Environmental Science and Pollution Research

https://doi.org/10.1007/s11356-020-09110-7

RESEARCH ARTICLE

Revisiting the environmental Kuznets curve hypothesis using

innovation: new evidence from the top 10 innovative economies
Sakir Gormus 1 & Mucahit Aydin 2

Received: 20 March 2020 / Accepted: 28 April 2020

# Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
The aim of this study is to examine the relationship between ecological footprint, economic growth, renewable energy consump-
tion, and innovation within the framework of the environmental Kuznets curve (EKC) hypothesis for the top 10 innovative
economies, namely, Denmark, Finland, Germany, Israel, Korea, The Netherlands, Sweden, Switzerland, the UK, and the USA, in
the period of 1990–2015. For this purpose, the long-term relationship between variables was examined with a panel cointegration
test. The results show that the variables in the EKC model move together in the long run. According to the long-run estimation
results, the EKC hypothesis is valid for Israel, but not for the other countries. The study also makes the following observations: (i)
For Korea, the USA, Finland, and the whole panel, innovation appears to reduce environmental pollution. (ii) Renewable energy
consumption reduces environmental pollution for Denmark, Germany, The Netherlands, and the USA. (iii) Globalization has an
impact on the reduction of environmental pollution for Germany and Switzerland. As a result, developing policies on the use of
more innovative technologies in the countries studied will have a positive impact on environmental pollution.

Keywords EKC hypothesis . Innovation . Renewable energy consumption . Globalization

Introduction international conferences have been organized, including

the 1997 Kyoto climate change conference and the 2015
A globalized world results in an increase in competition Paris climate conference. In these sessions, determinations
between economies. In this competitive environment, econ- are made and decisions are taken regarding climate change
omies try to achieve sustainable growth by concentrating on and environmental pollution. For example, it was stated in
comparatively advantageous areas. However, the competi- the 2009 Kyoto Protocol that the use of renewable energy to
tive environment that comes with globalization raises some reduce carbon emissions should be supported. Another de-
problems. Considering global warming, depletion of natural cision related to environmental pollution involved the pro-
resources, and increase in carbon emissions, the biggest tection of natural water resources and forests. In this frame-
problem is environmental pollution. This is a global prob- work, besides ensuring sustainable economic growth for
lem in which developed and developing economies are developed and developing economies, the sustainability of
jointly responsible. To overcome this global issue, many the environment is important.
Based on these discussions, researchers across different dis-
ciplines have started to look for solutions to reduce environ-
Responsible editor: Nicholas Apergis
mental problems arising from human activities. Developing
effective solutions for environmental degradation depends on
* Mucahit Aydin
[email protected]
the correct definition of pollution. Carbon emissions are used in
the literature to represent environmental pollution. However,
Sakir Gormus considering that environmental pollution depends not only on
[email protected]
greenhouse gas emissions but also on other factors, carbon
1
Faculty of Political Sciences, Department of Public Finance, Sakarya emissions alone are not an adequate indicator for environmental
University, Esentepe Campus, Serdivan, Sakarya, Turkey pollution. The ecological footprint variable calculated by Rees
2
Faculty of Arts and Sciences, Department of Statistics, Sakarya (1992) and Wackernagel (1994) is a more comprehensive indi-
University, Esentepe Campus, Serdivan, Sakarya, Turkey cator, together with the representation of environmental
 Environ Sci Pollut Res

