THE JOURNAL OF ENERGY

AND DEVELOPMENT

Ali Maâlej and Alexandre Cabagnols,

“CO2 Emissions and Growth: Exploring

the Nexus between Renewable Energies,
Economic Activity, and Technology,”
Volume 46, Number 1

Copyright 2022
 CO2 EMISSIONS AND GROWTH: EXPLORING THE
NEXUS BETWEEN RENEWABLE ENERGIES,
ECONOMIC ACTIVITY, AND TECHNOLOGY

Ali Ma^alej AND Alexandre Cabagnols*

Introduction

D uring the industrial revolution, fossil-based energies have been one of the
main factors of economic growth. As is noted by E. Wrigley “a necessary
condition for the move from a world where growth was at best asymptotic to one
in that it could be, at least for a period, exponential was dependent upon the dis-
covery and exploitation of a vast reservoir of energy that had remained untapped
in organic economies.”1 However, in the long run, fossil-based energies face two
main obstacles: fixed natural endowments and pollutant emissions. First, the finite
initial endowment of non-renewable energy resources explains why we are
“obliged to use lower-grade ores, drill deeper, and extend our supply network,”
which results in an “increase [of] our per capita use of energy and our per capita
impact on the environment.”2 Second, the use of non-renewable energy also is

*Ali Ma^alej is an Assistant Professor in Economics at the National Engineering School of Sfax
(Tunisia). He earned his Doctorate and HDR in economics from the University of Sfax. Dr Ma^alej’s
research and teaching interests focus on entrepreneurship education, economics and management of
innovation, and sustainable development. Dr. Ma^alej’s work has been published in Efficiency in Higher
Education and Entrepreneurship Education in addition to Environmental Economics.
Alexandre Cabagnols is a Senior Lecturer in Economics at Polytech Clermont-Ferrand (France). He
earned his Ph.D. in economics of innovation from the University of Lyon. Dr. Cabagnols’ research and
teaching interests surround entrepreneurship education, green innovation, and start-up innovation
strategies. Dr. Cabagnols’s works have been published in Entrepreneurship Education in addition to
Environmental Economics.

The Journal of Energy and Development, Vol. 46, Nos. 1-2

Copyright # 2022 by the International Research Center for Energy and Economic Development
(ICEED). All rights reserved.
1
2 THE JOURNAL OF ENERGY AND DEVELOPMENT

linked to a dramatic surge of carbon dioxide (CO2) emissions3 that, among other
anthropogenic greenhouse gases, is considered one of the main causes of global
warming and climate change.4 These environmental concerns and their far-
reaching implications are driving the calls for the development of a new energy
revolution based on renewable energy sources that would be neutral from a climate
change perspective.5 However, during the first stage of technological development
of renewable energies their adoption costs may be higher than the costs of fossil-
based energies.6 This could result in a political dilemma between two conflicting
targets: economic growth versus CO2 emission reduction. That dilemma could par-
ticularly affect lower-income economies that face strong financial and technologi-
cal constraints.7 For these countries, a compromise may be the promotion of
improved fossil-based energy technologies with higher returns.8 Conversely, the
level of technological competence of the wealthier countries could enable them to
take greater advantage of new sources of energy relative to less wealthy nations.
An additional and complicating factor in addressing the issue of climate change
is that fossil-based energies are not the only drivers of CO2 emissions from
human-based activities as non-fossil-based CO2 emissions also exist.9 Among
other sources, these emissions result from the cement industry, the building of
roads, the iron and steel industries, and from many other industrial chemical pro-
cesses. Consequently, CO2 emissions have three main drivers: (1) the quantity of
fossil energy production and consumption that depends on political choices con-
cerning the promotion of renewable energies in the national energy mix; (2) the
importance of non-fossil CO2 emissions that result from the global level of eco-
nomic activity (particularly in the industrial sector); and (3) the energy and CO2
efficiency of the industrial processes in use, which is determined by a nation’s
technological level. Considering these three drivers offers a new perspective con-
cerning the study of the possible dilemma between CO2 emission reduction poli-
cies and economic activity. The purpose of our research is to investigate this issue
on a sample of three wealthy northern European countries: Germany, Finland, and
Denmark.
Our paper is structured as follows. In the first section, we provide a literature
review of the research on the connection between gross domestic product (GDP),
energy use, CO2 emissions, and technology. In the second section, we describe our
methodology. In the third section, we report the results of the error correction
model (ECM) estimations and offer our conclusions in the final section.

Literature Review

Many countries are setting national targets to increase consumption of renew-

able energy in an attempt to reduce CO2 emissions. Several studies have shown
that increasing the use of renewable energies could reduce CO2 emissions and
 THE CO2 EMISSIONS AND GROWTH NEXUS 3

improve economic growth. M. Rahman and X. Vu compare the relationships

among renewable energy consumption, economic growth, and CO2 emissions in
their case studies of Australia and Canada for the period of 1960–2015.10 They use
autoregressive distributed lag (ARDL) bounds tests to analyze the long-run rela-
tionships among the selected variables. The causal relationships were examined
using vector error correction model (VECM) Granger causality tests. In Australia,
the ARDL tests indicate that economic growth increases CO2 emissions in both the
long and short run; however, trade and renewable energy consumption decrease
CO2 emissions only in the short run. For Australia, the VECM causality tests indi-
cate that in the short run, economic growth, trade, and renewable energy consump-
tion Granger cause CO2 emissions. In the long run, economic growth, trade, and
renewable energy consumption have a bidirectional causal relationship with CO2
emissions. In Canada, ARDL tests show that trade increases CO2 emissions both
in the long and short run, whereas economic growth and urban population increase
CO2 emissions only in the long run. VECM causality tests report a long-run bidi-
rectional causality among CO2 emissions, economic growth, and renewable energy
consumption.
In the case of France, J. Ang analyzes the dynamic relationship between CO2
emissions, energy consumption, and GDP over the period 1960–2000.11 The
author’s multivariate VECM analysis indicates that energy use is a key determinant
of CO2 emissions. The relationship between GDP and CO2 emissions would be
quadratic in the long run. These results support the environmental Kuznets curve
(EKC) hypothesis12 that states that in a first stage the level of environmental pollu-
tion increases with income, stabilizes in a second stage, and finally decreases in a
third stage.
E. Dogan and F. Seker investigate the EKC model and explore the effect of
GDP, the square of GDP, renewable energy, and non-renewable energy on the
level of CO2 emissions in 15 European Union (EU) countries over the period
1980–2012.13 The dynamic ordinary least squares estimator indicates that both
trade openness and renewable energy use contribute to improve the environment,
which is not the case of non-renewable energy. Furthermore, the results show that
the EKC hypothesis is valid for the EU countries since the elasticities of CO2 emis-
sions with respect to real income and the quadratic real income are positive and
negative, respectively. The Dumitrescu-Hurlin panel causality results report a bidi-
rectional causality between renewable energy and CO2 emissions and unidirec-
tional causality running from real GDP to CO2 emissions, from CO2 emissions to
non-renewable energy, and from trade openness to CO2 emissions. The causality is
absent between non-renewable energy and GDP, renewable energy, and trade
openness, respectively.
Furthermore, S. Shafiei and R. Salim analyze the effect of non-renewable and
renewable energy consumption on CO2 emissions for OECD countries over the
period 1980 to 2011.14 The authors investigated that question with the STIRPAT
4 THE JOURNAL OF ENERGY AND DEVELOPMENT

