Charfeddine 2019

Renewable Energy 139 (2019) 198e213
Contents lists available at ScienceDirect
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
Impact of renewable energy consumption and financial development

on CO2 emissions and economic growth in the MENA region: A panel
vector autoregressive (PVAR) analysis
Lanouar Charfeddine a, *, Montassar Kahia b, c
a
Department of Finance and Economics, College of Business and Economics, Qatar University, P.O.Box: 2713, Doha, Qatar
b
LAREQUAD & FSEGT, University of Tunis El Manar, Tunisia
c
Department of Finance & Economics, College of Business and Economics, Qassim University, Saudi Arabia
a r t i c l e i n f o a b s t r a c t
Article history: Unlike previous studies in the energy-environment literature, this study employed the panel vector
Received 24 March 2018 autoregressive (PVAR) model that was developed by Love and Zicchino [1] to examine the impact of
Received in revised form renewable energy and financial development on carbon dioxide (CO2) emissions and economic growth.
14 December 2018
Moreover, we used the impulse response function tool, which was developed in the same context, to
Accepted 2 January 2019
Available online 18 February 2019
better understand the reaction of the two main variables of interest, CO2 emissions and economic
growth, aftershocks on renewable energy and financial development variables. Finally, the analysis was
completed by the variance decomposition of all variables. These analyses were conducted for a 24
Keywords:
CO2 emissions
countries of the Middle East and North Africa (MENA) region from 1980 to 2015.
Economic growth Overall, the results show that both renewable energy consumption and financial development have a
Renewable energy slight influence and can only slightly explain CO2 emissions and economic growth. These results indicate
Financial development that both financial development and the renewable energy sectors in the MENA countries are still weak
Panel VAR regarding making contributions to environmental quality improvements and economic growth. Several
policies are proposed and discussed.
© 2019 Elsevier Ltd. All rights reserved.
1. Introduction and these outcomes clearly state that if no action is taken in term
of GHG emissions reduction, the average worldwide temperature
In the last few decades, a growing consensus has emerged be- will go up by more than 2 C.2 Following many environmental in-
tween scientific, energy and environmental scholars regarding the stitutions and environmental experts, if this threshold level of ris-
threatening effects of climate change on human life, human health ing temperature is surpassed, then the impact on nature and
and the quality of the environment for future generations [2e5]. To human life of these rapidly growing levels of GHG emissions will be
avoid a global environmental disaster, several researchers and very profound and will severely affect all aspects of life [10e12].
policymakers have highlighted the importance of reducing green- According to a recent report of the Medical Society Consortium on
house gas (GHG) emissions, which are considered to be the main Climate and Health, if the threshold of 2 C is surpassed, then there
source of climate changes [6e8]. For instance, following Stern et al. will be a rapid rise in seas levels, mass extinctions, super droughts,
[9]: “if no action is taken to curb GHG emissions, the concentration of increased wildfires, intense hurricanes, decreased crops and fresh
GHGs in the atmosphere could double as early as 2035 from its pre- water and the melting of the Arctic. This will also lead to a reduc-
industrial level”. This affirmation was confirmed by the outcomes tion in the quality of air and food and increases water contamina-
of the 2016 Paris Agreement conference of Parties (COP21 [138]),1 tion and health problems [13e15].
Empirically, several studies in the environmental literature
* Corresponding author.
2
E-mail addresses: lcharfeddine@qu.edu.qa (L. Charfeddine), kahia. https://medsocietiesforclimatehealth.org/reports/medical-alert/.
montassar2013@gmail.com (M. Kahia).
1
https://unfccc.int/sites/default/files/english_paris_agreement.pdf.
https://doi.org/10.1016/j.renene.2019.01.010
0960-1481/© 2019 Elsevier Ltd. All rights reserved.
L. Charfeddine, M. Kahia / Renewable Energy 139 (2019) 198e213 199
found that the change in climate is mainly due to the rapid increase energy consumption and financial development on two environ-
in GHG emissions from carbon dioxide (CO2)3 as well as from mental degradation and economic key indicators, specifically CO2
methane gas and nitrous oxide [16e18]. Many others studies, emissions and economic growth.
including Sathaye et al. [19]; have reported that most of the efforts, The contribution of the current research study to the energy and
such as the Kyoto Protocol and the United Nations Framework environmental empirical literature is three fold. First, the study
Convention on Climate Change (UNFCCC), remain insufficient to attempts to fill the gap in the empirical literature regarding the
respond to the climate change difficulties. As a consequence, many impact of renewable energy consumption and financial develop-
researchers and scholars have claimed that the enactment of new ment on economic growth and CO2 emissions in the MENA region.
laws, regulations and economic policies can help to improve envi- The examination of the MENA region as a case of study is particu-
ronmental quality [20e22]. However, several other studies have larly interesting since most of the MENA countries are very energy
shown that improving environmental quality via energy conser- intensive, they possess a high potential for energy production from
vation policies can lead to economic growth reduction [23e25].4 renewable energy (such as solar and wind), and most have a huge
Thus, the major challenge for policymakers is how to balance be- capacity of financial resources. Second, unlike previous studies that
tween the objective of reducing environmental degradation and have employed traditional econometric models, such as the vector
maintaining reasonable levels of economic growth [26e28]. Thus, error correction model (VECM) and/or the vector autoregressive
policymakers are challenged to make consistent and inexpensive (VAR) framework,5 this study makes use of a recent multivariate
energy sources available while concurrently curbing greenhouse econometric tool known as the panel data vector autoregressive
gas (GHG) emissions. (PVAR) technique.6Third, this study proposes and discusses several
From a practical point of view, several important policy options important policies designed to improve the impact of financial
have been advanced to better design an environmental strategy development and renewable energy on CO2 emissions and eco-
that can cut the increasing level of CO2 emissions. In particular, nomic growth.
from energy, environmental and economic perspectives, three The remaining sections of the paper are organized as follows.
propositions are of particular interest: (1) the promotion of Section 2 presents a review of the literature related to the impact of
renewable energy, (2) the development of the financial sector, and renewable energy consumption and financial development on CO2
finally, (3) the exploration of the interactions between financial emissions and economic growth. Section 3 presents the materials
development and renewable energy on economic growth and CO2 and empirical method used in this research study. Section 4 pre-
emissions. The first proposition consists of promoting and sents and discusses the empirical findings. Section 5 discusses the
expanding renewable energy sources, known to be clean, green policy implications of our results. Finally, Section 6 concludes the
energy and friendly to the environment. By doing this, renewable paper.
energy sources can help significantly to reduce CO2 emissions and
other types of pollutants (see also [29e31,27,32]). Concerning the 2. Literature review
second proposition, several empirical studies have shown that the
financial sector can play an important role in reducing CO2 emis- This section is devoted to reviewing the literature that have
sions by encouraging technological advances in the energy sector examined the impact of renewable energy consumption and
[33e37]. Finally, regarding the last proposition, the literature have financial development on CO2 emissions and economic growth.
shown that the interaction between renewable energy consump-
tion and financial development can play an important role in 2.1. Impact of renewable energy on CO2 emissions and economic
curbing CO2 emissions (see Refs. [33,38e40]). For instance, ac- growth
cording to Brunnschweiler [38]; the commercial banking and credit
markets can play a dynamic role in promoting and stimulating the A lively debate over the last two decades was sparked regarding
renewable energy sector. Additionally, Kim and Park [40] confirmed the rapid development of renewable energy and its impact on the
that the dependence of renewable sectors on debt and equity quality of the environment and economic growth [41e43,27,32,44].
financing has led to a disproportionately faster increase in coun- From a climate change perception, the use of renewable energy
tries with developed financial markets. sources has long been known for having a positive effect on envi-
Moreover, from an empirical point of view, most previous ronmental quality by reducing the level of GHG emissions in the
studies that have investigated the impact of renewable energy and atmosphere [6,45]. Moreover, following the OECD report of 2013,
financial development on CO2 emissions and economic growth the investment in renewable energy sources is also known to be
have employed the univariate class of models by using either a time less carbon intensive compared with that of traditional energy.
series or panel data analysis. In addition, only a few studies have Thus, by promoting the adoption of renewable energy, countries
simultaneously included renewable energy consumption and will improve the quality of the environment and develop a global
financial development to assess their impact on CO2 emissions and green and clean environmental framework. However, from an
economic growth. To our knowledge, there is no study in the economic perspective, the development of renewable energy
empirical literature that has examined this research question in a sources offers many energy and economic benefits [46e48]. These
multivariate panel context and that accounts for both the simul- economic benefits include but are not limited to solving several
taneous effects of renewable energy consumption and financial problems, such as energy security, energy mix portfolio diversifi-
development on CO2 emissions and economic growth. Hence, to fill cation, outflow of foreign currency, and unemployment, since the
these gaps, in this study, we used a multivariate panel approach, renewable energy sector is more labor intensive compared with the
known as the panel vector autoregressive model (PVAR), of Love
and Zicchino [1] to examine the simultaneous impact of renewable
5
Some other studies have used panel data techniques by use random and/or
fixed effects, the generalized method of moments (GMM) model, panel cointe-
gration tests and/or Granger-causality methods.
3 6
The major source of carbon dioxide (CO2) emissions is the burning of fossil This model has the obvious advantage of merging the conventional VAR
fuels, which represented 87% of the world's energy supply in 2012 [68]. method, which allows for treating all the considered variables in the regression as
4
This can hold in general if the growth or the feedback hypothesis between endogenous with the panel data procedure, which enables unobserved individual
economic growth and energy consumption cannot be rejected. heterogeneity.
200 L. Charfeddine, M. Kahia / Renewable Energy 139 (2019) 198e213
nonrenewable energy sector [49e51]. In particular, investing in development may deteriorate the environmental quality by
renewable energy will enable oil-importing countries to reduce increasing manufacturing activities, causing an increase in pollu-
their dependence on external oil [27]. However, for oil exporting tion levels and environmental degradation [35].
countries, it will enhance technology transfers and economic Empirically, there is no consensus among researchers regarding
diversification as well as maintain hydrocarbon export revenue the impact of financial development on CO2 emissions and eco-
[44]. nomic growth. As mentioned in the previous subsection, we focus
Overall, the empirical findings are mixed and seem to depend in on presenting and discussing a group of countries (see Table 2).8
the econometric methods applied (time series or panel data), the The majority of previous studies found evidence for different
economic characteristics of the country/countries analyzed and the types of causal relationship directions between financial develop-
period of study [52]. In the following, we present and discuss only ment, CO2 emissions and economic growth. For example, some
previous studies that have examined a group of countries,7 since in studies found evidence for a unidirectional causality from financial
our analysis we examined the case of the MENA group of countries. development to economic growth (Sadorsky [77] for 9 Central and
For instance, several research studies have shown evidence for a Eastern European countries and Al-Mulali and Che Sab [78] for Sub-
bidirectional causality association between renewable energy and Saharan African countries). Evidence for a unidirectional causality
CO2 emissions as well as between renewable energy and economic from financial development to CO2 emissions was found by many
growth (see Apergis and Payne [42,53] for 7 Central American researchers, including Salahuddin et al. [36]; Bekhet et al. [79];
countries and South America, and Al-Mulali et al. [41] studied this among others.
for Latin America and the Caribbean countries, whereas Moutinho To summarize, while there is some theoretical consensus
and Robaina [43] studied this for 20 European countries). The one- regarding the positive impact of renewable energy and financial
way causal relationship between economic growth, CO2 emissions development on environmental quality and economic growth,
and renewable energy was also confirmed by some empirical previous empirical studies have had difficulty in reaching this same
studies (see Sadorsky [54] for G7 countries). In addition, some other consensus. To contribute to this literature, we used the panel VAR
studies found that renewable energy consumption is an efficient econometric model. This PVAR model is discussed in detail in the
strategy toward sustainable growth and environmental quality next section.
improvements (see Bo €lük and Mert [45] examined this for 16 Eu-
ropean Union countries, Bhattacharya et al. [6] studied this for 85 3. Empirical methodology
developed and developing economies, and Ito [55] studied this for
42 developing countries). The idea that renewable energy harms 3.1. Data descriptions
economic growth was only found by a few studies, including Silva
et al. [56] for Denmark, Portugal and Spain. Annual data over the period of 1980e2015 was collected from
the World Bank Development Indicators (WDI) online databases
and the U.S. Energy Information Administration (EIA). Six variables
2.2. Impact of financial development on CO2 emissions and were used in this study to examine the impact of renewable energy
economic growth consumption and financial development on environmental degra-
dation and economic growth in the MENA region. These variables
Theoretically, there is strong consensus among researchers include: (1) labor force represented by the number of persons in
regarding the crucial role that financial development plays in millions, (2) capital measured by the gross capital formation as the
stimulating economic growth [57e59]. Currently, there is no doubt share of real GDP, (3) CO2 emissions per capita used to represent
that financial development is an indispensable pillar for economic environmental degradation, (4) the real GDP per capita, (5) the
growth because it allows accumulation of capital through pooling percentage of renewable energy consumption on the total energy
and mobilization of savings, improving the necessary information consumption, and (6) financial development represented by the
about investment activities and optimally allocating capital. The share of domestic credit to private sector scaled by the real GDP. All
financial sector also plays a crucial role in monitoring energy these variables were converted to the natural logarithm to reduce
emissions by encouraging technological advance in the energy nonnormality and heteroscedasticity [75,80]. The description of
supply sector for reducing the levels of emissions [35]. This shows these variables, their measure, symbol, predicted sign and eco-
that financial development, which shows the real availability of nomic explanation are presented in Table 3.
financial resources for productive activities and funding channels Regarding the list of the MENA countries examined in this paper,
for projects by banks and stock markets [60], can play a construc- it includes Lebanon, Algeria, Jordan, Israel, Egypt, Bahrain, Malta,
tive and crucial role in the struggle against environmental deteri- Iran, Tunisia, Morocco, Iraq, Turkey, Libya, Kuwait, Qatar, Oman,
oration mainly via the reduction of CO2 emissions. From this Armenia, Saudi Arabia, Cyprus, Mauritania, Syria, Georgia, Yemen
perspective, environmental degradation diminishes with financial and the United Arab Emirates.
improvement. In addition, financial development can usually begin
research and development (R&D) activities, attract foreign direct 3.2. PVAR specification
investment (FDI) and successively speed up economic activities to
influence the quality of the environment via investments in green- This analysis used the panel data vector autoregressive (PVAR)
associated projects [28,61e63]. A well-developed financial sector model developed by Love and Zicchino [1]; this model allowed us
reduces the borrowing cost, encourages investment activities and to account for unobserved individual heterogeneity for the entire
restrains the spread of energy emissions by boosting the efficiency
of the energy sector [62,64]; and [65]. However, financial
8
The presentation of the results for single countries regarding the impact of
financial development on CO2 emissions and economic growth are presented in
7
For the case of a single country, the results are mixed overall (see Ref. [66]; detail in Table 2. Overall, positive effects on the quality of the environment were
Jaforullah and King [68], for the USA; Tiwari [29] for India; Lin and Moubarak [67] found by Zhang [72] for China, Shahbaz et al. [73] for Malaysia, Shahzad et al. [76]
and Long et al. [69] for China; Bento and Moutinho [70] for Italy; Chunark et al. [71] for Pakistan and Charfeddine and Ben Khediri [75] for the UAE. Positive effects on
for Thailand; Belaïd and Youssef [16] for Algeria and Danish et al. [2] for Pakistan). the quality of the environment were found by Abbasi and Riaz [33] for the case of
More details about these studies are reported in Table 1. Pakistan and by Ozturk and Acaravci [74] for the case of Turkey.
Table 1
A summary of previous studies investigating the linkage amongst renewable energy consumption, CO2 emissions, and economic growth.
Authors Country/ Time period Methodology Results

