Energy Strategy Reviews

Energy Strategy Reviews 32 (2020) 100535
Contents lists available at ScienceDirect
Energy Strategy Reviews

journal homepage: http://www.elsevier.com/locate/esr
Renewable energy consumption and robust globalization(s) in OECD

countries: Do oil, carbon emissions and economic activity matter?
Hemachandra Padhan a, *, Purna Chandra Padhang g, Aviral Kumar Tiwari b, e, Rizwan Ahmed c,
Shawkat Hammoudeh d, f
a
Department of Humanities and Social Sciences (HSS), Indian Institute of Technology (IIT), Madras, India
b
Rajagiri Business School, Rajagiri Valley Campus, Kochi, India
c
Department of Finance University of Birmingham, Business School, United Kingdom
d
Lebow College of Business, Drexel University, USA
e
South Ural State University, Lenin Prospect 76, Chelyabinsk 454080, Russian Federation
f
Institute of Business Research, University of Economic Ho Chi Minh, Vietnam
g
XLRI, Xavier School of Management, Jamshedpur, Jharkhand, India
A R T I C L E I N F O A B S T R A C T
JEL classification: This study examines the impact of globalization(s) on renewable energy consumption in OECD countries by
Q42 endogenizing per capita GDP, oil prices and per capita carbon emissions. We use robust globalization(s) as
Q41 measured by the “classic”, “reconstructed” and “revisited” globalization indexes. The novel method of Machado
F64
and Silva Panel quantile regression (2019) approach is used to obtain robust findings for the renewable energy
C33
consumption-globalization nexus. The results confirm the presence of a long-run association between renewable
Keywords:
energy consumption with globalization(s), per capita GDP, oil prices and per capita carbon emissions. The
Renewable energy consumption
Globalization(s)
empirical results also describe that there are positive effects for per capita income, the real price of oil and carbon
OECD economies emissions per capita on the renewable energy consumption. In addition, a higher level of (overall, economics,
social and political) classic globalization promotes renewable energy consumption, while the “reconstructed”
and “revisited” economic globalization reduces the use of renewable energy consumption, and this finding is also
robust to different measures of economic globalization. Moreover, the panel quantile regression reveals that
renewable energy consumption increases the domestic economy in the middle (0.50) quantile group of the
population through importing more advanced technology and positive spilling over markets, while the lower
quantile group and the higher quantile group of the population are using non-renewable (coal, wood) energy
because of the livelihood practice that is based on coal and wood (for the lower quantiles group of the popu
lation) and for the sake of speedy growth (for the higher quantiles group of the population) that worsens the
environmental quality without caring for the contents of globalization.
1. Introduction supply-to-total energy is 13.5% for OECD countries which is higher than
in other region. Further, the annual growth rate of renewable energy
Renewable energy consumption seems to the main driver for eco increases to 3.2% over the period 2000–2018, rising 1.7% from the
nomic growth in the Organization of Economic Cooperation and 1990–2000 decade. But the share of renewable energy in total energy
Development (OECD) countries. According to International Energy mix drastically increases from 3.1% in 1990 to 35.7% in 2018. The
Agency (IEA, 2020),1 renewable energy provides a comprehensive pic higher growth rate of renewable energy was observed because of the
ture of historical and current markets tends for OECD countries. IEA increasing share in total renewable energy of solar energy (40.1%), wind
gives a new insight to overview renewable energy consumption and (19.9%), liquid biofuels (16.7%), biogas (8%) and solar thermal energy
waste in the world since 1990s. In 2018, the share of renewable energy (5.6%), respectively. Moreover, the OECD countries have used
* Corresponding author.
E-mail addresses: [email protected] (H. Padhan), [email protected] (P.C. Padhang), [email protected] (A.K. Tiwari), [email protected]
(R. Ahmed), [email protected] (S. Hammoudeh).
1
https://www.iea.org/reports/renewables-information-2019.
https://doi.org/10.1016/j.esr.2020.100535
Received 24 February 2020; Received in revised form 10 June 2020; Accepted 7 August 2020
Available online 17 September 2020
2211-467X/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
H. Padhan et al. Energy Strategy Reviews 32 (2020) 100535
renewable energy in the transportation sector to protect environmental international activities and foreign direct investment [14,15].
quality since the last three decades. Given the concerns about carbon Indeed, a higher degree of investment is required in technology in
emissions, climate change and global warming, the OECD countries order to shape renewable energy consumption by foreign firms. In
have increasingly emphasised diversifying the energy mix and securing addition, both import and export can provide better inputs to renewable
energy supply. This interest towards changing energy and more mixed energy consumption. In short, the developed countries or OECD coun
energy supported underpinned by heterogeneous government policies2 tries desire to use the low carbon energy-generating sources such as
worldwide opens a market for renewable energy consumption globally. renewable energy sources by implementing higher levels of technology,
Therefore, we are motivated to examine the impact of globalization(s) trade openness etc. At this stage, the OECD countries focus on the impact
on renewable energy consumption for the case of 30 OECD countries. of globalization(s) on environmental quality rather than on the eco
The novelty or contribution of the paper is that we use different nomic performance since they use less-carbon dioxide-emitting energy
patterns of globalization(s) of Konjunkturforschungsstelle (KOF; 2018) sources in the production processes. In addition, we use the control
to examine their impacts on renewable energy consumption. First, this is variables such as per capita income, the real oil price, and the level of
the first study that uses classic globalization(s),3 reconstructed economic carbon dioxide emissions in the relationship between the patterns of
globalization(s)4 and finally the revisited economic globalization(s)5 in globalization(s) and their impact on renewable energy consumption to
examining the renewable energy consumption-globalization nexus. obtain the robust empirical findings. We use real gross domestic product
Second, we use the novel method of the Machado and Silva Panel (GDP) because as income increases energy consumption also increases,
quantile regression (2019)6 approach to analyse the relationship be underlying the view that GDP also is considered as a factor that impacts
tween globalization(s) and renewable energy consumption with the renewable energy consumption. Oil prices affect the renewable energy
objective of obtaining robust findings for the renewable energy- through fossil fuel consumption. Moreover, the carbon emissions also
globalization nexus. This quantile regression gives more explicit re use a factor effecting renewable energy consumption [8]. We use this set
sults than the ordinary least square method because (i) it captures in of control variables as per the literature based on Cheng et al., [2,3];
dividual heterogeneity and distributional heterogeneity that are Gozgor [16]; Sarkodie and Adams [17].
neglected by the ordinary least square method [1–4]. (ii) This method The remainder of the paper is organized as follows. Section 2 ex
describes the entire conditional distribution of the dependent variable. plains the related literature on the determinants of renewable energy as
(iii) It gives not only one solution to each quantile but also s robust well as the relationship between the energy demand and the levels of
outlying observation of the exogenous variables [1,5–7]. nuanced globalization. Section 3 reviews the theoretical framework on
Further, it is necessary to understand the main components and renewable energy. Section 4 fits the data with the empirical models and
factors related to renewable energy consumption due to the ongoing explains the econometric methodology used in the empirical analysis.
issues on climate change and global warming. According to Ref. [8], the Section 5 reports and discusses the empirical findings. Section 6 pro
demand for renewable energy consumption would rise in future glob vides the conclusion and policy implications.
ally. Indeed, according to Ref. [9,10]; and [11], globalization(s) can
have both positive and negative impacts on energy consumption. Ac 2. Review of related studies
cording to Rahman and Miah [12]; globalization impacts negatively
energy consumption for 26 developing countries. After 1991, global We review the literature on the relationships between renewable
ization makes a change in human lives of the world and their energy consumption and economic growth, CO2 remissions, oil prices
socio-economic and political factors through giving more importance to and nuances of globalization.
foreign direct investment, capital flows and international relation in
term of trade. According to Dreher [13]; for enhancing economic per 2.1. Renewable energy consumption and economic growth nexus
formance in the long run, a higher degree of globalization is necessary in
developing and developed economies. Further, a higher degree of According to Sener et al., [18]; the interest in the literature on the
globalization not only makes better performance of economic structure drivers of renewable energy deployment had been relatively weak until
but also modifies the nature of energy consumption [9]. Therefore, this 2010.8 With the Kyoto Protocol’s first commitment, increasing crude oil
change in energy consumption can be decreased or increased depending prices and rising volatility of traditional energy prices have sparked a
on the patterns of globalization(s). Globalization and its relationship stronger interest in this topic. A fast-developing strand of the literature
with energy consumption are well addressed in the “Pollution haven examines the relationship between renewable energy consumption and
hypothesis7”. This impact of globalization can be realized through economic growth [19] and [20–23]). However, those studies do not
have a consensus on the direction and the magnitude of the relationship
as they have used different data sets, time periods and methodologies.
2 Compared to studies on the traditional renewable energy-growth nexus,
Tax benefits, rebates and tariffs.
3 the new determinants of renewable energy consumption have motivated
Economic, social, political, and overall globalization developed by Dreher
[13]. a new research in this field. In one of the earliest studies, Sadorsky [24]
4
The reconstructed economic globalization comprises economic globaliza analyzes the determinants of the renewable energy in the G7 countries
tion index by adding the real trade openness instead of the nominal trade using panel cointegration techniques. That study finds that real GDP per
openness that enhances the economic globalization index of Dreher [13] by capita and the per capita carbon dioxide emissions are the main drivers
using new variables such as the trade partner diversification, international debt, of the renewable energy consumption in the long run, while the oil price
and international reserves (Gozgor [87]. has a negative but a relatively smaller impact on this consumption. In a
5
This revisited economic globalization index is developed by Gygli et al., subsequent study, Sadorsky [25] shows that real per capita income has a
(2019) which enriches the economic globalization index of Dreher [13] by positive effect on per capita renewable energy consumption in 18
adding the trade partner diversification, foreign debt, international reserves, etc
emerging economies by using panel cointegration, and the evidence is in
to the relationship.
6 line with Sadorsky [24]. This implies that higher economic growth is a
This method is more advanced than the traditional quantile regression such
as Koenker and Bassett [99]; Koenker [102]; Lamarche [103]; Galvao [105] and
key factor in terms of increasing the share of renewable energy in total
Canay [104] because traditional methods suffer from under and energy consumption. Salim and Rafiq [26] analyse the determinants of
over-estimating of coefficients.
7
Pollution-intensive productions in developed countries make strict envi
8
ronmental regulations that should be transferred to those developing countries See Sener et al. [18], for a brief literature review that investigates the
with the less-environmental regulations [135]. drivers and barriers of renewable energy deployment in various countries.
2
renewable energy consumption in Brazil, China, India, Indonesia, the 2.4. Renewable energy consumption and Globalization(s) relationship
Philippines, and Turkey. They show that renewable energy consumption
is influenced by income and pollutant emissions in Brazil, China, India, Indeed, globalization enables countries to decrease energy con
and Indonesia. However, per capita income is the only determinant of sumption by importing technology into energy consumption and pro
the renewable energy demand in the Philippines and Turkey in the long duction activities [61]. Economic globalization, which enhances the
run. In line with the previous studies, Salim and Rafiq [26] also observe level of technology, increases productivity in energy sources (i.e.,
that the crude oil price has a negative impact on renewable energy energy-saving) and decreases the cost of energy. At this stage, the impact
consumption. Destek and Sinha [27]; found an inverted U-shaped rela of trade openness, as a benchmark indicator of economic globalization,
tionship between renewable energy consumption and economic growth on energy demand has been investigated in the energy economics
in 24 OECD countries. Further, A positive relationship between eco literature. For example, Baek et al. [61]; Copeland and Taylor [62]; Lean
nomic growth and renewable energy consumption was also found by and Smyth Lean and Smyth [63]; Ozturk and Acaravci [64]; and Shah
Belaid and Zrelli [28] for 9 Mediterranean countries, Rahman and baz et al. [65], found that trade openness decreases energy consumption
Velayutham [29] for South Asia, Razmi, et al. [30] for 130 developing by providing imported technology and stimulating environmental
countries, Fan and Hao [31] for Chinese provinces and Chen and Stengos quality. However, according to Copeland [66]; trade openness increases
[32] for Iran. Alam and Murad [33] also found that economic growth the per capita energy consumption. In addition, Aissa et al. [67] and
promotes renewable energy consumption in the long-run and has an Narayan and Smyth [68] found a validity of the neutral hypothesis (no
adverse effect in the short-run for 25 OECD countries. significant relationship) between exports (the main indicator of eco
nomic globalization) and energy consumption in the Middle East and
2.2. Renewable energy consumption CO2 and CO2 emissions linkages North Africa countries and the African countries, respectively.
Furthermore, Dedeoglu and Kaya [69–71] confirmed a validity of the
Further, Marques et al. [34] show that the lobby of the traditional feedback effect (bidirectional causality) between the exports (the main
energy sources, the energy self-sufficiency, the carbon dioxide emis indicator of economic globalization) and energy consumption in the
sions, and the per capita income are important determinants of the long run. For instance, Shahbaz et al. [72] examine the causal linkage
renewable energy demand in 24 European countries. Omri and Nguyen between energy consumption and the trade openness in 91 low-income,
[8] examine the determinants of the renewable energy consumption in middle-income, and high-income economies. The authors found a
64 countries and find that per capita carbon dioxide emissions have a U-shaped relationship between energy consumption and trade openness
positive effect on the renewable energy consumption. In a comprehen in the low- and middle-income countries, but there is an “inverted
sive study, Aguirre and Ibikunle [35] illustrate that carbon dioxide U-shaped” relationship between the related variables in the high-income
emissions, energy consumption, GDP per capita, ratification of the Kyoto economies.
protocol, high electricity usage rates in the industry sector are the There are only a few studies that have used the KOF globalization
fundamental drivers of renewable energy growth in 38 countries. Par index of Dreher [13]. For example, Shahbaz et al. [9], examine the ef
amati et al. [36], found that carbon dioxide emissions have a negative fects of globalization indexes of Dreher [13] on energy consumption in
impact on renewable energy demand in 20 emerging market economies. India. The authors found that the level of globalization reduces energy
Chen [37] shows that there is a positive relationship between economic consumption through the channels of energy efficiency and less carbon
development and renewable energy consumption. In addition, the level dioxide-emitting technologies. Further, Shahbaz et al. [10], demonstrate
of carbon dioxide emissions, per capita exports, per capita imports, and that globalization increases energy consumption in most of the devel