methodology (Stochastic Impacts by Regression on Population, Affluence and

Technology). STIRPAT was proposed by T. Dietz and E. Rosa and R. York et al.;15
it is a stochastic version of the IPAT equation (Impact 5 Population, Affluence,
Technology). Using the Augmented Mean Group (AMG) estimator, S. Shafiei and
R. Salim show that the effect of renewable energy consumption on CO2 emissions
is negative and significant, but the effect of non-renewable energy consumption on
CO2 emissions is positive and significant in the long run. The findings also show
that the effect of total population size, GDP per capita, industrialization and urbani-
zation on CO2 emissions is positive and significant. Furthermore, the empirical
results support the EKC hypothesis for the relationship between urbanization and
CO2 emissions in the long run.
Moreover, K. Saidi and A. Omri examine the impact of renewable energy con-
sumption on both economic growth and CO2 emissions in 15 major renewable
energy-consuming countries during the period 1990–2014.16 Using Fully Modified
Ordinary Least Square (FMOLS) estimations, the results for the growth function
show that renewable energy consumption has a positive and significant impact on
economic growth for the United States, China, Germany, Spain, Italy, Japan, Swe-
den, and Portugal. However, the impact is not significant for Brazil, India, United
Kingdom, France, Canada, Denmark, and Poland. Moreover, the results for the
environmental function show that renewable energy consumption has a negative
and significant impact on CO2 emissions for five countries—the United States,
Germany, United Kingdom, Sweden, and Portugal. The impact is positive in China
and not significant in Spain, Brazil, Italy, India, Japan, France, Canada, Denmark,
and Poland. Concerning the effect of economic growth on CO2 emissions, the
impact would be positive in the United States, Spain, Italy, Japan, and Sweden. On
the contrary, the impact of growth on CO2 emissions would be negative in China,
Germany, Brazil, India, Poland, and Portugal.
In addition, L. Dauda et al. study the effects of innovation, economic growth,
and energy consumption on CO2 emissions for 18 developed and developing coun-
tries over the period of 1990 to 2016.17 Using a comparative approach, they
grouped the countries into three subpanels: the Group of Six or G6 (Canada,
France, Germany, Italy, Japan, and United Kingdom); MENA (Bahrain, Egypt,
Morocco, Tunisia, Algeria, Israel, and United Arab Emirates); and BRICS (Brazil,
Russia, India, China, and South Africa). To estimate the long-run relationship
between the interest variables, the panel fully modified ordinary least square
(FMOLS) and panel dynamic ordinary least square (DOLS) methods are used. The
findings show that the effect of economic growth on CO2 emissions is negative
and significant in G6 and MENA countries. However, the effect is positive and sig-
nificant in BRICS countries only using the FMOLS estimation. Moreover, the
effect of innovation on CO2 emissions is negative and significant in G6 developed
countries but is positive and significant in BRICS counties only using the FMOLS
estimation and is positive and significant in MENA countries only using the DOLS
 THE CO2 EMISSIONS AND GROWTH NEXUS 5

estimation. Furthermore, the findings show that energy consumption significantly

increases CO2 emissions in all panel levels.
M. Rahman and E. Velayutham explore the relationship between renewable and
non-renewable energy consumption and economic growth for a panel of five South
Asian countries over the period of 1990–2014.18 The results of the FMOLS estima-
tion method for each country (separately) indicates that the effect of renewable
energy consumption on economic growth is positive and significant in Bangladesh,
India, and Nepal. The effect is not significant in Pakistan and Sri Lanka. Moreover,
the effect of non-renewable energy consumption on economic growth is positive
and significant in India, while negative and significant in Bangladesh. The
Dumitrescue-Hurlin panel causality tests reveal that there is a unidirectional causal-
ity running from economic growth to renewable energy consumption. In conclu-
sion, the authors notice that “both types of energy consumption positively and
significantly affect the economic growth of South Asia. However, the influence of
renewable energy consumption on growth is much stronger than that of non-
renewable energy.”19
The opposite conclusion is drawn in the research by S. Adams et al. concerning
a sample of 30 Sub-Saharan African countries over the period 1980–2012.20 The
explanation suggested by the authors is “that many of the countries in the region
have started investing in the development and use of renewable energy technolo-
gies and consequently they are yet to reap the full benefits.”21 This is similar to S.
Kahouli-Brahmi’s conclusion, which suggests that the adoption of these emerging
renewable technologies may be subject to initially low but progressively increasing
returns to adoption related to learning dynamics.22 In low-tech countries, where the
energy infrastructure is still under development, non-renewable mature technolo-
gies would enjoy stronger scale effects whereas the learning dynamics of emerging
technologies would be slower to appear. The consequence would be a higher con-
tribution to GDP growth of non-renewable energy than renewable energy.
A. Elia indicates that the adoption cost of such emerging technologies is deter-
mined by numerous drivers that are not only related to R&D activities but also to
“learning drivers such as market dynamics and learning by interacting across dif-
ferent stakeholders and geographical related to the learning capacity of the coun-
tries in which these technologies are used.”23 Consequently, we could expect that
the adoption of renewable energies should have a more positive impact on GDP
growth in countries with strong innovative ecosystems than in those with weaker
innovative ones.
The econometric results reported in these different studies show that it is diffi-
cult to draw general conclusions concerning the relationship between GDP, energy
consumption, renewable energies, and CO2 emissions. The structural connections
between these variables depend on the geographic area, the level of technological
competence, and the kind of technology under consideration (renewable or not).
Consequently, we have decided to study that question on a small number of
6 THE JOURNAL OF ENERGY AND DEVELOPMENT

northern European countries with a high level of technological competencies and

relatively homogeneous. Given their high level of technological competencies, we
expect that in these countries the adoption of renewable energies should be less dif-
ficult than in lower-tech countries. Therefore, the development of renewable
energy should have a positive impact on GDP and a negative impact on CO2 emis-
sions. We also expect that higher levels of technological activity should be associ-
ated with lower CO2 emissions since the direction of technological innovation is
often oriented toward cost-reducing innovations on existing processes that often
result in lower energy use.