pollution variables such as destroyed fertile lands, polluted nat- Considering that the increase in income in the EKC hy-
ural water resources, and forest destruction. The ecological pothesis decreases the environmental pollution depends on
footprint consists of six subcomponents: fishing grounds, car- exceeding the threshold value, the focus point in the research
bon footprint, forest, built-up land, grazing land, and cropland. of this hypothesis is to exceed this target, which is called the
These components have been created to provide comprehensive threshold value. At this point, the tools that will support the
information about the environment. Situations where the eco- environmental pollution-reducing effect of increasing income
logical footprint, which can be defined as the measurement of should be determined well. When the threshold value is
the biological area required for human needs, is higher than the exceeded in the EKC hypothesis, the most important role in
current biocapacity express the ecological openness. The eco- reducing the environmental pollution associated with income
logical openness, on the other hand, means that continuous increase is technological development (Andreoni and
human needs cannot be met—in other words, environmental Levinson 2001). This means that even if higher economic
destruction cannot be sustained. In this respect, ecological foot- activity requires more energy use, innovative energy use pol-
print is an effective variable that can be used in determining icies can reduce environmental pollution as a result of less
environmental policies. energy need (Fernández et al. 2018). Developments such as
Another approach to be considered for solving environ- technological innovations, increased competition, and com-
mental problems is determining the causes of pollution. mercial collaborations have paved the way for innovative
Studies in the current literature focus on the fact that economic ideas. Innovation is defined as the application of a new or
growth causes environmental pollution. To widen the field of improved product (goods or services) in new business prac-
investigation, the impact on the environment of macroeco- tices or workplace organization (OECD, E. 2005). The rela-
nomic variables that cause economic growth should be exam- tionship between innovation and economic growth has been
ined. Energy consumption, which is one of the most important studied by many economists. According to Schumpeter
inputs of industrial production, has an effect on economic (1934), economic growth is driven by innovation. Solow
growth (Aydin 2019a, b, c). The energy consumption- (1957) stated that innovation and technological development
economic growth relationship that began with the industrial are determinants of economic growth. According to Romer
revolution has become increasingly studied as markets have (1986) and Nelson (1996), there is a positive relationship be-
become globalized. This relationship is closely followed by tween innovation and economic growth. From an environ-
energy producers and consumers, especially economies. The mental perspective, economic growth is closely related to en-
reason for this is that sustainable growth is directly related to vironmental pollution. The effect of innovation on the rela-
how energy resources are used effectively. Effectively used tionship between environment and economic growth is ex-
energy sources are crucial in terms of sustainability of re- plained by endogenous growth models (Fernández et al.
sources and economic activities. However, energy sources that 2018). According to these models, the rate of replacement of
are not used effectively cannot achieve sustainable economic the sources causing environmental pollution with environ-
growth. The relationship between economic growth and ener- mentally friendly sources increases with innovation. Using
gy use is linked to levels of environmental pollution. This more environmentally friendly resources has positive effects
issue can be addressed from two different perspectives. First, on the environment. Moreover, these models imply that in-
increased economic activity will lead to higher energy con- creasing income will increase environmental protection
sumption and more polluting emissions. Second, growth trends, and thus emissions will decrease. The main
based on innovative technology will create less environmental determinant in the reduction of emissions here is the
pollution. In the literature, the effects of economic growth and innovation that comes with technological development. In
energy consumption on environmental pollution can be exam- this context, one of the variables that affects economic
ined within the framework of the environmental Kuznets growth in the EKC hypothesis is innovation. Aldieri et al.
curve (EKC) hypothesis developed by Grossman and (2019) state that innovation is related to environmental condi-
Krueger (1991). According to this hypothesis, economic tions. Patent applications, defined as innovative ideas, aim to
growth will increase environmental pollution through the im- use environmentally friendly energy sources. Moreover, in-
pact of fossil fuels used, until it reaches a certain income level, creasing innovation and knowledge has an important role in
but when the threshold value is exceeded, economic growth reducing environmental pollution (Azevedo et al. 2018). The
will decrease with the use of renewable resources (Aydin impact of innovation on reducing environmental pollution ap-
2019a, b, c). Although the EKC hypothesis initially examined pears in the long run (Jaffe and Stavins 1994). What this
the relationship between growth and environment, variables means is that investments in innovation in the long run are
such as energy consumption, globalization, urbanization, hu- expected to reduce environmental pollution through the use of
man capital, financial development, and trade openness, clean resources. It is therefore necessary to include innovation
which affect growth ultimately, are included in the EKC mod- in environmental pollution reduction policies (Dauda et al.
el (Pata and Aydin 2020). 2019). Innovation can be represented by the number of patent
Environ Sci Pollut Res

applications (Khattak et al. 2020). Figure 1 shows the relation- The studies in which the ecological footprint is used to
ship between the GDP and innovation for the 10 most inno- represent environmental pollution in the EKC model in-
vative economies. The figure reveals that GDP per capita and clude the following. Caviglia-Harris et al. (2009) analyzed
the number of patent applications move together over a period the ecological footprint economic growth relationship for
of time. Accordingly, there is a relationship between the two 146 countries in the framework of the EKC hypothesis,
variables. using data from 1961–2000. According to the results of
In this study, the relationship between innovation and en- the study using panel data methods, the EKC hypothesis
vironment for the top 10 innovative economies is examined is not valid in the countries examined. Hervieux and
with the help of the EKC hypothesis. The reason for choosing Darné (2015) examined the same relationship using data
the 10 most innovative economies was to reveal the impact of from 1961 to 2007 for seven Latin-American countries.
innovation on the environment more clearly. The study con- The results confirm that the EKC hypothesis is not valid
tributes to the existing literature in two ways. First, the EKC in the countries studied. Al-mulali et al. (2015a) analyzed
hypothesis with the innovation variable is included for the first the validity of the EKC hypothesis for 93 countries for the
time for the top 10 innovative economies. Second, contrary to period 1980–2008, using ecological footprint. Their results
the existing studies in the literature, the relationship between show that the validity of the EKC hypothesis increases with
innovation and ecological footprint is examined, not the rela- the growing level of the countries’ incomes. Ozturk et al.
tionship between innovation and carbon emissions. (2016) examined the validity of the EKC hypothesis using
The second part of the study contains a summary of the data from the period 1988 to 2008 for 144 countries. They
relevant literature. Data, model, and methodology are intro- found a negative relationship between the ecological foot-
duced in the third part. The fourth part of the study presents print and the explanatory variables of the model. The study
the empirical results. The study is completed with conclusions also concludes that the validity of the EKC hypothesis is
and policy recommendations. higher in high-income countries than in low-income coun-
tries. Charfeddine and Mrabet (2017) examined the ecolog-
ical footprint economic growth relationship for 15 MENA
countries using 1995–2007 data. According to the results,
Literature review the EKC hypothesis is valid in the countries studied.
Ulucak and Bilgili (2018) examined the impact of econom-
There are many studies in the literature where the EKC hy-
ic growth on the ecological footprint in the 1961–2013 pe-
pothesis is tested. In these studies, in addition to the economic
riod for low-, middle-, and high-income countries within
growth independent variable, variables such as globalization,
the framework of the EKC hypothesis. The number of
urbanization, human capital, financial development, and trade
countries in which the EKC hypothesis is valid is higher
openness are also used in the EKC model. However, the num-
than the number of countries where the results are not valid.
ber of studies in which the innovation variable is included in
Destek and Sarkodie (2019) examined the same relation-
the EKC equation is very low. Most of the studies in which the
ship for newly industrialized countries in the period 1977-
EKC hypothesis has been tested use carbon emissions to rep-
2013. The results confirm the validity of the EKC hypoth-
resent environmental pollution, while few studies use the eco-
esis in China, India, South Korea, Thailand, and Turkey.
logical footprint.