changes in the urbanization also explain the renewable energy con oped countries using techniques that validated the globalization-driven
sumption at the different Chinese regions. Moreover [38,39], and Yu energy consumption hypothesis. However, there is a negative relation
et al. [40], showed that renewable energy consumption reduces the ship between the related variables in the case of the United Kingdom and
carbon emissions in China. The negative relationship between renew the United States. Moreover, Shahbaz et al. [11], showed that global
able energy consumption and CO2 emission is also found by Akram et al. ization promotes energy consumption in Ireland and the Netherlands.
[41] for developing countries, Destek and Aslan [42] for G-7 countries To conclude the literature review on the relationship between energy
and Liu et al. [43] for the UK economy. Cheng et al. [32] indicated that consumption and globalization are found to be mixed at best. The usage
reducing the cost of carbon emissions increases renewable energy con of the narrowly defined economic globalization (e.g. exports, imports,
sumption for the case of China. Indeed, Xu and Buyya [44], found that FDI, and trade openness) and different econometric techniques are
carbon emissions can be reduced through renewable energy consump among the main reasons for the mixed findings. Kutan et al. [73],
tion at varied power plants at different locations including California, confirm a positive relationship between FDI on renewable energy con
Virginia and Dublin. sumption in the BRICS countries. Moreover, Gozgor et al. [74] found
that globalization positively impacts renewable energy consumption. As
2.3. Renewable energy consumption and oil prices relationship per the given literature, researchers have taken globalization such as
trade openness, FDI, export, and import but none of the studies has
According to Gourevitch [45]; Ikenberry [46]; and Geller et al. [47]; highlighted different patterns of globalization and assessed their impact
oil prices serve as a measure of impacts on different energy consump on renewable energy consumption. Therefore, by conducting this
tion. As they increase, the demand for energy increases through tech research, we are able to fulfil the gap.
nological innovation and carbon emissions cost reductions. For the
relationship between real oil prices and renewable energy consumption, 3. Theoretical framework on renewable energy
the most noticeable literature includes Henriques and Sadorsky [48];
Broadstock et al. [49]; Sadorsky [50]; Wen et al. [51], Inchauspe et al. This section explains the theoretical setting of renewable energy
[52]; Reboredo, [53]; Reboredo et al. [54]; Shah et al. [55] and Kyritsis consumption which is based on Rafiq and Alam [75]’s renewable energy
and Serletis, [56]. Apergis and Payne [57], showed a long-run rela consumption determinants in the emerging economies and Sardosky
tionship between real oil prices and renewable energy for South Amer (2009a)’s seminal work that is based on renewable energy consumption
ica. Cheon and Urpelainen [58], indicated that oil prices increase the in the G7 economies. According to the theoretical understanding, a
renewable energy consumption through technological innovation. Ac standard approach on renewable energy consumption can be postulated
cording to Nilsson [59] and Hsiaoet al. [60]; oil prices have a positive based on its price and gross domestic product [76,77]. There are two
and significant impact on renewable energy consumption through the approaches of renewable energy consumption, first, the demand side
innovation of new technology. approach of renewable energy consumption started by J.M. Keynes in
3
1936 that is based on a macroeconomics growth framework for the 4.2. Econometric methodology
demand for goods and services. By focusing on Keynes’s theory, the
demand side approach focuses on the effective demand and consump In this paper, we use the quantiles via the moments regression
tion that determine the output level. Its major issues are that the effec analysis which was presented by the Machado and Silva Panel Quantile
tive demand is generated by governments, the households and the regression (2019). The study considers a situations under which it is
businesses and is not based on the free market setup. This theory is desirable to estimate the regression quantiles by estimating the condi
mainly applied to analyzing the relationship between energy prices, tional means.
renewable energy consumption and gross domestic product that stands The conventional regression analysis (also known as the mean
as an energy demand model [25,76–81]. Moreover [82,83], considers regression) primarily examines the influence of explanatory variables on
pollutant emissions as a variable that links energy demand and eco the conditional means of the explained variables. Therefore, the mean
nomic activities. But later, the study was not considered as a demand regression simply underscores the central tendency of the conditional
side approach due to lack of energy prices. distribution, which overlooks the effect of explanatory variables for the
Second, according to the supply side approach of the renewable whole distribution of the variables. Additionally, outliers are easily
energy consumption, the growth of the economy can be effectively affected by the conventional regression. Therefore, in contrast to the
fostered through capital investments, flexible tax reductions and dwin mean regression, the current paper focuses on the quantile regression
dling regulations. Based on this approach, individuals can gain from a analysis to recognise the different stages of the variables. Furthermore,
high output supply supplemented with minimizing costs, which will the quantile regression not only considers the influence of outliers but
raise the level of employment. This concept also suggests that output also reduces the possible unobserved heterogeneity and covariates,
supply is a major component of economic growth that explains demand which adds to relatively robust results. Moreover, it is also common that
and consumption as a subordinate consequence. This theory also shows empirical studies adopt panel data to control for unrecognised hetero
that tax reductions in a particular economy set up enhance economic geneity and covariates by including the fixed effects (e.g., Ref. [88,89];
growth by motivating saving, investment, and work. Finally, it evaluates and [90]. We have adopted the methodology of the Machado and Silva
energy use (renewable and non-renewable) in a standard production Panel Quantile regression (2019) which focuses on the conditional
function for economic activities [84–86]. location-scale model and is also widely measured by Koenker and Bas
sett [91]; Gutenbrunner and Jureckova [92]; Koenker and Zhao [93]; He
4. Data, model and methodology [94]; and Zhao [95]. Moreover, the methodology suggests an estimator
of the conditional quantiles accessed by merging the estimates of the
4.1. Data location and scale functions, both of these estimates defined by condi
tional expectations of suitably defined variables.
We work with panel data9 covering the period 1970–2015 for 30 The main advantage of this method is that it authorizes the use of
OECD countries10. This data includes total renewable energy con methodologies that are specifically effective in the estimation of con
sumption in million tonnes oil equivalent (Mtoe) (LNREC), CO2 emis ditional means; for instance, differencing out individual effects in panel
sions per capita in kg (LNCO2PC), per capita GDP in constant 2010 USD data models, and at the same time giving information on how the re
(LNRGDPC) used as the proxy of economic growth, real oil prices in gressors influence the whole conditional distribution. The informational
million tonnes oil equivalent (Mtoe) (LNROP), economic globalization advantages are the most interesting component of the quantile regres
(LNEGI), social globalization (LNSGI), political globalization (LNPGI), sion (e.g., the influential papers by Refs. [96]; and [97]. Moreover, our
and overall globalization (LNOGI) developed by Dreher, [13]; recon model approach mainly estimates the regression quantiles that do not
structed economic globalization (LNRC_EGI) generated by Gozgor, [87]. cross, an essential requisite usually overlooked in empirical studies (for
First, we use the classical globalization index (economic, social, political e.g. Refs. [94]; and [98].
and overall globalization) of Dreher [13]. Second, we use the recon Moreover, it is important to note that our estimators in the model
structed economic globalization index of Gozgor [16] that recalculates need firmer assumptions on the presence of moments than those
the economic globalization index by using the real trade openness required for the validity of the Koenker and Bassettís [99] estimator. On
instead of the nominal trade openness that enhances the economic the other side, our estimators analyse the same conditional quantiles
globalization index of Dreher [13] by using new variables, such as the under the normal asymmetric loss function, and these estimators are
trade partner diversification, international debt, and international re fundamentally robust. The structure of the model is restrictive and es
serves. We also use the revisited economic globalization index devel timates that covariates only affect the distribution of interest through
oped by Gygli et al., (2019) that enriches the economic globalization known location and scale functions. On the other hand, researchers
index of Dreher [13] by adding new variables such as trade partner usually plan to develop even robust estimations; however, in our
diversification, foreign debt, international reserves, etc. Moreover, all approach the mode has the potential to test the assumption that the
the data are taken from both the World development indicators (WDI, covariance only affects the location and scale functions, and conse
World Bank, 2020) and the BP Statistical Review of World Energy quently it is essential to measure whether or not our approach is robust
(2018), and the KOF index of globalization (KOF, 2018). Taking the in a specific function.
above discussions into account, this study examines an empirical model The suggested methodology is not intended to be a replacement of
for renewable energy consumption that is consistent with the broad the existing well-established estimation approaches constructed on the
literature related to the impact of relevant control variables on this check function.11 Rather, our suggested method can be considered as a
energy consumption. The classical control variables used in this study supplement tool that can enhance those practices and allow the esti
such as CO2 emissions, economic growth and real oil prices are based on mation of the regression quantiles in a more generalised way. For
[2,3]; Gozgor [87]; Sarkodie and Adams [17]. instance, the advantage of this approach when using panel data to es
timate regression quantiles is that it includes individual effects. More
over, the drawback of the quantile regressions with individual effects
faced supplementary parameters (for instance, Ref. [100]; and [101]
9
Data will be shared as per the request of the authors.
10
Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark,
11
Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Japan, Korea, The check function is a loss function that retrieves the τ-th sample quantile
Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovakia, Spain, (for more details, the reader can refer to Quantile Regression (2005), Section
Sweden, Switzerland, Turkey, United Kingdom, United States. 1.3, by Roger Koenker).
4
and the more recent literature presents the challenges faced by the where U = (Y− (∝ + X β))
′
. These settings structure the grounding of the
σ (δ+ Z′ γ)
models (e.g., Ref. [102–108]; and Powell, 2017). However, these ap
estimation method called the Method of Moments Quantile Regression –
proaches did not obtain any exceptional prominent acceptance because
MM-QR). The equation (MC1) indicates similarity to the Restricted
of their difficulty in computation or their depending on very restrictive
Quantile Regression of He [94] and Zhao [95] but the suggested model
assumptions for the fixed effects impact on the quantile regression.
examines distinctive moment conditions. Moreover, this model is very
Although, based on the restrictive assumption, this methodology has the
effective which makes it very simple to adopt the quantile regression in a
benefit of having an easy implementation for complex and large prob
broader range of models. Especially, the MC1 is used for estimation of
lems and also has the benefit of showing how the individual effects affect
panel data models with fixed effects and MC2 has the benefit of exam
the whole distribution, instead of just the location shifters (e.g., Refs.
ining the structural quantile functions described by Ref. [111,112].
[102–104]. Moreover, the methodology can be used in the estimations
of cross-sectional models with endogenous variables [109–112].
5. Findings and discussion
4.3. Model estimation Fig. 1 displays the Q-Q plots for all the variables under the study.
Basically, a Q-Q plot or a quantile-quantile plot is a graphical interpre
The basic idea of the paper is to examine the estimation of the con tation that explains the distributions of the data relative to the theo
ditional quantiles of the random variable Y’s dependence on the distri retical distribution such as the normal distribution [113]. Moreover, the
butional conditional K-vector of covariates X that relate to the location- Q-Q plot appears as a scatterplot constructed by plotting two groups of
scale family. quantiles against one another. For instance, for both groups of quantiles
( (theoretical quantile and sample quantile) representing the same dis
´ (1)
′
Y = ∝ + X β + σ δ + Z γ)U,
tribution, we should observe that the points create a line that is
approximately straight. From the following figure, we observe that all
where ∝, β , δ, γ´′ ) ∈ R2(k+1) are unknown parameters and Z is a k-vector our variables are roughly plotted on the straight theoretical line.
′
of known differentiable (with probability 1) transformations of the Moreover, the deviations from the straight line are minimal and this is a
components of X with elements lgiven by fair indication of the normal distribution.
We examine the descriptive statistics in Table 1. Evidently, the dis
Zl = Zl (X), l = 1, ……, K, (2)
tributions of all the variables are moderately skewed except those of
σ ( .) is called the C2 function such that P{σ (δ + Z γ) > 0} = 1, where U
′ LNOGI, LNEGI, LNSGI and LNPGI which are highly negatively skewed.
is an unrecognised random variable, independent of X, with a density The positive excess kurtosis values indicate that the distributions have
function fu (.) bounded away from 0 and normalized to prove the fatter tails. Moreover, the correlation analysis indicates the levels of two
moment conditions or more variables fluctuate together. In our analysis, LNEGI has a strong
positive correlation with LNOGI, while LNRC_EGI has a strong positive
E (U) = 0 E(|U|) = 1. (3) correlation with LNEGI.
Before analysing the panel quantile regression models, we have
Equation (1) is known as the linear heteroscedasticity model where σ
examined four panel unit root and stationarity tests for all the variables.
(.) is a unique function and Z = X. The model has been widely used in the
We have conducted three panel unit root tests including the Levin et al.
literature (e.g., Ref. [91–95]. Further, adopting Equation (3) and given
[114], test, the Im et al. [115], test and Breitung t-stat test (2005). We
the exogeneity of the regressors, the vector parameters examined under
have also carried out cross-sectional dependency analysis and results are
this study can be identified from the following form of moment condi
reported in Appendix A1. Based on the significant evidence of
tions (for analysis purpose i.i.d is considered an assumption).
cross-sectional dependency, we have demeaned the data and applied
E [RX] = 0 (4) panel unit root tests that are already reported in Table 2. I must also say
that Breitung t-stat (2005) test, which considers the cross-sectional
E [R] = 0 (5) dependence. The first generation unit-root tests are used intensively in
[( ( ] ( ) the panel data studies. Levin et al. [114],’s test depends on the pooled
E |R| − σ δ + Z ´γ)) Dσγ = 0 MC1 (6) data, while Im et al. [115],’s analysis is obtained as an average of the
Augmented Dickey-Fuller (ADF) test statistics. Levin et al. [114],’s
[( ( ]
unit-root test allows for heterogeneity in the intercept, while Im et al.
E |R| − σ δ + Z ´γ)) Dσδ = 0 (7)
[115], accept potential heterogeneity in both the intercept and the slope
[( ( ] terms for the cross-section units. However, these tests do not consider
E I R ≤ q (τ)σ δ + Z ´γ)) τ = 0 (8) cross-sectional dependence problems due to unobserved common fac
tors. Therefore, considering the potential cross-sectional dependence,
′
we use Breitung t-stat (2005) which captures the cross-sectional
where = Y − ( ∝ + X´β) = σ(δ + Z γ)U, Dσγ = ∂σ(δ+Z , Dσδ = ∂σ δ+∂δZ σ,
′
′ γ)
∂γ
dependence. Table 2 gives the results of those panel units’ root and
the location-scale model determines the scale function σ (.), we can