Data and Method

This study explores the effect of renewable energy consumption and innovation
on GDP and CO2 emissions in three developed countries—Germany, Finland, and
Denmark—for the period 1990–2015. It uses annual data from the World Bank
Indicators.24 Eight variables are studied:
- Y: gross domestic product per capita (constant 2010 U.S. $),
- K: gross fixed capital formation per capita (constant 2010 U.S. $),
- CO2: CO2 emissions (metric tons per capita),
- Tec: energy use as measured by total final energy consumption (kilograms of
oil equivalent per capita),
- NRE: non-renewable energy consumption (kilograms of oil equivalent per
capita),25
- RE: renewable energy consumption (kilograms of oil equivalent per capita),
- RE100: renewable energy consumption (% of total final energy
consumption),26
- PAT: number of patents per thousand of population (residents and non-
residents).27
Following the literature (O. Awodumi and A. Adewuyi; S. Mahjabeen et al.;
and K. Saidi and A. Omri),28 we consider four models: two specifications of the
growth function and two specifications of the environmental function.
Alternative specifications of the growth function:

Yit 5 f ðKit ; TEcit ; RE100it ; PAT it Þ (1)

Yit 5 f ðKit ; NREit ; REit ; PAT it Þ (2)

Alternative specifications of the environmental function:

CO2it 5 f ðYit ; TEcit ; RE100it ; PAT it Þ (3)

 THE CO2 EMISSIONS AND GROWTH NEXUS 7

CO2it 5 f ðYit ; NREit ; REit ; PAT it Þ (4)

Equations 1 and 2 are the “growth functions.” They explore the marginal effects
of investment, total energy consumption, percentage of renewable energy in total
energy consumption, non-renewable energy, renewable energy consumption, and
number of patents per thousand of population on economic growth. C. Lee and C.
Chang maintain that energy is an essential factor of production.29 A. Ma^alej and
A. Cabagnols show that this positive impact is particularly significant in countries
poorly endowed in oil resources.30 In regard to European countries, F. Adedoyin
et al. show that non-renewable and renewable energy have both a positive and sig-
nificant influence on economic growth only in the long run.31 In addition to energy
use, innovation also is considered as a key driver of growth (J. Schumpeter, R.
Maradana et al., R. Pradhan, and F. Adedoyin et al.).32 Therefore, we expect that
Tce, RE, and PAT should have long-run positive impacts on GDP growth.
Equations 3 and 4 are the “environmental functions.” They examine the mar-
ginal effects of economic growth, total energy, percentage of renewable energy in
total energy consumption, non-renewable energy, renewable energy consumption,
and number of patents per thousand of population on CO2 emissions. Several varia-
bles affect CO2 emissions, such as economic growth (J. Ang and M. Rahman and
X. Vu),33 energy consumption (S. Sarkodie and S. Adams and A. Ma^alej and
A. Cabagnols),34 nonrenewable energy consumption (J. B. Ang; S. Sarkodie and
S. Adams; and F. Belaïd and M. Zrelli),35 renewable energy consumption (S. Sar-
kodie and S. Adams; M. Rahman and X. Vu; O. Awodumi and A. Adewuyi; and
F. Belaïd and M. Zrelli),36 and innovation (Y. Yu and Y. Du and L. Dauda et al.).37
The variables involved in these four equations are a priori strongly intercon-
nected. Our goal is to disentangle their relationships so that we could distinguish
between short- and long-run relationships and quantify their impacts. Therefore,
we have decided to perform a cointegration analysis that is described by E. Nkoro
and A. Uko as:
The analysis of long-term relationships between integrated variables and the repara-
metrization of the relationship between the variables considered in an error correction
model (ECM). The reparameterized result gives the short-term dynamics and the long-
term relationship of the variables considered.38

More precisely, our study employs ARDL estimation for co-integration.

E. Nkoro and A. Uko show that:
The adoption of the ARDL cointegration technique does not require pretests for unit
roots unlike other techniques. Consequently, ARDL cointegration is preferable when
dealing with variables that are integrated of different order, I(0), I(1) or combination
of both and, robust when there is a single long run relationship between the underlying
variables in a small sample size.39
8 THE JOURNAL OF ENERGY AND DEVELOPMENT

Moreover, according to M. Shrestha and G. Bhatta:

A dynamic error correction model (ECM) can be derived from ARDL through a sim-
ple linear transformation. Also, the ECM integrates the short-run dynamics with the
long-run equilibrium without losing long-run information and avoids problems such as
spurious relationship resulting from non-stationary time series data.40

Along the same lines, we will use the following error correction version of the
ARDL model. The ARDL specification of the four equations previously mentioned
is detailed as follows:

Model 1 : DlnYt 5 a0 1 d1 lnYt1 1d2 lnKt1 1 d3 lnTEct1 1 d4 lnRE100t1

XP X P
1 d5 lnPAT t1 1 b1j DlnYtj 1 b2j DlnKtj
XP XP j51 j51
XP
1 b3j DlnTEctj 1 j51
b4j DlnRE100 tj 1 b DlnPAT tj 1 «t
j 5 1 5j
j51

Model 2 : DlnYt 5 a0 1 d1 lnYt1 1d2 lnKt1 1 d3 lnNREt1 1 d4 lnREt1

X P XP XP
1 d5 lnPAT t1 1 b1j DlnYtj 1 b2j DlnKtj 1 b3j DlnNREtj
XP j51
XPj 5 1 j51
1 b DlnREtj 1
j 5 1 4j
b DlnPAT tj 1 «t
j 5 1 5j

Model 3 : DlnCO2t 5 a0 1 d1 lnCO2t1 1d2 lnYt1 1 d3 lnTEct1

XP XP
1 d4 lnRE100t1 1 d5 lnPAT t1 1 b1j DlnCO2tj 1 b2j DlnYtj
XP XP j51
XP
j51
1 b3j DlnTEctj 1 j51
b4j DlnRE100 tj 1 b DlnPAT tj 1 «t
j 5 1 5j
j51

Model 4 : DlnCO2t 5 a0 1 d1 lnCO2t1 1d2 lnYt1 1 d3 lnNREt1 1 d4 lnREt1

XP X P
1 d5 lnPAT t1 1 b1j DlnCO2tj 1 b2j DlnYtj
X
P XP j51
XP j51
1 b3j DlnNREtj 1 j51
b 4j DlnRE tj 1 b DlnPAT tj 1 «t
j 5 1 5j
j51

where lnY, lnCO2, lnTEc, lnNRE, lnRE, lnRE100, and lnPAT denote the loga-
rithmic transformation of Y, CO2, TEc, NRE, RE, RE100, and PAT. The intercept
is denoted by a and D is the lag operator. The error term of each equation at time t
is represented as «t . bs capture the short-term dynamics and the ds measure the
long-term relationships.
 THE CO2 EMISSIONS AND GROWTH NEXUS 9

The study employs the Phillips-Perron test to examine the presence of unit root
among the data series. According to S. Sarkodie and S. Adams:
The Phillips-Perron (Phillips and Perron, 1988) unit root test accounts for serial corre-
lation by implementing Newey-West standard errors, as such produces more accurate
unit root tests.41

ARDL bounds tests are used to examine the long-run relationship between the
interest variables in models 1 through 4. According to M. Rahman and X. Vu,
The joint F-statistic is employed to test the null hypothesis of no cointegration. If the
calculated F-statistic is below the lower limit of the critical value, the null hypothesis
is not rejected. If the calculated F-statistic is higher than the upper limit of the critical
value, the null hypothesis is rejected. If the calculated F-statistic is between the lower
limit and upper limit of the critical values, a conclusion cannot be made.42

