Received: 19 January 2020 Revised: 27 February 2020 Accepted: 17 March 2020

DOI: 10.1002/sd.2057

RESEARCH ARTICLE

An assessment of the environmental sustainability corridor:

Investigating the non-linear effects of environmental taxation
on CO2 emissions

Recep Ulucak1 | Danish2 | Yacouba Kassouri1

1
Faculty of Economics and Administrative
Sciences, Department of Economics, Erciyes Abstract
University, Kayseri, Turkey The effectiveness of environmental taxes for achieving carbon abatement has been
2
School of Economics and Trade, Guangdong
subject to important debates on the sustainable development pathway. Theoretically,
University of Foreign Studies, Guangzhou,
China it is expected that environmental taxation (ERT) can reduce carbon emissions by
enhancing innovation and energy efficiency that are the main drivers of environmen-
Correspondence
Danish, School of Economics and Trade, tal sustainability. The purpose of the study is to investigate the non-linear effects of
Guangdong University of Foreign Studies,
ERT on carbon dioxide (CO2) emissions by controlling environmental technologies,
Guangzhou 510006, China.
Email: [email protected] patens and economic growth within the limitation of the available longest data set
covering the period of 1994–2015. The study employs the panel smooth transition
regression approach, which determines the threshold levels endogenously, by
observing the lower and higher integration levels of countries through globalization
as a transition variable. The empirical findings recommend that ERT decreases CO2
emissions at higher levels of globalization. The non-linear effect of environmental-
related technologies and patents on CO2 emissions is also favourable at higher levels
of globalization.

KEYWORDS

CO2 mitigation, environmental-related technologies, environmental taxes, globalization, PSTR,

sustainable development

1 | I N T RO DU CT I O N pollution, is one of the crucial challenges (Du, Liu, Lei, & Huang, 2018)
in terms of environmental sustainability (Ulucak, Yücel, &
Sustainable development that is based on economic, societal and Koçak, 2019). Besides, climate change stemming from greenhouse
environmental aspects is the best alternative for countries to maxi- gases has become a serious issue that needs priority base arrange-
mize the welfare with limited resources against unlimited wants. ments. The rise in global average temperature brings about
Because, it is not possible to establish a prosperous life standard by unpredictable, harsh and unplanned environmental and climate
only achieving sustainable economic conditions such as higher per changes, which adversely influence human lives. On the other hand,
capita income level, price stability, current account balance, full global warming is an undesired by-product of economic activities for
employment. Also, social improvements such as justice, equity, ecosystems (Devi & Gupta, 2019). In particular, air pollution, as the
democracy, cultural and educational attainments are required to be main driver of temperature rise, seriously affects and increases the
able to increase the welfare level in addition to a healthy environment risk of public health globally, in fast industrializing countries including
that is the main source of life (Bilgili & Ulucak, 2018). However, the BRICS (Brazil, Russia, India, China and South Africa) economies
environmental dimension of sustainable development has become a (Lu et al., 2019). The BRICS countries include large economies and
vital issue that should be taken into serious consideration. The upris- have practiced rapid economic growth over two decades, which cre-
ing environmental concerns, particularly the widespread haze ates serious pressure on the environment due to the fact that the rise

Sustainable Development. 2020;1–9. wileyonlinelibrary.com/journal/sd © 2020 ERP Environment and John Wiley & Sons Ltd. 1
2 ULUCAK ET AL.