stationarity tests. The results demonstrate that the null hypothesis of the
examine that information and develop the identification on the substi
presence of unit root could not be rejected for all of the variables in the
tute set of moment conditions.
selected levels. However, the unit root hypothesis for all variables based
E [UX] = 0 (9) on the first difference is completely rejected at the 1% level of signifi
cance. Therefore, the first difference sequence is important in our
E [U] = 0 (10) empirical analysis as well (e.g., Frankel and Rose, 1996; Oh, 1996;
[ ] Lothian, 1997; Taylor, 1996). Moreover, our results are robust to
E (|U| − 1 ) Dσγ = 0 (11) cross-sectional dependence.
[ ] The foregoing discussion implies that we should adopt an effective
E (|U| − 1 ) Dσδ = 0 (MC2) (12) and useful statistical inference, especially when the underlying as
sumptions are violated since depending on a robust standard error is
E [I(U < q (τ)) − τ] = 0 (13) usually common. On the basis of the evidence currently available, it
seems fair to suggest that most commonly well-known covariance matrix
5
Fig. 1. The Q-Q plot constructed by plotting two groups of quantiles against one another.
Table 1
Summary statistics and correlations.
LNREC LNRGDPC LNROP LNCO2PC LNOGI LNEGI LNSGI LNPGI LNRC_EGI LNREV_EGI
Mean 0.365827 10.15203 3.863516 2.072413 4.268262 4.200170 4.181932 4.422157 4.163246 4.126055
Median 0.531485 10.31078 3.774327 2.116788 4.345536 4.284343 4.319608 4.487385 4.233560 4.199472
Maximum 5.584399 11.42538 4.776669 3.153484 4.526195 4.577328 4.537708 4.589186 4.547128 4.501493
Minimum − 6.907755 8.272306 2.408746 0.118882 3.440380 3.102938 2.975821 3.837861 3.232802 3.254210
Std. Dev. 2.078497 0.668041 0.576984 0.572280 0.221775 0.271079 0.325246 0.157163 0.270977 0.259287
Skewness − 0.629483 − 0.806743 − 0.095849 − 0.735041 − 1.239142 − 1.235129 − 1.302143 − 1.420951 − 0.932351 − 0.884059
Kurtosis 3.963194 3.014016 2.225298 3.456151 4.075741 4.604448 3.974588 4.186218 3.510316 3.247932
lnrec lnrgdpc lnrop Lnco2pc lnogi lnegi lnsgi lnpgi lnrc_egi Lnrevegi
lnrec 1.000
lngdpc 0.3729 1.000
lnrop 0.3044 0.1478 1.000
lnco2pc 0.2431 0.6617 − 0.0020 1.000
lnogi 0.3591 0.7280 0.1960 0.5630 1.000
lnegi 0.2952 0.6457 0.1840 0.4663 0.9172 1.000
lnsgi 0.3445 0.7194 0.1838 0.5916 0.9564 0.8355 1.000
lnpgi 0.3504 0.5707 0.1484 0.4296 0.8011 0.5912 0.6777 1.000
lnrc_egi 0.1674 0.5609 0.1597 0.3568 0.8604 0.9486 0.7824 0.5376 1.000
lnrevegi 0.2826 0.7089 0.2206 0.4497 0.8902 0.9438 0.8025 0.6152 0.9035 1.000
estimators are developed by Eicker [116]; Huber [117] and White renewable energy as the level of globalization is examined through three
[118]. Further research in this area may include Arellano [119]; Froot distinct indexes of this measure including “classic”, “reconstructed” and
[120] and Rogers [117] which specify that it is adept to some extent to “revisited” (e.g., see Ref. [13,16]; Gygli et al., 2019). Likewise, a higher
ease the assumption of independently distributed residuals. Moreover, level of globalization also includes (overall, economic, social and po
their generalised estimator specifies consistent standard errors on the litical) factors which affect the renewable energy consumption during
condition that the residuals are correlated within but uncorrelated be the reconstructed and revisited economic globalization. In Table 3 of
tween clusters. Newey and West [122] developed another useful method Model 1, the results of the Driscoll-Kraay test has a significantly stronger
to find heteroscedasticity and autocorrelation established on some lag and positive impact on total renewable energy consumption as a per cent
consistent standard errors. They have included GMM as well as the of GDP (LNREC), in comparison to other fixed effect models. In the other
addition of White’s estimator and is known as the Newey-West estimator words, real capita income to GDP (LNRGPDC), real oil prices (LNROP),
with a consideration of zero lag length, similar to the White estimator. CO2 emissions per capita in kg (LNCO2PC) and overall globalization
Driscoll and Kraay [123] introduced a non-parametric covariance ma (LNOGI) have a positive and significant effect on total renewable energy
trix estimator which gives heteroscedasticity and autocorrelation consumption (e.g., Ref. [124–126].
consistent standard errors. Moreover, it specifies a robust to spatial and In Table 4 of Model 2, we have included economic globalization in
temporal dependence. place of overall globalization along with the other common variables.
From Tables 3–8, we investigate alternative fixed effect models. We The results of the estimated White fixed effects and Rogers fixed effects
have analysed the OLS, White, Rogers, Newey-West, Driscoll-Kraay tests have a strong and positively significant influence on total renewable
and OLS MOM models. From these tables, we have examined the asso consumption. Such as real capita income to GDP (LNRGPDC), real oil
ciation between renewable energy consumption and the control vari prices (LNROP), and Economic Globalization (LNEGI) have a positive
ables such as globalization, real GDP, oil prices and carbon emissions. and significant effect on total renewable energy consumption under the
We place major emphasis on the effect of nuanced globalization on model assumptions of the White FE and Rogers FE. However, the CO2
6
Table 2 emissions per capita in kg (LNCO2PC). The results indicate that the OLS
Unit root tests. and White and Rogers models yield a strong and positive significant
Variables Levin, Lin & Chu t Im, Pesaran and Shin Breitung t-stat impact of real capita income to GDP (LNRGPDC), real oil prices
(2002) W-stat (2003) (2005) (LNROP), and social globalization (LNSGI) on total renewable energy
Statistic P-value Statistic P-value Statistic P-value consumption (LNREC), in comparison with the other models and coin
cide with the studies of [127]. However, the CO2 emissions per capita in
lnrec 3.052 0.998 1.132 0.871 3.443 0.999
lngdpc 1.626 0.948 1.280 0.899 2.399 0.991
kg (LNCO2PC) results are similar to those of Model 2, since they indicate
lnrop 1.422 0.922 0.284 0.388 2.279 0.988 a positive but insignificant influence on total renewable energy
lnco2pc 1.033 0.150 3.668 0.999 3.342 0.999 consumption.
lnogi 2.217 0.986 1.561 0.997 3.715 0.999 In Table 6 of Model 4, we have considered Political Globalization
lnegi 2.587 0.995 3.534 0.999 3.652 0.999
(LNPGI) in addition to real capita income to GDP (LNRGPDC), real oil
lnsgi 0.971 0.834 0.363 0.358 1.012 0.844
lnpgi 3.631 0.999 0.809 0.209 1.117 0.986 prices (LNROP), CO2 emissions in kg per capita (LNCO2PC). The results
lnrc_egi 2.483 0.998 1.013 0.997 0.836 0.999 demonstrate a strong and significantly positive association of real capita
lnrevegi 10.458 0.999 1.102 0.864 2.905 0.998 income to GDP (LNRGPDC), real oil prices (LNROP), and Political
Δlnrec − 8.511 0.000 − 19.172 0.000 − 8.792 0.000 Globalization (LNPGI) with total renewable energy consumption
Δlngdpc − 16.006 0.000 − 18.294 0.000 − 10.379 0.000
Δlnrop − 31.486 0.000 − 23.957 0.000 − 9.882 0.000
(LNREC) in the following fixed effects models the OLS, White, Rogers
Δlnco2pc − 29.095 0.000 − 7.213 0.000 − 4.657 0.000 and Driscoll-Kraay and support the evidence of the related research also
Δlnogi − 17.008 0.000 − 2.790 0.002 − 11.519 0.000 examined by Ref. [128]. However, CO2 emissions in kg per capita (LN
Δlnegi − 16.134 0.000 − 4.753 0.000 − 4.930 0.000 CO2PC) indicate a positive but insignificant influence towards total
− 3.397 0.000 − 17.089 0.000 − 12.712 0.000
Δlnsgi
renewable energy consumption.
Δlnpgi − 22.151 0.000 − 7.630 0.000 − 15.951 0.000
Δlnrc_egi − 16.026 0.000 − 3.651 0.000 − 6.397 0.000 In Table 7 of Model 5, we have measured economic globalization by
Δlnrevegi − 19.842 0.000 − 17.446 0.000 − 12.334 0.000 the reconstructed globalization index (LNRC_EGI). Along with the in
clusion of the common globalization variables such as real capita income
to GDP (LNRGPDC), real oil prices (LNROP), CO2 emissions per capita in
emissions per capita in kg (LN CO2PC) are positive but they show an kg (LN CO2PC), the results indicate that the White and the Rogers model
insignificant influence on total renewable energy consumption and the shows a positive and significant influence of the real capita income on
results are into line with the study of Lin and Moubarak [21]. GDP (LNRGPDC) and the real oil prices (LNROP) on total renewable
By considering Table 5 of Model 3, this model includes social glob energy consumption. Moreover, a negative and significant influence of
alization (LNSGI) along with the earlier three common variables: real the reconstructed economic globalization index (LNRC_EGI) is observed
capita income to GDP (LNRGPDC), real oil prices (LNROP) and CO2 on total renewable energy consumption (LNREC) as well. The negative
Table 3
Model1 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnogi)
OLS White Rogers Newey-West Driscoll-Kraay OLSMO
lnrgdpc 0.7041*** (4.92) 0.7041*** (5.19) 0.7041*** (5.12) 0.7041 (1.12) 0.7041*** (2.72) − 1.0728 (− 0.80)
lnrop 0.8910*** (3.02) 0.8910*** (8.72) 0.8910*** (9.62) 0.8910*** (5.64) 0.8910*** (5.51) 3.6959*** (4.37)
lnco2pc 0.0416 (0.33) 0.0416 (0.31) 0.0416 (0.34) 0.0416 (0.07) 0.0416 (0.18) 0.0784 (0.12)
lnogi 1.5133*** (2.77) 1.5133*** (4.05) 1.5133*** (4.06) 1.5133 (1.12) 1.5133** (2.23) 3.8717 (1.46)
Constant − 16.6824*** (− 8.95) − 16.6824*** (− 13.16) − 16.6824*** (− 14.63) − 16.6824*** (− 3.70) − 16.6824*** (− 8.03) − 17.6918* (− 1.93)
R2 0.2168 0.2168 0.2168 0.2168
N 1079 1079 1079 1079 1079 1079
Table 4
Model2 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnegi)
lnrgdpc 0.8621*** (7.79) 0.8621*** (6.55) 0.8621*** (6.48) 0.8621 (1.41) 0.8621*** (3.46) − 1.3839 (− 1.05)
lnrop 0.9305*** (2.86) 0.9305*** (9.11) 0.9305*** (10.01) 0.9305*** (5.71) 0.9305*** (5.72) 4.1453*** (7.24)
lnco2pc 0.1145 (0.96) 0.1145 (0.85) 0.1145 (0.94) 0.1145 (0.19) 0.1145 (0.50) − 0.6102 (− 0.91)
Lnegi 0.5848 (1.43) 0.5848** (2.12) 0.5848** (2.30) 0.5848 (0.62) 0.5848 (1.25) − 0.0153 (− 0.01)
Constant − 14.5847*** (− 9.89) − 14.5847*** (− 13.02) − 14.5847*** (− 15.11) − 14.5847*** (− 3.55) − 14.5847*** (− 8.20) 1.4253 (0.14)
R2 0.2081 0.2081 0.2081 0.2081
N 1079 1079 1079 1079 1079 1079
Notes. Robust standard errors are in parentheses., ***p < 0.01, **p < 0.05, *p < 0.1. lnegi captures economic globalization.
Table 5
Model3 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnsgi)
lnrgdpc 0.7699*** (4.95) 0.7699*** (5.74) 0.7699*** (5.70) 0.7699 (1.25) 0.7699*** (3.03) − 1.1593 (− 0.97)
lnrop 0.9047*** (2.98) 0.9047*** (8.84) 0.9047*** (9.77) 0.9047*** (5.81) 0.9047*** (5.61) 3.1039*** (5.30)
lnco2pc 0.0184 (0.13) 0.0184 (0.13) 0.0184 (0.14) 0.0184 (0.03) 0.0184 (0.08) − 0.2059 (− 0.47)
Lnsgi 0.8622** (2.10) 0.8622** (3.30) 0.8622** (3.18) 0.8622 (0.91) 0.8622* (1.75) 0.3718 (0.10)
Constant − 14.5011*** (− 11.78) − 14.5011*** (− 13.75) − 14.5011*** (− 14.52) − 14.5011*** (− 3.40) − 14.5011*** (− 7.85) 1.2438 (0.14)
R2 0.2128 0.2128 0.2128 0.2128
N 1079 1079 1079 1079 1079 1079
Notes. Robust standard errors are in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. lnsgi represents social globalization.
7
Table 6
Model4 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnpgi)
lnrgdpc 0.7215*** (4.68) 0.7215*** (5.96) 0.7215*** (6.23) 0.7215 (1.34) 0.7215*** (3.34) 0.9420 (0.47)
lnrop 0.8911*** (3.11) 0.8911*** (8.90) 0.8911*** (9.94) 0.8911*** (5.60) 0.8911*** (5.71) 1.5247* (1.93)
lnco2pc 0.0621 (0.55) 0.0621 (0.47) 0.0621 (0.52) 0.0621 (0.11) 0.0621 (0.28) − 0.1806 (− 0.52)
Lnpgi 2.7504*** (3.13) 2.7504*** (6.51) 2.7504*** (5.87) 2.7504* (1.91) 2.7504*** (3.34) 2.4036 (0.64)
Constant − 22.6059*** (− 6.10) − 22.6059*** (− 13.23) − 22.6059*** (− 12.25) − 22.6059*** (− 3.70) − 22.6059*** (− 6.89) − 23.7914** (− 1.99)
R2 0.2350 0.2350 0.2350 0.2350
N 1079 1079 1079 1079 1079 1079
Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. . lnpgi represents political globalization.
Table 7
Model5 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnrc egi)
lnrgdpc 1.0966*** (10.00) 1.0966*** (8.57) 1.0966*** (8.76) 1.0966* (1.93) 1.0966*** (4.67) − 0.9740 (− 0.78)
lnrop 0.9785** (2.69) 0.9785*** (9.58) 0.9785*** (10.53) 0.9785*** (5.92) 0.9785*** (6.02) 4.0017*** (11.73)
lnco2pc 0.1193 0.1193 (0.88) 0.1193 (0.98) 0.1193 (0.20) 0.1193 (0.52) − 0.5442 (− 0.78)
(1.05)
lnrc_egi − 0.4644 (− 1.04) − 0.4644* (− 1.83) − 0.4644** (− 2.05) − 0.4644 (− 0.59) − 0.4644 (− 1.10) − 0.1958 (− 0.22)
Constant − 12.7665*** (− 7.80) − 12.7665*** (− 11.35) − 12.7665*** (− 12.61) − 12.7665*** (− 2.90) − 12.7665*** (− 6.80) − 2.1356 (− 0.20)
R2 0.2073 0.2073 0.2073 0.2073
N 1079 1079 1079 1079 1079 1079
Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. lnrc_egi represents reconstructed economic globalization.
Table 8
Model6 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnrevegi)
lnrgdpc 1.1484*** [8.46] 1.1484*** [8.17] 1.1484*** [8.02] 1.1484* [1.80] 1.1484*** [4.29] − 0.3339 [-0.25]
lnrop 0.9903** [2.71] 0.9903*** [9.62] 0.9903*** [10.57] 0.9903*** [6.04] 0.9903*** [6.06] 0.4646* [1.85]
lnco2pc 0.0966 [0.95] 0.0966 [0.71] 0.0966 [0.77] 0.0966 [0.16] 0.0966 [0.41] − 0.9327*** [-3.66]
lnrevegi − 0.0004 [-1.04] − 0.0004* [-1.90] − 0.0004* [-1.90] − 0.0004 [-0.50] − 0.0004 [-1.03] − 0.0001 [-0.08]
Constant − 14.959*** [-12.62] − 14.959*** [-12.04] − 14.959*** [-11.70] − 14.9593*** [-2.82] − 14.9593*** [-6.30] 5.0264 [0.53]
R2 0.2075 0.2075 0.2075 0.2075
N 1079 1079 1079 1079 1079 1079
Robust standard errors are in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. . lnrevegi represents reconstructed economic globalization.
influence of reconstructed economic globalization index enhanced the distribution. In addition, the panel quantile regression outcomes are
carbon emissions and negatively affects renewable energy (e.g., robust and help in the interpretation of the results of the variables. They
Ref. [128,129]; Khan et al., 2019). On the other hand, CO2 emissions per are also more adequate than those of the OLS regression. In Table 9, we
capita in kg (LN CO2PC) have an insignificant influence on total have used similar explanatory variables as in Table 3 in our panel
renewable energy consumption in all fixed effects models. quantile regression. We have noticed that real capita income to GDP
In Table 8 of Model 6, we have adopted a variable for economic (LNRGPDC), real oil prices (LNROP) and overall globalization (LNOGI)
globalization by measuring it through the revisited economic global are clearly significant and positive at the lower and higher quantiles.
ization index (LNREVEGI), in addition to including the other common This finding indicates that total renewable energy consumption strongly
globalization variables such as real capita income to GDP (LNRGPDC), relying on real capita income to GDP (LNRGPDC), real oil prices
real oil prices (LNROP), CO2 emissions per capita in kg (LN CO2PC). The (LNROP) and overall globalization (LNOGI) in the OECD countries
outcome has a resemblance with that of Model 5 in the sense that the under consideration (e.g., Refs. [124,125]; Zhao et al., 2018). On the
White and Rogers approach has a strong positive and significant effect other hand, the coefficient of CO2 emissions per capita in kg (LN CO2PC)
on the real GDP per capita (LNRGPDC) and real oil prices (LNROP) but is negative and significant at the lower quantile (5th); but it becomes
has a negative and significant influence of the revisited economic insignificant at the 10th to 80th quantiles and then turns significantly
globalization index (LNREVEGI) on total renewable energy consump positive at the higher quantiles. This finding implies that a higher
tion (LNREC), therefore our results are brough into line with (e.g., Refs. renewable energy consumption leads to lower carbon emissions in the
[128–130]. On the other hand, CO2 emissions per capita in kg higher renewable energy consuming countries, whereas the opposite
(LNCO2PC) haves an insignificant influence on total renewable energy holds true in the lower-renewable energy consuming countries. Our
consumption in the fixed effect models except the OLSMO. results align with the arguments of Nguyen and Kakinaka [131].
The following analyses presented in Tables 9–14 are based on the In Table 10 of Model 2, we have used the same common globalization
quantiles via the moments regression analysis presented by Machado variables, i.e. real capita income to GDP (LNRGPDC), real oil prices
and Silva Panel Quantile regression (2018). The quantile regression (LNROP), and CO2 emissions per capita in kg (LNCO2PC) and also
estimates provide one solution to each quantile. By using this method included the additional variable called economic globalization (LNEGI).
ology, we can determine the total renewable energy consumption The results for the common globalization variables are similar to those
throughout the conditional distribution, specifically in the OECD in Table 9 of Model 1 (e.g., Refs. [124–126]. However, economic glob
countries with consideration of the most and the least renewable energy alization is negative and significant at the higher quantiles (i.e., 90th)
consumption. Moreover, the other advantage of this methodology is that but significantly positive at the 2nd to 7th quantiles. These results imply
it helps one understand what happens at the extremes of the that economic globalization increases renewable energy consumption in
8
the middle quantile level by enabling the importation of advanced