Countries
Category 1: Time series analysis

Menyah and USA 1960e2007 Toda-Yamamoto causality CO2/RE;
Wolde- Y/RE
Rufael [66]
Tiwari [29] India 1960e2009 Structural Vector Autoregressive (SVAR) The consumption of RE sources stimulates economic growth and
approach decreases CO2 emissions.
Lin and China 1977e2011 VECM-Granger causality RE4Y and CO2/RE (Long-Run);
Moubarak RE… CO2 and RE …Y (Short-Run)
[67]
Jaforullah USA 1965e2012 Cointegration test/VECM Granger causality test RE/CO2; RE…Y; Increasing consumption from RE sources is an
and King effective solution to mitigate CO2 emissions.
[68]
Long et al. China 1952e2012 Cointegration analysis/Granger causality Y/ RE; RE/CO2; Renewable energy consumption has positive
[69] analysis/Static and dynamic regression analysis/ effect on economic growth and a negative influence on carbon
Impulse response function analysis emissions.
Bento and Italy 1960e2011 ARDL model/Toda-Yamamoto test Y/ RE; RE reduces the level of CO2 emissions per capita in the short
Moutinho and long-run
[70]
Chunark et al. Thailand Power Asia-Pacific Integrated Model/Computable RE will mitigate CO2 emissions and spur economic growth.
[71] development General Equilibrium (AIM/CGE)
plan 2015 (2015
e2036)
laïd and
Be Algeria 1980e2012 ARDL model/VECM Granger causality test RE/CO2(Long-run);
Youssef RE/Y (Long-run)
[16]
Danish et al. Pakistan 1970e2012 ARDL model/FMOLS/DOLS/VECM Granger RE4CO2; RE…Y; RE plays a vital role in curbing CO2 emissions
[2] causality test
Category 2: Panel data analysis
Sadorsky [54] G7 countries 1980e2005 FMOLS and DOLS Y/RE;
CO2/RE
Silva et al. USA, Denmark, 1960e2004 Structural Vector Autoregressive (SVAR) Except for USA, an increase in using RE sources have economic costs
[56] Portugal and methodology in terms of GDP per capita (i.e, it harms economic growth) and
Spain decrease the CO2 emissions per capita.
Apergis and 7 Central 1980e2010 FMOLS/Regime-wise RE4Y
Payne [53] American Granger-causality Tests RE4CO2
countries
Bo€ lük and 16 EU (European 1990e2008 Panel fixed effect model renewable energy consumption appreciably reduces
Mert [45] Union) countries CO2 emissions (around 1/2)
Al-Mulali Latin America 1980e2010 Cointegration test/VECM Granger causality test RE4CO2;
et al. [41] and the RE4Y
Caribbean
countries
Apergis and South America 1980e2010 FMOLS/VECM Granger causality test RE4CO2;
Payne [42] RE4Y
Moutinho 20 European 1991e2010 Cointegration Analysis/The Innovative RE4CO2;
and countries Accounting Approach/VECM Granger causality RE4Y
Robaina test
[43]
Bhattacharya 85 developed 1991e2012 The system-GMM, FMOLS model The increase of renewable energy consumption has an important
et al. [6] and developing positive and negative effect on economic productivity andCO2
economies emissions, respectively.
Ito [55] 42 developing 2002e2011 GMM/PMG RE contributes to environmental quality improvements and has a
countries positive effect on economic growth.
Notes: 1)/, 4, and … denote unidirectional causality relationships, bidirectional causality relationships, and no causality relationships, respectively; 2) CO2, RE and Y are
abbreviations for carbon dioxide emissions, renewable energy consumption, and GDP (economic growth), respectively.
Source: Authors' tabulation.
series via the introduction of fixed effects that improve the coher- suitable technique to investigate macroeconomic dynamics. First,
ence and the consistency of the measurement. Furthermore, the being neutral with regard to a particular theory of growth or
PVAR approach has several practical benefits that make it a more development, the statistical model of PVAR is more based on the
contemporary movements of a series than on a particular concept
of macroeconomics,9 which, if not approved, can be distorting [81].
9
Despite the absence of an explicit economic theory behind the PVAR model, the Second, in line with the interdependence realities, the present
two equations of interest in our study, the CO2 emissions and economic growth PVAR model does not make a distinction between endogenous and
equations, are based in two important theories. For instance, the economic growth exogenous variables; instead, all variables are mutually treated as
equation can be viewed as an augmented version of the neoclassical Cobb-Douglas
production function, where the renewable energy consumption is added as an
endogenous. Each PVAR variable relies not only on its historical
additional factor of production (see Refs. [133e135]). After that some other vari- realization but also on other variables, indicating a real simulta-
ables are introduced, such as financial development and CO2 emissions, to obtained neity between the variables and their treatment. Third, PVAR pro-
the main equation of interest. For the second equation, CO2 emissions equation, it vides a model for endogenous and exogenous shocks, which are
can be viewed as a derivation of the STIRPAT model (see Ref. [136].
Table 2
A summary of previous studies investigating the linkage amongst financial development, CO2 emissions, and economic growth.
Authors Country/Countries Time Methodology Results

period
Category 1: Time series analysis

Zhang [72] China 1980 Cointegration theory, Variance FD leads to CO2 emissions
e2009 decomposition, Granger causality test
Shahbaz et al. Malaysia 1971 ARDL, VECM FD4Y;
[73] e2011 FD4CO2;
FD lowers environmental quality
Islam et al. Malaysia 1971 ARDL VECM FD support the long term economic growth
[139] e2009
Ozturk and Turkey 1960 Bounds F. test FD/Y (Short-run);
Acaravci [74] e2007 FD/CO2 (Long-run);
FD has no significant impact on per capita carbon emissions
Komal and Pakistan 1972 System GMM estimation technique FD has a positive and significant effect on Y.
Abbas [141] e2012
Charfeddine UAE 1975 Cointegration with structural breaks, FD/CO2;
and Ben e2011 VECM Granger causality FD4Y
Khediri [75] FD deteriorates environmental quality
Abbasi and Riaz Pakistan 1971 An augmented VAR approach, Granger GDP per capita had statistically significant impact on CO2; FD…CO2 (Short-run for
[33] e2011 causality the full time sample); FD/CO2 (Short-run for the reduced time sample).
1988 The financial variables may mitigate emissions where greater liberalization
e2011 degree and financial sector development came to pass
Shahzad et al. Pakistan 1971 ARDL bounds test for cointegration FD/CO2;
[76] e2011 procedure FD contributes to CO2 emissions (positive)
Category 2: Panel data analysis
Tamazian et al. BRIC 1992 Panel model FD reduces CO2 emissions
[64] e2004
Sadorsky [3,77] 9 Central and 1996 Panel GMM (generalized method of Deposit money bank assets/Y (positive);
Eastern European e2006 moments) Financial system deposits/Y (positive);
countries Liquid liabilities/Y (positive).
Al-Mulali and Sub Saharan 1980 Panel Granger causality FD/Y;
Che Sab [78] African countries e2008 CO24FD;
FD lowers environmental quality
Al-Mulali and GCC countries 1980 DOLS, VECM Granger causality FD4 Y
Lee [140] e2009 FD reinforces economic growth
Salahuddin Gulf Cooperation 1980 DOLS, FMOLS, Dynamic Fixed Effect FD…Y;
et al. [36] Council (GCC) e2012 model(DFE), Granger causality FD/CO2;
countries FD improves Environmental quality
Bekhetet al. GCC countries 1980 Cointegration tests, the ARDL approach, CO2/FDin Saudi Arabia and Qatar; FD/CO2 in UAE, Oman and Kuwait; FD…CO2
[79] e2011 VECM Granger causality test in Bahrain; FD/Y in Saudi Arabia and Qatar; Y/FD in UAE and Oman; FD…Y in
Kuwait and Bahrain.
Nasreen et al. South Asian 1980 Multivariate framework. Bounds test for Financial stability gets better environmental quality; financial stability has a
[37] countries e2012 cointegration and Granger causality positive and significant impact on economic growth
approach
Notes: 1)/, 4, and … denote unidirectional causality relationships, bidirectional causality relationships, and no causality relationships, respectively; 2) CO2, FD, and Y are
abbreviations for carbon dioxide emissions, financial development and GDP (economic growth), respectively.
unquestionably the most important sources of macroeconomic indicates the vector that determines the specific effects of the
dynamics for small open economies. In addition, PVAR is relatively country found in this regression, dt represents the dummy variables
uncomplicated for coherent and efficient estimations for both for the country's specific time and εit denotes the residual vector.
cases: a single country or a panel composed of different countries. The matrix form of the PVAR model reported in (1) can also be
Finally, PVAR has obvious realistic value as a practical instrument rewritten in six equations, Eqs. (2)e(7), as follows:
for exploring the mutual impact of the consumption of renewable
energy sources and financial development on CO2 emissions and
economic growth in the MENA region countries and for providing
strategic advice.
The proposed panel VAR model is given by, X
p
Xp