Fig. 1 Innovation and GDP per

capita for top 10 innovative
economies (1990–2015)
 Environ Sci Pollut Res

Studies in the literature that tested the extended EKC hy- that the EKC hypothesis is valid only for six countries and the
pothesis with renewable energy include the following. López- impact of globalization varies by country. Haseeb et al. (2018)
Menéndez et al. (2014) examined the environmental investigated the EKC hypothesis for BRICS countries for the
pollution-renewable energy relationship in the 1996–2010 pe- period 1995–2014. The results of the study indicate that the
riod for 27 countries from the European Union within the EKC hypothesis is valid and the effect of globalization on
framework of the EKC hypothesis. The results confirm the environmental pollution is not significant. Khan and Ullah
significant effect of renewable energy on environmental pol- (2019) tested the EKC hypothesis for Pakistan in the 1975–
lution. Jebli and Youssef (2015) analyzed the validity of the 2014 period. According to the results, the EKC hypothesis is
EKC hypothesis for Tunisia for the period 1980–2009, using valid and globalization has a significant and positive effect on
renewable energy. The results confirm the existence of an environmental pollution. Destek (2019) examined the validity
inverted U-shaped relationship between economic growth of the EKC hypothesis for Central and Eastern European coun-
and environmental pollution. On the other hand, the results tries in the 1995–2015 period. The results display that the EKC
show that renewable energy consumption has a positive effect hypothesis is valid and globalization increases environmental
on environmental pollution. Bölük and Mert (2015) tested the pollution. Zafar et al. (2019) tested the EKC hypothesis for the
validity of the EKC hypothesis using data from 1961–2010 period 1990–2014 in OECD countries. According to the results,
for Turkey. According to the results, the EKC hypothesis is the EKC hypothesis is valid and globalization reduces
valid for Turkey and renewable energy has a positive effect on environmental pollution for these countries. Suki et al. (2020)
environmental pollution. Bilgili et al. (2016) examined the investigated the validity of the EKC hypothesis for Malaysia in
validity of the EKC hypothesis for the 19 OECD countries the 1970–2018 period. The results confirm the existence of an
for the period 1977–2010, using renewable energy consump- inverted U-shaped relationship between economic growth and
tion. The results indicate that the EKC hypothesis is valid for environmental pollution and globalization increases environ-
the panel and renewable energy consumption has a negative mental pollution for Malaysia.
impact on environmental pollution. Zoundi (2017) investigat- Studies in the literature that tested the extended EKC hy-
ed the validity of the EKC hypothesis using data from the pothesis with innovation include the following. He and Jiang
period 1980–2012 for the 25 African countries. The results (2012) examined the environmental pollution-innovation re-
show that the EKC hypothesis is not valid in these countries. lationship for China with the help of the EKC hypothesis. The
On the other hand, the negative impact of renewable energy results show that innovation reduces environmental pollution
consumption on environmental pollution has been confirmed. in China. Baiardi (2014) examined the impact of innovation
Pata (2018) tested the validity of the EKC hypothesis using on environmental pollution for 20 Italian regions. The results
data from 1974–2014 for Turkey. According to the results, the confirm a significant effect of innovation on the EKC model.
EKC hypothesis is valid for Turkey. However, renewable en- Alvarez-Herranz et al. (2017) analyzed the validity of the
ergy consumption is not at a desirable level to reduce environ- EKC hypothesis using data for the period 1990–2012 for 17
mental pollution. Erdogan et al. (2020) examined the validity OECD countries. The results confirm the existence of an
of the EKC hypothesis for the 25 OECD countries in the inverted N-shaped relationship between economic growth
1990–2014 period. The results indicate that the EKC hypoth- and environmental pollution. Furthermore, a positive effect
esis is not valid for the OECD countries in the period studied. of innovation on environmental pollution has been confirmed.
On the other hand, the significant effect of renewable energy Balsalobre-Lorente et al. (2019) investigated the impact of
consumption on environmental pollution has been confirmed. innovation on environmental pollution using the 1995–2016
Pata and Aydin (2020) tested the validity of the EKC hypoth- data for 16 OECD countries under the EKC hypothesis. The
esis for six hydropower energy-consuming countries in the results show that innovation has a positive effect on environ-
1965–2016 period, using hydropower energy consumption. mental pollution. Dauda et al. (2019) analyzed the impact of
The results of the study show that the EKC hypothesis is not innovation on environmental pollution using the 1990–2016
valid and hydropower energy consumption is not used effec- data for different regions. The results suggest that innovation
tively enough to reduce environmental pollution in these reduces environmental pollution in G6 countries while in-
countries. creasing it in MENA and BRICS countries. In another study
Studies in the literature that tested the extended EKC hy- conducted for BRICS countries, Khattak et al. (2020) exam-
pothesis with globalization include the following. Shahbaz ined the same relationship for the period 1980–2016 and
et al. (2015) examined the validity of the EKC hypothesis for reached the conclusion that innovation increase carbon emis-
India using data from 1970 to 2012. The results display that the sions in all countries studied, with the exception of Brazil.
EKC hypothesis is valid for India, but globalization has no When the literature is examined, the relationship between in-
significant effect on environmental pollution. Shahbaz et al. novation and ecological footprint within the framework of
(2016) tested the validity of the EKC hypothesis for 19 EKC hypothesis is absent. This study seeks to fill the gap in
African countries in the 1971–2012 period. The results show the literature in this sense.
Environ Sci Pollut Res