in economic growth leads to natural resources extraction and rising transition regression (PSTR) to produce robust results, that is, a unique
energy demand (Danish & Wang, 2019). Despite the aggressive feature of the study is the use of PSTR approach for investigation of
growth in the last two decades, BRICS countries have faced crucial the causal relationship between environmental taxes, ENRT, patents,
challenges in the pathway of their developments (Danish, Baloch, & per capita GDP and CO2 emissions.
Bo, 2019). Overcoming these environmental challenges, even if the The remaining of the study is organized as follows: the second
BRICS countries manage to sustain the dramatic economic growth, part explains materials and methods; the third part presents and dis-
does not seem possible without taking environmental measures such cusses empirical results; finally, the study is concluded.
as environmental taxation (ERT). The experiences from recent years
indicate that carbon emission mitigation has become a bigger chal-
lenge that cannot only be controlled through economic growth or 2 | M A T E R I A L S A N D M ET H O D S
decreasing energy intensity or energy demand. Therefore, together
with taxes on the energy production sector, environmental charges on 2.1 | Theoretical framework
polluting industries, technology transfers through trade and FDI can
be alternative tools to decrease degradation and improve the quality From a theoretical perspective, the theory of ERT holds that environ-
in the environment (Lin & Jia, 2019). mental charges and taxation contribute to environmental protection
The detrimental effects of environmental issues motivate (Vatn, 2015). Although carbon emission mitigation may be achieved in
policymakers to combat them through effective environmental regula- several ways, ERT and charges can encourage the implementation of
tion measures, in order to reach carbon dioxide (CO2) mitigation tar- environmentally friendly mechanisms into the market and minimize
gets, such as shaped by the Kyoto protocol (Abdullah & the energy consumption that is the major driver of CO2 emissions
Morley, 2014). Taxation is one of the best alternative tools for this through improvements in energy efficiency and clean technologies
purpose with additional benefits for governments. For instance, public (Calderón et al., 2016). Besides, shocks in such taxation have the
debt and its dramatical accumulation are important issues of economic potential to be able to improve the energy mix in favour of renewable
sustainability. Similarly, pollution and environmental degradation sources which are the best alternative of emission abatement by stim-
mostly stemming from carbon emissions is another issue of sustain- ulating directed technological change and green growth (Danish &
ability from the environmental aspect. In this respect, surplus needs Ulucak, 2020; Niu, Yao, Shao, Li, & Wang, 2018). Pollutant charges
government to sustain their debt budget can be balanced through tax- also influence the regional degradation that has different dimensions
ation that reduces emissions and thus environmental quality can be by industrial or sectoral areas or countries. On the other hand, the tra-
improved (Halkos & Papageorgiou, 2018). On the other hand, it is con- ditional economic perspective has a concern about taxes in general
cluded that higher taxes and charges for pollution may be helpful to because of their contractionary pressure on economic growth. How-
establish an efficient way of utilizing resources and encouraging eco- ever, modern growth theories have shown that environmental regula-
nomic growth (Shi, Qiao, Shao, & Wang, 2019). Therefore, environ- tions including taxes are the main driver for the transition to clean
mental taxes can be considered as a fundamental tool to regulate technology that can establish a sustainable development path for
markets, and externalities stemming from pollution-intensive activities economies (Acemoglu, Akcigit, Hanley, & Kerr, 2016). Also, revenues
can be decreased by loading the costs to their producers (Wang from such implementations may provide additional funds to update
et al., 2019). In today's deteriorating environment, research on the outdated infrastructure that consumes more resources and produces
role of ERT or similar applications in the reduction of CO2 emissions is more waste and pollution, to improve human capital accumulation
vitally important (Shi et al., 2019). Besides, the promotion of that is a crucial determinant of long-run economic growth and envi-
environmental-related technologies (ENRT), innovation and carbon ronmental awareness as long as they are well managed. However, the
pricing is central for regulating emission-intensive activities are crucial main purpose of ERT, in reality, is to minimize environmental degrada-
drivers to put successful environmental policies into effect for carbon tion especially to mitigate carbon emissions as is focused by the Kyoto
abatement (Hashmi & Alam, 2019). However, empirical evidence from Protocol and Paris Agreement. The link between economic growth
the current literature, covering these inter-linkages, is relatively inade- and carbon tax comes from current production activities mostly based
quate. Based on the motivation given above, this study aims at inves- on carbon-intensive industries in many countries and it is accepted
tigating the linkages between ERT and carbon emission taking per that such taxes will lead to decrease production. From this viewpoint,
capita GDP, patents and energy-related technologies into account for a decrease in production is expected to decrease carbon emissions.
BRICS countries (Brazil, Russia, China, India and South Africa). The critical point here is that short-run observations may be mislead-
The possible contribution of this paper comes from several ing while long-run experiences should be a focal point. Considering
aspects: (i) This is the first attempt, to the best of our knowledge, to traditional and modern perspectives on the issue, empirical investiga-
investigate how carbon emissions react to ERT in BRICS countries at tions on the long-run behaviour of ERT will shed light on the under-
different levels of globalization. (ii) To investigate the ERT and CO2 standing of how carbon emissions react to ERT. Empirical research
emissions nexus, several important variables are added including has focused on whether ERT decreases CO2 emissions by controlling
ENRT, patents and per capita GDP to avoid omitted bias. (iii) Non- economic growth and some related variables. Based on the
linear behaviour of study variables is controlled through panel smooth econometrical estimations, recent studies mostly show that ERT
ULUCAK ET AL. 3