DLnðLit Þ ¼ m1i þ a1j DLn Litj þ b1j DLn Kitj
Yit ¼ mi þ AðLÞYit þ ai þ dt þ εit (1) j¼1 j¼1
X
p
Xp

where Yit is a vector of the endogenous stationary series10 (L, K, FD, þ c1j DLn FDitj þ d1j DLn RECitj
REC, GDP, CO2), and mi represents the matrix of country-specific j¼1 j¼1
fixed effects. The subscripts, which are defined as i and t, refer to X
p
Xp
country and time, respectively. AðLÞ represents the matrix poly- þ e1j DLn GDPitj þ f1j DLn CO2itj þ a1i
j¼1 j¼1
nomial in the lag operator with AðLÞ ¼ A1 L1 þ A2 L2 þ … þ Ap Lp , ai
þ d1t þ ε1it
(2)
10
All the variables are converted to the natural logarithm to obtain reliable and
consistent results and more stationary behavior [80].
Table 3
Variables description, expected sign and economic explanation.
Variables Measure Data source Symbol Predicted Economic impacts on

sign on
CO2 GDP CO2 GDP
CO2 emissions Measured in metric tons per capita WDI CO2 NA NA NA NA

Real GDP (Economic growth) Measured in per capita US$ WDI GDP NA NA NA NA
Renewable energy consumption Measured in percentage of EIA REC e ± Is negative if renewable energy helps MENA Is positive if renewable energy can
total energy consumption countries considerably in reducing the level efficiently ensure a sustainable growth. Is
of carbon emissions and improve the negative if renewable energy is too
environmental quality. Positive otherwise. expensive causing adverse effects on
economic growth.
Domestic credit to private sector Measured in proportion of GDP WDI FD ± ± Is positive if financial development Is positive if financial development boosts
(Financial development) enhances manufacturing activities leading economic growth through efficient
to an increase in CO2 emissions. Negative if strategies. Negative if the financial sector is
financial development permits firms experiencing a period of crisis and financial
investing in eco-friendly associated projects instability that adversely affects economic
contributing to environmental quality growth.
improvements.
The gross capital formation (Capital) Measured in proportion of GDP WDI K ± ± Is positive if the used capital is employed in Is positive if the level of stock capital is high
activities that are not friendly to the and provides a reduced transaction costs.
environment. Negative otherwise. Negative otherwise.
Labor forces Measured in number of persons PWT90 L ± ± Is positive if the labor forces increase the Is positive if labor forces are educated,
engaged in millions CO2 emissions level and contribute to skilled and more productive. Negative
environmental deterioration. Negative otherwise.
otherwise.
Notes: CO2 and GDP are the main dependent variables of our study. Hence, they have no predicted sign and economic impacts (x).NA means not applicable as the associated variables are the dependent variables.

þ
þ
þ
þ
þ
þ
p
p
DLnðKit Þ ¼ m2i þ
þ
þ
j¼1
j¼1
X
X
p
p
DLnðFDit Þ ¼ m3i þ
þ
þ
j¼1
j¼1
X
X
p
p
DLnðRECit Þ ¼ m4i þ
p
p
p
DLnðGDPit Þ ¼ m5i þ
j¼1
j¼1
X
X
j¼1
j¼1
j¼1
X
X
X
p
p
p
DLn CO2itj ¼ m6i þ
j¼1
j¼1
j¼1
X
p
L. Charfeddine, M. Kahia / Renewable Energy 139 (2019) 198e213
X
X
p
j¼1
X
þ d2t þ ε2it
j¼1
X
p
þ d3t þ ε3it
j¼1
X
þ d4t þ ε4it

þ d5t þ ε5it

þ a6i þ d6t þ ε6it

c2j DLn FDitj þ

c3j DLn FDitj þ
a2j DLn Litj þ
e2j DLn GDPitj þ

c4j DLn FDitj þ
p
a3j DLn Litj þ
c5j DLn FDitj þ

e3j DLn GDPitj þ
j¼1
X
a4j DLn Litj þ

p
e4j DLn GDPitj þ

p
c6j DLn FDitj þ

a5j DLn Litj þ
e5j DLn GDPitj þ

j¼1
j¼1
p
j¼1
X
p
X
p
p
a6j DLn Litj þ

j¼1
e6j DLn GDPitj þ

X
j¼1
j¼1
j¼1
X
X
X
p
p
j¼1
j¼1
X
j¼1
j¼1
X
j¼1
X
X
X
p
j¼1
j¼1
X
X

b2j DLn Kitj
d2j DLn RECitj

b3j DLn Kitj

d3j DLn RECitj

b4j DLn Kitj
b5j DLn Kitj

d4j DLn RECitj

d5j DLn RECitj

b6j DLn Kitj

d6j DLn RECitj

f6j DLn CO2itj

f2j DLn CO2itj þ a2i
formation criterion (SIC) is used to determine the optimal

During the estimation process of this model, the Schwarz in-
203
(7)
(6)
(5)
(4)
(3)
autoregressive lag-length, j, where j 2ð1:::; pÞ. To surmount the 3.3.1. Pesaran [96] diagnostic test for panel cross-section
constraints on the parameters because of their violation in practice, dependence
fixed effects (indicated by ai ) are incorporated in the specification It is well known that the Lagrange multiplier test statistic of
to allow for individual heterogeneity in the levels of the entire Breusch and Pagan [97] for cross-sectional dependence does not
series. The PVAR model allows the inclusion of country fixed ef- have a high power and is not appropriately centered for a fixed T. To
fects, designated by mi, which accounts for all constant time factors overcome these limits, Pesaran [96] proposed a modified Lagrange
unobserved at the country level. However, the inclusion of country multiplier (CDLM ) statistic test for cross-sectional dependence.
fixed effects causes an estimation challenge, which occurs in any The CDLM statistic is specified as follows:
specification that incorporates lags of the dependent variables: the sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
fixed effects are associated with the regressors, and hence, the X X
N1 N
1
mean-differencing method frequently applied to remove the fixed CDLM ¼ r 2ij 1
Tb (8)
N ðN 1Þ i¼1 j¼iþ1
effects of countries may generate biased and subjective coefficients.
One technique to overcome this challenge is to apply the forward
where rij refers to the product-moment correlation coefficient of
mean-differencing or explicitly ‘Helmert procedure’ [82], which
consequently conserves the orthogonality between transformed the disturbances.11 Under the null assumption of absence of cross-
and lagged independent variables [1], allowing the lagged inde- sectional dependence with the first T /∞ and then N /∞ (H0:
pendent variables to be used to more consistently estimate the rij ¼ rji ¼ 0 for isj versus H1: rij ¼ rji for some isj), CDLM asymp-
coefficient using the GMM system. Another advantage of the PVAR totically follows a normal distribution. Nonetheless, this check is
model is allowing for common time effects, di , that are included in likely to show evidence of considerable and significant size dis-
the regression (1) to capture any global macroeconomic shock that tortions when N is large compared with T. To deal with this issue,
possibly will have an effect on all countries correspondingly. To deal Pesaran [96] subsequently developed a new test for cross-sectional
with time effects, we distinguished all variables before inclusion in dependence (CD) that can be performed where N is large and T is
the model, and these variables correspond to setting dummies in small. This test is based on the pairwise correlation coefficients of
the system. Furthermore, evaluating the effect of orthogonal the OLS residuals obtained from standard augmented Dickey-Fuller
shocks, describing the impact of the shock of one variable to [98] specifications of each variable in the panel rather than their
another variable while keeping all other variables invariant, is the squares utilized in the LM test. The CD statistic is derived as follows:
principal advantage of the PVAR approach. This is achieved with the sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
use of the panel impulse response functions (IRFs), which describe 2T X X
N1 N
CD ¼ r ij
b (9)
the reaction of one variable in response to changes in another N ðN 1Þ
i¼1 j¼iþ1
variable in the system, since all other shocks are maintained equal
to zero. To examine the impulse response functions, we required an Under the null assumption of the absence of cross-sectional
estimation of their confidence intervals, which were generated dependence (H0) with T /∞ and then N /∞ in any order, CD
using Monte Carlo simulations with 1000 bootstraps. In addition, asymptotically follows a normal distribution. Pesaran [96] proves
we evaluated the variance decomposition, which shows the per- that the CD test is likely to have a superior small sample of prop-
centage change in a variable which is explained by the shock to erties (for both N and T small). Once the cross-sectional depen-
another variable accumulated over time. The variance decomposi- dence is confirmed by rejecting the null hypothesis, then the next
tion shows the magnitude of the total effect. We declare the total step will consists of using the appropriate panel units root tests (i.e.,
accumulated effect over 10 annual periods, but longer horizons (20 Pesaran [89] and Smith et al. [91]. In view of the existence of a
periods) provided equivalent results [1]. cross-sectional dependence in the panel, we applied the panel unit
root test of the second generation of Pesaran [89] and Smith et al.
[91]; to establish the integration degree of our candidate variables.
3.3. Panel unit root test 3.3.2. Pesaran [89] panel unit root test
Pesaran [89] proposed a cross-sectional augmented version of
Before proceeding in advance with the PVAR framework, the the IPS-test of Im, Pesaran and Shin [86] that supposes indepen-
first step of the estimation process consists in examining the data dence across units. This is motivated by the fact that the time series
properties of all the series in term of stationarity. In fact, two is found the simultaneously interrelated in many macroeconomic
generations of unit root tests exist for the panel data: the first applications, considering regional or country data [99]. Pesaran
generation admits the presence of cross-sectional independence, [89] reexamined the hypothesis of cross-sectional independence
and the second generation acknowledges the existence of cross- and proposed a single-factor model that takes into account re-
sectional dependence [83]. The first generation include the panel siduals for their heterogeneous loading factors and recommended
unit root tests of Levin et al. [84]; Im et al. [85,86], Maddala and Wu augmenting the conventional ADF specification with the transverse
[87]; and Hadri [88]; whereas the second, it includes Pesaran [89]; averages of the lagged levels and the first differences of the indi-
Moon and Perron [90], Smith et al. [91] and Carrion-i-Silvestre at al. vidual variables. The following regression represents the cross-
[92]. Nevertheless, Panel unit root tests of the first-generation may sectional augmented ADF (CADF):
possibly lead to erroneous and incorrect results because of the
presence of size distortions if significant levels of positive residual DYit ¼ pi þ bi Yi;t1 þ ui Y t1 þ 4i DY t þ εit (10)
cross-section dependence exists and if it is not taken into consid-
eration [87,93,94]. Hence, testing for the cross-sectional depen- 1
P
N
1
P
N
where Y t ¼ N Yit , DY t ¼ N DYit and εit is the regression
dence in a panel study is crucial to select the appropriate estimator. i¼1 i¼1
As in Salim and Rafiq [95] and Salahuddin et al. [36]; to explore the error. The Pesaran [89] panel unit root test is based on the average
presence of a cross-sectional dependence, we performed the
diagnostic test given by Pesaran [96] for cross-sectional
11
dependence. See De Hoyos and Sarafidis [137] for more details.
of the individual ADF statistics augmented in the cross-section. Table 4

Pesaran [89] constructed a modified version of the IPS-test as Results of cross-sectional dependence (CD) test of Pesaran [96].
follows Variables Lag length (p)
X p ¼1 p ¼2 p ¼3 p ¼4
1 N
CIPS ¼ CADFi (11) L 4.10***
[0.000] 3.78***
[0.000] 3.23***
[0.000] 2.65*** [0.004]
N K 3.12*** [0.000] 3.04*** [0.000] 2.54*** [0.002] 2.16*** [0.009]
i¼1
FD 5.67*** [0.000] 5.30*** [0.000] 5.14*** [0.000] 4.53*** [0.000]
REC 3.42*** [0.003] 3.16*** [0.005] 2.88*** [0.006] 2.28*** [0.008]
where CADFi represents the cross-sectionally augmented Dickey-
GDP 5.87*** [0.000] 5.25*** [0.000] 5.12*** [0.000] 4.91*** [0.000]
Fuller statistic for the i-the cross-sectional unit given by the t-ra- CO2 5.82*** [0.000] 5.45*** [0.000] 5.17*** [0.000] 4.70*** [0.000]
tio of bi in the CADF Eq. (10). The CIPS critical values were simulated
Notes: The null hypothesis indicates cross-sectional independence. The CD statistic
by Pesaran for different sample sizes. follows N (0, 1) distributions. [.]: Probabilities. *** and ** denotes the significance at
1% and 5% level, respectively.
3.3.3. Smith et al. [91] panel unit root test