Data, model, and methodology N −1 N 2

CD ¼ T ∑ ∑ b
ρij ð3Þ
i¼1 j¼iþ1
Data
where b ρij shows the correlation between errors calculated
This study examines the validity of the EKC hypothesis for from Eq. 1. Pesaran (2004) proposed a test statistic as follows,
the top 10 innovative economies, namely, Denmark, Finland, since the test statistic calculated in Eq. 3 can give deviant
Germany, Israel, Korea, The Netherlands, Sweden, results in large samples:
Switzerland, the UK, and the USA, for the period 1990– sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
2015. In the study, gdp and gdp2 denote per capita GDP and 1 N −1 N 2
CDLM1 ¼ ∑ ∑ Tb ρij −1 ð4Þ
its square (constant 2010 US $), respectively, ef denotes per N ðN −1Þ i¼1 j¼iþ1
capita ecological footprint (global hectares per person), inv
denotes per capita innovation (patent applications, residents), This test statistic was obtained by making some corrections
ren denotes per capita renewable energy consumption (total to the test statistic given in Eq. 3. In this way, cross-section
final consumption, ktoe), and glob denotes overall globaliza- dependency can be tested in large samples. However, where
tion index derived from three sub-indexes, namely, economic the observation (N) is greater than the time (T), Eq. 4 is reg-
globalization, social globalization, and political globalization. ulated by Pesaran (2004) and the following test statistic is
gdp and inov data are taken from the World Bank database; obtained.
ren and glob data are taken from the International Energy sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !
Agency (IEA 2020) database and Dreher (2006), respectively; 2T N −1 N
CDLM2 ¼ ∑ ∑ b ρ ð5Þ
ef data is taken from Global Footprint Network database. All N ðN −1Þ i¼1 j¼iþ1 ij
the data used have been converted into a logarithmic form.
For all three test statistics, the null hypothesis shows that
Model there is no cross-section dependency, while the alternative
hypothesis refers to cross-section dependency.
The model used for the test of the EKC hypothesis was as In the study, delta tests recommended by Pesaran and
follows: Yamagata (2008) were used as homogeneity tests. The test
statistic for the first of the delta tests, an extended version of
lne f t ¼ β0 þ β1 lngdpit þ β2 lngdp2it þ β3 lnrenit the method proposed by Swamy (1970), is obtained as fol-
þ β4 lninovit þ β5 lnglobit þ εit ð1Þ lows:
!
whereβ1, β2, β3, β4, and β5 are the coefficients of lngdpt, pffiffiffiffi N −1 e
S−k
b
Δ¼ N pffiffiffiffiffi ð6Þ
lngdpt2, lnrenit, lninovit, and lnglobit, respectively. εt is the error 2k
term. There are some conditions to determine the validity of the
EKC hypothesis. First, cointegration relationship should exist, where e S shows the modified Swamy statistics. Delta test
then the coefficient of lngdpt should be positive and the coeffi- statistics have been modified as follows under the assumption
cient of lngdpt2should be negative and statistically significant in of normally distributed errors:
Eq. (1). The formula used for the turning point is as follows: 0  1
B −1 e C    
pffiffiffiffiB N S−E ezbit
β
− 2β1
Y* ¼ e ð2Þ C
b adj ¼
2
Δ NB B sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
C E ez ¼ k; var ez
C
@ A bit bit
var ez
Methodology bit

¼ 2k ðT −k−1Þ=T þ 1 ð7Þ
Cross-sectional dependence and slope homogeneity tests
For both test statistics, the null hypothesis shows homoge-
In studies in which panel data analysis is used, cross-section neity, while the alternative hypothesis represents the heteroge-
dependency tests should be carried out to determine the ap- neous slope.
propriate analysis method. In this study, three different test
statistics were calculated for the cross-sectional dependence.
In addition to calculating cross-section dependency for vari- Panel unit root test
ables, it was also calculated for EKC model. The test statistics
proposed by Breusch and Pagan (1980) were used for testing Peseran (2007) proposed unit root test statistics that can be
the cross-sectional dependence for Eq. 1. used in the case of cross-section dependency in panel data
 Environ Sci Pollut Res

analysis. Based on the ADF unit root test, this test statistic has respectively. The null hypothesis of the LM test indicates
been expanded to take into account cross-sectional dependen- cointegration for all cross-section units.
cy as follows:

Δyit ¼ ai þ bi yi;t−1 þ ci yt−1 þ d i Δyt þ eit ð8Þ Empirical results

N
where yt and Δyt are defined as N −1 ∑ yit and In this study, the cross-sectional dependence of variables and
i¼1 EKC model was tested using the Breusch and Pagan (1980)
N LM, Pesaran (2004) scaled LM, and Pesaran (2004) CD tests,
N −1 ∑ Δyit , respectively. This test statistic, known as cross-
i¼1 and the results are reported in Table 1. The results show the
sectional augmented ADF (CADF), is calculated for each sec- existence of cross-section dependence in both variables and
tion unit. Using these test statistics, CIPS panel unit root test EKC model, according to all three tests. For this reason, in the
statistics are calculated as follows: rest of the study unit root and cointegration tests that take into
account the cross-section dependency were used. Table 1 also
CIPS ðN ; T Þ ¼ N −1 ∑Ni¼1 t i ðN ; T Þ ð9Þ shows the results of homogeneity test. The results show that
the model has heterogeneous slopes, by rejecting the null hy-
where ti(N, T) shows the CADF test statistics calculated for
pothesis of homogeneous slopes.
each cross-section unit. For the CIPS panel unit root test, the
Table 2 shows the unit root test results for the variables.
unit root null hypothesis is tested against the alternative hy-
According to these results, all variables have a unit root at
pothesis which shows stationary.
level, while they are stationary at first difference with different
significance levels. In other words, the integration degree of
Panel cointegration tests Conventional cointegration tests do
all variables is one, that is I(1). In the continuation of the
not consider cross-sectional dependence. However, in the lit-
empirical analysis, the long-run relationship between ecolog-
erature, the recently developed cointegration tests do take this
ical footprint and explanatory variables was examined. This
situation into consideration. Westerlund (2008) proposed two
also involved testing the cointegration relationship for the
test statistics for the panel cointegration test, which takes into
EKC model.
account the cross-sectional dependence:
Table 3 shows the panel cointegration tests results.
n  2 T 2 According to the results, there is a cointegration for the EKC
DH g ¼ ∑ b
Si ϕ bi ∑ be
e i −ϕ ð10Þ
i¼1
it−1
t¼2
model. Here, ecological footprint and explanatory variables
 2 n T 2 move together in the long term. Long-run coefficients for
DH p ¼ b e ϕ
S n ϕ− b ∑ ∑ be ð11Þ the EKC model were estimated using the mean group (MG)
it−1
i¼1 t¼2 and augmented mean group (AMG) estimators.
Table 4 shows the long-term coefficients for the EKC mod-
where DHg denotes panel statistics and is obtained by sum-
el. Analyzed by country, the results are as follows: (i)
ming individual terms before being put together. The DHp
Renewable energy consumption for Denmark is statistically
group represents the average statistic and is created by multi-
significant and reduces environmental pollution; (ii)
plying and then adding various terms. The null hypothesis of
Globalization and renewable energy consumption are statisti-
the Durbin Hausman test is that there is no cointegration,
cally significant for Germany and have a reducing effect on
while the alternative hypothesis is cointegration in the panel.
Westerlund and Edgerton (2007) proposed a bootstrap LM Table 1 Cross-section dependence and slope homogeneity tests results
panel cointegration test based on McCoskey and Kao (1998).
The McCoskey and Kao (1998) methodology does not give Variables CDLM1 CDLM2 CD
confident results in small samples and cross-sectional depen- lnef 209.2685* 17.31542* 5.602933*
dence. To overcome these problems, Westerlund and lninov 450.7750* 42.77244* 3.046984*
Edgerton (2007) use the bootstrap scheme (Aydin 2019a, b, lngdp 1060.205* 107.0120* 32.54377*
c). The following LM statistic is used to test the cointegration lnren 540.3711* 52.21670* 19.16041*
between variables: lnglob 1093.277* 110.4981* 33.06062*
Model 414.7519* 38.97527* 8.302120*
1 N T −2 2
LM þ
N ¼ ∑ ∑ω b S ð12Þ Slope homogeneity Test statistics P value
N T 2 i¼1 t¼1 i it b
Δ 9.298* 0.000
b adj
Δ 10.543* 0.000
where N and T are indexes of the cross-sectional units and
b i are the partial
time series, respectively. Additionally, Sit and ω *indicates the rejection of the null hypothesis at 1% significance level
sum process of the residuals and long-run variance,
Environ Sci Pollut Res

Table 2 CIPS panel unit namely the EKC hypothesis, is valid for Israel. Accordingly,
root test results Variables Level First differences
the increase of GDP per capita for these countries will increase
lnef − 1.992 − 3.844* environmental pollution up to a certain turning point, while
lninov − 1.302 − 3.404* increasing the GDP per capita will reduce environmental pollu-
lngdp − 2.003 − 2.514** tion when the threshold is exceeded. The turning point for Israel
lngdp2 − 1.997 − 2.496** was calculated using AMG result, and the result is shown in Fig.
lnren − 1.729 − 3.615* 2 (in Appendix). According to the result, per capita GDP for the
lnglob − 2.152 − 3.429* Israel has exceeded the turning point.

* and ** indicate the rejection of the null

hypothesis at 1% and 5% significance
levels
Conclusion and policy recommendations