technology. However, in the lower and higher quantiles, economic
globalization increases non-renewable (coal, wood) energy consump
− 8.475***
0.8662***
1.0601***
1.0241***
(0.1064)
(0.0997)
(0.1364)
(0.2637)
(1.3039)
0.1918*
tion because the livelihood strategy dominates in the lower quantiles
1079
0.95
(11)
while speedy growth strategy rules in the higher quantiles (e.g.,
Ref. [132].
In Table 11 of Model 3, along with our common globalization vari
ables i.e. real capita income to GDP (LNRGPDC), real oil prices (LNROP),
− 5.463***
0.3830***
1.0828***
1.1691***
(0.1329) CO2 emissions per capita in kg (LN CO2PC), we have included social
(0.0415)
(0.1891)
(0.3649)
(0.9218)
0.5324
globalization (LNSGI). The results indicate that social globalization has a
1079
(10)
0.9
positive and significant impact at the lower to middle quantiles (10%–

50%) and then the impact becomes positively insignificant. However, at
0.2021 (0.1578)
0.4028 (0.5783) the higher quantile (95%), it turns again to be positive and significant.
− 12.962***
This finding indicates that the social globalization factors have a strong
0.8603***
1.0265***
(0.1439)
(0.1132)
(1.8596) influence on the total renewable consumptions as the social norms of