decreases CO2 emissions (Hashmi & Alam, 2019; He, Ya, Chengfeng, switch from one regime to another. Such restriction is not always feasi-
Yuan, & Xiao, 2019; Mardones & Baeza, 2018; Niu et al., 2018; Shi ble in real-world because most emerging countries are heterogeneous
et al., 2019; Wang, Wang, Shi, & Wei, 2019; Yuan, Zhang, Wang, and react to ERT and related environmental technologies differently at
Wang, & Zuo, 2017). On the other hand, ERT is also handled by some different time periods; so, there can be the infinite number of small
studies to establish a green transformation of economies and sustain- regimes between the two extreme regimes, making the sudden change
able development (Albrecht, 2006; Danish, Ulucak, Khan, Baloch, & across regimes quite impossible (Zhang, Hu, & Chang, 2019). Therefore,
Li, 2019). However, the non-linear behaviour of ERT in the mitigation the PSTR approach is flexible enough to estimate coefficients of the
of environmental pollution by considering global integration develop- regression without any extant information about the structural change
ments has not been considered in the literature. To reveal differential of the transition variable. Second, the PSTR approach is powerful to
effects of taxation, the study follows a model structure presented as- consider main econometric problems such as heterogeneity and time
is in Equation (1): variability by allowing the parameters to change smoothly on panel
units and time dimension (González & Teräsvirta, 2006; Wang &
lnCO2it = a0 + a1 lnY it + a2 lnENRTit + a3 lnERTit + a4 lnPTNTit Wei, 2019). Therefore, the PSTR model is more suitable to consider
ð1Þ
+ a6 lnGLOBit + ωit , heterogeneous effects arising the ongoing environmental policies based
on environmental technologies and/or pollution taxation in the investi-
where CO2 is carbon dioxide emission, Y is per capita GDP, ENRT rep- gation of the economic growth-environmental degradation nexus
resents environmental-related technologies; ERT is environmental taxa- (Tiba, 2019; Wang, Yan, & Komonpipat, 2019).
tion, PTNT is patents and GLOB is globalization in time ‘t’ for cross In this paper, it is claimed that globalization through exports,
section i. Finally, ω is the error term. The expected sign of Y is positive imports, FDI and trade openness is likely to be the most relevant fac-
but it may decrease the environmental pollution after reaching a thresh- tor explaining the cross-country heterogeneity in the relationship
old level due to the well-known composition and technique effects that between environmental taxes and pollution in emerging countries.
are the theoretical underpinnings of the environmental Kuznets curve Finally, in a recent paper, De Vos and Westerlund (2019) revealed that
approach (Grossman & Krueger, 1995). Despite the composition and there is strong evidence on consistency loss for parameters when tak-
technique effect income may not reduce CO2 emissions without addi- ing cross-sectional averages in nonlinear estimations. The PSTR
tional environmental measures (Danish et al., 2019). Similarly, scale, approach is superior to tackle such issue as well as taking account
composition and technique effects are highlighted to explain the rela- potential endogeneity and non-linearity in the data generating pro-
tionship between globalization and environmental pollution cess. More specifically, the PSTR approach divides our sample into
(Copeland & Taylor, 2013). Also, uncertainty or negligence of environ- two extreme regimes delimited by the estimated threshold level of
mental policies may increase pollution (Jiang, Zhou, & Liu, 2019). Fur- globalization with smooth and gradual changes in the value of the
thermore, they may promote to transfer carbon-intensive industries estimated coefficients across regimes. After setting the level of global-
from developed countries with strict environmental regulations to ization as the transition variables, we estimate the two extreme
developing countries in which environmental policies are not stringent regimes. One can specify the corresponding two-regime PSTR model
enough through foreign direct investment (Danish, Wang, & with a single transition function as follows:
Wang, 2018; Sarkodie & Strezov, 2019). Meanwhile, environmental
regulations ensure the proper implication of environmental patents CO2it = μi + δ0 Xit + δ1 Xit  G ðSit : γ; cÞ + ɛit with γ > 0 ð2Þ
(Cheng, Ren, Wang, & Yan, 2019). It is clear that environmental regula-
tions measures encourage the environmental-related technological pro- where i = 1,2,3 …… N and t = 1,2, …… T being the number of cross
gress which increases efficient energy use and reduces carbon section under study and the time dimensions of the panel, respec-
emissions (Guo, Qu, & Tseng, 2017). Innovation in renewable energy tively. μi denotes the fixed effect factor (unobserved heterogeneous
leads to enhance green energy use with less carbon emission intercept). CO2it denotes the measure of carbon dioxide emissions.
(Ganda, 2019; Jordaan et al., 2017). Hence, environmental policy- Xit = (X1t, X2t...) includes environmental related-technologies, environ-
related factors are included in the model such as environmental-related mental taxes, patents and real per capita GDP. ɛit is assumed to be i.i.d
technologies, ERT and patents related to the environment. error term. G(Sit; γ; c) is a smooth continuous transition function of Sit,
which is the transition variable corresponding to the level of globaliza-
tion in our study. Following (González et al., 2005), the transition
2.2 | PSTR approach function is assumed to be as is Equation (3):
Y
n  
GðSit ;γ; cÞ = ½1 + expð− γ Sit − c j  − 1 ð3Þ
Although several methods have been propesed to model non-linear
j=1
estimation of panel data, this study follows the PSTR approach devel-
oped by González, Teräsvirta, & Van Djik (2005) for several reasons. where γ represents the smoothness parameter that acts for the
First, PSTR models permit to model for both sharp and smooth transi- speed of transition in terms of smoothness or sharpness and cj,
tion between regimes, in contrast to previous non-linear models j = 1,2…. n is an n-dimensional vector of location. The transition func-
(e.g., Markov switching models and panel threshold) assuming abrupt tion is restricted between zero and one that implies two critical stages
4 ULUCAK ET AL.