In this study, the effect of innovation on environmental pollu-

environmental pollution; (iii) Innovation coefficient is statis-
tion was examined for emerging countries in the period of
tically significant for Korea, and the variable has an environ-
1990–2015 within the framework of the EKC hypothesis. For
mental pollution-reducing effect; (iv) Renewable energy con-
this purpose, cross-sectional dependence tests were performed,
sumption for The Netherlands is statistically significant and
and the existence of cross-sectional dependence for models and
reduces environmental pollution; (v) The coefficient of glob-
series was determined. The integration degrees of the series
alization variable is statistically significant for Switzerland,
were then examined using panel unit root tests that take into
and the variable reduces environmental pollution; (vi) For
account the cross-sectional dependence. Within the framework
the Finland, the coefficient of the innovation variable is statis-
of the EKC model, long-term relationships between ecological
tically significant and has a reducing effect on environmental
footprint and explanatory variables were examined with panel
pollution; (vii) For the USA, innovation and renewable energy
cointegration tests. According to the results, long-term relation-
consumption are statistically significant and reduce environ-
ship was determined in the countries examined for the EKC
mental pollution. Finally, when the whole panel is analyzed,
model. Finally, long-run coefficients were estimated to deter-
the coefficient of the innovation is statistically significant and
mine the validity of the EKC hypothesis.
reduces environmental pollution.
When the results obtained from the study are analyzed
This study also investigated the effect of explanatory vari-
within the framework of the EKC hypothesis, it shows that
ables on environmental pollution. Accordingly, the innovation
the hypothesis is valid for Israel. For Israel, the threshold value
is significant for Finland, Korea, and the USA and reduces
was calculated and the result is shown in Fig. 2. Accordingly,
environmental pollution for all these countries. Renewable
it is seen that the GDP value of Israel exceed the threshold
energy consumption is significant for Denmark, Germany,
value. The results show that economic growth in Israel de-
The Netherlands, and the USA, and has an environmental
creases environmental pollution. Accordingly, the effects of
pollution-reducing effect in all these countries. Globalization
environmental protection policies for Israel emerged positive-
is significant for Germany and Switzerland and reduces envi-
ly. In other countries, environmental policies are not sufficient
ronmental pollution in both countries.
to reduce environmental pollution. Other results from the
When the long-run coefficients are analyzed within the
study are as follows: (i) Innovation is statistically significant
framework of the EKC hypothesis, lngdp is positive for Israel,
for Korea, the USA, Finland, and the whole panel, and ap-
while lngdp2 is negative. This result shows that with the impact
pears to reduce environmental pollution; (ii) Renewable ener-
of globalization and innovation, the inverted U shape relation-
gy consumption is statistically significant for Denmark,
ship between environmental pollution and economic growth,
Germany, the Netherlands, and the USA, and it also has an
environmental pollution-reducing effect; (iii) Globalization
Table 3 Panel cointegration tests results
has been achieved significantly for Germany and
Tests Value P value Switzerland, and has been found to have a negative impact
on environment pollution. Our panel results are consistent
Durbin Hausman with He and Jiang (2012), Baiardi (2014), Alvarez-Herranz
Durbin Hausman Group 6.308* 0.000 et al. (2017), Balsalobre-Lorente et al. (2019), and Dauda et al.
Durbin Hausman Panel 3.528* 0.000 (2019) for innovation; consistent with Pata (2018) and Al-
LM Bootstrap Mulali et al. (2015b) for renewable energy consumption; con-
Lm Statistic 17.633 0.976 sistent with Shahbaz et al. (2016), Khan and Ullah (2019), and
*indicates the rejection of the null hypothesis at 1% significance levels.
Zafar et al. (2019) for globalization.
The null hypothesis of the LM Bootstrap test indicates cointegration in According to country-based results, innovation has a posi-
the model. The bootstrap is based on 1000 replications tive impact on environmental pollution in Korea, the USA,
 Environ Sci Pollut Res

Table 4 Robustness tests: long-

run estimates Countries lninov lnren lngdp lngdp2 lnglob EKC

Denmark MG 0.048 − 0.193** − 81.269* 3.777* 0.118 Not valid

AMG 0.032 − 0.212** − 83.933* 3.901* − 0.135 Not valid
Finland MG − 0.619* 0.034 15.257 − 0.644 1.452 Not valid
AMG − 0.547** 0.224 13.169 − 0.575 0.089 Not valid
Germany MG − 0.000 − 0.115* 30.468 − 1.405 − 1.463* Not valid
AMG 0.157 − 0.117* 25.711 − 1.181 − 1.566* Not valid
Israel MG − 0.002 − 0.036 70.486** − 3.470** − 0.104 Valid
AMG 0.045 − 0.033 56.774*** − 2.797*** − 0.449 Valid
Korea MG − 0.642*** 0.041 28.336 − 1.270 − 3.715 Not valid
AMG − 0.311*** − 0.404 − 1.969 0.189 − 3.292 Not valid
Netherlands MG 0.030 − 0.26*** − 97.806** 4.631** 0.969 Not valid
AMG 0.041 − 0.167** − 79.357 3.757 0.255 Not valid
Sweden MG − 0.004 − 0.060 4.531 − 0.190 0.594 Not valid
AMG − 0.192 0.469 34.372 − 1.602 − 1.969 Not valid
Switzerland MG − 0.103 0.052 9.566 − 0.376 − 0.644** Not valid
AMG − 0.193 0.061 − 23.511 1.102 − 1.249** Not valid
UK MG − 0.157 0.007 − 48.705*** 2.392*** 1.859 Not valid
AMG − 0.163 − 0.006 − 47.834 2.351 1.807 Not valid
USA MG − 0.192** − 0.06*** 7.918 − 0.294 0.675 Not valid
AMG − 0.224** − 0.06*** 6.990 − 0.248 0.471 Not valid
Panel MG − 0.164** − 0.036 − 6.121 0.312 − 0.025 Not valid
AMG − 0.149** − 0.002 − 9.958 0.489 − 0.603 Not valid

*, **, and *** indicate the rejection of the null hypothesis at 1%, 5% and 10% significance levels, respectively