1079 globalization increase the dependency on renewable energy consump
0.8
(9)
tion in OECD countries. Those results relate with study of Marques et al.,
[127]. Moreover, the common globalization variables illustrate similar
0.0925 (0.1475)
results as demonstrated earlier in Table 9 (e.g., Refs. [124–126].

− 14.573***
0.8555***
0.8383***
In Table 12 of Model 4, we have included political globalization

0.8956**
(0.1451)
(0.1109)
(0.4258)
(1.4155)
(LNPGI) along with three common globalization variables i.e. real capita
1079
0.7
(8)
income to GDP (LNRGPDC), real oil prices (LNROP), CO2 emissions per
capita in kg (LNCO2PC), using quantiles via moments regression analysis
as presented in the Machado and Silva Panel Quantile regression (2018).
− 17.145***
The outcome indicates that political globalization is strongly and posi

0.9230***
0.9440***
1.2211***
− 0.0692
(0.1048)
(0.0882)
(0.1192)
(0.3695)
(1.1859)
tively significant in the lowest to highest quantiles (5%–95%). It clearly

1079
indicates that political globalization promotes strongly total renewable

0.6
(7)
energy consumption in the 30 OECD countries over the period 1970 to

2015. Recently, OECD countries emphasize renewable energy programs
and politicians have shown interest and shifting their reliance towards
− 19.582***
0.9013***
0.9378***
1.8253***
renewable energy [133]. The other explanatory variables present similar

− 0.1656
(0.1144)
(0.0924)
(0.1249)
(0.3700)
(1.2769)
results as mentioned earlier in Table 9 (e.g., Ref. [124–126].

1079
0.5
(6)
In Table 13 of Model 5, we incorporate the reconstructed economic

globalization (RC_EGI) along with the common globalization variables i.
e. real capita income to GDP (LNRGPDC), real oil prices (LNROP), CO2
Machado and Silva Panel Quantile regression (2019).Model1 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnogi)
emissions in kg per capita (LNCO2PC). The results indicate that the

− 19.510***
0.7187***
0.8355***
2.1967***
reconstructed economic globalization is negatively significant at the

− 0.0710
(0.1223)
(0.1124)
(0.1422)
(0.3975)
(1.4134)
lower quantiles (5%, 10% and 40%); however, it is actively negatively

1079
0.4
(5)
significant at higher quantiles i.e. (i.e., 80%, 90% and 95%). It dem
onstrates that the reconstructed economic globalization reduces the
0.0123 (0.1485)
influence of renewable energy consumption in OECD countries (e.g.,

− 18.267***
Ref. [128,129]; Khan et al., 2019). The other explanatory variables

0.3594***
0.8263***
2.6137***
(0.1316)
(0.1206)
(0.3951)
(1.4032)
illustrate similar results as demonstrated earlier in Table 9 (e.g.,

1079
Ref. [124–126].
0.3
(4)
In Table 14 of Model 6, we replace the old measure of globalization

by the revisited economic globalization (REVEGI) variable and present
0.0785 (0.1531)
0.0306 (0.1570)
the results measuring the effect on total renewable energy consumption

− 17.343***
0.9104***
2.8484***
in this table. The revisited economic globalization has a resemblance

(0.1209)
(0.4112)
(1.3285)
with the results of the reconstructed economic globalization. The lower

1079
0.2
(3)
quantiles (5% and 10%) have, respectively, a negatively significant and

a strongly negatively significant impacts at higher quantiles (i.e., 70%,
80%, 90% and 95%). Similarly, the impact of the revisited economic
− 22.095***
globalization reduces the effect of renewable energy consumption in

0.9263***
3.0907***
− 0.4882
(0.2603)
(0.2242)
(0.3254)
(1.0300)
(2.9644)
OECD countries (e.g., Ref. [128,129]; Khan et al., 2019). Moreover, the
0.4621*
Robust standard errors are in parentheses.

1079
common globalization variables show similar results as indicated earlier

0.1
(2)
in Table 9 (e.g., Ref. [124–126].

***p < 0.01, **p < 0.05, *p < 0.1.
6. Conclusions and policy implications

− 1.9788***
− 30.329***
1.4582***
1.1815***
(0.4488)
(0.3306)
(0.5712)
(1.6356)
(3.8818)
2.7392*
The main aim of this study is to examine the impact of real capita
1079
0.05
(1)
income or GDP, real oil price, CO2 emissions and the control variables of
globalization on total renewable energy consumption, using panel data
Observations
models for 30 OECD countries over the period 1970–2015. We use the
VARIABLES
Constant
control variables including per capita income, the real oil price, and the
lnco2pc
lnrgdpc
Table 9
lnrop
lnogi
level of carbon dioxide emissions in the relationship between the pattern

of globalization(s) and its impact on renewable energy consumption to
9
H. Padhan et al.
Table 10
Machado and Silva Panel Quantile regression (2019).Model2 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnegi)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
VARIABLES 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95
lnrgdpc 1.7672*** 1.0359*** 0.4565*** 0.3910** 0.6781*** 0.8337*** 0.9342*** 0.9696*** 0.9399*** 0.3423** 0.1632**
(0.2435) (0.2705) (0.1519) (0.1667) (0.1335) (0.1052) (0.1020) (0.1412) (0.1330) (0.1487) (0.0819)
lnrop 0.9880*** 0.8954*** 0.8588*** 0.8295*** 0.7526*** 0.8940*** 0.9287*** 0.8799*** 1.0196*** 1.0618*** 0.9943***
(0.2683) (0.1516) (0.1147) (0.1271) (0.1085) (0.0849) (0.0861) (0.1123) (0.1123) (0.0873) (0.0646)
lnco2pc − 1.4668** − 0.1803 0.2843* 0.2802* 0.0470 (0.1306) − 0.0215 − 0.0072 0.0771 (0.1522) 0.1895 (0.1557) 1.1848*** 1.2863***
(0.6659) (0.2900) (0.1495) (0.1604) (0.1157) (0.1200) (0.2155) (0.0895)
lnegi − 0.7648 − 0.4120 0.9041*** 1.5232*** 1.6807*** 1.4481*** 1.0076*** 0.6124* 0.1321 (0.4290) − 0.5224** − 0.0051
(0.7291) (0.4282) (0.2284) (0.2801) (0.2714) (0.2509) (0.2624) (0.3382) (0.2645) (0.2215)
Constant − 18.461*** − 13.595*** − 13.136*** − 14.369*** − 16.675*** − 17.314*** − 16.354*** − 14.623*** − 12.533*** − 5.0689*** − 4.8630***
(2.9146) (1.7248) (1.0340) (1.2626) (1.0330) (1.0093) (1.0318) (1.3498) (1.6394) (0.8424) (1.0341)
Observations 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079

***p < 0.01, **p < 0.05, *p < 0.1.
10
Table 11
Machado and Silva Panel Quantile regression (2019).Model3 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnsgi)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
VARIABLES 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95
lnrgdpc 1.5309*** 0.4135 (0.2817) 0.1677 (0.1571) 0.5030*** 0.8970*** 1.0715*** 1.1076*** 1.1505*** 0.8656*** 0.3758** 0.2387 (0.1613)
(0.5534) (0.0844) (0.0851) (0.1082) (0.0960) (0.1446) (0.1522) (0.1713)
lnrop 1.0494** 0.9284*** 0.8801*** 0.9659*** 0.8369*** 0.9733*** 0.9122*** 0.9154*** 1.0281*** 1.0808*** 0.7119***
(0.4075) (0.2338) (0.1270) (0.1136) (0.1158) (0.1004) (0.0882) (0.1102) (0.1162) (0.0895) (0.1067)
lnco2pc − 1.7297*** − 0.4135 − 0.0023 − 0.1109 − 0.2551* − 0.2417* − 0.0145 0.0089 (0.1486) 0.1961 (0.1558) 1.1668*** 1.0103***
(0.6048) (0.3487) (0.1828) (0.1403) (0.1498) (0.1424) (0.1241) (0.1786) (0.1586)
lnsgi 2.2662 (1.4648) 1.8832*** 1.8372*** 1.5699*** 1.2347*** 0.8281*** 0.3119 (0.2616) 0.0136 (0.2844) 0.2158 (0.4048) 0.3292 (0.2612) 0.8027***
(0.7273) (0.3488) (0.2422) (0.2972) (0.2927) (0.2027)
Constant − 28.915*** − 16.390*** − 13.626*** − 15.429*** − 16.777*** − 16.954*** − 15.050*** − 13.911*** − 12.183*** − 6.2768*** − 7.2878***
(2.7906) (1.9349) (1.2192) (0.9999) (1.0973) (1.0510) (0.8951) (1.1346) (1.2814) (0.9223) (1.3185)
Observations 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079

***p < 0.01, **p < 0.05, *p < 0.1.
H. Padhan et al.
Table 12
The Machado and Silva Panel Quantile regression (2019).Model4 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnpgi)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
VARIABLES 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95
lnrgdpc 1.1358*** 0.5941*** 0.3741** 0.6104*** 0.8564*** 1.0468*** 0.9315*** 0.8582*** 0.5998*** 0.0879 0.1018 (0.1287)
(0.2619) (0.1563) (0.1496) (0.1153) (0.0964) (0.1175) (0.1003) (0.1020) (0.1383) (0.1249)
lnrop 0.9243*** 0.9127*** 0.9103*** 0.8305*** 0.9223*** 1.0026*** 0.9093*** 0.7474*** 0.8455*** 0.9600*** 0.8573***
(0.2052) (0.1724) (0.1302) (0.1160) (0.1117) (0.0992) (0.0887) (0.0941) (0.1235) (0.1290) (0.1080)
lnco2pc − 1.7227*** − 0.6709*** 0.0707 (0.1842) 0.0269 (0.1426) 0.0079 (0.1320) − 0.0080 0.0628 (0.1086) 0.0472 (0.1188) 0.2369 (0.1677) 1.2849*** 1.0581***
(0.3915) (0.2383) (0.1360) (0.2181) (0.1408)
lnpgi 6.5028*** 6.3676*** 4.4043*** 3.6803*** 3.2948*** 1.7310*** 1.8336*** 2.3183*** 2.3810*** 1.2184** 1.8322***
(0.9881) (1.0596) (0.6991) (0.5637) (0.6193) (0.5914) (0.2875) (0.3535) (0.6129) (0.4898) (0.4436)
Constant − 43.076*** − 38.081*** − 27.803*** − 25.990*** − 26.630*** − 21.467*** − 20.228*** − 20.580*** − 18.571*** − 9.886*** − 11.282***
(3.2864) (3.9090) (2.7251) (2.3573) (2.6091) (2.3473) (1.2075) (1.4066) (2.3599) (2.2735) (2.0415)
Observations 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079 1079
Notes. Robust standard errors are in parentheses.