in the model. Based on the transition variable (KOF globalization in (10). This test can be carried out by following an LM-type test pro-
levels of countries over time), the relationship between cedure with an F-distribution as below:
environmental-related developments, economic growth, and environ-
mental degradation is defined by the parameter δ0 in the first regime ðSSR0 −SSR1 Þ∕mK ~
LMF = FðmK, TN−N−K − mðK + 1ÞÞ ð7Þ
SSR0 =ðTN−N−mK Þ
(when G(.) = 0) and the effects of ERT, ENRT, patents and per capita
GDP on (X-elasticity of) environmental degradation represented by
CO2 emissions equal δ0 + δ1 when G(.) = 1. However, when Sit takes a where SSR0 and SSR1 correspond to the panel sum of squared resid-
higher level, the logistic function becomes a gradual shift of parame- uals, respectively. K is the number of regressors, T and N stand for
ters, δ0 changes to δ0 + δ1. For instance, the effects of environmental number of years and countries, m represents the order of the logistic
charges on CO2 vary across country and time given the value taken transition function in (7). (ii) Specify the number r of transition func-
by the transition function as below: tions and determine the number of extreme regimes that are equal to
r + 1. (iii) the parameters in the selected PSTR model is estimated
∂CO2it (González et al., 2005; González, Teräsvirta, van Dijk, & Yang, 2017;
= δ0 + δ1 GðSit ; γ; cÞ ð4Þ
∂X it
Hansen, 1999). (iv) finally the validity of the estimated PSTR model is
checked by conducting the test to verify that there is no remaining
PSTR models capture both the panel transition regression heterogeneity.
approach developed by Hansen (1999) and the linear panel data esti-
mation under fixed effects. Note that, when γ approaches zero, the
PSTR estimation turns into the linear panel data estimation under 2.3 | Data
fixed effects, while when γ goes to infinity, the PSTR model trans-
forms to the PTR model with sharp transition. By generalizing the The study used the data from 1994 to 2015 for Brazil, China, India
PSTR with more than two regimes, the following procedure can be and South Africa. The variables of the model are CO2 emission, per
cansidered for the estimation: capita GDP, patents, environmental-related technologies, and envi-
ronmental taxation. The data on CO2 emission are measured in mil-
Xr  
j
CO2it = μi + δ0 Xit + δ X Gj
j = 1 j it
Sit ; γ;c j + εit ð5Þ lion metric tons, while data on real GDP are measured in billions of
constant 2005 U.S. dollars. Both CO2 emissions and per capita
GDP are sourced from the World Bank. Patent data present a num-
 