Finally, to prevent spurious and biased inferences in the four ARDL models for
the three countries, we used several diagnostic tests to analyze the independence of
the residuals: Breusch-Godfrey LM test for autocorrelation, White test for hetero-
scedasticity, and Lagrange Multiplier Jarque-Bera Normality Test. We examined
the stability of the four ARDL models using the CUSUM Squared Plots.43

Results and Analysis

We will now report our study’s results. Table 1 shows that the level of renew-
able energy use per capita in Finland (lnRE 5 7.56) is higher than Denmark
(lnRE 5 6.13) and Germany (lnRE 5 5.31). Moreover, the share of renewable
energy consumption to total final energy consumption is higher in Finland
(lnRE100 5 3.42) than both Denmark (lnRE100 5 2.57) and Germany
(lnRE100 5 1.6). However, the level of the CO2 emissions is a little higher in Fin-
land (2.36) relative to both Germany (2.30) and Denmark (2.22). The number of
patents per thousand of population is higher in Germany (lnPAT 5 20.38) than
Finland (lnPAT 5 20.71) and Denmark (lnPAT 5 21.13). This suggests that inno-
vation is not necessarily well intended to promote renewable energy and replace
non-renewable energy. As shown by the standard deviation results, volatility was
highest in Denmark for lnNRE (1.18), lnRE (1.98) for Finland, and lnPAT (1.55)
for Germany. Conversely, Germany was the least volatile economy in terms of
lnTEc (0.03).
Table 2 shows the results of the Phillips-Perron unit root tests for Germany,
Finland, and Denmark. Our findings indicate that the variable lnPAT is stationary
at its level for Denmark. While the remaining variables are all stationary at first
level and are integrated at I(1) in all three countries. We employ the ARDL bounds
tests to examine the cointegration of the interest variables and to estimate the error
10 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 1
DESCRIPTIVE STATISTICAL ANALYSIS a
Germany Finland Denmark
Variables Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
lnY 10.55 0.10 10.59 0.17 10.90 0.10
lnK 8.97 0.07 9.11 0.20 9.22 0.18
lnCO2 2.30 0.07 2.36 0.12 2.22 0.22
lnTEc 8.31 0.03 8.74 0.07 8.16 0.08
lnNRE 8.24 0.08 8.37 0.09 7.99 1.18
lnRE 5.31 0.66 7.56 1.98 6.13 0.44
lnRE100 1.60 0.69 3.42 0.15 2.57 0.51
lnPAT 20.38 1.55 20.71 0.43 21.13 0.15

a
Std. Dev. 5 standard deviation; lnY 5 logarithmic transformation of Y: gross domestic product per capita
(constant 2010 U.S. $); lnK 5 logarithmic transformation of K: gross fixed capital formation per capita (constant
2010 U.S. $); lnCO2 5 logarithmic transformation of CO2: CO2 emissions (metric tons per capita); lnTEc 5
logarithmic transformation of TEc: energy use as measured by total final energy consumption (kilograms of oil
equivalent per capita); lnNRE 5 logarithmic transformation of NRE: non-renewable energy consumption (kilograms
of oil equivalent per capita); lnRE 5 logarithmic transformation of RE: renewable energy consumption (kilograms of
oil equivalent per capita); lnRE100 5 logarithmic transformation of RE100: renewable energy consumption (% of
total final energy consumption); and lnPAT 5 logarithmic transformation of PAT: number of patents per thousand of
population (residents and non-residents).

Table 2
UNIT ROOT TESTINGa
Germany Finland Denmark
Test statistic: Z(t) Test statistic: Z(t) Test statistic: Z(t)
Variables
At level At 1st difference At level At 1st difference At level At 1st difference
lnY 25.58 25.31** 20.85 23.44** 22.02 23.61**
lnK 21.38 24.48** 21.32 23.11* 21.38 22.94*
lnCO2 20.75 28.82** 21.17 25.86** 1.06 27.48**
lnTEc 22.08 28.78** 21.64 26.11** 20.17 27.13**
lnNRE 20.24 29.36** 20.04 25.20** 2.17 26.10**
lnRE 0.39 24.33** 20.94 28.33** 0.17 25.56**
lnRE100 0.45 25.45** 0.57 26.94** 1.80 25.56**
lnPAT 21.76 22.95* 22.16 27.35** 25.76**

a
**denotes significance at the 1-percent level; *denotes significance at the 5-percent level; lnY 5 logarithmic
transformation of Y: gross domestic product per capita (constant 2010 U.S. $); lnK 5 logarithmic transformation of
K: gross fixed capital formation per capita (constant 2010 U.S. $); lnCO2 5 logarithmic transformation of CO2: CO2
emissions (metric tons per capita); lnTEc 5 logarithmic transformation of TEc: energy use as measured by total final
energy consumption (kilograms of oil equivalent per capita); lnNRE 5 logarithmic transformation of NRE: non-
renewable energy consumption (kilograms of oil equivalent per capita); lnRE 5 logarithmic transformation of RE:
renewable energy consumption (kilograms of oil equivalent per capita); lnRE100 5 logarithmic transformation of
RE100: renewable energy consumption (% of total final energy consumption); and lnPAT 5 logarithmic
transformation of PAT: number of patents per thousand of population (residents and non-residents).
 THE CO2 EMISSIONS AND GROWTH NEXUS 11

Table 3
BOUNDS TESTING TO COINTEGRATION a
Germany Finland Denmark
Estimated Optimal Optimal Optimal
equation lag length F-statistics lag length F-statistics lag length F-statistics
M1 (1,0,0,0,0) 28.89** (1,1,0,0,0) 5.98** (1,1,0,0,0) 4.28*
M2 (1,0,0,0,0) 27.52** (1,1,0,0,0) 6.05** (1,1,0,0,0) 4.26*
M3 (1,0,0,0,1) 24.23** (1,0,1,0,0) 11.57** (1,0,1,0,0) 14.90**
M4 (1,0,0,0,0) 55.01** (1,0,0,2,1) 88.01** (1,0,0,0,0) 127.37**
Significance level Critical Values (0.1-0.01) Upper bounds I(1)
Lower bounds I(0)
10% 2.45 3.52
5% 2.86 4.01
1% 5.06 5.06

a
**denotes significance at the 1-percent level; *denotes significance at the 5-percent level.