***p < 0.01, **p < 0.05, *p < 0.1.
11
Table 13
Machado and Silva Panel Quantile regression (2019).Model5 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnrc egi)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
VARIABLES 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95
lnrgdpc 1.8406*** 1.1203*** 0.6821*** 0.5947*** 1.0062*** 1.1484*** 1.3151*** 1.3103*** 1.0911*** 0.5867*** 0.9069***
(0.2030) (0.2058) (0.1643) (0.1598) (0.1168) (0.1201) (0.1016) (0.1170) (0.1320) (0.1252) (0.1079)
lnrop 0.8467*** 0.9326*** 0.9311*** 0.8980*** 0.8982*** 0.9626*** 0.9221*** 1.0738*** 0.9739*** 1.2447*** 1.0267***
(0.2278) (0.1611) (0.1335) (0.1326) (0.1217) (0.0980) (0.0850) (0.0997) (0.0905) (0.1198) (0.1004)
lnco2pc − 1.1643* − 0.0870 0.3596** 0.3753** 0.0171 (0.1267) − 0.0789 − 0.0348 0.0375 (0.1338) 0.2397* 0.9282*** 0.7401***
(0.6336) (0.2265) (0.1804) (0.1637) (0.1335) (0.1206) (0.1410) (0.1769) (0.1032)
lnrc_egi − 1.7671** − 0.9742** 0.0673 (0.3237) 0.2899 (0.3067) 0.5196* 0.3540 (0.2768) − 0.2767 − 0.7250** − 1.2700*** − 2.1051*** − 2.8734***
(0.8981) (0.3852) (0.2914) (0.2665) (0.3032) (0.3902) (0.4786) (0.2321)
Constant − 14.976*** − 12.477*** − 12.421*** − 11.787*** − 15.607*** − 15.988*** − 14.674*** − 13.112*** − 8.0944*** − 1.1761 0.4685 (1.2372)
(3.1672) (1.2390) (1.3706) (1.3643) (0.9715) (1.1597) (1.0978) (1.2096) (1.5638) (1.7488)
Observations 1079 1079 1079 1,07z9 1079 1079 1079 1079 1079 1079 1079

***p < 0.01, **p < 0.05, *p < 0.1.
obtain the robust empirical findings. We use real gross domestic product
(GDP) because as income increases energy consumption also increases,
underscoring that GDP is considered as a factor that impacts renewable
− 0.0016***
− 14.520***
1.3244***
1.1293***
0.5166***
energy consumption. Oil prices affect renewable energy through their
(0.1647)
(0.1203)
(0.1319)
(0.0002)
(1.3665)
impacts on fossil fuels. Moreover, the carbon emissions also ia a factor
1079
0.95
(11) affecting renewable energy consumption [8]. We use this set of common
control variables following the literature as in Ref. [2,3,16,17].
The main contribution of the paper is based on the different con
− 0.0014***
− 12.278***
figurations of globalization of Konjunkturforschungsstelle (KOF; 2018)
0.9428***
1.2336***
0.7730***
(0.1523)
(0.1315)
(0.1950)
(0.0002)
(1.3544)
as it uses classic globalisation, reconstructed economic globalisation and
1079
(10)
revisited economic globalisation and examines thier influence on

0.9
renewable energy consumption. Moreover, we have applied the novel

Panel quantile regression technique developed by Machado and Silva
0.0467 (0.1520)
regression [134] to examine the impact of globalisation(s) and renew

− 0.0010***
− 14.136***
1.2547***
1.0233***
able energy consumption to achieve robust results for the

(0.1391)
(0.1187)
(0.0003)
(1.3571)
1079 energy-globalization nexus. Moreover, our methodology of quantile
0.8
(9)
regression provides more explicit results than the ordinary least square
method because it takes in account the heterogeneity and distributional
heterogeneity which are neglected by the ordinary least square method.
− 0.0007***
− 16.926***
Further, this method explains the entire conditional distribution of the

1.4792***
1.0375***
− 0.0867
(0.1220)
(0.0967)
(0.1362)
(0.0002)
(1.1093)
dependent variable and also is robust to outlying observation of the

1079
exogenous variables. The empirical results demonstrate the existence of

0.7
(8)
a long-run relationship between the renewable energy consumption and

the control variables. Specifically, the results show a positive and sig
nificant relationship between per capita income, real price of oil and
− 15.556***
1.3028***
0.9220***
carbon emissions per capita and renewable energy.

− 0.0436
− 0.0002
(0.0786)
(0.0366)
(0.1204)
(0.0001)
(0.5452)
Higher levels of globalization (overall, economics, social and politi

1079
0.6
(7)
cal) widely stimulate renewable energy consumption. On one hand, the

results show that the reconstructed and revisited economic globalization
0.0002 (0.0002)
reduces renewable energy consumption, and this conclusion is also

− 14.300***
robust to different measures of economic globalization. They also

1.0992***
0.9546***
− 0.0019
(0.1155)
(0.0943)
(0.1338)
(1.0106)
confirm that renewable energy consumption boosts the domestic econ

1079
omy in the middle section thrugh importing modern and advanced

0.5
(6)
technology. On the other hand, the lower and higher sections rely more
Machado and Silva Panel Quantile regression (2019).Model6 : ln rec = f (lnrgdpc, lnrop, lnco2 pc, lnrevegi)
on non-renewable (coal, wood) energy as a result of the livelihood

0.0080 (0.1211)
approach that depends on wood and coal (lower quantiles section) to

− 12.885***
0.9346***
0.8356***
0.0006***
generate energy and of the accelerated growth (higher quantiles sec

(0.1275)
(0.1189)
(0.0002)
(1.0992)
tion), respectively, which deteriorates the quality of the environment

1079
0.4
(5)
without caring about the essence of globalization.

Based on the results of the study, the following policy implications
0.0003 (0.0003)
may be adopted to help improve environmental quality in OECD coun

− 10.347***
tries. Policymakers should focus on enhancing the growth of renewable

0.5537***
0.8643***
0.4191**
(0.1851)
(0.1331)
(0.1672)
(1.5752)
energy segment in a cost-effective way. Therefore, OECD countries

1079
should compare the cost of reducing carbon emissions with the capital
0.3
(4)
cost of renewable energy. Moreover, the policymakers of OECD coun

tries should plan and enforce growth-oriented policies and agendas to
0.0001 (0.0003)
attain a consistent drop in emissions. For instance, as their income rises,

− 11.932***
0.6475***
0.9536***
OECD countries can design environmental awareness, implement firm

(0.1869)
(0.1370)
(0.1816)
(1.6518)
0.3481*
environmental protocols, and adopt clean technologies for renewable

1079
0.2
(3)
energy consumption that is eco-friendly. It is also important to

strengthen technological innovations and knowledge transfer across
OECD to improve renewable energy efficiency.
− 15.895***
Further, the host countries should measure the per capita income
1.1175***
0.9268***
− 0.0007*
− 0.1072
(0.2372)
(0.1881)
(0.2455)
(0.0004)
(1.9714)
before assessing the introduction of renewable energy consumption into

1079
the country. Each OECD country should consume energy efficiently and
0.1
(2)
establish energy development programs to educate citizens how to shift

***p < 0.01, **p < 0.05*p < 0.1.
from fossil fuels (such as oil and coal) towards more green renewable
energy system. Moreover, OECD governments should provide subsidies
− 21.631***
− 1.2142**
− 0.0013**
1.8341***
0.9257***
to moderate the usage costs of renewable energy consumption and

(0.2343)
(0.2945)
(0.6118)
(0.0006)
(1.8332)
educate consumers how to start from the grass root level to attain the
1079
0.05
(1)
benefits of renewable energy consumption (such as reducing air pollu

tion, decrease the dependence on coal and other fossil fuels), and
Observations
welcome industries’ initiatives related to renewable energy resources.

VARIABLES
Our findings suggest that higher carbon-emitting countries could take

Constant
Table 14
lnco2pc
lnrgdpc
lrevegi
lnrop
advantage of higher economic growth and ultimately increase the

population as well. Therefore, higher CO2-emitting countries in OECD
12
should not only address economic growth but should also pay attention CRediT authorship contribution statement
to growing population in order to reduce carbon emissions. In addition,
OECD countries should also enhance globalization through more trade Hemachandra Padhan: Conceptualization, Resources, Data cura
openness which increases the reliance on renewable energy consump tion, Writing - original draft, Writing - review & editing. Aviral Kumar
tion. Finally, the most important implication of our findings is that the Tiwari: Methodology, Software, Validation, Formal analysis, Investi
design of uniform renewable energy consumption policies equally across gation, Resources, Data curation, Writing - original draft, Writing - re
OECD countries are likely to lead to different levels of renewable energy view & editing. Rizwan Ahmed: Investigation, Resources, Data
consumption requirements. Therefore, renewable energy consumption curation, Writing - original draft, Writing - review & editing. Shawkat
parameters should be measured and developed differently across OECD Hammoudeh: Resources, Data curation, Supervision, Writing - original
countries. draft, Writing - review & editing, Visualization.
Funding Declaration of competing interest
This paper does not have any funding from any institutions, agencies The authors declare that they have no known competing financial
or commercial sectors. interests or personal relationships that could have appeared to influence
the work reported in this paper.
Appendix A. Cross sectional dependence test
Variables Breusch-Pagan LM Pesaran scaled LM Bias-corrected scaled LM Pesaran CD
Lnrec 10059.25*** 326.2925*** 325.9592*** 98.53908***

Lngdpc 15615.43*** 514.6649*** 514.3315*** 123.5521***
Lnrop 17108.00*** 565.2676*** 564.9343*** 129.7603***
lnco2pc 7089.907*** 225.6225*** 225.2891*** 28.47142***
Lnogi 15332.66*** 505.0781*** 504.7448*** 122.2097***
Lnegi 13957.28*** 458.4483*** 458.1150*** 113.2989***
Lnsgi 13665.48*** 448.5554*** 448.2221*** 115.5948***
Lnpgi 9655.907*** 312.6180*** 312.2847*** 91.13016***
lnrc_egi 13179.78*** 432.0885*** 431.7552*** 108.0582***
Lnrevegi 15332.27*** 505.0649*** 504.7315*** 121.8918***
Note: ***p < 0.01.
Appendix B. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.esr.2020.100535.