where G j Sitj ;γ j ; c j with j = 1, …r is used to symbolize the logistic ber of attractive properties compared to other alternative metrics
function drawn by Equation (3) and ‘r + 1’ reflects the regime changes of innovation. Patent data refer to the number of patents on envi-
in the PSTR estimation. However, in case ‘r = 0’, the procedure ronmental technologies and it was taken in percentages.
becomes a linear panel estimation, while ‘r = 1’ and ‘r = 2’ imply two Environment-related technologies refer to the development of
and three regime changes, respectively. environment-related technologies and it is measured in percent-
The estimation procedure of the PSTR model initially follows the ages. Environmental taxes are considered to measure revenues
panel fixed effects approach to eliminate the cross-section effects and from various activities that have a relation with environmental pur-
then applies the non-linear least-squares (NLLS) for the estimation of poses and cover a wide range of applications from energy products
the transformed model. Following González et al. (2005), the following to vehicle fuels. The data on patents, environmental technologies
estimation procedures can be tracked for PSTR models: and taxes can be gathered from the databank of the Organization
(i) the null hypothesis of linearity against the alternative smooth for Economic Co-operation and Development (OECD). The globali-
transition is checked. If linearity is rejected, the appropriate transition zation is an index that is widely used in the literature in recent
variable Sit and the form of the transition function are selected. times and known as the KOF index, provided by the Swiss Eco-
González et al. (2005) propose the following procedure to nomic Institute.
check them.
H0 : γ = 0 or H0 : δ0 = δ1. In order to overcome the uni-
dentified nuisance parameters problem under the null hypothesis 3 | RESULTS AND DISCUSSION
as suggested by Luukkonen et al., (1988), the transition function G
(Sit; γ; c) is replaced with its ‘first-order Taylor approximation’ 3.1 | Cross-sectional dependence
around γ = 0 and to check the equivalent hypothesis as is in
Equation (6). This section examines the non-linear relationship between the
study variables and presents the obtained results. The first step is
to determine whether there exists a cross-correlation among panel
CO2it = μi + δ0 X it + δ0 X it KOFit + δ0 Xit KOFit + ::… + δm Xit KOFit m + εit ð6Þ
sections. Thus, cross-sectional dependence (CSD) is initially
Then, testing linearity of the relationship between environmental checked. Several tests are employed for testing the existence of
charges and carbon emissions is equivalent to testing H0 : δ0 = δ1 = …δm CSD. To this end, the LM (Lagrange Multiplier) procedure proposed
ULUCAK ET AL. 5

by Breusch and Pagan (1980), CD test proposed by Pesaran (2004) 3.4 | Result of non-linearity tests
and LMadj test proposed by Pesaran, Ullah, and Yamagata (2008) are
performed. The findings from these CSD tests are shown in Table 1 Having established the non-linear relationship between CO2 emis-
which suggests the null hypothesis of ‘there is no cross-sectional sions and its underlying determinants, now, we question whether
dependence’ is strongly rejected. Hence, the study variables have there is any remaining non-linearity. To this end, we follow González
CSD issue and this result shows that any shock occurs in one of the et al. (2017) and determine the number of transition functions, r, or,
countries affects the others. the number of extreme regimes that are equal to r + 1. Table 4 pre-
sents the results of no remaining non-linearity tests. Again, the three
tests of no remaining non-linearity do not reject the null hypothesis of
3.2 | Panel unit root test one transition process. Our findings suggest a single threshold level of
globalization between the two extreme regimes.
Next step is to confirm the stationarity of the data series. As the vari-
ables have CSD issue, so stationarity level of the series is examined by
Moon and Perron's (2004) panel unit root test, which is capable of 3.5 | PSTR estimation results
handling CSD issue and results are shown in Table 2. From results, the
null hypothesis meaning non-stationarity is strongly rejected at the Before discussing the empirical results, it is important to provide readers
conventional level of significance. This indicates that the variables in with relevant information on the regime change process. In light of the
the PSTR model are stationary at the level. logistic transition function with globalization as the threshold variable,
^ denotes the estimated number of common factors;
Notes: m the effects of technology, patent and taxation on CO2 emissions will
Ta_bar and Tb_bar refer to the unit root test statistics proposed by vary with the level of globalization. For instance, study variables may
Moon and Perron (2004). The null hypothesis of the two tests H0 have different coefficients with different signs for lower and higher
assumes the existence of the unit root for all panel units. ρpool is the levels of globalization. Our results show that r = 1, suggesting that there
modified pooled estimates of the autoregressive parameter. P-values are two regimes, which can be defined according to the estimated level
are between brackets. of threshold c = 1.814 that corresponds nearly 65 as an index value of
globalization measurement. Also, the slope (γ) value (368.30) points out
a sharp transition between regime 1 and regime 2. Thus, regime 1 refers
3.3 | Results of linearity test to low globalization levels (lower than c), while regime 2 corresponds to
high globalization levels (higher than c).
The first step in the specification of the PSTR model is to test the Table 5 presents the results of the PSTR regression model. Let us
null hypothesis of linearity. Table 3 reports the results of the linear- first comment on the results concerning our key environmental regu-
ity tests. The results show that the null hypothesis r = 0 (linearity) is lation measures. We find that environment-related technological pro-
strongly rejected in favour of the alternative of logistic PSTR speci- gress and environment-related taxation contribute to CO2 emissions,
fication at the conventional level of significance. It implies that while the effect of the patent is not statistically significant in the
environmental technologies, patents, ERT, economic growth and lower regime of globalization. In other words, environment-related
globalization impact CO2 emissions differently, depending on the technological progress and taxation have an increasing effect on car-
degree of globalization. We can thus proceed to the estimation of bon emissions in the first regime. This output for environment-related
the PSTR model to investigate this relationship more deeply. This technological progress and regulations can be explained by the
result implies that the specification of the transition variable is rebound effect and the green paradox that underline how technologi-
more convincingly featured by one threshold with two extreme cal progress and regulations may result in higher CO2 emission
regimes. Once the linearity hypothesis has been tested and speci- (H. Wang & Wei, 2019). According to the rebound effect mechanism,
fied, we proceed to estimate the remaining non-linearity in the next technological progress would promote people to increase the demand
sub-section. for goods and services since it decreases their costs by requiring