correction model in all three countries. The ARDL bounds tests are applicable irre-
spective of whether the regressor variables are I(0), I(1), or mutually
cointegrated.44
The ARDL bounds tests is used to examine the long-run relationship between
the interest variables in the four models and for the three countries. Table 3 indi-
cates the results of the ARDL bound test and shows that our calculated F-statistics
are larger than the upper critical bound at 5 percent, when we treat the predicted
variables. Indeed, the variables of interest are cointegrated for a long-run relation-
ship in four equations and in three countries.
For the three countries, table 4 shows that the four ARDL models have no serial
correlation (Breusch Godfrey LM test), have a constant variance free from hetero-
skedasticity problem (White test), and the errors have a normal distribution
(Lagrange Multiplier Jarque-Bera Normality Test). Furthermore, figures 1 through
3 for Germany, Finland, and Denmark, respectively, reveal that the CUSUM
Squared Plots from the ARDL regressions show that the plots are within the
95-percent confidence band graphs and then confirming the stability of the ARDL
models in the four models and for the three countries.
In the output section ADJ in table 5, we report the negative speed-of-adjustment
coefficients: they measure how strongly the dependent variable reacts to a devia-
tion from the equilibrium relationship in one period or, in other words, how
quickly such an equilibrium distortion is corrected.
In the long run, the effect of the variable lnK on lnY is positive and significant
in Germany and Denmark. It is particularly strong in Germany, which illustrates
the importance of investment in that country. On the contrary, in Finland, the effect
of lnk on lnY is only seen in the short run.
12 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 4
DIAGNOSTIC TEST AND MODEL VALIDATION
Breusch-Godfrey LM Test for Autocorrelation
Germany Finland Denmark
Lags Prob. Lags Prob. Lags Prob.
(1) chi2 df chi2 (1) chi2 df chi2 (1) chi2 df chi2
M1
1 2.16 1 0.14 1 0.18 1 0.66 1 0.26 1 0.60
M2
1 2.60 1 0.106 1 0.13 1 0.71 1 0.29 1 0.59
M3
1 0.25 1 0.61 1 1.67 1 0.19 1 2.67 1 0.102
M4
1 0.44 1 0.50 1 0.87 1 0.34 1 0.52 1 0.47
White’s Test for Heteroskedasticity
Germany Finland Denmark
chi2() Prob.chi2 chi2() Prob.chi2 chi2() Prob.chi2
M1 24.36 0.22 25 0.40 25 0.40
M2 23.88 0.24 25 0.40 25 0.40
M3 24.59 0.21 25 0.40 25 0.40
M4 24.42 0.22 24 0.40 20.58 0.42
Lagrange Multiplier Jarque-Bera Normality Test
Germany Finland Denmark
LM Prob. LM DF Prob. LM DF Prob.
Test DF Chi2 chi2 Test Chi2 chi2 Test Chi2 chi2
M1 0.12 2 0.93 1.74 2 0.41 4.38 2 0.11
M2 0.18 2 0.91 1.80 2 0.40 4.41 2 0.11
M3 0.34 2 0.84 4.13 2 0.12 0.83 2 0.66
M4 0.39 2 0.82 0.85 2 0.65 0.13 2 0.93

In the long run, energy use (lnTEc) increases economic growth only in Ger-
many and Finland; it is not significant in Denmark. That result indicates that, at
least in Germany and Finland, energy is an important driver of growth.
The results concerning the effect of renewable and non-renewable energy help
us understand what kind of energy source is the most beneficial for growth. We
observe that the effect of the consumption of non-renewable energy per capita
(lnNRE) is not statistically significant at the 5-percent level in the three countries.
It confirms the results of C. Chen et al.45 who found that non-renewable energy
 THE CO2 EMISSIONS AND GROWTH NEXUS 13

Figure 1
GERMANY: CUSUM SQUARED PLOT FOR ARDL REGRESSION

M1 – Germany M2 – Germany

1 1

0 0

1996 t 2015 1996 t 2015

M3 – Germany M4 – Germany

1 1

0 0

1996 tt 2015 1996 t 2015

has no significant effect on economic growth in developed countries. However, it

is in contradiction with the findings of F. Adedoyin et al.46 who report that the
effect of non-renewable energy on economic growth in European Union countries
is positive and significant in the long term but is negative and insignificant in the
short run.
Concerning the consumption of renewable energy (lnRE) and the share of
renewable energy in the total energy consumption (lnRE100), we report a positive
and significant effect on lnY in the three countries. It reveals that the impact of
energy on growth differs depending on its source: renewable or not. C. Chen et al.47
find that the relationship between renewable energy consumption and economic
14 THE JOURNAL OF ENERGY AND DEVELOPMENT

Figure 2
FINLAND: CUSUM SQUARED PLOT FOR ARDL REGRESSION

M1 – Finland M2 – Finland

1 1

0 0

1996 t 2015 1996 t 2015

M3 – Finland M4 – Finland

1 1

0 0

1996 t 2015 1996 t 2015

growth depends on the amount of renewable energy used. The greater the amount
of renewable energy, used the stronger its effect on economic growth. That result
contradicts E. Dogan et al.48 who concluded that the share of renewable energy
consumption to total energy consumption drives economic growth only for very
low-income OECD countries. The authors maintain that the effect of renewable
energy on economic growth decreases with the country’s income level. Further-
more, E. Dogan and F. Seker49 found an absence of causality between renewable
energy and GDP in European Union countries.
The effect of innovation on economic growth is positive and significant only in
Denmark whereas we expected a positive impact in all of the countries studied.
However, this is a result also obtained by R. Maradana et al.50 who examined the
 THE CO2 EMISSIONS AND GROWTH NEXUS 15

Figure 3
DENMARK: CUSUM SQUARED PLOT FOR ARDL REGRESSION

M1 – Denmark M2 – Denmark

1 1

0 0

1996 t 2015 1996 t 2015

M3 – Denmark M4 – Denmark

1 1

0 0

1996 t 2015 1996 t 2015

long-run relationship between innovation and economic growth in the European

Economic Area countries. They show that patents do not Granger-cause per-capita
economic growth in Germany and Denmark. However, for Finland and the panel
of European Economic Area countries, they report a bidirectional causality
between innovation and per-capita economic growth.51 This unexpected result may
be explained by the weakness of the patent indicator.52 For example, F. Adedoyin
et al.53 show that innovation in European Union countries, defined by R&D expen-
ditures, drives economic growth in the presence of renewable and non-renewable
energy consumption only in the long run.
In table 6, we examine the effect of total energy consumption, non-renewable
energy, renewable energy consumption, and innovation on CO2 emissions
16 THE JOURNAL OF ENERGY AND DEVELOPMENT

Table 5
ESTIMATE THE ERROR CORRECTION MODEL: GROWTH FUNCTION M1-M2 a
Germany Finland Denmark
Variables Eq. 1 Eq. 2 Eq. 1 Eq. 2 Eq. 1 Eq. 2
D.lnY Coeff. Coeff. Coeff. Coeff. Coeff. Coeff.
ADJ
lnY L1. 20.71*** 20.72*** 20.18** 20.18** 20.44*** 20.44***
Long Run
lnK 0.49*** 0.51*** 0.04 0.04 0.22*** 0.22***
lnTEc 0.41** 2.08** 0.19
lnNRE 0.22 0.64* 0.04
lnRE 0.12*** 1.36** 0.12***
LnRE100 0.10*** 1.06* 0.12***
lnPAT 0.03 0.02 0.30 0.29 0.19** 0.18**
Short Run
lnK D1. 0.27*** 0.26*** 0.21** 0.21***
cons 1.79 2.46 22.08** 20.97 3.20*** 3.55***

a
*** denotes significance at the 1-percent level; ** denotes significance at the 5-percent level; * denotes
significance at the 10-percent level; Coeff. 5 coefficients; lnY 5 logarithmic transformation of Y: gross domestic
product per capita (constant 2010 U.S. $); lnK 5 logarithmic transformation of K: gross fixed capital formation per
capita (constant 2010 U.S. $); lnCO2 5 logarithmic transformation of CO2: CO2 emissions (metric tons per capita);
lnTEc 5 logarithmic transformation of TEc: energy use as measured by total final energy consumption (kilograms of
oil equivalent per capita); lnNRE 5 logarithmic transformation of NRE: non-renewable energy consumption
(kilograms of oil equivalent per capita); lnRE 5 logarithmic transformation of RE: renewable energy consumption
(kilograms of oil equivalent per capita); lnRE100 5 logarithmic transformation of RE100: renewable energy
consumption (% of total final energy consumption); and lnPAT 5 logarithmic transformation of PAT: number of
patents per thousand of population (residents and non-residents).