!
The bootstrap panel unit root test proposed by Smith et al. [91] P
ai ; G ¼ 1 P T b
N N
Gt ¼ N1 b a i ; whereas in the other two
employs a ‘sieve’ sampling scheme to account for both the time SEðb
aiÞ
a N b
a ð1Þ
i¼1 i¼1 i
series and cross-sectional dependencies of the data via bootstrap
blocks. The specific tests that we performed are denoted as t, LM, tests Pt ¼ b a ; P ¼ Ta
t b ; the alternative hypothesis is that the
SEðb
aÞ
max and min. Then, t refers to the bootstrap version of the famous panel is cointegrated as a whole [108].13
P
N
panel unit root test from Im et al. [86]; LM ¼ N1 LMi represents
i¼1 4. Results and discussion
the mean of the individual Lagrange multiplier (LMi ) statistics test
initiated by Solo [100]; max specifies the test from Leybourne [101]; 4.1. Panel unit root and panel cointegration tests
PN
and min ¼ N 1 mini is an alternative of the individual Lagrange
i¼1 The results of the CD tests are reported in Table 4 which show
multiplier (LMi) [more powerful] with mini ¼ min (LMfi, LMri), that the null hypothesis of cross-section independence is rejected
where LMfi and LMri are based on forward and backward re- with a high significance, e.g. at 1% level regardless of the number of
gressions.12 All four tests are derived with a unit root under the null lags used (lag length ðpÞ ¼ 1; 2; 3 and 4). The result indicates that
assumption and heterogeneous autoregressive roots under the the MENA countries examined in this study are cross-sectionally
alternative, which implies that a rejection should be taken as evi- dependent, which can be explained by the high economic and
dence in favor of stationarity for at least one country. financial links between many of these countries. Moreover, this
result suggests that panel unit root tests that account for cross-
sectional dependence, such as the tests of Perasan [89] and Smith
3.4. Westerlund [102] cointegration test for panel-dependent units et al. [91] are those that should be used.
The results of the cross-sectional dependence panel unit root
In most cases, researchers make use of the panel cointegration tests of Pesaran [89] and Smith et al. [91] are reported in Table 5.
method proposed by Pedroni [103,104]. Nevertheless, the suppo- Both types of tests show strong evidence that all the six variables
sition of cross-sectional independence suggested by Pedroni are integrated with first order, e.g., Ið1Þ is at the 1% level of signif-
[103,104] is extremely restraining [105]. Considering the existence icance. This means that the first difference is sufficient to make all
of cross-sectional dependence in the panel, the cointegration pro- the series stationary, a necessary condition that is required before
cedure proposed by Westerlund [102] was conducted. Unlike the estimating a PVAR model for these series.
residual-based cointegration tests, this test avoids the problem of However, since all the series under study are integrated with
common factor restriction. In fact, the common factor restriction first order and are cross-sectionally dependent, then it is natural to
occurs under the condition that the long-run cointegrating vector examine whether there is evidence for a long-run relationship
of the variable levels is identical to the short-run adjustment pro- between all the variables. To test for possible existence of a long-
cess of the variables for their first differences [106]. This common run relationship, we employed the Westerlund [102] panel coin-
factor limitation has been used as a reasonable justification for the tegration tests to test the null hypothesis of cointegration against
failure of the null assumption in many studies when cointegration the alternative of absence of cointegration. The results of the
has been robustly suggested in theory, such as in Ho [107]. In fact, Westerlund [102] tests are reported in Table 6, which show that the
Westerlund [102] developed four panel cointegration tests (Gt , Ga , null hypothesis of no cointegration could not be rejected by all four
Pt , Pa ), which are based on structural dynamics rather than residual tests calculated. Therefore, the best way to investigate the possible
and, therefore, require no common limiting factors. The idea is to existence of a significant relationship between all these variables is
verify the null hypothesis of noncointegration by assuming that the to estimate a model for all the variables in first differences. Moti-
error correction term in a conditional error correction specification vated by the recent development of the PVAR model, we estimated
of the panel is equivalent to zero. All tests are normally distributed and discuss in the next section the results of the PVAR model.
and are general enough to account for unit-specific trend, the
short-term dynamics specific to unity, the slope parameters, and 4.2. PVAR estimation results
the cross-sectional dependence and to recommend p-values that
are robust against cross-sectional dependencies that pass through This subsection describes the results of the estimation of the
bootstrapping. Two tests can be considered to check the alternative
hypothesis that at least one unit is cointegrated.
13
For further details regarding the derivation of these statistics, see Westerlund
12
See Smith et al. [91] for more details. [102].
Table 5
Results of panel root tests.
Variables CIPS(2007) Smith et al. [91]
t LM max min
L 3.37 (1.000) 2.30 (0.065) 6.06 (0.303) 2.77 (0.073) 4.67 (0.193)
DðLÞ 8.46*** (0.000) 4.80*** (0.000) 12.44*** (0.000) 4.63*** (0.000) 11.31*** (0.000)
K 2.25 (0.264) 2.13 (0.344) 6.52 (0.186) 2.03 (0.054) 5.05 (0.164)
DðKÞ 15.35*** (0.000) 4.72*** (0.000) 13.67*** (0.000) 4.56*** (0.000) 12.03*** (0.000)
FD 1.13 (0.871) 2.69 (0.170) 6.16 (0.230) 2.71 (0.072) 4.62 (0.183)
DðFDÞ 13.57*** (0.000) 4.45*** (0.000) 13.80*** (0.000) 4.29*** (0.000) 12.53*** (0.000)
REC 0.40 (0.344) 2.23 (0.291) 6.81 (0.177) 2.62 (0.063) 6.31 (0.052)
DðRECÞ 14.41*** (0.000) 4.78*** (0.000) 12.31*** (0.000) 4.39*** (0.000) 12.61*** (0.000)
GDP 1.09 (0.138) 2.81 (0.120) 7.21 (0.101) 2.60 (0.069) 6.45 (0.067)
DðGDPÞ 10.82*** (0.000) 5.69*** (0.000) 16.86*** (0.000) 5.91*** (0.000) 16.43*** (0.000)
CO2 0.94 (0.173) 2.16 (0.311) 6.12 (0.237) 2.80 (0.073) 5.11 (0.091)
DðCO2 Þ 20.21*** (0.000) 5.30*** (0.000) 14.53*** (0.000) 4.70*** (0.000) 12.83*** (0.000)
*** **
Notes:Probability values are reported in parentheses. Panel root test includes intercept and trend. and denotes the significance at 1% and 5% level, respectively. Dð:Þ
denotes the first differences.
Table 6 lagged values. Finally, the results show that financial development
Westerlund [102] ECM panel cointegration tests, 1980e2012. does not positively contribute to capital accumulation, which in-
Statistic Value Robust P-value dicates that financial development in the MENA countries is not
Gt 2.657 0.150
well developed and is still in its first stage of development.
Ga 11.093 0.055 Third, regarding the financial development equation, as ex-
Pt 11.088 0.125 pected, the first and second lagged values of financial development
Pa 8.152 0.355 have positive and significant effects on the current value of finan-
Notes: We allow for a constant and deterministic trend in the cointegration rela- cial development at 1% and 10% level of significance, respectively.
tionship. P-values are robust critical values obtained through bootstrapping pro- Moreover, as expected by the economic theory, we found evidence
cedure with 400 replications. Results for H0: No cointegration. **Critical values at the for significant positive effect of economic growth (the first lagged
5% significance level.
value) on financial development. Additionally, the employment
level (second lag) has a negative effect on the level of financial
PVAR model reported in Table 7.14 First, for the labor equation, the development at the 1% level of significance. This result is inter-
results reported in column 2 of Table 7 show that only four co- esting since it shows that the availability of the labor force slows
efficients are statistically significant at the conventional level of down the process of financial development in the MENA region.
significance. In particular, the results show that both the first and Finally, unexpectedly, CO2 emissions have a positive impact on
second lags of the labor are positively correlated with its current financial development; however, this result is only significant at the
level. The estimated coefficients associated with the first and sec- 10% level.
ond lags of labor are equal to 0.196 and 0.371, and they are signif- Fourth, for the equation related to the renewable energy con-
icant at the 5% and 1% levels, respectively. Additionally, the first lag sumption variable, only the first and second lagged values of
of renewable energy consumption growth positively determines financial development are significant with a positive sum of co-
the current level of labor at the 5% level of significance, which in- efficients equal to 0.52. This result is particularly interesting
dicates that increasing renewable energy consumption will induce because it indicates that the expansion and development of
a decrease in the level of unemployment. Finally, environmental financial activities boosts the renewable energy sector via capital
degradation, which is represented by CO2 emissions, has a signifi- lending and providing equity financing and debt for funding green
cant and negative impact at the 1% level of significance on the projects [40,62,64,110]. This result also confirms the results of Al-
current labor level. This result indicates that improving the envi- Mulali et al. [41] for Latin America and the Caribbean countries,
ronmental quality in the region will help to reduce the level of those of Lin et al. [110] for China, and those of Kim and Park [40] for
unemployment. These results can be considered as one of the a panel of 30 countries that were selected from four continents,
spillovers effects of investing in renewable energy sources [109]. namely, Asia, the Americas, Europe and Oceania.
Second, for the capital equation, the results show that only three Fifth, for the economic growth equation, the results show that
variables have a significant impact on the level of capital. These are five coefficients are significant if the 10% level of significance is
the coefficients associated with the second lags of labor, capital and considered. First, the results show that the first and second lags of
financial development, which are significant at 10%, 5% and 10% the labor force variable strongly determine positively the level of
levels, respectively. All variables have a negative impact on the economic growth at 1% level of significance. Second, regarding the
current level of capital, e.g., all estimated coefficients have negative lagged values of the renewable energy consumption, our results
signs. These results indicate that labor and capital are substitute show that only the second lag which positively determines the
factors and that capital is not used in an efficient way since there current level of economic growth, a result that is in line with pre-
are significant negative relationships between its current and vious studies in the field, e.g., Lin and Moubarak [67] for China,
Jaforullah and King [68] for the USA and Danish et al. [2] for
Pakistan. Third, concerning our second main variable, financial
14 development, we found also that only the first lag, which de-
Before beginning the results and discussion, note that while the order of var-
iables is not important in this step of the PVAR estimation, it is particularly termines economic growth at the 10% level of significance but it has
determinant when conducting the impulse response analysis. The order used in a negative sign. Several reasons can explain this latter result. For
both the estimation and response impulse analysis is L, K, FD, REC, GDP and CO2 instance, the MENA countries seem to do not have comprehensive
emissions. Since the order is not important in the model estimation step, we
financial incentive frameworks and policy schemes that positively
discuss the argumentation and the order employed in the next subsection.
Table 7
Results of panel VAR model.
Response to Response of
DðLt Þ DðKt Þ DðFDt Þ DðRECt Þ DðGDPt Þ DðCO2 t Þ
DðLt1 Þ 0.196** (2.163) 0.155 (0.409) 0.311 (1.084) 0.416 (1.212) 1.034*** (3.183) 0.028 (0.131)
DðLt2 Þ 0.371*** (4.868) 0.614* (1.774) 0.743*** (3.171) 0.113 (0.277) 0.946*** (3.460) 0.196 (1.005)
DðKt1 Þ 0.002 (0.381) 0.187 (1.560) 0.121 (1.506) 0.074 (1.452) 0.009 (0.257) 0.117*** (2.955)
DðKt2 Þ 0.003 (0.840) 0.158** (2.553) 0.047 (1.560) 0.022 (0.847) 0.023 (0.625) 0.078** (2.087)
DðFDt1 Þ 0.011 (1.474) 0.041 (0.516) 0.204*** (3.451) 0.348** (2.072) 0.067* (1.892) 0.033 (0.861)
DðFDt2 Þ 0.009 (1.286) 0.137* (1.830) 0.095* (1.797) 0.173*** (2.595) 0.055 (1.295) 0.007 (0.276)
DðRECt1 Þ 0.012** (2.101) 0.029 (0.722) 0.053 (1.307) 0.111 (1.503) 0.005 (0.402) 0.068** (1.986)
DðRECt2 Þ 0.001 (0.164) 0.040 (0.932) 0.008 (0.297) 0.003 (0.054) 0.063* (1.757) 0.008 (0.215)
DðGDPt1 Þ 0.009 (0.376) 0.101 (0.427) 0.330** (2.235) 0.126 (0.421) 0.298** (1.980) 0.138* (1.871)
DðGDPt2 Þ 0.038 (1.496) 0.101 (0.427) 0.076 (0.657) 0.288 (0.890) 0.077 (0.795) 0.166** (2.157)
DðCO2 t1 Þ 0.001 (0.056) 0.055 (0.495) 0.008 (0.152) 0.057 (0.825) 0.002 (0.062) 0.193 (1.615)
DðCO2 t2 Þ 0.026*** (3.431) 0.114 (1.513) 0.119* (1.805) 0.009 (0.169) 0.023 (0.658) 0.097 (1.433)
Notes: Six-variables VAR model is estimated by GMM. Country-time and fixed effects are removed prior to estimation. Reported numbers show the coefficients of regressing
the column variables on lags of the row variables. Heteroskedasticity adjusted t-statistics are in parentheses. ***, ** and * indicates significance at 1%, 5% and 10% with critical
values of 2.576, 1.96 and 1.645, respectively. D(.) denotes the first differences.
influence their economic growth [111e113]. Moreover, most of the variable only with a lag. Based on this recommendation and the
MENA region countries are in their first stage of financial devel- economic theory as well as previous studies, the following order
opment [114]. Furthermore, during the last two decades, most of was used in this study: L; K; FD; REC; GDP; and CO2 emissions. The
the MENA countries have been subject to financial instability, dis- two variables, L and K, are factors of production, and they should
turbances and crises, such as the 2007e2008 subprime crisis, the appear before all the variables in the model. The CO2 emission is the
sociopolitical events and the spread of conflicts related to the Arab least exogenous variable since it is influenced by all other variables;
Spring. All these factors combined together make the financial thus, it should appear in the last position. FD, REC and GDP are
systems to not perform as well as expected. taken in this order since FD is a determinant for REC and GDP and
Finally, for the CO2 emissions equation, the results are more since REC can also influence GDP. For these last three variables, a
interesting. First, as expected, the coefficients of the two lags of similar order was adopted by Tiwari [29].
economic growth are positively correlated with CO2 emissions, Figs. 1 and 2 (a and b)15 report the IRFs of CO2 emissions and the