TABLE 1 Results for cross-sectional

Variables LM [Prob] CDLM [Prob] CD [Prob] LMadj [Prob]
dependence
lnCO2 79.436 [0.000] 20.044 [0.000] 8.676 [0.000] 19.949 [0.000]
lnGDP 126.514 [0.000] 33.634 [0.000] 11.247 [0.000] 33.539 [0.000]
lnPTNT 113.129 [0.000] 29.770 [0.000] 10.623 [0.000] 29.675 [0.000]
lnERT 14.854 [0.021] 20.101 [0.000] 3.365 [0.000] 11.306 [0.001]
lnENT 60.163 [0.000] 14.481 [0.000] 1.856 [0.037] 14.385 [0.000]
lnKOF 121.280 [0.000] 32.124 [0.000] 32.028 [0.000] 11.008 [0.000]
6 ULUCAK ET AL.

TABLE 2 Results of Moon and

lnCO2 lnGDP lnPTNT lnERT lnENT lnKOF
Perron's (2004) panel unit root test
^
m 4 3 3 3 3 3
Ta_bar −2.711 −8.043 −3.951 −7.852 −3.145 −7.923
[0.003] [0.000] [0.000] [0.000] [0.000] [0.000]
Tb_bar −2.599 −2.655 −1.675 −3.031 −1.718 −5.117
[0.004] [0.000] [0.046] [0.001] [0.042] [0.000]

ρpool 0.969 0.989 0.881 0.575 0.785 0.856

TABLE 3 Linearity tests results

fewer sources, even though it provides the production with less emis-
Threshold variable KOF index Statistic P-values sion (Ulucak & Koçak, 2018; Zhang & Lin Lawell, 2017). On the other
H0: linear model vs. H1: PSTR model with at least one threshold variable hand, the green paradox takes attention to how inappropriate regula-
(r = 1) tions stimulate the over-extraction of resources and to increase CO2
Wald tests (LM) 14.190 0.028 emissions (Van Der Werf & Di Maria, 2012) and suggests that it is
Fisher tests (LM) 2.507 0.029 possible to display an upward trending pattern in CO2 emissions in
LR tests (LR) 15.543 0.016 response to the implementation of environmental regulation policies.