(Equation 3 and Equation 4). We observe a negative speed-of-adjustment coeffi-

cient in the output section ADJ. The long-run analysis shows that economic growth
does not significantly affect CO2 emissions in Finland and Denmark. However, in
Germany there is a positive and significant impact of GDP on CO2 emissions. J.
Ang observes a similar result for France: a long run positive inverted-U shape
impact of GDP growth on CO2 emissions (although in the short run the causality
goes from CO2 to GDP).54 In a different context, A. Ma^alej and A. Cabagnols55
find that the effect of economic growth on CO2 emissions is positive and signifi-
cant in Middle East and North African countries with important oil endowments.
On the contrary, L. Dauda et al.56 find that economic growth decreases the level of
CO2 emissions in developed G6 countries.
The effect of energy use (lnTEc) on CO2 emissions is positive and significant in
Germany and Denmark, which is in line with the results of several other studies
(L. Dauda et al., J. Ang, and A. Ma^alej and A. Cabagnols).57 The effect is not sig-
nificant in Finland, which could be due to the important share of renewable energy
in that country.
 THE CO2 EMISSIONS AND GROWTH NEXUS 17

Table 6
ESTIMATE THE ERROR CORRECTION MODEL: ENVIRONMENTAL FUNCTION M3-M4 a
Germany Finland Denmark
Variables Eq. 3 Eq. 4 Eq. 3 Eq. 4 Eq. 3 Eq. 4
D.lnCO2 Coeff. Coeff. Coeff. Coeff. Coeff. Coeff.
ADJ
lnCO2 L1. 20.87*** 20.88*** 20.53*** 21.02*** 21.38*** 21.09***
Long Run
lnY 0.51*** 0.71*** 0.06 0.08 20.02 0.08
lnTEc 1.12*** 0.93 1.54***
lnNRE 0.87*** 1.22*** 1.16***
lnRE 20.06*** 20.55** 20.05
LnRE100 20.09*** 21.10*** 20.19***
lnPAT 20.19*** 20.24*** 20.19** 20.23*** 0.03 20.14**
Short Run
lnTEc D1. 1.01*** 20.59*
lnRE D1. 0.62***
lnRE LD. 0.20
lnPAT D1. 20.11*
cons 210.80*** 210.86*** 21.50 24.81*** 213.22*** 28.70***

a
*** denotes significance at the 1-percent level; ** denotes significance at the 5-percent level; * denotes
significance at the 10-percent level; Coeff. 5 coefficients; lnY 5 logarithmic transformation of Y: gross domestic
product per capita (constant 2010 U.S. $); lnK 5 logarithmic transformation of K: gross fixed capital formation per
capita (constant 2010 U.S. $); lnCO2 5 logarithmic transformation of CO2: CO2 emissions (metric tons per capita);
lnTEc 5 logarithmic transformation of TEc: energy use as measured by total final energy consumption (kilograms of
oil equivalent per capita); lnNRE 5 logarithmic transformation of NRE: non-renewable energy consumption
(kilograms of oil equivalent per capita); lnRE 5 logarithmic transformation of RE: renewable energy consumption
(kilograms of oil equivalent per capita); lnRE100 5 logarithmic transformation of RE100: renewable energy
consumption (% of total final energy consumption); and lnPAT 5 logarithmic transformation of PAT: number of
patents per thousand of population (residents and non-residents).

The impact of non-renewable energy on CO2 emissions is positive and signifi-

cant in the three countries. This result is consistent with the findings of E. Dogan
and F. Seker58 who noticed the importance of the effect of non-renewable energy
on environmental degradation in EU countries. Conversely, renewable energy use
significantly decreases the level of CO2 emissions in Germany and Finland. In
Denmark, the effect of renewable energy is negative on CO2 emissions and signifi-
cant only in Equation 2 (not in Equation 3). Our results are consistent with the
findings of K. Saidi and A. Omri.59 In a sample of 15 countries, these authors
observed that the increase of renewable energy reduces the level of CO2
emissions in Germany (and most of the other countries in their sample), but not in
Denmark.
In line with our expectations, the effect of innovation on CO2 emissions is sig-
nificant and negative in Germany and Finland. In Denmark, the effect of
18 THE JOURNAL OF ENERGY AND DEVELOPMENT

innovation on CO2 emissions is negative and significant only in Equation 3. Our

results agree with those of L. Dauda et al. that found innovation reduces CO2 emis-
sions in developed G6 countries.60

Conclusion

This study examined the relationships between GDP, renewable and non-
renewable energy consumption, innovation, and CO2 emissions over the period
1995–2015 in three north European countries: Germany, Finland, and Denmark.
ARDL bounds tests are employed to analyze the long-run relationships between
these variables. We observe that energy use is an important factor of economic
growth in the long run in Germany and Finland. However, it has no significant
impact in Denmark. From a political point of view, it is interesting to notice that a
higher proportion of renewable sources of energy in the total energy consumption
has no harmful consequences on growth. On the contrary, we observe a significant
and positive impact in Germany and Finland, and no significant influence in Den-
mark. The gradual increase of renewable energy in these countries should not ham-
per growth; on the contrary, it may have a positive impact. Concerning the
reduction of CO2 emissions, the impact of renewable energy may not be so obvious.
Indeed, we observe that the determinants of CO2 emissions are not the same in these
three countries.
- In Germany, GDP growth has a strong and positive impact on CO2 emissions
whereas the coefficient associated with renewable energy is a small size
(although negative and significant). In that country, the impact of the techno-
logical activity on CO2 emissions reduction seems stronger than that of
renewable energy. Consequently, in the case of Germany, if the diffusion of
renewable energy is positive for growth it may have a final mitigated impact
on CO2 emissions: its direct negative impact on CO2 emissions may be offset
by its indirect positive impact on GDP. In that country, the reduction of CO2
emissions may be more sensitive to the promotion of innovations in technolo-
gies and industries where energy and non-energy CO2 emissions are high.
- In Finland, the impact of renewable energy on GDP is not strongly significant
(10 percent) but simultaneously, GDP growth has no side effect on CO2 emis-
sions. Consequently, in that country, the promotion of renewable energies
should have a very strong impact on the reduction of CO2 emissions. Techno-
logical activities also have a negative impact on CO2 emissions. It points out
the relevance of complimentary targeted technological policies devoted to the
reduction of CO2 emissions related to energy or non-energy use.
- In Denmark, despite no significant impact of the level of energy use on GDP,
we notice a positive impact of the proportion of renewable energies in the energy
mix. The reason may be the substitution of oil imports by the development of
 THE CO2 EMISSIONS AND GROWTH NEXUS 19