where the first lag coefficient is significant at 10% and the second at economic growth with 5% errors bands. Figs. 1(a) and 2(a) illustrate
5%. Additionally, the coefficients of the first and second lags of the the reaction of CO2 emissions and economic growth to one standard
capital variable are also positively correlated with CO2 emissions at deviation shock in financial development. Figs. 1(b) and 2(b) show
the 1% and 5% levels of significance, respectively. This clearly in- the reaction of CO2 emissions and economic growth to one standard
dicates that the capital in the MENA region is employed in activities deviation shock in renewable energy consumption.
that are not friendly to the environment. Regarding the financial
development variable, the results show that it does not explain the
level of CO2 emissions in the MENA region whatever the lag 4.3.1. Impact of shocks of FD and REC on CO2 emissions
considered. Finally, concerning our second main variable of inter- For financial development, since in the PVAR estimation, the
est, the renewable energy consumption variable, the results show financial development does not determine CO2 emissions, then the
that its first lag has a negative impact on CO2 emissions at con- interpretation of the impulse response function to one standard
ventional level, e.g. 5%. This implies that renewable energy plays a deviation shock on FD should be considered carefully. The inter-
vital role in curbing CO2 emissions and in improving the environ- pretation of the results here is to help understanding the expected
mental quality in the MENA region since the sign of the coefficient impact if the FD significantly determines CO2 emissions with the
associated with renewable energy is negative. These results same obtained sign. The results show that the effect of one standard
confirm those of Apergis and Payne [42] for South America, Mou- deviation shock in the growth of financial development on the CO2
tinho and Robaina [43] for 20 European countries and Bhattacharya emissions growth was instantaneously positive but then negative
et al. [6] for a panel of 85 developed and developing economies. from the first year to year 5 and equal to zero from year 6e10. The
results show also that the maximum negative impact occurs in the
first year with a value equal to 0.0135. Overall, the impact of
4.3. Impulse-response functions (IRFs) discussion financial development on CO2 emissions is not very high, and it
decreases quickly, indicating that the contribution of financial in-
This subsection presents and discusses the results of the stitutions to efficiently improving environmental quality, in the
impulse-response functions (IRFs). Precisely, we focus our analysis MENA countries, is very feeble and weak. This result is expected
on the two main variables of interest, renewable energy con- since most of the MENA countries' financial sectors are not well
sumption and financial development. As discussed above, an developed. In fact, the MENA economies have underdeveloped and
important step when conducting the IRFs analysis is to select the inefficient financial and banking systems, which are characterized
appropriate order of variables. The IRFs in the PVAR model, by the lack of good allocation of financial resources [116]. In addi-
developed by Love and Zicchino [1]; are based on the Cholesky tion, the financial liberalization policies implemented by these
decomposition of the matrix of variance-covariance residues to countries is also characterized by their inefficiency at both markets
ensure the orthogonalization of shocks as suggested by Sims [115]. and legal frameworks as well as an inappropriate institutional
Sims [115] recommended that the variables that appear earlier in structure. All these factors led to the outbreak of financial and
the order are more exogenous and affect the following variables
simultaneously or even with a lag, whereas the variables appearing
15
later in the systems are more endogenous and affect the preceding Obtained through simulations of Monte Carlo with 1000 replications.
Fig. 1. Reaction of CO2 emissions to FD and REC one standard deviation shock.
Fig. 2. Reaction of GDP to FD and REC one standard deviation shock.
banking crises that have marginalized the role of their financial growth of one standard deviation shock to financial development
systems and have contributed to their instability [117]. and to the growth of renewable energy consumption, respectively.
For renewable energy consumption, the results show that the The results show that the impact on economic growth as a result of
response of the growth of CO2 emissions to one standard deviation one standard deviation shock to financial development was nega-
shock in the growth of renewable energy consumption was negative during the first two years and became positive for the rest of
tive at s ¼ 0 and in the first year (s ¼ 1). Then, it is weakly positive the period (up to ten). Additionally, the impact remained positive
in year 2, and the impact of shocks die completely after 3 years. but very weak at the end of the period. Moreover, the negative
Regarding the overall effect, the cumulative effect is negative impact of the shock of financial development on economic growth
indicating that renewable energy consumption can be employed as is explained by the fact that the MENA countries are experiencing a
an effective tool to reduce emissions derived from carbon dioxide period of financial instability, increases in asymmetric information
(CO2). Hence, renewable energies may help to reduce the level of and several financial problems, inducing a rise in uncertainty which
CO2 emissions and environmental degradation in the short-term. has lead to an economic recession. In fact, the spread of conflicts in
This outcome is in accordance with some previous empirical find- the MENA region has led to the increase in financial instability,
ings, which have confirmed the fundamental role of renewable which in turn has caused a rise in asymmetric information. This
energy in the reduction of CO2 emissions. It also supports the idea affect the performance of companies and creates financial hardship,
that enhancing the development of renewable energy sources is an and which may adversely influence economic growth [112,119].
efficient way to moderate CO2 emissions [68,69,71,2,6,55]; among
others). However, our finding is not consistent with other empirical 4.3.3. However, the results indicated that the response of economic
results that have concluded that an increase in renewable energy growth to one standard deviation shock in the growth of renewable
consumption does not contribute to reductions in environmental energy consumption was always positive and dies after eight years
degradation [66,118]; among others). In this regard, Apergis et al. Moreover, the results show that the impact became very weak
[118] argued that the renewable energy supply cannot be relied on
from the sixth year. This result is an evidence that the renewable
to meet the peaks in demand. Consequently, a back up to traditional
energy sector in the MENA countries is in its initial phase of
energy sources is required, and this would typically make use of
development, which requires higher costs and cannot satisfy rising
fossil fuels. Importantly, most of the previous studies have focused
energy needs. In fact, barriers and obstacles related to the high cost
on exploring the effects and/or responses of CO2 emissions to
of renewable energy and energy shortages due to population
renewable energy consumption shocks in the long-term and not in
growth, all of which cause major growth in term of energy demand
the short-term.
in the MENA countries, have an adverse influence on the economic
growth. Explicitly, the production of renewable energy sources is
4.3.2. Impact of FD and REC shocks on economic growth more expensive than the production from traditional sources. In
Fig. 2(a) and (b) show the results of the impact on economic reality, most renewable energy technologies in the MENA region
are still in their early stages in term of commercial development, have a direct impact on environmental quality and economic
and much infrastructure is required to be built for this develop- growth. Based on the results discussed above, several economic,
ment. Furthermore, the labor forces in the MENA region are poorly energy and environmental policies can be recommended. In this
skilled, and this deficit could unexpectedly increase the incurred study, we classify these policies into two subsections: (1) policies
costs and lower the productivity and efficiency of renewable energy suggested to boost economic growth and (2) policies to improve
projects. All the aforementioned reasons slow down the short-term environmental quality.
economic growth in the region [120,121].
5.1. Policies to boost economic growth
4.4. Variance decomposition analysis
5.1.1. Policies related to the financial sector
An analysis of the financial system of the MENA countries re-
Although impulse responses can give details regarding the in-
veals many deficiencies and shows that the system is overall not as
fluence of variations in one variable on other variables, they do not
well developed as it should be, given the fact that most of the
specify the magnitude and degree of these effects. As a result, the
countries are well endowed with natural resources. In this study,
variance decomposition technique was performed to determine
since we found a nonsignificant impact at 5% level of significance of
this. The variance decomposition provides information about the
financial development on economic growth, then it is important
variation in percentages in the dependent series that are attribut-
that the MENA region should rapidly develop their financial and
able not only to their own shocks but also to shocks generated by
banking systems via implementation of more efficient financial
the other variables. Table 8 shows the results of the variance
reforms. The proposed reforms should be designed to improve both
decomposition obtained from the orthogonalized impulse response
the level and quality of financial services to achieve expected re-
coefficient matrices. The variance decomposition shows that labor
sults behind the development of the financial system [57e59]. We
explains approximately 26% of the variations in economic growth
believe that this can be achieved by encouraging competition in the
and 1.82% of the fluctuations in CO2 emissions, capital explains
financial system and expanding its structure and be providing a
approximately 0.98% of the fluctuations in economic growth and
legal framework and a sound institutional environment to exert
3.12% of the variations in CO2 emissions, financial development
positive effects on the sector. Moreover, it is important to improve
explains approximately 8.72% of the changes in economic growth
the quality of financial allocation to increase the productivity and
and 3.21% of the fluctuations in CO2 emissions, and renewable
yield of the sector. An important point here is that governments
energy consumption explains approximately 0.61% of the variations
and policymakers should show a real desire to improve and
in economic growth and 0.53% of the changes in CO2 emissions. The
develop financial systems for both local and international investors
last fluctuations confirm that the influences of both renewable
[117].
energy consumption and financial development are relatively
In addition, the MENA economies should improve the process of
small, and each has little ability to explain the two variables of
credit allocation via the privatization of national banks, the rein-
interest (CO2 emissions and economic growth). This confirms that
forcing of credit regulation and an increase in competition in the
the renewable energy sector in the countries of the MENA region is
banking sector. Moreover, a well-developed regulatory infrastruc-
in its early developmental stage, and there is still a strong depen-
ture is required as well as the implementation of procedures to
dence on conventional and traditional energy sources to satisfy the
lower the excessive volatility of stock prices. These measures will
intensifying demand. With regard to the financial sector, the MENA
allow the stock markets of the MENA region to spur economic
economies are characterized by weaker financial institutions and
growth [114].
instable periods. More explicitly, there is a “resource curse” related
Overall, it is evident that improving the performance of the
to the financial sector that results to less contributions via in-
financial system in the region is essential to allow the financial
vestments in projects, for example, green and eco-friendly projects,
sector to be a real stimulator of economic growth.
and this curse directly impacts economic growth [122,123]. In this
regard, financial development strongly contributes to environ-
5.1.2. Policies related to the renewable energy sector
mental quality improvement and economic growth in the region
All the MENA countries are well located geographically,
when a certain threshold level of financial development is attained.
providing them with a high potential for producing renewable
energy via solar and wind [121]. However, since the empirical re-
5. Policy implications sults show evidence for a positive effect of renewable energy con-
sumption on economic growth only at the 10% level of significance,
Improving environmental quality without reducing economic it is important to understand these unexpected results of the slight
growth is one of the most challenging tasks for MENA's govern- efficiency of renewable energy consumption in stimulating eco-
ments and policymakers. The success of new policies and strategies nomic growth in the MENA region. In particular, we recommend
mainly rely on a comprehensive understanding of the variables that that the MENA governments and policymakers not only promote
renewable energy consumption but also facilitate financial funding
and stimulate the use of renewable energy sources in efficient ways
Table 8
by consumers and industries. These measures will allow them to
Variance decomposition of the panel VAR model (%).
fully benefit from the advantages of renewable energy sources. In
DðLÞ DðKÞ DðFDÞ DðRECÞ DðGDPÞ DðCO2 Þ addition, the MENA countries should restructure the renewable
93.88 1.10 1.54 1.18 26 1.82 energy market, develop a legal and policy framework related to this
0.08 92.85 2.40 1.17 0.98 3.12 sector, and try to develop the required skills and information
2.02 2.51 85.58 3.44 8.72 3.21
[124,125].
0.82 0.29 0.47 91.05 0.61 0.53
2.21 2.74 9.12 3.02 63.39 6.61 In fact, it is likely that investing in renewable energy certainly
0.99 0.51 0.89 0.14 0.30 84.71 leads to an enhancement in economic growth [126,127]. The
Notes: The results are based on the orthogonalized impulse-responses. Percent of
rationale behind this association lies in the concept of job creation
variation in the column variable (10 periods ahead) explained by the row variable. [128,129]. For instance, encouraging investment in renewable en-
Dð:Þ denotes the first differences. ergy sources would make it possible to create novel business
opportunities; for example, mitigation credit trading, wetland 6. Conclusion, limits and extension