and Finland. Various policies have been implemented to in- aims to have an energy-free energy system by 2050 (IRENA,
crease innovation in these countries. For example, in Korea IEA and REN21 2018); and (iv) the Energiewende plan,
many incentive policies are implemented for R&D centers by which has operated for nearly ten years in Germany, aims to
the Korean Ministry of Science, ICT and Future Planning. make the country’s energy system more efficient and mainly
These incentives include taxation (income tax reduction for uses renewable resources (IEA 2020). Finally, the results of
foreign technology experts), human resources (special re- the study show that globalization has an effect on reducing
search expert system), finance (R&D guarantee), and technol- environmental pollution in Germany and Switzerland.
ogy support (industry-academia-research center cluster sup- Germany began networking globally in the 1990s and con-
port project). The USA supports innovative ideas through tinues to grow strongly. Switzerland is one of the most inte-
the Global Innovation Fund, which is supported by the US grated economies in the world, although it is not in the EU
Agency for International Development and has an office in (Weiß et al. 2018). Overall, promoting market integration by
Washington, D.C. The Finnish Innovation Fund works for reducing or removing trade barriers increases environmental
sustainable everyday living based on an economy where wel- quality in both countries.
fare is not based on the overconsumption of fossil fuels, When the countries are analyzed as a whole, it is seen
through many programs including the carbon-neutral circular that innovation has a positive effect on the environment.
economy. In addition, renewable energy reduces environmen- Figure 3 (in Appendix) shows the values for the ecological
tal pollution in Denmark, Germany, the Netherlands, and the footprint and innovation variables in 1990 and 2015 for the
USA. In these countries, promoting renewable energy con- top ten innovative economies. The charts show that innova-
sumption through additional taxes imposed on fossil fuel use tion has increased more than twofold from 1990 to 2015,
will have a positive impact on the environment. For example: while environmental pollution has decreased by 16%.
(i) US renewable energy sector investments broke a new re- Accordingly, it can be seen that one of the effects of the
cord in 2019, exceeding $55 billion (IEEFA 2019); (ii) the decrease in environmental pollution is innovation. In this
Dutch central government plans to make 16% of all energy sense, in addition to existing policies, increasing investments
used in the Netherlands sustainable by 20231; (iii) Denmark for innovation, providing incentives, and improving techno-
logical infrastructure will reduce environmental pollution.
1
https://www.government.nl/topics/renewable-energy
Environ Sci Pollut Res

Appendix

Fig. 2 Turning Point of EKC Hypothesis for Israel

Fig. 3 Innovation and Ecological

Footprint for the top 10 Innovaon Ecological Footprint
innovative economies (1990;215)

600000 65

400000 60
55
200000
50
0 45

1990 2015 1990 2015

References Aydin M (2019b) Renewable and non-renewable electricity

consumption–economic growth nexus: evidence from OECD coun-
Aldieri L, Bruno B, Vinci CP (2019) Does environmental innovation tries. Renew Energy 136:599–606
make us happy? An empirical investigation. Socio Econ Plan Sci Aydin M (2019c) The effect of economic growth on obesity for the most
67:166–172 obese countries: new evidence from the obesity Kuznets curve. Eur J
Al-Mulali U, Weng-Wai C, Sheau-Ting L, Mohammed AH (2015a) Health Econ 20(9):1349–1358
Investigating the environmental Kuznets curve (EKC) hypothesis Azevedo VG, Sartori S, Campos LM (2018) CO2 emissions: a quantita-
by utilizing the ecological footprint as an indicator of environmental tive analysis among the BRICS nations. Renew Sust Energ Rev 81:
degradation. Ecol Indic 48:315–323 107–115
Al-Mulali U, Saboori B, Ozturk I (2015b) Investigating the environmen- Baiardi D (2014) Technological progress and the environmental Kuznets
tal Kuznets curve hypothesis in Vietnam. Energy Policy 76:123–131 curve in the twenty regions of Italy. BE J Econ Anal Policy 14(4):
Alvarez-Herranz A, Balsalobre-Lorente D, Shahbaz M, Cantos JM 1501–1542
(2017) Energy innovation and renewable energy consumption in Balsalobre-Lorente, D, Shahbaz, M, Jabbour, C J C, Driha, O M (2019)
the correction of air pollution levels. Energy Policy 105:386–397 The role of energy innovation and corruption in carbon emissions:
Andreoni J, Levinson A (2001) The simple analytics of the environmental evidence based on the EKC hypothesis In Energy and environmen-
Kuznets curve. J Public Econ 80(2):269–286 tal strategies in the era of globalization (pp 271-304) Springer, Cham
Aydin M (2019a) The effect of biomass energy consumption on economic Bilgili F, Koçak E, Bulut Ü (2016) The dynamic impact of renewable
growth in BRICS countries: a country-specific panel data analysis. energy consumption on CO2 emissions: a revisited environmental
Renew Energy 138:620–627 Kuznets curve approach. Renew Sust Energ Rev 54:838–845
 Environ Sci Pollut Res

