy Consumption and Trade: Environmental Kuznets Curve Analysis for Sub-saharan
Africa Countries. Online at. http://mpra.ub.uni-muenchen.de/54300/.
Ben Jebli, Mehdi, Ben Youssef, Slim, 2013. Combustible Renewables and Waste
Consumption, Exports and Economic Growth: Evidence from Panel for Selected
MENA Countries. MPRA Paper 47767. University Library of Munich, Germany.
Ben Mbarek, M., Ben Ali, N., Feki, R., 2014. Causality relationship between CO2
emissions, GDP and energy intensity in Tunisia. Environ. Dev. Sustain 16,
1253e1262.
Ben Mbarek, M., Khairallah, R., Feki, R., 2015. Causality relationships between
renewable energy, nuclear energy and economic growth in France. Environ.
Syst. Decis. 35 (1), 133e142.
Bildirici, M.E., 2013. Economic growth and biomass energy. Biomass Bioenergy 50,
19e24.
BP, 2013. Statistical Review of World Energy.
Breitung, J., 2000. The local power of some unit root tests for panel data. Adv. Econ.
15, 161e177.
Canning, D., Pedroni, P., 2008. Infrastructure, long-run economic growth and causality tests for cointegrated panels. Manch. School 76, 504e527.
Chang, T., Gatwabuyege, F., Gupta, R., Inglesi-Lotz, R., Manjezi, N.C., SimoKengne, B.D., 2014. Causal relationship between nuclear energy consumption
and economic growth in G6 countries: evidence from panel Granger causality
tests. Prog. Nucl. Energy 77, 187e193.
Elliot, D., 2007. Nuclear or Not? Does Nuclear Power Have a Place in Sustainable
Energy Future? Palgrave Macmillan, Houndmills, Basingstoke.
Engle, R.F., Granger, C.W.J., 1987. Cointegration and error correction: representation,
estimation, and testing. Econometrica 55, 251e276.
Ferguson, C.D., 2007. Nuclear Energy: Balancing Benets and Risks. Council of
Foreign Relations, CRS No. 28.
Fiore, K., 2006. Nuclear energy and sustainability: understanding ITER. Energy
Policy 34, 3334e3341.
Granger, C.W.J., 1969. Investigating causal relations by econometric models and
cross spectral methods. Econometrica 37, 424e438.
Groen, J.J.J., Kleibergen, F., 2003. Likelihood-based cointegration analysis in panels
of vector error-correction models. J. Bus. Econ. Statistics 21 (2), 295e318.
Hadri, K., 2000. Testing for stationarity in heterogeneous panel data. Econ. J. 3,
148e161.
Hoffert, M.I., Caldeira, K., Benford, G., Criswell, D.R., Green, C., Herzog, H., Jain, A.K.,

373

Kheshgi, H.S., Lackner, K.S., Lewis, J.S., Lightfoot, H.D., Manheimer, W.,
Mankins, J.C., Mauel, M.E., Perkins, L.J., Schlesinger, M.E., Volk, T., Wigley, T.M.L.,
2002. Advanced technology paths to global climate stability: energy for a
greenhouse planet. Science 298, 981e987.

rie, 2005. Une synthe
se des tests de racine unitaire
Hurlin, Christophe, Mignon, Vale