Note: Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. lnrgdpce represents the log returns of GDP per capit. lncop refer
to log return of oil prices, ICO2 denotes log returns of CO2 emissions and lnorgi captures overall globalization.
References [14] J. Baek, Y. Cho, W.W. Koo, The environmental consequences of globalization: a
country-specific time-series analysis, Ecol. Econ. 68 (8–9) (2009) 2255–2264.
[15] F. Belaid, M.H. Zrelli, Renewable and non-renewable electricity consumption,
[1] A. Abadie, J. Angrist, G. Imbens, Instrumental variables estimates of the effect of
environmental degradation and economic development: evidence from
subsidized training on the quantiles of trainee earnings, Econometrica 70 (1)
Mediterranean countries, Energy Pol. 133 (2019) 110929.
(2002) 91–117.
[16] B. Bhattacharya, S.R. Paramati, I. Ozturk, S. Bhattacharya, The effect of
[2] M. Aguirre, G. Ibikunle, Determinants of renewable energy growth: a global
renewable energy consumption on economic growth: evidence from top 38
sample analysis, Energy Pol. 69 (2014) 374–384.
countries, Appl. Energy 162 (2016) 733–741.
[3] M.S.B. Aissa, M.B. Jebli, S.B. Youssef, Output, renewable energy consumption
[17] J. Breitung, The local power of some unit root tests for panel data, in: B.H. Baltagi
and trade in Africa, Energy Pol. 66 (2014) 11–18.
(Ed.), Advances in Econometrics, Volume 15: Nonstationary Panels, Panel
[4] R. Akram, F. Chen, F. Khalid, Z. Ye, M.T. Majeed, Heterogeneous effects of energy
Cointegration, and Dynamic Panels, JAY Press, Amsterdam, 2000, pp. 161–178.
efficiency and renewable energy on carbon emissions: evidence from developing
[18] D.C. Broadstock, H. Cao, D. Zhang, Oil shocks and their impact on energy related
countries, J. Clean. Prod. 247 (2020) 119122.
stocks in China, Energy Econ. 34 (6) (2012) 1888–1895.
[5] M.M. Alam, M.W. Murad, The impacts of economic growth, trade openness and
[20] M. Buchinsky, Changes in the US wage structure 1963-1987: application of
technological progress on renewable energy use in organization for economic co-
quantile regression, Econometrica: Journal of the Econometric Society (1994)
operation and development countries, Renew. Energy 145 (2020) 382–390.
405–458.
[6] A. Alper, O. Oguz, The role of renewable energy consumption in economic
[21] I.A. Canay, A simple approach to quantile regression for panel data, Econom. J.
growth: evidence from asymmetric causality, Renew. Sustain. Energy Rev. 60
14 (3) (2011) 368–386.
(2016) 953–959.
[22] G. Chamberlain, Quantile regression, censoring, and the structure of wages, in:
[7] J.B. Ang, CO2 emissions, energy consumption, and output in France, Energy Pol.
Advances in Econometrics: Sixth World Congress, vol. 2, 1994, pp. 171–209.
35 (10) (2007) 4772–4778.
[24] C. Chen, M. Pinar, T. Stengos, Renewable energy consumption and economic
[9] N. Apergis, J.E. Payne, Renewable energy consumption and growth in Eurasia,
growth nexus: evidence from a threshold model, Energy Pol. 139 (2020) 111295.
Energy Econ. 32 (6) (2010) 1392–1397.
[25] C. Cheng, X. Ren, Z. Wang, C. Yan, Heterogeneous impacts of renewable energy
[10] N. Apergis, J.E. Payne, Renewable energy consumption and economic growth:
and environmental patents on CO2 emission-Evidence from the BRIICS, Sci. Total
evidence from a panel of OECD countries, Energy Pol. 38 (1) (2010) 656–660.
Environ. 668 (2019) 1328–1338.
[11] N. Apergis, J.E. Payne, Renewable energy, output, carbon dioxide emissions, and
[26] C. Cheng, X. Ren, Z. Wang, Y. Shi, The impacts of non-fossil energy, economic
oil prices: evidence from South America, Energy Sources B Energy Econ. Plann.
growth, energy consumption, and oil price on carbon intensity: evidence from a
10 (3) (2015) 281–287.
panel quantile regression analysis of EU 28, Sustainability 10 (11) (2018) 4067.
[12] M. Arellano, Practitioners’ corner: computing robust standard errors for within-
[28] A. Cheon, J. Urpelainen, Oil prices and energy technology innovation: an
groups estimators, Oxf. Bull. Econ. Stat. 49 (4) (1987) 431–434.
empirical analysis, Global Environ. Change 22 (2) (2012) 407–417.
[13] J. Asafu-Adjaye, The relationship between energy consumption, energy prices
[29] V. Chernozhukov, I. Fernández Val, A. Galichon, Quantile and probability curves
and economic growth: time series evidence from Asian developing countries,
without crossing, Econometrica 78 (3) (2010) 1093–1125.
Energy Econ. 22 (6) (2000) 615–625.
[30] V. Chernozhukov, C. Hansen, An IV model of quantile treatment effects,
Econometrica 73 (1) (2005) 245–261.
13
[31] V. Chernozhukov, C. Hansen, Instrumental quantile regression inference for sustainable economic development, Emerg. Mark. Finance Trade 54 (8) (2018)
structural and treatment effect models, J. Econom. 132 (2) (2006) 491–525. 1761–1777.
[32] V. Chernozhukov, C. Hansen, Instrumental variable quantile regression: a robust [81] E. Kyritsis, A. Serletis, Oil prices and the renewable energy sector, Energy J. 40
inference approach, J. Econom. 142 (1) (2008) 379–398. (2019) (The New Era of Energy Transition).
[33] B.R. Copeland, Policy endogeneity and the effects of trade on the environment, [82] C. Lamarche, Robust penalized quantile regression estimation for panel data,
Agric. Resour. Econ. Rev. 34 (1) (2005) 1–15. J. Econom. 157 (2) (2010) 396–408.
[34] B.R. Copeland, M.S. Taylor, North-South trade and the environment, Q. J. Econ. [83] T. Lancaster, The incidental parameter problem since 1948, J. Econom. 95 (2)
109 (3) (1994) 755–787. (2000) 391–413.
[35] B.R. Copeland, M.S. Taylor, Trade, growth, and the environment, J. Econ. Lit. 42 [84] H.H. Lean, R. Smyth, Multivariate Granger causality between electricity
(1) (2004) 7–71. generation, exports, prices and GDP in Malaysia, Energy 35 (9) (2010)
[36] O. Damette, P. Delacote, On the economic factors of deforestation: what can we 3640–3648.
learn from quantile analysis? Econ. Modell. 29 (6) (2012) 2427–2434. [86] A. Levin, C.-F. Lin, C.-S.J. Chu, Unit root tests in panel data: asymptotic and finite-
[37] D. Dedeoğlu, H. Kaya, Energy use, exports, imports and GDP: new evidence from sample properties, J. Econom. 108 (1) (2002) 1–24.
the OECD countries, Energy Pol. 57 (2013) 469–476. [88] B. Lin, M. Moubarak, Renewable energy consumption – economic growth nexus
[38] M.A. Destek, Investigation on the Role of Economic, Social, and Political for China, Renew. Sustain. Energy Rev. 40 (2014) 111–117.
Globalization on Environment: Evidence from CEECs, Environmental Science and [89] F. Liu, S. Tait, A. Schellart, M. Mayfield, J. Boxall, Reducing carbon emissions by
Pollution Research, 2019, pp. 1–14. integrating urban water systems and renewable energy sources at a community
[39] M.A. Destek, A. Sinha, Renewable, non-renewable energy consumption, economic scale, Renew. Sustain. Energy Rev. 123 (2020) 109767.
growth, trade openness and ecological footprint: evidence from organisation for [90] J.A. Machado, J.S. Silva, Quantiles via moments, J. Econom. 213 (1) (2019)
economic Co-operation and development countries, J. Clean. Prod. 242 (2020) 145–173.
118537. [91] A.C. Marques, J.A. Fuinhas, J.P. Manso, Motivations driving renewable energy in
[40] M.A. Destek, A. Aslan, Disaggregated renewable energy consumption and European countries: a panel data approach, Energy Pol. 38 (2010) 6877–6885.
environmental pollution nexus in G-7 countries, Renew. Energy 151 (2020) [92] L.M. Marques, J.A. Fuinhas, A.C. Marques, Augmented energy-growth nexus:
1298–1306. economic, political and social globalization impacts, Energy Procedia 136 (2017)
[43] A. Dreher, Does globalization affect growth? Evidence from a new index of 97–101.
globalization, Appl. Econ. 38 (10) (2006) 1091–1110. [94] P.K. Narayan, R. Smyth, Multivariate Granger causality between electricity
[44] J.C. Driscoll, A.C. Kraay, Consistent covariance matrix estimation with spatially consumption, exports and GDP: evidence from a panel of Middle Eastern
dependent panel data, Rev. Econ. Stat. 80 (4) (1998) 549–560. countries, Energy Pol. 37 (1) (2009) 229–236.
[47] W. Fan, Y. Hao, An empirical research on the relationship amongst renewable [96] W.K. Newey, K.D. West, Hypothesis testing with efficient method of moments
energy consumption, economic growth and foreign direct investment in China, estimation, Int. Econ. Rev. (1987) 777–787.
Renew. Energy 146 (2020) 598–609. [97] J. Neyman, E.L. Scott, Consistent estimates based on partially consistent
[48] C.A. Flores, A. Flores-Lagunes, D. Kapetanakis, Lessons from quantile panel observations, Econometrica: Journal of the Econometric Society (1948) 1–32.
estimation of the environmental Kuznets curve, Econom. Rev. 33 (8) (2014) [98] F. Nicolli, F. Vona, Energy market liberalization and renewable energy policies in
815–853. OECD countries, Energy Pol. 128 (2019) 853–867.
[50] A.F. Galvao Jr., Quantile regression for dynamic panel data with fixed effects, [99] M.M. Nilsson, The Effect of Oil Prices on Patents in Renewable Energy: A DTC
J. Econom. 164 (1) (2011) 142–157. Approach, 2019.
[51] A.F. Galvao, K. Kato, Smoothed quantile regression for panel data, J. Econom. [100] K.H. Nguyen, M. Kakinaka, Renewable energy consumption, carbon emissions,
193 (1) (2016) 92–112. and development stages: some evidence from panel cointegration analysis,
[52] A.F. Galvao, L. Wang, Efficient minimum distance estimator for quantile Renew. Energy 132 (2019) 1049–1057.
regression fixed effects panel data, J. Multivariate Anal. 133 (2015) 1–26. [101] W. Oh, K. Lee, Energy consumption and economic growth in Korea: testing the
[53] H. Geller, P. Harrington, A.H. Rosenfeld, S. Tanishima, F. Unander, Polices for causality relation, J. Pol. Model. 26 (8–9) (2004) 973–981.
increasing energy efficiency: thirty years of experience in OECD countries, Energy [102] A. Omri, D.K. Nguyen, On the determinants of renewable energy consumption:
Pol. 34 (5) (2006) 556–573. international evidence, Energy 72 (2014) 554–560.
[54] K.H. Ghali, M.I. El-Sakka, Energy use and output growth in Canada: a [103] I. Ozturk, A. Acaravci, The long-run and causal analysis of energy, growth,
multivariate cointegration analysis, Energy Econ. 26 (2) (2004) 225–238. openness and financial development on carbon emissions in Turkey, Energy
[55] P. Gourevitch, The second image reversed: the international sources of domestic Econ. 36 (2013) 262–267.
politics, Int. Organ. 32 (4) (1978) 881–912. [107] M.J. Pickl, The renewable energy strategies of oil majors–From oil to energy?
[56] G. Gozgor, M.K. Mahalik, E. Demir, H. Padhan, The impact of economic Energy Strategy Reviews 26 (2019) 100370.
globalization on renewable energy in the OECD countries, Energy Pol. 139 (2020) [109] S. Rafiq, K. Alam, Identifying the determinants of renewable energy
111365. consumption in leading renewable energy investor emerging countries, in: 39th
[57] G. Gozgor, Robustness of the KOF index of economic globalisation, World Econ. Australian Conference of Economists, Held, 2010, pp. 27–29.
41 (2) (2018) 414–430. [111] S.M. Rahman, M.D. Miah, The impact of sources of energy production on
[59] G. Gozgor, A new approach to the renewable energy-growth nexus: evidence from globalization: evidence from panel data analysis, Renew. Sustain. Energy Rev.
the USA, Environ. Sci. Pollut. Control Ser. 25 (17) (2018) 16590–16600. 74 (2017) 110–115.
[60] C. Gutenbrunner, J. Jurecková, Regression rank scores and regression quantiles, [112] M.M. Rahman, E. Velayutham, Renewable and non-renewable energy
Ann. Stat. (1992) 305–330. consumption-economic growth nexus: new evidence from South Asia, Renew.
[64] X. He, Quantile curves without crossing, Am. Statistician 51 (2) (1997) 186–192. Energy 147 (2020) 399–408.
[66] C.Y.L. Hsiao, W. Lin, X. Wei, G. Yan, S. Li, N. Sheng, The impact of international [113] S.F. Razmi, B.R. Bajgiran, M. Behname, T.E. Salari, S.M.J. Razmi, The
oil prices on the stock price fluctuations of China’s renewable energy enterprises, relationship of renewable energy consumption to stock market development and
Energies 12 (24) (2019) 4630. economic growth in Iran, Renew. Energy 145 (2020) 2019–2024.
[67] P.J. Huber, The behaviour of maximum likelihood estimates under nonstandard [114] J.C. Reboredo, M.A. Rivera-Castro, A. Ugolini, Wavelet-based test of co-
conditions, Proceedings of the fifth Berkeley Symposium on Mathematical movement and causality between oil and renewable energy stock prices, Energy
Statistics and Probability 1 (1) (1967) 221–233. Econ. 61 (2017) 241–252.
[68] G.J. Ikenberry, The irony of state strength: comparative responses to the oil [115] J.C. Reboredo, Is there dependence and systemic risk between oil and renewable
shocks in the 1970s, Int. Organ. 40 (1) (1986) 105–137. energy stock prices? Energy Econ. 48 (2015) 32–45.
[69] K.S. Im, M.H. Pesaran, Y. Shin, Testing for unit roots in heterogeneous panels, [116] S. Ruhul, S. Rafiq, A.K. Hassan, Causality and dynamics of energy consumption
J. Econom. 115 (1) (2003) 53–74. and output: evidence from non-OECD Asian countries, J. Econ. Dev. 33 (2)
[70] J. Inchauspe, R.D. Ripple, S. Trück, The dynamics of returns on renewable energy (2008) 1–26.
companies: a state-space approach, Energy Econ. 48 (2015) 325–335. [117] W. Rogers, Regression standard errors in clustered samples, Stata Technical
[72] C. Kalayci, The impact of economic globalization on CO2 emissions: the case of Bulletin 13 (1993) 19–23.
NAFTA countries, Int. J. Energy Econ. Pol. 9 (1) (2019) 356. [118] P. Sadorsky, Renewable energy consumption, CO2 emissions and oil prices in
[74] K. Kato, A.F. Galvao Jr., G.V. Montes-Rojas, Asymptotics for panel quantile the G7 countries, Energy Econ. 31 (3) (2009) 456–462.
regression models with individual effects, J. Econom. 170 (1) (2012) 76–91. [119] P. Sadorsky, Renewable energy consumption and income in emerging
[75] M.K. Khan, J.Z. Teng, M.I. Khan, M.O. Khan, Impact of globalization, economic economies, Energy Pol. 37 (10) (2009) 4021–4028.
factors and energy consumption on CO2 emissions in Pakistan, Sci. Total Environ. [120] P. Sadorsky, Trade and energy consumption in the Middle East, Energy Econ. 33
688 (2019) 424–436. (5) (2011) 739–749.
[76] R. Koenker, Quantile regression for longitudinal data, J. Multivariate Anal. 91 (1) [122] P. Sadorsky, Correlations and volatility spillovers between oil prices and the
(2004) 74–89. stock prices of clean energy and technology companies, Energy Econ. 34 (1)
[77] R. Koenker, G. Bassett Jr., Regression quantiles. Econometrica, Journal of the (2012) 248–255.
Econometric Society (1978) 33–50. [123] R.A. Salim, S. Rafiq, Why do some emerging economies proactively accelerate
[78] R. Koenker, G. Bassett Jr., Robust tests for heteroscedasticity based on regression the adoption of renewable energy? Energy Econ. 34 (4) (2012) 1051–1057.
quantiles, Econometrica: Journal of the Econometric Society (1982) 43–61. [124] M. Salman, X. Long, L. Dauda, C.N. Mensah, S. Muhammad, Different impacts of
[79] R. Koenker, Q. Zhao, L- estimation for linear heteroscedastic models, Journal title export and import on carbon emissions across 7 ASEAN countries: a panel
of Nonparametric Statistics 3 (3–4) (1994) 223–235. quantile regression approach, Sci. Total Environ. 686 (2019) 1019–1029.
[80] A.M. Kutan, S.R. Paramati, M. Ummalla, A. Zakari, Financing renewable energy
projects in major emerging market economies: evidence in the perspective of
14
[125] S.A. Sarkodie, S. Adams, Renewable energy, nuclear energy, and environmental [135] T. Vasylieva, O. Lyulyov, Y. Bilan, D. Streimikiene, Sustainable economic
pollution: accounting for political institutional quality in South Africa, Sci. Total development and greenhouse gas emissions: the dynamic impact of renewable
Environ. 643 (2018) 1590–1601. energy consumption, GDP, and corruption, Energies 12 (17) (2019) 3289.
[126] H. Sasana, J. Aminata, Energy subsidy, energy consumption, economic growth, [137] X. Wen, Y. Guo, Y. Wei, D. Huang, How do the stock prices of new energy and
and carbon dioxide emission: Indonesian case studies, Int. J. Energy Econ. Pol. 9 fossil fuel companies correlate? Evidence from China, Energy Econ. 41 (2014)
(2) (2019) 117. 63–75.
[127] S.E.C. Sener, J.L. Sharp, A. Anctil, Factors impacting diverging paths of renewable [138] H. White, A heteroskedasticity-consistent covariance matrix estimator and a
energy: a review, Renew. Sustain. Energy Rev. 81 (2018) 2335–2342. direct test for heteroskedasticity, Econometrica: Journal of the Econometric Society
[128] I.H. Shah, C. Hiles, B. Morley, How do oil prices, macroeconomic factors and (1980) 817–838.
policies affect the market for renewable energy? Appl. Energy 215 (2018) 87–97. [139] M. Wiley, J.F. Wiley, Univariate data visualization, in: Advanced R Statistical
[129] M. Shahbaz, A. Lahiani, S. Abosedra, S. Hammoudeh, The role of globalization in Programming and Data Models, Apress, Berkeley, CA, 2019, pp. 1–31.
energy consumption: a quantile cointegrating regression approach, Energy Econ. [140] M. Xu, R. Buyya, Managing renewable energy and carbon footprint in multi-cloud
71 (2018) 161–170. computing environments, J. Parallel Distr. Comput. 135 (2020) 191–202.
[130] M. Shahbaz, H.H. Lean, A. Farooq, Natural gas consumption and economic [142] D. Yan, Y. Kong, X. Ren, Y. Shi, S. Chiang, The determinants of urban
growth in Pakistan, Renew. Sustain. Energy Rev. 18 (2013) 87–94. sustainability in Chinese resource-based cities: a panel quantile regression
[131] M. Shahbaz, H. Mallick, M.K. Mahalik, P. Sadorsky, The role of globalization on approach, Sci. Total Environ. 686 (2019) 1210–1219.
the recent evolution of energy demand in India: implications for sustainable [143] S. Yu, X. Hu, L. Li, H. Chen, Does the development of renewable energy promote
development, Energy Econ. 55 (2016) 52–68. carbon reduction? Evidence from Chinese provinces, J. Environ. Manag. 268
[132] M. Shahbaz, S. Nasreen, C.H. Ling, R. Sbia, Causality between trade openness and (2020) 110634.
energy consumption: what causes what in high, middle and low income [144] Q. Zhao, Restricted regression quantiles, J. Multivariate Anal. 72 (1) (2000)
countries? Energy Pol. 70 (2014) 126–143. 78–99.
[133] M. Shahbaz, S.J.H. Shahzad, M.K. Mahalik, P. Sadorsky, How strong is the causal [145] X. Zhao, D. Luo, Forecasting fossil energy consumption structure toward low-
relationship between globalization and energy consumption in developed carbon and sustainable economy in China: evidence and policy responses, Energy
economies? A country-specific time-series and panel analysis, Appl. Econ. 50 (13) strategy reviews 22 (2018) 303–312.
(2018) 1479–1494. [146] H. Zhu, L. Duan, Y. Guo, K. Yu, The effects of FDI, economic growth and energy
[134] U. Soytas, R. Sari, The relationship between energy and production: evidence consumption on carbon emissions in ASEAN-5: evidence from panel quantile
from Turkish manufacturing industry, Energy Econ. 29 (6) (2007) 1151–1165. regression, Econ. Modell. 58 (2016) 237–248.
15