banking and threatened species banking. All these actions would
make new opportunities for employment and hence boost eco- This study examined the impact of renewable energy con-
nomic growth. sumption and financial development on CO2 emissions and eco-
nomic growth in the MENA region by employing the PVAR
technique that was recently developed by Love and Zicchino [1].
5.2. Policies to improve environmental quality The most important results of the paper can be summarized in
four important points. First, we found that renewable energy con-
5.2.1. Policies related to the financial sector sumption has a negative impact on CO2 emissions, indicating the
As discussed earlier, in the empirical literature, there is a possibility of improving the environmental quality in the MENA
growing consensus among researchers on the fact that a well- region by promoting the renewable energy sector. However, the
developed financial system should be certainly associated with results showed an absence of any negative significant impacts of
environment protection and safety. However, since our empirical financial development on environmental degradation. Second, we
results show an absence of the expected positive impact of financial also found a weak positive impacts, significant at 10% level of sig-
development on environmental quality, it is first important for nificance, of renewable energy consumption on economic growth.
MENA region policymakers to understand and analyze the reasons However, the results show evidence of unexpected results of a
behind this unexpected result. In particular, the MENA govern- negative significant impact of financial development on economic
ments and policymakers should continue the process of financial growth. Third, the results show that the labor force is the main
reforms by promoting investments in research and development driver of economic growth for the MENA region. Additionally, while
for innovative technologies with regard to green energy in which we found that the level of capital harms the quality of the envi-
the MENA countries have a comparative advantage (such as ronment, it is not a significant contributor to economic growth at all
biomass, biogas, hydro, solar and solid waste) [62,64]. level of significance. Finally, the results show that financial devel-
Therefore, through the process of its development, the financial opment has a negative significant impact on renewable energy
system should support the quality of the environment by orienting consumption, indicating that the financial sector does not play its
banking operations to the support of green energy and low carbon expected role in improving the renewable energy sector.
projects. For instance, the banking system should encourage in- Regarding the impulse response results, our empirical findings
vestments in energy-proficient technologies by providing interest showed that the response of economic growth to one standard
discounts and taking into consideration the carbon-related re- shock on renewable energy consumption displays a positive sign
quirements of their financial products. Moreover, financial orga- but its effect is decreasing overtime until period 8 where it will
nizations could propose special financial services for investment in completely die. Additionally, the CO2 emissions response to one
eco-friendly associated projects [40,41]. Moreover, a good coordi- standard deviation shock on renewable energy consumption show
nation among several stakeholders, such as research institutes, a negative impact, indicating that increasing the consumption of
different government ministries, industry and universities is widely renewable energy sources is considered an effective and consistent
required to implement successful mitigation strategies. solution to restraining emissions mainly derived from carbon di-
oxide (CO2). Regarding financial development, the results provide
evidence that the response of both economic growth and CO2
5.2.2. Policies related to the renewable energy sector emissions to one standard deviation shock on financial develop-
Given the crucial role of renewable energy in reducing CO2 ment is negative during the first four periods for CO2 emissions and
emissions, governments should plan and put into practice effective weakly positive from s ¼ 5 to s ¼ 9 and dies completely after that.
and valuable support policies to stimulate the investment process For economic growth, the impact is negative during the first three
in new technologies and to create energy expertise in an attempt to periods and positive from period four to ten with a decreasing
improve environmental quality [27,32,130,131]. One of the most behavior.
important measures to support both renewable energy and energy Furthermore, the variance decomposition analysis confirmed
efficiency schemes in the MENA countries is the excessive use of the that both renewable energy consumption and financial develop-
Clean Development Mechanism (CDM), whose development is ment have a slight influence and can slightly explain the two var-
certainly associated with environmental protection and safety. This iables of interest (i.e., CO2 emissions and economic growth).
is motivated by the fact that the MENA region is an attractive CDM Additionally, the financial sector of the MENA countries is still too
destination since it is not only wealthy in renewable energy sources weak to contribute to environmental quality improvements and
but also has a robust gas and oil industry. Furthermore, substantial economic growth; however, this can occur once a certain threshold
changes are required so that the MENA countries can exploit their of financial development is reached.
renewable energy potential and position themselves as eco- The main shortcoming of this study include ignoring the het-
friendly countries [132]. The main changes are supposed to occur erogeneity among the MENA countries. Thus, a possible extension
at the following levels. First, setting mandatory renewable energy of this investigation may be to segment the entire sample of
targets within a specified time period. Second, the creation of a countries into different subsamples depending on the evolution of
renewable energy agency, which is a crucial step to push renewable the renewable energy share to assess the main determinant of
energy development and to increase the transparency of the sys- economic growth and environment degradation for each analyzed
tem by avoiding overlaps. Finally, the gradual removal of subsidies MENA subgroup. Other possible extensions of this study may
from traditional (conventional) energy technologies by continuing include incorporating other institutional and sociodemographic
to adjust energy prices, including the externalities of energy pro- indicators and other proxies of environmental degradation.
duction and the exclusion of subsidies in particular, is necessary to
ensure a more level playing field for renewables. Certainly, such a
procedure should be smoothly executed with regards to local so- Appendix A. Supplementary data
cioeconomic constraints. Consequently, a set of practical and real-
istic incentives and policies that support more low-carbon finance Supplementary data to this article can be found online at
is an essential mission for conserving MENA's resources. https://doi.org/10.1016/j.renene.2019.01.010.
References 1793e1806.
[30] K. Van den Bergh, E. Delarue, W. D'haeseleer, Impact of renewables
deployment on the CO2 price and the CO2 emissions in the European elec-
[1] Inessa Love, L. Zicchino, Financial development and dynamic investment
tricity sector, Energy Policy 63 (2013) 1021e1031.
behavior: evidence from panel VAR, Q. Rev. Econ. Finance 46 (2006) €
[31] F.C. Ozbugday, B.C. Erbas, How effective are energy efficiency and renewable
190e210.
energy in curbing CO2 emissions in the long run? A heterogeneous panel
[2] Danish, B. Zhang, B. Wang, Z. Wang, Role of renewable energy and non-
data analysis, Energy 82 (2015) 734e745.
renewable energy consumption on EKC: evidence from Pakistan, J. Clean.
[32] M. Kahia, M.S. Ben Aïssa, M. Kadria, L. Charfeddine, Modelling the treatment
Prod. 156 (2017) 855e864.
effect of renewable energy policies on economic growth: evaluation from
[3] P. Sadorsky, Some future scenarios for renewable energy, Futures 43 (2011a)
MENA countries, J. Clean. Prod. 149 (2017b) 845e855.
1091e1104.
[33] F. Abbasi, K. Riaz, CO2 emissions and financial development in an emerging
[4] X.-C. Yuan, Y.-M. Wei, B. Wang, Z. Mi, Risk management of extreme events
economy: an augmented VAR approach, Energy Policy 90 (2016) 102e114.
under climate change, J. Clean. Prod. 166 (2017) 1169e1174.
€necke, H. Spielmann, U.M. Azeiteiro, F. Alves, M.L. De Car- [34] J. Frankel, A. Rose, An estimate of the effect of common currencies on trade
[5] W.L. Filho, J. Bo
and income, QJE Con (2002) 437e466.
valho, G.J. Nagy, Climate change and health: an analysis of causal relations on
[35] A.L. Jensen, Beverton and Holt life history invariants result from optimal
the spread of vector-borne diseases in Brazil, J. Clean. Prod. (2018), https://
trade off of reproduction and survival, Can. J. Fish. Aquat. Sci. 53 (1996)
doi.org/10.1016/j.jclepro.2017.12.144.
820e822.
[6] M. Bhattacharya, S.A. Churchill, S.R. Paramati, The dynamic impact of
[36] M. Salahuddin, J. Gow, I. Ozturk, Is the long-run relationship between eco-
renewable energy and institutions on economic output and CO2 emissions
nomic growth, electricity consumption, carbon dioxide emissions and
across regions, Renew. Energy 111 (2017) 157e167.
financial development in Gulf Cooperation Council Countries robust? Renew.
[7] M. Gabriela, How much can CO2 emissions be reduced if fossil fuel subsidies
Sustain. Energy Rev. 51 (2015) 317e326.
are removed? Energy Econ. 64 (2017) 91e104.
[37] S. Nasreen, S. Anwar, I. Ozturk, Financial stability, energy consumption and
[8] K. Perez, C.G.-A. Marcela, I. Alfredo, Energy and GHG emission efficiency in
environmental quality: evidence from South Asian economies, Renew. Sus-
the Chilean manufacturing industry: sectoral and regional analysis by DEA
tain. Energy Rev. 7 (2017) 1105e1122.
and Malmquist indexes, Energy Econ. 66 (2017) 290e302.
[38] C. N. Brunnschweiler, Finance for renewable energy: an empirical analysis of
[9] N. Stern, S. Peters, V. Bakhshi, A. Bowen, C. Cameron, S. Catovsky, D. Crane,
developing and transition economies, Environ. Dev. Econ. 15 (2010)
S. Cruickshank, S. Dietz, N. Edmonson, S.-L. Garbett, L. Hamid, G. Hoffman,
241e274.
D. Ingram, B. Jones, N. Patmore, H. Radcliffe, R. Sathiyarajah, M. Stock,
[39] T.D. Corsatea, S. Giaccaria, R.L. Ar antegui, The role of sources of finance on
C. Taylor, T. Vernon, H. Wanjie, D. Zenghelis, Stern Review: the Economics of
the development of wind technology, Renew. Energy 66 (2014) 140e149.
Climate Change, HM Treasury, London, 2006.
[40] J. Kim, K. Park, Financial development and deployment of renewable energy
[10] IPCC (International Panel on Climate Change Working Group II), Climate
technologies, Energy Econ. 59 (2016) 238e250.
Change 2007: Impacts, Adaptation and Vulnerability. Published for the In-
[41] U. Al-Mulali, C.F. Tang, I. Ozturk, Estimating the environment kuznets curve
ternational Panel on Climate Change, 2007.
hypothesis: evidence from Latin America and the caribbean countries,
[11] NIEHS (National Institutes of Environmental Health Science), A Human
Renew. Sustain. Energy Rev. 50 (2015) 918e924.
Health Perspective on Climate Change, 2010. http://www.niehs.nih.gov/
[42] N. Apergis, J.E. Payne, Renewable energy, output, carbon dioxide emissions,
health/docs/climater eport2010.pdf.
and oil prices: evidence from South America, Energy Sources Part B Energy
[12] O. Deschenes, Temperature, human health, and adaptation: a review of the
Econ. Plan. Policy 10 (3) (2015) 281e287.
empirical literature, Energy Econ. 46 (2014) 606e619.
[43] V. Moutinho, M. Robaina, Is the share of renewable energy sources deter-
[13] B. Li, S. Wu, Effects of local and civil environmental regulation on green total
mining the CO2 kWh and income relation in electricity generation? Renew.
factor productivity in China: a spatial Durbin econometric analysis, J. Clean.
Prod. 153 (2017) 342e353.
[44] M. Kahia, M.S. Ben Aïssa, L. Charfeddine, Impact of renewable and non-
[14] Z.-N. Lu, H. Chen, Y. Hao, J. Wang, X. Song, T.M. Mok, The dynamic rela-
renewable energy consumption on economic growth: new evidence from
tionship between environmental pollution, economic development and
the MENA Net Oil Exporting Countries (NOECs), Energy 116 (2016) 102e115.
public health: evidence from China, J. Clean. Prod. 166 (2017) 134e147.
[45] G. Bo €lük, M. Mert, Fossil & renewable energy consumption, GHGs (green-
[15] L. Guo, Y. Qu, M.L. Tseng, The interaction effects of environmental regulation
house gases) and economic growth: evidence from a panel of EU (European
and technological innovation on regional green growth performance,
Union) countries, Energy 74 (2014) 439e446.
J. Clean. Prod. 162 (2017) 894e902.
laïd, M. Youssef, Environmental degradation, renewable and non- [46] H. Dai, X. Xie, Y. Xie, J. Liu, T. Masui, Green growth: the economic impacts of
[16] F. Be
large-scale renewable energy development in China, Appl. Energy 162
renewable electricity consumption, and economic growth: assessing the
(2016) 435e449.
evidence from Algeria, Energy Policy 102 (2017) 277e287.
[47] I.A. Nienhueser, Y. Qiu, Economic and environmental impacts of providing
[17] IPCC, Climate Change 2014 Synthesis Report Summary for Policymakers,
renewable energy for electric vehicle charging e a choice experiment study,
2014.
€ € Ozdemir,
€ Appl. Energy 180 (2016) 256e268.
[18] S. Ozokcu, O. Economic growth, energy, and environmental Kuz-
[48] D. Spiegel-Feld, B. Rudyk, G. Philippidis, Allocating the economic benefits of
nets curve, Renew. Sustain. Energy Rev. 72 (2017) 639e647.
renewable energy between stakeholders on Small Island Developing States
[19] J. Sathaye, P.R. Shukla, N.H. Ravindranath, Climate change, sustainable
(SIDS): arguments for a balanced approach, Energy Policy 98 (2016)
development and India: global and national concerns, Curr. Sci. India 90
744e748.
(2006) 314e325.
[49] J. Blazejczak, F.G. Braun, D. Edler, W.-P. Schill, Economic effects of renewable
[20] L.M. Cardenas, C.J. Franco, I. Dyner, Assessing emissions-mitigation energy
energy expansion: a model-based analysis for Germany, Renew. Sustain.
policy under integrated supply and demand analysis: the Colombian case,
Energy Rev. 40 (2014) 1070e1080.
J. Clean. Prod. 112 (2016) 3759e3773.
[50] M. Ortega-Izquierdo, P. del Río, Benefits and costs of renewable electricity in
[21] V. Moutinho, C. Varum, M. Madaleno, How economic growth affects emis-
Europe, Renew. Sustain. Energy Rev. 61 (2016) 372e383.
sions? An investigation of the environmental Kuznets curve in Portuguese
[51] Y. Mu, C. Wang, W. Cai, The economic impact of China's INDC: distinguishing
and Spanish economic activity sectors, Energy Policy 106 (2017) 326e344.
the roles of the renewable energy quota and the carbon market, Renew.
[22] X. Zhang, Y. Wang, How to reduce household carbon emissions: a review of
experience and policy design considerations, Energy Policy 102 (2017)