Note: LM and LR denote Lagrange multiplier and likelihood ratio tests for
Therefore, the underlying factors of the rebound effect and the green
linearity. paradox may be possible reasons to explain why people use more
resources and increase CO2 emissions in the sample countries. Similar
results are also obtained by Wang and Wei (2019) for OECD coun-
tries and emerging economies. Their study confirms the rebound
effect for technological progress and the green paradox for the envi-
TABLE 4 Remaining non-linearity results
ronmental regulation variables by using PSTR approach.
Threshold variable KOF index Statistic P-values
In the second regime with a higher degree of globalization,
H0: PSTR with r = 1 vs. H1: PSTR with at least r = 2 environment-related technological progress, environment-related tax-
Wald tests (LM) 0.803 0.902 ation, and patent mitigate CO2 emissions. Specifically, we highlight
Fisher tests (LM) 0.100 0.996 that environment-related technological progress, environment-related
LR tests (LR) 0.807 0.992 taxation and patent reduce CO2 emissions in the second regime char-
acterized by a higher degree of globalization. Based on these findings,
Note: LM and LR denote Lagrange multiplier and likelihood ratio tests for
linearity. one may claim that globalization is recognized as one of the best path-
ways to increase the effectiveness of environment-related technologi-
cal progress, environment-related taxation and patent in our panel
members. Our study sheds new light on this ongoing debate by
reporting that globalization plays a vital role in reversing the rebound
TABLE 5 Results of panel smooth transition regression model
regression effect and the green paradox in emerging countries. We find strong
evidence that countries in our sample can overcome possible unde-
Variable Regime 1 P-value Regime 2 P-value
sired outcomes of environmental regulations by increasing the level of
lnERT 2.602c 0.052 −0.351 0.783
globalization. This indicates that the effectiveness of environmental-
lnENTX 0.371b 0.025 −0.398b 0.033 related technologies and environmental charges for achieving carbon
lnPTNT 0.228 0.308 −0.299c 0.091 abatement is sensitive to the low and high globalization regimes,
lnGDP −0.168c 0.034 −0.453a 0.007 which firmly provides evidence for the non-linear globalization thresh-
lnKOF −0.011b 0.024 −0.303c 0.072 old effect of environmental technologies and environmental charges.
Constant −0.070 c
0.056 0.016 0.145 We estimate the optimal level of globalization for the sample coun-
Threshold (c) 1.814 tries and provide the following insights: we find strong evidence on
Slope (γ) 368.30 the difference in the impact of environmental regulation measures

AIC −3.360 captured by environment-related technologies, taxation and patents

on CO2 emissions.
R-squared 0.76
Shedding light on the other determinants of CO2 emissions, one
LR 4.957
may argue that the ongoing rapid economic growth in sample coun-
RSS 1.853
tries has caused serious environmental damage. However, empirical
a, b, c
Note: shows the significance level at 1%, 5% and 10%, respectively. results show that the coefficients of GDP are negative and statistically
ULUCAK ET AL. 7