national energy production industries. From a statistical point of view, the impact
of these renewable energies on the reduction of CO2 emissions is not as obvious
as in the case of Germany and Finland (both equations 3 and 4 do not give two
significant estimates). Concerning the impact of the technological activities, we
face the same statistical instability of the results.
The empirical evidence reported in this paper indicate that in the case of Ger-
many, Finland, and Denmark the promotion of renewable energies is not harmful
for growth; rather it would promote GDP in Germany and Denmark and have no
impact in Finland. Concerning the consequences of renewable energies on CO2
emissions, we observe that their direct impact is always negative and significant
(in equation 3 at least) but that technologies also play a significant role. Although
relatively similar, these three wealthy European countries also exhibit some impor-
tant differences concerning the sensitivity of their CO2 emissions to growth, to
renewable energy, and to technological activities. It calls for the promotion of
country-specific environmental policies that take into account: (1) the indirect
influence of renewable energies on CO2 emissions via their impact on GDP
growth; (2) the different sources of CO2 emissions that are not only related to
energy use; and (3) the specificities of national innovation systems in which renew-
able technologies are developed and/or implemented.
This type of structural analysis specific to each country goes against the current
econometric approaches that implicitly suppose that a single model can handle the
diversity of nations. As indicated in the literature review, many empirical works
report contradictory results for the same country or group of countries. Small differ-
ences in the sample, in the methodology, or in the model suffice to produce opposite
conclusions. Therefore, the conclusions of the literature show very low robustness.
Thus, we should also look at how we can enhance the methodology used in this
study. It might be more relevant to build models that are less ambitious in terms of
generality, but more relevant to the countries under consideration. Following, for
example, A. Elia,61 we could start each national analysis with an in-depth mono-
graph of its energy and innovation system. That qualitative work carried out during
that initial stage would provide the relevant material for a second stage of modeling,
quantification, and estimation. Depending on the institutional arrangements specific
to each country, different models could emerge that truly enrich our understanding
of the topic and provide effective conclusions relevant for policy making.

NOTES
1
E. A. Wrigley, “Energy and the English Industrial Revolution,” Philosophical Transactions of
the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 371, no. 1986 (2013),
p. 9.

P. R. Ehrlich and J. P. Holdren, “Impact of Population Growth,” Science, vol. 171 (1971), p.
2

1213.
20 THE JOURNAL OF ENERGY AND DEVELOPMENT
3
C.-C. Lee and C.-P. Chang, “Energy Consumption and Economic Growth in Asian Econo-
mies: A More Comprehensive Analysis Using Panel Data,” Resource and Energy Economics, vol.
30, no. 1 (2008), pp. 50–65.
4
S. Yao, S. Zhang, and X. Zhang, “Renewable Energy, Carbon Emission and Economic
Growth: A Revised Environmental Kuznets Curve Perspective,” Journal of Cleaner Production,
vol. 235 (2019), pp. 1338–352.

C. Zou, Z. Qun, Z. Guosheng, and X. Bo, “Energy Revolution: From a Fossil Energy Era to a
5

New Energy Era,” Natural Gas Industry B, vol. 3, no. 1 (2016), pp. 1–11.

S. Kahouli-Brahmi, “Testing for the Presence of Some Features of Increasing Returns to

6

Adoption Factors in Energy System Dynamics: An Analysis via the Learning Curve Approach,”
Ecological Economics, vol. 68 (2009), pp. 1195–212.

A. Sinha, M. Shahbaz, and T. Sengupta, “Renewable Energy Policies and Contradictions in

7

Causality: A Case of Next 11 Countries,” Journal of Cleaner Production, vol. 197, no. 1 (2018),
pp. 73–84 and J. Blazquez, R. Fuentes-Bracamontes, C. A. Bollino, aand N. Nezamuddin, “The
Renewable Energy Policy Paradox,” Renewable and Sustainable Energy Reviews, vol. 82 (2018),
pp. 1–5.

M. M. Rahman and X.-B. (Benjamin) Vu, “The Nexus between Renewable Energy, Economic
8

Growth, Trade, Urbanisation and Environmental Quality: A Comparative Study for Australia and
Canada,” Renewable Energy, vol. 155 (2020), pp. 617–27 and L. Dauda, X. Long, C. N. Mensah,
and M. Salman, “The Effects of Economic Growth and Innovation on CO2 Emissions in Different
Regions,” Environmental Science and Pollution Research, vol. 26 (2019), pp. 15028–5038.

D. Cui, Z. Deng, and Z. Lui, “China’s Non-Fossil Fuel CO2 Emissions from Industrial Proc-
9

esses,” Applied Energy, vol. 254, issue C (2019).

10
M. M. Rahman and X.-B. (Benjamin) Vu, op. cit.

J. B. Ang, “CO2 Emissions, Energy Consumption, and Output in France,” Energy Policy,
11

vol. 35 (2007), pp. 4772–778.

12
The EKC is named for Simon Kuznets who hypothesized that income inequality first rises
and then falls as economic development proceeds (A. Ma^alej and A. Cabagnols, “Energy Con-
sumption, CO2 Emissions and Economic Growth in MENA Countries,” Environmental Econom-
ics, vol. 11, no. 1 (2020), pp. 133–50).

E. Dogan and F. Seker, “Determinants of CO2 Emissions in the European Union: The Role
13

of Renewable and Non-Renewable Energy,” Renewable Energy, vol. 94 (2016), pp. 429–39.

S. Shafiei and R. A. Salim, “Non-Renewable and Renewable Energy Consumption and CO2
14

Emissions in OECD Countries: A Comparative Analysis,” Energy Policy, vol. 66 (2014), pp.
547–56.

T. Dietz and E. A. Rosa, “Effects of Population and Affluence on CO2 Emissions,” Proceed-
15

ings of the National Academy of Sciences, USA (Ecology), vol. 94 (1997), pp. 175–79 and R.
York, E. Rosa, and T. Dietz, “STIRPAT, IPAT and ImPACT: Analytic Tools for Unpacking the
Driving Forces of Environmental Impacts,” Ecological Economics, vol. 46, no. 3 (2003), pp.
351–65.
 THE CO2 EMISSIONS AND GROWTH NEXUS 21
16
K. Saidi and A. Omri, “The Impact of Renewable Energy on Carbon Emissions and Eco-
nomic Growth in 15 Major Renewable Energy-consuming Countries,” Environmental Research,
vol. 186 (2020).
17
L. Dauda et al., op. cit.
18
M. M. Rahman and E. Velayutham, “Renewable and Non-Renewable Energy Consumption-
Economic Growth Nexus: New Evidence from South Asia,” Renewable Energy, vol. 147 (2020),
pp. 399–408.
19
M. M. Rahman and E. Velayutham, op. cit., p. 406.