[52] A.O. Adewuyi, O.B. Awodumi, Renewable and non-renewable energy-
116e124.
growth-emissions linkages: review of emerging trends with policy impli-
[23] N. Apergis, C.F. Tang, Is the energy-led growth hypothesis valid? New evi-
cations, Renew. Sustain. Energy Rev. 69 (2017) 275e291.
dence from a sample of 85 countries, Energy Econ. 38 (2014) 24e31.
[53] N. Apergis, J.E. Payne, Renewable energy, output, CO2 emissions, and fossil
[24] P.-Y. Chen, S.-T. Chen, C.-S. Hsu, C.-C. Chen, Modeling the global relationships
fuel prices in Central America: evidence from a nonlinear panel smooth
among economic growth, energy consumption and CO2 emissions, Renew.
transition vector error correction model, Energy Econ. 42 (2014) 226e232.
[54] P. Sadorsky, Renewable energy consumption and income in emerging
[25] M.A. Destek, A. Aslan, Renewable and non-renewable energy consumption
economies, Energy Policy 37 (10) (2009) 4021e4028.
and economic growth in emerging economies: evidence from bootstrap
[55] K. Ito, CO2 emissions, renewable and non-renewable energy consumption,
panel causality, Renew. Energy 111 (2017) 757e763.
and economic growth: evidence from panel data for developing countries,
[26] L. Charfeddine, A.Y. Al-Malk, K.A. Korbi, Air Pollution in Qatar: Causes and
Int. Econ. 151 (2017) 1e6.
Challenges Air Pollution in Qatar: Causes and Challenges, WP/CBE. 1, 2016,
[56] S. Silva, I. Soares, C. Pinho, The Impact of Renewable Energy Sources on
pp. 1e7.
Economic Growth and CO2 Emissions - a SVAR Approach, Universidade do
[27] M. Kahia, M.S. Ben Aïssa, L. Charfeddine, Renewable and non-renewable
Porto, Faculdade de Economia do Porto, 2012, p. 407.
energy use - economic growth nexus: the case of MENA Net Oil Importing
[57] R.I. McKinnon, Money and Capital in Economic Development, Brookings
Countries, Renew. Sustain. Energy Rev. 71 (2017a) 127e140.
Institution, Washington DC, 1973.
[28] L. Charfeddine, A.Y. Al-Malk, K.A. Korbi, Is it possible to improve environ-
[58] J.A. Schumpeter, The Theory of Economic Development, Harvard University
mental quality without reducing economic growth: evidence from the Qatar
Press, Cambridge, MA, 1911.
economy, Renew. Sustain. Energy Rev. 82 (2018) 25e39.
[59] E.S. Shaw, Financial Deepening in Economic Development, Oxford University
[29] A.K. Tiwari, A structural VAR analysis of renewable energy consumption, real
Press, New York, 1973.
GDP and CO2 emissions: evidence from India, Econ. Bull. 31 (2) (2011)
[60] P. Sadorsky, The impact of financial development on energy consumption in
emerging economies, Energy Policy 38 (2010) 2528e2535. [93] P.G.J. O'Connell, The overvaluation of purchasing power parity, J. Int. Econ. 44
[61] J. Frankel, D. Romer, Does trade cause growth? Am. Econ. Rev. 89 (3) (1999) (1998) 1e19.
379e399. [94] A. Banerjee, M. Marcellino, C. Osbat, Testing for PPP: should we use panel
[62] A. Tamazian, B.B. Rao, Do economic, financial and institutional developments methods? Empir. Econ. 30 (2005) 70e91.
matter for environmental degradation? Evidence from transitional econo- [95] R.A. Salim, S. Rafiq, Why do some emerging economies proactively accelerate
mies, Energy Econ. 32 (1) (2010) 137e145. the adoption of renewable energy? Energy Econ. 34 (2012) 1051e1057.
[63] F. Hayat, M.D.S. Pirzada, A.A. Khan, The validation of Granger causality [96] M.H. Pesaran, General Diagnostic Tests for Cross Section Dependence in
through formulation and use of finance-growth-energy indexes, Renew. Panels; CESifo Working Paper No. 1229, Cambridge University, United
Sustain. Energy Rev. 81 (2) (2018) 1859e1867. Kingdom, 2004.
[64] A. Tamazian, J. Chousa, K. Vadlamannati, Does higher economic and financial [97] T. Breusch, A. Pagan, The Lagrange multiplier test and its application to
development lead to environmental degradation: evidence from BRIC model specification in econometrics, Rev. Econ. Stud. 47 (1980) 239e253.
countries? Energy Policy 37 (2009) 246e253. [98] D. Dickey, W. Fuller, Distribution of the estimators for autoregressive time
[65] L. Charfeddine, The impact of energy consumption and economic develop- series with a unit root, J. Am. Stat. Assoc. 74 (1979) 427e431.
ment on ecological footprint and CO2 emissions evidence from a Markov [99] M. Christidou, T. Panagiotidis, Purchasing power parity and the european
switching equilibrium correction model, Energy Econ. 65 (2017) 355e374. single currency: some new evidence, Econ. Modell. 27 (2010) 1116e1123.
[66] K. Menyah, Y. Wolde-Rufael, CO2 emissions, nuclear energy, renewable en- [100] V. Solo, The order of differencing in ARIMA models, J. Am. Stat. Assoc. 79
ergy and economic growth in the US, Energy Policy 38 (2010) 2911e2915. (1984) 916e921.
[67] B. Lin, M. Moubarak, Renewable energy consumption e economic growth [101] S. Leybourne, Testing for unit roots using forward and reverse Dickey fuller
nexus for China, Renew. Sustain. Energy Rev. 40 (2014) 111e117. regressions, Oxf. Bull. Econ. Stat. 57 (4) (1995) 559e571.
[68] M. Jaforullah, A. King, Does the use of renewable energy sources mitigate [102] J. Westerlund, Testing for panel cointegration with multiple structural
CO2 emissions? A reassessment of the US evidence, Energy Econ. 49 (2015) breaks, Oxf. Bull. Econ. Stat. 68 (2007) 101e132.
711e717. [103] P. Pedroni, Critical values for cointegration tests in heterogeneous panels
[69] X. Long, E.Y. Naminse, J. Du, J. Zhuang, Nonrenewable energy, renewable with multiple regressors, Oxf. Bull. Econ. Stat. 61 (1999) 653e670.
energy, carbon dioxide emissions and economic growth in China from 1952 [104] P. Pedroni, Panel cointegration: asymptotic and finite sample properties of
to 2012, Renew. Sustain. Energy Rev. 52 (2015) 680e688. pooled time series tests with an application to the ppp hypothesis: new
[70] J.P.C. Bento, V. Moutinho, CO2 emissions, non-renewable and renewable results, Econom. Theor. 20 (2004) 597e627.
electricity production, economic growth, and international trade in Italy, [105] A. Banarjee, J. Carrion -i- Silvestre, Cointegration in Panel Data and with
Renew. Sustain. Energy Rev. 55 (2016) 142e155. Breaks and Cross-Section Dependence, European Central Bank, 2006.
[71] P. Chunark, B. Limmeechokchai, S. Fujimori, T. Masui, Renewable energy Working paper no. 591.
achievements in CO2 mitigation in Thailand's NDCs, Renew. Energy 114 [106] J. Kremers, N. Ericsson, J. Dolado, The power of cointegration tests, Oxf. Bull.
(2017) 1294e1305. Econ. Stat. 54 (1992) 325e348.
[72] Y.-.J. Zhang, The impact of financial development on carbon emissions: an [107] T. Ho, A panel cointegration approach to the saving-investment correlation,
empirical analysis in China, Energy Policy 39 (2011) 2197e2203. Empir. Econ. 27 (2002) 91e100.
[73] M. Shahbaz, S.A. Solarin, H. Mahmood, M. Arouri, Does financial develop- [108] D. Persyn, J. Westerlund, Error-correctionebased cointegration tests for
ment reduce CO2 emissions in Malaysian economy? A time series analysis, panel data, STATA J. 8 (2008) 232e241.
Econ. Modell. 35 (2013) 145e152. [109] M. Prasad, S. Munch, State-level renewable electricity policies and re-
[74] I. Ozturk, A. Acaravci, The long-run and causal analysis of energy, growth, ductions in carbon emissions, Energy Policy 45 (2012) 237e242.
openness and financial development on carbon emissions in Turkey, Energy [110] B. Lin, O.E. Omoju, J.U. Okonkwo, Factors influencing renewable electricity
Econ. 36 (2013) 262e267. consumption in China, Renew. Sustain. Energy Rev. 55 (2016) 687e696.
[75] L. Charfeddine, K. Ben Khediri, Financial development and environmental [111] M. Beser, N.O. Kilic, The influence of Arab spring effect on economic growth
quality in UAE: cointegration with structural breaks, Renew. Sustain. Energy in the Middle East countries: structural break panel data analysis, Eurasian J.
Rev. 55 (2016) 1322e1335. Econ. Financ. 5 (3) (2017) 133e141.
[76] S.J.H. Shahzad, R.R. Kumar, M. Zakaria, M. Hurr, Carbon emission, energy [112] S. Ghosh, Political transition and bank performance: how important was the
consumption, trade openness and financial development in Pakistan: a Arab spring? J. Comp. Econ. 44 (2016) 372e382.
revisit, Renew. Sustain. Energy Rev. 70 (2017) 185e192. [113] IMF (International Monetary Fund), Toward New Horizons: Arab Economic
[77] P. Sadorsky, Financial development and energy consumption in Central and Transformation amid Political Transitions (Washington), 2015.
Eastern European frontier economies, Energy Policy 39 (2011b) 999e1006. [114] K. Gazdar, M. Cherif, Institutions and the financeegrowth nexus: empirical
[78] U. Al-Mulali, B.C.N. Che Sab, The impact of energy consumption and CO2 evidence from MENA countries, Borsa Istanb. Rev. 15 (2015) 137e160.
emission on the economic growth and financial development in the Sub [115] C.A. Sims, Macroeconomics and reality, Econometrica 48 (1980) 1e48.
Saharan African countries, Energy 39 (2012) 180e186. [116] P. Demetriades, S. Law, Finance, institutions and economic development, Int.
[79] H.A. Bekhet, A. Matar, T. Yasmin, CO2 emissions, energy consumption, eco- J. Finance Econ. 11 (2006) 245e260.
nomic growth, and financial development in GCC countries: dynamic [117] M. Kenza, G.N. Salah Eddine, The effect of the financial sector development
simultaneous equation models, Renew. Sustain. Energy Rev. 70 (2017) on growth: the case of the MENA countries, Arab Econ. Bus. J. 11 (2016)
117e132. 72e85.
[80] B. Vogelvang, in: first ed.Econometrics: Theory and Applications with [118] N. Apergis, J.E. Payne, K. Menyah, Y. Wolde-Rufael, On the causal dynamics
EViews, Prentice Hall, UK, 2005, pp. 14e16. between emissions, nuclear energy, renewable energy, and economic
[81] A.P. Kireyev, Comparative Macroeconomic Dynamics in the Arab World; A growth, Ecol. Econ. 69 (2010) 2255e2260.
Panel Var Approach, IMF Working Papers 00/54, International Monetary [119] P. Richard, Financial Market Instability and CO2 Emissions. GREDI Working
Fund, 2000. Paper No. 10-20, University of Sherbrooke, 2010.
[82] M. Arellano, O. Bover, Another look at the instrumental variable estimation [120] J. Corfee-Morlot, V. Marchal, C. Kauffmann, C. Kennedy, F. Stewart,
of error component models, J. Econom. 68 (1995) 29e51. C. Kaminker, G. Ang, Towards a Green Investment Policy Framework: the
[83] L. Barbieri, Panel Unit Root Tests: A Review, Universit a Cattolica del Sacro Case of Low-Carbon, Climate-Resilient Infrastructure, OECD, Paris, 2012.
Cuore, Dipartimento di Scienze Economiche e Sociali-Piacenza, 2010. Avail- [121] OECD, Renewable Energies in the Middle East and North Africa: Policies to
able from: http://www3.unicatt.it/unicattolica/dipartimenti/DISES/allegati/ Support Private Investment, OECD Publishing, 2013.
LBarbie ri43.pdf. [122] A. Barajas, R. Chami, S. Yousefi, The impact of financial development on
[84] A. Levin, C.F. Lin, C.S.J. Chu, Unit root tests in panel data: asymptotic and growth in the Middle East and North Africa, in: Middle East and Central Asia
finite sample properties, J. Econom. 108 (2002) 1e24. Regional Economic Outlook. Chapter 3.3 IMF: Washington DC, 2011.
[85] K.S. Im, M.H. Pesaran, Y. Shin, Testing for Unit Roots In Heterogeneous [123] H. Hamadi, C. Bassil, Financial development and economic growth in the
Panels, Mimeo, Department of Applied Economics, University of Cambridge, MENA region, Comp. Econ. Stud. 57 (2015) 598e622.
1997. [124] T.M. Sooriyaarachchi, I.-T. Tsai, S. El Khatib, A.M. Farid, T. Mezher, Job crea-
[86] K.S. Im, M.H. Pesaran, Y. Shin, Testing for unit roots in heterogeneous panels, tion potentials and skill requirements in, PV, CSP, wind, water-to-energy and
J. Econom. 115 (2003) 53e74. energy efficiency value chains, Renew. Sustain. Energy Rev. 52 (2015)
[87] G. Maddala, S. Wu, A comparative study of unit root tests and a new simple 653e668.
test, Oxf. Bull. Econ. Stat. 61 (1999) 631e652. [125] A.K. Aliyu, B. Modu, C.W. Tan, A review of renewable energy development in
[88] K. Hadri, Testing for unit roots in heterogeneous panel data, Econom. J. 3 Africa: a focus in South Africa, Egypt and Nigeria, Renew. Sustain. Energy
(2000) 148e161. Rev. 81 (2018) 2502e2518.
[89] M.H. Pesaran, A simple panel unit root test in the presence of cross section [126] M. Bhattacharya, S.R. Paramati, I. Ozturk, S. Bhattacharya, The effect of
dependence, J. Appl. Econom. 22 (2007) 265e312. renewable energy consumption on economic growth: evidence from top 38
[90] H.R. Moon, B. Perron, Testing for a unit root in panel with dynamic factors, countries, Appl. Energy 162 (2016) 733e741.
J. Econom. 122 (2004) 81e126. [127] B. Atems, C. Hotaling, The effect of renewable and nonrenewable electricity
[91] V. Smith, S. Leybourne, T.-H. Kim, More powerful panel unit root tests with generation on economic growth, Energy Policy 112 (2018) 111e118.
an application to the mean reversion in real exchange rates, J. Appl. Econom. [128] P. Dvor t, D.V. Horst, B. Franta
ak, S. Martina l, K. Ture , Renewable energy
ckova
19 (2004) 147e170. investment and job creation; a cross-sectoral assessment for the Czech Re-
[92] J.L. Carrion-i-Silvestre, T. del Barrio-Castro, E. Lopez-Bazo, Breaking the public with reference to EU benchmarks, Renew. Sustain. Energy Rev. 69
panels: an application to GDP per capita, Econom. J. 8 (2005) 159e175. (2017) 360e368.
[129] T. Bulavskaya, F. Reyne s, Job creation and economic impact of renewable [136] R. York, E.A. Rosa, T. Dietz, STIRPAT, IPAT and ImPACT: analytic tools for
energy in The Netherlands, Renew. Energy 119 (2018) 528e538. unpacking the driving forces of environmental impact, Ecol. Econ. 46 (2003)
[130] M. Pe rez de Arce, E. Sauma, J. Contreras, Renewable energy policy perfor- 351e365.
mance in reducing CO2 emissions, Energy Econ. 54 (2016) 272e280. [137] R.E. De Hoyos, V. Sarafidis, Testing for cross-sectional dependence in panel-
[131] J. Jarke, G. Perino, Do renewable energy policies reduce carbon emissions? data models, STATA J. 6 (4) (2006) 482e496.
On caps and inter-industry leakage, J. Environ. Econ. Manag. 84 (2017) [138] COP21, Paris agreement, in: Conference of Parties. December 12th, 2015.
102e124. https://unfccc.int/sites/default/files/english_paris_agreement.pdf.
[132] E. Menichetti, A. El Gharras, How Can Renewable Energy Help Contribute to [139] F. Islam, M. Shahbaz, A.U. Ahmed, Md.M. Alam, Financial development and
the Development of the MENA Countries?, 2017. Future Notes 6, MENARA energy consumption nexus in Malaysia: A multivariate time series analysis,
Papers, http://www.iai.it/sites/default/files/menara_fn_6.pdf. Econ. Model. 30 (2013) 435e441.
[133] B. Beaudreau, Engineering and economic growth, Energy Econ. 16 (2005) [140] U. Al-mulali, J.Y.M. Lee, Estimating the impact of the financial development
211e220. on energy consumption: Evidence from the GCC (Gulf Cooperation Council)
[134] C.C. Lee, C.P. Chang, P.F. Chen, Energyeincome causality in OECD countries countries, Energy 60 (2013) 215e221.
revisited: the key role of capital stock, Energy Econ. 30 (2008) 2359e2373. [141] R. Komal, F. Abbas, Linking financial development, economic growth and
[135] J.D. Hamilton, Nonlinearities and the macroeconomic effects of oil prices, energy consumption in Pakistan, Renew. Sustain. Energy Rev. 44 (2015)
Macroecon. Dyn. 15 (S3) (2011) 364e378. Cambridge University Press. 211e220.