Bölük G, Mert M (2015) The renewable energy, growth and environmen- Nelson RR (1996) The sources of economic growth Cambridge. Harvard
tal Kuznets curve in Turkey: an ARDL approach. Renew Sust Energ University Press, MA
Rev 52:587–595 OECD, E (2005) Oslo manual: guidelines for collecting and interpreting
Breusch TS, Pagan AR (1980) The lagrange multiplier test and its appli- innovation data. Paris 2005, Sp, 46
cations to model specification in econometrics. Rev Econ Stud Ozturk I, Al-Mulali U, Saboori B (2016) Investigating the environmental
47(1):239–253 Kuznets curve hypothesis: the role of tourism and ecological foot-
Caviglia-Harris JL, Chambers D, Kahn JR (2009) Taking the “U” out of print. Environ Sci Pollut Res 23(2):1916–1928
Kuznets: a comprehensive analysis of the EKC and environmental Pata UK (2018) Renewable energy consumption, urbanization, financial
degradation. Ecol Econ 68(4):1149–1159 development, income and CO2 emissions in Turkey: testing EKC
Charfeddine L, Mrabet Z (2017) The impact of economic development hypothesis with structural breaks. J Clean Prod 187:770–779
and social-political factors on ecological footprint: a panel data anal- Pata UK, Aydin M (2020) Testing the EKC hypothesis for the top six
ysis for 15 MENA countries. Renew Sust Energ Rev 76:138–154 hydropower energy-consuming countries: evidence from Fourier
Dauda L, Long X, Mensah CN, Salman M (2019) The effects of econom- bootstrap ARDL procedure. J Clean Prod 264(121699):1–10
ic growth and innovation on CO2 emissions in different regions. Pesaran MH (2004) General diagnostic tests for cross section dependence
Environ Sci Pollut Res 26(15):15028–15038 in panels. CESifo Working Paper Series No. 1229; IZA Discussion
Destek MA (2019) Investigation on the role of economic, social, and Paper No. 1240
political globalization on environment: evidence from CEECs. Pesaran MH, Yamagata T (2008) Testing slope homogeneity in large
Environ Sci Pollut Res:1–14 panels. J Econ 142(1):50–93
Destek MA, Sarkodie SA (2019) Investigation of environmental Kuznets Rees WE (1992) Ecological footprints and appropriated carrying capac-
curve for ecological footprint: the role of energy and financial de- ity: what urban economics leaves out. Environ Urban 4(2):121–130
velopment. Sci Total Environ 650:2483–2489 Romer PM (1986) Increasing returns and long-run growth. J Polit Econ
Dreher A (2006) Does globalization affect growth? Evidence from a new 94(5):1002–1037
index of globalization. Appl Econ 38(10):1091–1110. Schumpeter JA (1934) The theory of economic development: An inquiry
Erdogan S, Okumus I, Guzel AE (2020). Revisiting the environmental into profits, capital, credit, interest and the business cycle, Harvard
Kuznets curve hypothesis in OECD countries: the role of renewable, University Press, Cambridge, MA
non-renewable energy, and oil prices. Environ Sci Pollut Res Int Shahbaz M, Mallick H, Mahalik MK, Loganathan N (2015) Does glob-
Fernández YF, López MF, Blanco BO (2018) Innovation for sustainabil- alization impede environmental quality in India? Ecol Indic 52:379–
ity: the impact of R&D spending on CO2 emissions. J Clean Prod 393
172:3459–3467 Shahbaz M, Solarin SA, Ozturk I (2016) Environmental Kuznets curve
Grossman, G M, Krueger, A B (1991) Environmental impacts of a North hypothesis and the role of globalization in selected African coun-
American free trade agreement (No w3914) National Bureau of tries. Ecol Indic 67:623–636
Economic Research Solow RM (1957) Technical change and the aggregate production func-
Haseeb A, Xia E, Baloch MA, Abbas K (2018) Financial development, tion. Rev Econ Stat 39:312–320
globalization, and CO2 emission in the presence of EKC: evidence
Suki NM, Sharif A, Afshan S, Suki NM (2020) Revisiting the environ-
from BRICS countries. Environ Sci Pollut Res 25(31):31283–31296
mental Kuznets curve in Malaysia: the role of globalization in sus-
He YW, Jiang JR (2012) Technology innovation based on environmental
tainable environment. J Cleaner Production 121669
Kuznets curve hypothesis. In Adv Mater Res 573:831–835
Swamy PA (1970) Efficient inference in a random coefficient regression
Hervieux MS, Darné O (2015) Environmental Kuznets curve and eco-
model. Econometrica 38(2):311–323
logical footprint: a time series analysis. Econ Bull 35(1):814–826
Ulucak R, Bilgili F (2018) A reinvestigation of EKC model by ecological
IEA (2020) International Energy Agency, Germany 2020 Energy Policy
footprint measurement for high, middle and low income countries. J
Review
Clean Prod 188:144–157
IEEFA (2019) Institute for Energy Economics and Financial Analysis
Wackernagel M (1994) Ecological footprint and appropriated carrying
(IEEFA)
capacity: a tool for planning toward sustainability (Doctoral disser-
IRENA, IEA and REN21 (2018) ‘Renewable energy policies in a time of
tation, University of British Columbia)
transition’
Jaffe AB, Stavins RN (1994) The energy-efficiency gap What does it Weiß J, Sachs A, Weinelt H (2018) Globalization Report. Who Benefits
mean? Energy Policy 22(10):804–810 Most from Globalization
Jebli MB, Youssef SB (2015) The environmental Kuznets curve, eco- Westerlund J (2008) Panel cointegration tests of the Fisher effect. J Appl
nomic growth, renewable and non-renewable energy, and trade in Econom 23(2):193–233
Tunisia. Renew Sust Energ Rev 47:173–185 Westerlund J, Edgerton DL (2007) A panel bootstrap cointegration test.
Khan D, Ullah A (2019) Testing the relationship between globalization Econ Lett 97(3):185–190
and carbon dioxide emissions in Pakistan: does environmental Zafar MW, Saud S, Hou F (2019) The impact of globalization and finan-
Kuznets curve exist? Environ Sci Pollut Res 26(15):15194–15208 cial development on environmental quality: evidence from selected
Khattak SI, Ahmad M, Khan Z U Khan A (2020) Exploring the impact of countries in the Organization for Economic Co-operation and
innovation, renewable energy consumption, and income on CO2 Development (OECD). Environ Sci Pollut Res 26(13):13246–
emissions: new evidence from the BRICS economies. Environ Sci 13262
Pollut Res 1-16 Zoundi Z (2017) CO2 emissions, renewable energy and the environmen-
López-Menéndez AJ, Pérez R, Moreno B (2014) Environmental costs tal Kuznets curve, a panel cointegration approach. Renew Sust
and renewable energy: re-visiting the environmental Kuznets curve. Energ Rev 72:1067–1075
J Environ Manag 145:368–373
McCoskey S, Kao C (1998) A residual-based test of the null of Publisher’s note Springer Nature remains neutral with regard to jurisdic-
cointegration in panel data. Econ Rev 17(1):57–84 tional claims in published maps and institutional affiliations.