S. Adams, E. K. M. Klobodu, and A. Apio, “Renewable and Non-Renewable Energy,

20

Regime Type and Economic Growth,” Renewable Energy, vol. 125 (2018), pp. 755–67.
21
S. Adams et al., op. cit., p. 762.
22
S. Kahouli-Brahmi, op. cit., p. 1207.
23
A. Elia, “Impacts of Innovation on Renewable Energy Technology Cost Reductions,” Renew-
able and Sustainable Energy Reviews, vol. 31 (2021), p. 1.
24
World Bank Development Indicators are available at https://databank.worldbank.org/source/
world-development-indicators.
25
M. M. Rahman and E. Velayutham, op. cit.

E. Dogan, B. Altinoz, M. Madaleno, and D. Taskin, “The Impact of Renewable Energy Con-
26

sumption to Economic Growth: A Replication and Extension of Inglesi-Lotz (2016),” Energy Eco-
nomics, vol. 90, article no. 104867, (2020).
27
R. P. Maradana, R. P. Pradhan, S. Dash, D. B. Zaki, K. Gaurav, M. Jayakumar, and A. K.
Sarangi, “Innovation and Economic Growth in European Economic Area Countries: The Granger
Causality Approach,” IIMB Management Review, vol. 31 (2019), pp. 268–82 and R. P. Pradhan,
M. B. Arvin, M. Nair, and S. E. Bennett, “The Dynamics Among Entrepreneurship, Innovation,
and Economic Growth in the Eurozone Countries,” Journal of Policy Modeling, vol. 42, no. 5
(2019), pp. 1106–122.
28
O. B. Awodumi and A. O. Adewuyi, “The Role of Non-Renewable Energy Consumption in
Economic Growth and Carbon Emission: Evidence from Oil Producing Economies in Africa,”
Energy Strategy Reviews, vol. 27 (2020), pp. 1–13; S. Z. A. Shah Mahjabeen, S. Chughtai, and B.
Simonetti, “Renewable Energy, Institutional Stability, Environment and Economic Growth Nexus
of D-8 Countries,” Energy Strategy Reviews, vol. 29 (2020); and K. Saidi and A. Omri, op. cit.
29
C.-C. Lee and C.-P. Chang, op. cit., p. 51.
30
A. Ma^alej and A. Cabagnols, op. cit., p. 140.
31
F. F. Adedoyin, F. V. Bekun, and A. A. Alola, “Growth Impact of Transition from Non-
renewable to Renewable Energy in the EU: The Role of Research and Development,” Renewable
Energy, vol. 159, issue C (2020), pp. 1139–145.
32
J. A. Schumpeter, Capitalism, Socialism and Democracy (New York: Harper and Brothers,
1932); R. P. Maradana et al., op. cit.; R. P. Pradhan et al., op. cit.; and F. Adedoyin et al., op. cit.
22 THE JOURNAL OF ENERGY AND DEVELOPMENT
33
J. B. Ang, op. cit., and M. M. Rahman and X.-B. (Benjamin) Vu, op. cit.

S. A. Sarkodie and S. Adams, “Renewable Energy, Nuclear Energy, and Environmental Pol-
34

lution: Accounting for Political Institutional Quality in South Africa,” Science of the Total Envi-
ronment, vol. 643 (2018), pp. 1590–1601, and A. Ma^alej and A. Cabagnols, op. cit.
35
J. B. Ang, op. cit.; S. A. Sarkodie and S. Adams, op. cit.; and F. Belaïd and M. H. Zrelli,
“Renewable and Non-Renewable Electricity Consumption, Environmental Degradation and Eco-
nomic Development: Evidence from Mediterranean Countries,” Energy Policy, vol. 133 (2019).
36
S. A. Sarkodie and S. Adams, op. cit.; M. M. Rahman and X.-B. (Benjamin) Vu, op. cit.; O.
B. Awodumi and A. O. Adewuyi, op. cit.; and F. Belaïd and M. H. Zrelli, op. cit.

Y. Yu and Y. Du, “Impact of Technological Innovation on CO2 Emissions and Emissions

37

Trend Prediction on ‘New Normal’ Economy in China,” Atmospheric Pollution Research, vol. 10,
no. 1 (2019), pp. 152–61 and L. Dauda et al., op. cit.

E. Nkorol and A. K. Uko, “Autoregressive Distributed Lag (ARDL) Cointegration Tech-

38

nique: Application and Interpretation,” Journal of Statistical and Econometric Methods, vol. 5, no.
4 (2016), p. 66.
39
Ibid., p. 63.

M. B. Shrestha and G. R. Bhatta, “Selecting Appropriate Methodological Framework for

40

Time Series Data Analysis,” The Journal of Finance and Data Science, vol. 4 (2018), p. 79.
41
S. A. Sarkodie and S. Adams, op. cit., p. 1594.
42
M. M. Rahman and X.-B. (Benjamin) Vu, op. cit., p. 620, and P. C. B. Phillips and P. Per-
ron, “Testing for a Unit Root in Time Series Regression,” Biometrika, vol. 75 (1988), pp. 335–46.

R. L. Brown, J. Durbin, and J. M. Evans, “Techniques for Testing the Constancy of Regres-
43

sion Relationships over Time,” Journal of the Royal Statistical Society, vol. 37, no. 2 (1975), pp.
149–92.

M. H. Pesaran and Y. Shin, “An Autoregressive Distributed Lag Modeling Approach to

44

Cointegration Analysis,” in S. Strom. A. Holly, and P. Diamond, eds., Centennial Volume of Ran-
gar Frisch (Cambridge: Cambridge University Press, 1999) and M. H. Pesaran, R. J. Smith, and
Y. Shin, “Bounds Testing Approaches to the Analysis of Level Relationships,” Journal of Applied
Econometrics, vol. 16, (2001), pp. 289–326.

C. Chen, M. Pinar, and T. Stengos, “Renewable Energy Consumption and Economic Growth
45

Nexus: Evidence from a Threshold Model,” Energy Policy, vol. 139, issue C (2020), p. 9.
46
F. F. Adedoyin et al., op. cit., p. 15.
47
C. Chen et al., op. cit., p. 1.
48
E. Dogan et al., op. cit., p. 6.
49
E. Dogan and F. Seker, op. cit., p. 439.
50
R. P. Maradana et al., op. cit.
51
Ibid., p. 277.
 THE CO2 EMISSIONS AND GROWTH NEXUS 23
52
Organization for Economic Co-operation and Development (OECD), OECD Patent Statistics
Manual (Paris: OECD, 2009).
53
F. F. Adedoyin et al., op. cit., p. 15.
54
J. B. Ang, op. cit., p. 4776.
55
A. Ma^alej and A. Cabagnols, op. cit., p. 139.
56
L. Dauda et al., op. cit.
57
L. Dauda et al., op. cit.; J. B. Ang, op. cit.; and A. Ma^alej and A. Cabagnols, op. cit.
58
E. Dogan and F. Seker, op. cit., p. 435.
59
K. Saidi and A. Omri, op. cit., p. 8.
60
L. Dauda et al., op. cit.
61
A. Elia, op. cit.