Charfeddine 2019

Uploaded by

Copyright:

Available Formats

Charfeddine 2019

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Charfeddine 2019

Uploaded by

Copyright:

Available Formats

Renewable Energy 139 (2019) 198e213

Contents lists available at ScienceDirect

Impact of renewable energy consumption and ﬁnancial development

Authors Country/ Time period Methodology Results

Category 1: Time series analysis

Authors Country/Countries Time Methodology Results

Category 1: Time series analysis

Variables Measure Data source Symbol Predicted Economic impacts on

CO2 GDP CO2 GDP

CO2 emissions Measured in metric tons per capita WDI CO2 NA NA NA NA

þ a6i þ d6t þ ε6it

e2j DLn GDPitj þ

a3j DLn Litj þ

c5j DLn FDitj þ

a4j DLn Litj þ

e4j DLn GDPitj þ

c6j DLn FDitj þ

e5j DLn GDPitj þ

a6j DLn Litj þ

e6j DLn GDPitj þ

d2j DLn RECitj

b3j DLn Kitj

d3j DLn RECitj

b4j DLn Kitj

b5j DLn Kitj

d4j DLn RECitj

d5j DLn RECitj

b6j DLn Kitj

d6j DLn RECitj

f6j DLn CO2itj

f3j DLn CO2itj þ a3i

f4j DLn CO2itj þ a4i

f5j DLn CO2itj þ a5i

formation criterion (SIC) is used to determine the optimal

of the individual ADF statistics augmented in the cross-section. Table 4

follows Variables Lag length (p)

3.3.3. Smith et al. [91] panel unit root test

Variables CIPS(2007) Smith et al. [91]

DðLt Þ DðKt Þ DðFDt Þ DðRECt Þ DðGDPt Þ DðCO2 t Þ

Fig. 2. Reaction of GDP to FD and REC one standard deviation shock.

You might also like