Energy Strategy Reviews

Uploaded by

Copyright:

Available Formats

Energy Strategy Reviews

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

Energy Strategy Reviews

Uploaded by

Copyright:

Available Formats

Energy Strategy Reviews 32 (2020) 100535

Contents lists available at ScienceDirect

Energy Strategy Reviews

Renewable energy consumption and robust globalization(s) in OECD

the middle quantile level by enabling the importation of advanced

positive and significant impact at the lower to middle quantiles (10%–

(1.8596) influence on the total renewable consumptions as the social norms of

results as demonstrated earlier in Table 9 (e.g., Refs. [124–126].

In Table 12 of Model 4, we have included political globalization

The outcome indicates that political globalization is strongly and posi­

tively significant in the lowest to highest quantiles (5%–95%). It clearly

indicates that political globalization promotes strongly total renewable

energy consumption in the 30 OECD countries over the period 1970 to

renewable energy [133]. The other explanatory variables present similar

results as mentioned earlier in Table 9 (e.g., Ref. [124–126].

In Table 13 of Model 5, we incorporate the reconstructed economic

emissions in kg per capita (LNCO2PC). The results indicate that the

reconstructed economic globalization is negatively significant at the

lower quantiles (5%, 10% and 40%); however, it is actively negatively

influence of renewable energy consumption in OECD countries (e.g.,

Ref. [128,129]; Khan et al., 2019). The other explanatory variables

illustrate similar results as demonstrated earlier in Table 9 (e.g.,

In Table 14 of Model 6, we replace the old measure of globalization

the results measuring the effect on total renewable energy consumption

in this table. The revisited economic globalization has a resemblance

with the results of the reconstructed economic globalization. The lower

quantiles (5% and 10%) have, respectively, a negatively significant and

globalization reduces the effect of renewable energy consumption in

Robust standard errors are in parentheses.

common globalization variables show similar results as indicated earlier

in Table 9 (e.g., Ref. [124–126].

6. Conclusions and policy implications

level of carbon dioxide emissions in the relationship between the pattern

Robust standard errors are in parentheses.

Robust standard errors are in parentheses.

Notes. Robust standard errors are in parentheses.

Robust standard errors are in parentheses.

revisited economic globalisation and examines thier influence on

renewable energy consumption. Moreover, we have applied the novel

regression [134] to examine the impact of globalisation(s) and renew­

able energy consumption to achieve robust results for the

Further, this method explains the entire conditional distribution of the

dependent variable and also is robust to outlying observation of the

exogenous variables. The empirical results demonstrate the existence of

a long-run relationship between the renewable energy consumption and

carbon emissions per capita and renewable energy.

Higher levels of globalization (overall, economics, social and politi­

cal) widely stimulate renewable energy consumption. On one hand, the

reduces renewable energy consumption, and this conclusion is also

robust to different measures of economic globalization. They also

confirm that renewable energy consumption boosts the domestic econ­

omy in the middle section thrugh importing modern and advanced

on non-renewable (coal, wood) energy as a result of the livelihood

approach that depends on wood and coal (lower quantiles section) to

generate energy and of the accelerated growth (higher quantiles sec­

tion), respectively, which deteriorates the quality of the environment

without caring about the essence of globalization.

may be adopted to help improve environmental quality in OECD coun­

tries. Policymakers should focus on enhancing the growth of renewable

The outcome indicates that political globalization is strongly and posi

regression [134] to examine the impact of globalisation(s) and renew

Higher levels of globalization (overall, economics, social and politi

confirm that renewable energy consumption boosts the domestic econ

generate energy and of the accelerated growth (higher quantiles sec

may be adopted to help improve environmental quality in OECD coun

cost of renewable energy. Moreover, the policymakers of OECD coun

benefits of renewable energy consumption (such as reducing air pollu

Lnrec 10059.25* 326.2925* 325.9592* 98.53908*