es de panel. Economie

vision 169, 253e294.
sur donne
Pre
Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels.
J. Econ. 115, 53e74.
Johansen, S., Juselius, K., 1990. Maximum likelihood estimation and inference on
cointegration with applications to the demand for money. Oxf. Bull. Econ. Stat
52 (2), 169e210.
Johansen, S., 1991. Estimation and hypothesis testing of cointegration vectors in
Gaussian vector autoregressive models. Econometrica 59, 1551e1580.
Johansen, S., 1995. Likelihood-based Inference in Cointegrated Vector Autoregressive Models. Oxford University Press, Oxford.
Kao, C., 1999. Spurious regression and residual-based tests for cointegration in
panel data. J. Econ. 90, 1e44.
Larsson, R., Lyhagen, J., Lothgren, M., 2001. Likelihood-based cointegration tests in
heterogeneous panels. Econ. J. 4, 109e142.
LAU, E., CHYE, X.-H., CHOONG, C.H.-K., 2011. Energy-growth causality: a panel
analysis. In: International Conference on Applied Economics e ICOAE 2011.
Lee, Chiu, 2011. Oil prices, nuclear energy consumption, and economic growth:
new evidence using a heterogeneous panel analysis. Energy Policy 39,
2111e2120.
Lee, C.C., 2005. Energy consumption and GDP in developing countries: a cointegrated panel analysis. Energy Econ. 27, 415e427.
Leit~
ao, N.C., 2014. Economic growth, carbon dioxide emissions, renewable energy
and globalization. Int. J. Energy Econ. Policy 4, 391e399.
Levin, A., Lin, C.F., Chu, C., 2002. Unit root tests in panel data: asymptotic and nitesample properties. J. Econ. 108, 1e24.
Maddala, G.S., Wu, S., 1999. A comparative study of unit root tests with panel data
and a new simple test. Oxf. Bull. Econ. Statistics 61, 631e652.
Menyah and Rufael, 2010. CO2 emissions, nuclear energy, renewable energy and
economic growth in the US. Energy Policy 38, 2911e2915.
Narayan, Paresh, K., Smyth, Russell, 2009. Multivariate granger causality between
electricity consumption, exports and GDP: evidence from a panel of Middle
Eastern countries. Energy Policy 37, 229e236.
Naser, H., 2015. Can nuclear energy stimulate economic growth? evidence from
highly industrialised countries. Int. J. Energy Econ. Policy 5, 164e173.
Nelson, C.R., Plosser, C.I., 1982. Trends and random walks in macroeconomic time
series: some evidence and implications. J. Monetary Econ. 10, 139e162.
Ocal, O., Aslan, A., 2013. Renewable energy consumptioneeconomic growth nexus
in Turkey. Renew. Sustain. Energy Rev. 28, 494e499.
Pao, H.-T., Fu, H.-C., 2013. Renewable energy, non-renewable energy and economic
growth in Brazil. Renew. Sustain. Energy Rev. 25, 381e392.
Pedroni, P., 1995. Panel Cointegration, Asymptotic and Finite Sample Properties of
Pooled Time Series Tests with an Application to the PPP Hypothesis. Working
Paper in Economics. Indiana University, 92e013.
Pedroni, P., 1997. Panel Cointegration, Asymptotic and Finite Sample Properties of
Pooled Time Series Tests with an Application to the PPP Hypothesis: New
Results, Working Paper in Economics. Indiana University.
Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels
with multiple regressors. Oxf. Bull. Econ. Statistics 61, 653e670.
Pedroni, P., 2000. Fully Modied OLS for Heterogeneous Cointegrated Panels. Advances in Econometrics (Book Series), 15, pp. 93e130.
Pedroni, P., 2004. Panel cointegration. Asymptotic and nite sample properties of
pooled time series tests with an application to the PPP hypothesis. Econ.
Theory 20, 597e625.
Pedroni, P., 2001. Purchasing power parity tests in cointegrated panels. Rev. Econ.
Statistics 83 (4), 727e731.
Rohatgi, U., Jo, J.H., Lee, J.C., Bari, R.A., 2002. Impact of the nuclear option on the
environment and the economy. Nucl. Technol. 137, 252e264.
Sadorsky, P., 2009a. Renewable energy consumption, CO2 emissions and oil prices in
the G7 countries. Energy Econ. 31, 456e462.
Sadorsky, p., 2009. Renewable energy consumption and income in emerging
economies. Energy Policy 37, 4021e4028.
Saidi, K., Hammami, S., 2015. The impact of energy consumption and CO2 emissions
on economic growth: fresh evidence from dynamic simultaneous-equations
models. Sustain. Cities Soc. 14, 178e186.
Sebri, M., Abid, M., 2012. Energy use for economic growth: a trivariate analysis from
Tunisian agriculture sector. Energy Policy 48, 711e716.
Service, R.F., 2005. News focus, is it time to shoot for the Sun? Science 309, 548.
Tiwari, A.K., 2011. A structural VAR analysis of renewable energy consumption, real
GDP and CO2 emissions: evidence from India. Econ. Bull. 31, 1793e1806.
Toda, H.Y., Yamamoto, T., 1995. Statistical inference in vector autoregressions with
possibly integrated process. J. Econom 66, 225e250.
Toth, F.L., Rogner, H.H., 2006. Oil and nuclear power: past, present, and future.
Energy Econ. 28, 1e25.
Tugcu, T., 2013. Disaggregate energy consumption and total factor productivity: a
cointegration and causality analysis for the Turkish economy. Int. J. Energy
Econ. Policy 3, 307e314.
Vaillancourt, K., Labriet, M., Loulou, R., Waaub, J., 2008. The role of nuclear energy in
long-term climate scenarios: an analysis with the World-TIMES model. Energy
Policy 36, 2296e2307.
WDI, 2014. World Development Indicators. World Bank. http://data.worldbank.org.

374

K. Saidi, M. Ben Mbarek / Progress in Nuclear Energy 88 (2016) 364e374

Wolde-Rufael, Y., Menyah, K., 2010. Nuclear energy consumption and economic
growth in nine developed countries. Energy Econ 32 (3), 550e556.
Yoo, S.H., 2006. The causal relationship between electricity consumption and economic growth in the ASEAN countries. Energy Policy 34, 3573e3582.

Yoo, S.H., Jung, K.O., 2005. Nuclear energy consumption and economic growth in
korea. Prog. Nucl. Energy 46, 101e109.
Yoo, S.H., Ku, S.J., 2009. Causal relationship between nuclear energy consumption
and economic growth: a multi-country analysis. Energy policy 37, 1905e1913.