significant across regimes, although the magnitude is higher in the sec- The findings of the study call for important policy implications.
ond regime. It implies that further globalization improves environmental The implementation of strict environmental regulation policies is not
awareness and promote clean growth by mitigating the growth effect of enough to reduce carbon emissions in emerging countries. The inter-
CO2 emissions. For the last exogenous variable, which is the degree of action and integration level of countries measured by the globalization
globalization, we determine the threshold value of 1.814 between the index is an important contributor in achieving a desirable result for
low globalization regime and the high globalization regime. We find that environmental sustainability. Results confirm that a higher level of
globalization is the major tool for reducing carbon emissions under both globalization empowers environmental-related technologies, patents
regimes, although the magnitude of the coefficient is higher in the sec- and taxes to decrease CO2 emissions. This mechanism most probably
ond regime of a higher degree of globalization. For the periods when a stems from increasing awareness, public pressure and demand toward
country displays a higher degree of globalization, environmental regula- establishing a green future through globalization. On the other hand,
tion policies are more efficient in mitigating CO2 emissions. Based on at a lower level of globalization, empirical findings of the study point
these findings, it is evident that the transmission channel of the globaliza- out to the green paradox and rebound effect underlining that more
tion level plays a key role in better understanding the effectiveness of CO2 emission may be released since inappropriate regulations or the
environmental regulation policies. technology may promote more resource usage and contribute to pol-
Contrary to the theoretical expectation, economic growth lowers lution. Therefore, countries with lower integration levels for global
CO2 emissions in both regimes although the magnitude is higher in the developments should expand their openness to interactions and inte-
second regime. The reason for this outcome can be explained on the gro- grations of worldwide issues and policies in terms of economic, social
unds of the diffusion and transfer of cleaner technologies through global- and environmental aspects. Considering the adverse effect of environ-
ization as is highlighted by (Copeland & Taylor, 2013) and the mental technology and taxation cautioned by the rebound effect and
enforcement of international environmental agreements such as the green paradox, they should also control overuse of resources and pol-
Kyoto Protocol, Paris agreement and the increasing environmental lution levels by implementing additional control policies such as allo-
awareness. The degree to which a country is globalized is important to cating quotas for resources, wastes and emissions to be able to
further mitigate the growth effects of CO2 emissions. Globalization may establish a sustainable development corridor. Future studies may con-
bring a set of institutional, economic, and political reforms for environ- tribute to the related literature by employing alternative environmen-
mental sustainability in emerging countries (Bilgili, Ulucak, Koçak, & tal indicators and control variables such as ecological footprint,
_Ilkay, 2020) and thus it may stimulate the effectiveness of environmental environmental sustainability index, environmental performance index
regulation policy actions and improve environmental quality. energy consumption, renewable energy technologies, total factor pro-
ductivity and also using different panel samples or single country
experiences.
4 | C O N CL U S I O N
OR CID
The effectiveness of environmental-related technologies, patents, and Recep Ulucak https://orcid.org/0000-0001-9938-0063
taxes for achieving carbon abatement has been subject to important Danish https://orcid.org/0000-0003-1046-1655
debates on the sustainable development pathway. Considering this
concern, the current study is an attempt to explore the role of such ENDNOTES
environmental factors in pollution reduction. The CO2 emission is 1
Fouquau, Hurlin, and Rabaud (2008) report that the PSTR model limits
taken as the dependent variable and a proxy for environmental pollu- the potential endogeneity bias in the estimation procedure. Based on this
argument, our estimators are robust to endogeneity bias.
tion; environmental-related technologies, environmental related taxes,
and patents are independent variables of the investigation. Various
2
“Since PSTR models allow for as many values of the impact coefficient,
lying between δ0 and δ0 + δ1 as the number of country-time observations,
econometric methodologies are employed on panel data of BRICS
some scholars only interpret the signs of δ0 and δ1 rather than their values.
countries covering the period from 1994 to 2015. Results of several 3
Russia was excluded due to data unavailability.
CSD tests confirmed the CSD across sample countries used in the
study and panel unit root test proposed by Moon and Perron (2004)
confirmed the stationarity of the study variables. Then, the PSTR RE FE RE NCE S
model was performed and robust results that reveal the effects of Abdullah, S., & Morley, B. (2014). Environmental taxes and economic
growth: Evidence from panel causality tests. Energy Economics, 42,
two regimes according to lower and higher levels of globalization
27–33. https://doi.org/10.1016/j.eneco.2013.11.013
were obtained. Recalling the rebound effect and the green paradox Acemoglu, D., Akcigit, U., Hanley, D., & Kerr, W. (2016). Transition to clean
for the lower degree of globalization, results show that technology. Journal of Political Economy, 124(1), 52–104. https://doi.
environmental-related technologies and environmental related taxes org/10.1086/684511
Albrecht, J. (2006). Green tax reforms for industrial transformation: Over-
increase CO2 emissions but the impact of the patents on CO2 emis-
coming institutional inertia with consumption taxes. Sustainable Devel-
sions is not statistically significant. However, in a higher degree of opment, 14(5), 300–311. https://doi.org/10.1002/sd.286
globalization, environmental-related variables used in the study Aydin, C., Esen, Ö., & Aydin, R. (2019). Is the ecological footprint related to
reduce pollution. the Kuznets curve a real process or rationalizing the ecological
8 ULUCAK ET AL.

consequences of the affluence? Evidence from PSTR approach. Ecolog- González, A., & Teräsvirta, T. (2006). Simulation-based finite sample linear-
ical Indicators, 98, 543–555. https://doi.org/10.1016/J.ECOLIND. ity test against smooth transition models*. Oxford Bulletin of Economics
2018.11.034 and Statistics, 68, 797–812. https://doi.org/10.1111/j.1468-0084.
Bilgili, F., & Ulucak, R. (2018). The nexus between biomass – Footprint and 2006.00457.x
sustainable development. Reference Module in Materials Science and González, A., Teräsvirta, T., van Dijk, D., & Yang, Y. (2017). Panel smooth
Materials Engineering, 2, 175–192. https://doi.org/10.1016/B978-0- transition regression models. Paper presented at the SSE/EFI Working
12-803581-8.10600-9 Paper Series in Economics and Finance.
Bilgili, F., Ulucak, R., Koçak, E., & _Ilkay, S. Ç. (2020). Does globalization González, A., Teräsvirta, T., & Van Djik, D. (2005). Panel smooth transition
matter for environmental sustainability? Empirical investigation for regression models. Paper presented at the SSE/EFI Working Paper
Turkey by Markov regime switching models. Environmental Science and (No. 604).
Pollution Research, 27(1), 1087–1100. https://doi.org/10.1007/ Grossman, G. M., & Krueger, A. B. (1995). Economic growth and the envi-
s11356-019-06996-w ronment. The Quarterly Journal of Economics, 110(2), 353–377. https://
Breusch, T. S., & Pagan, A. R. (1980). The Lagrange multiplier test and its doi.org/10.2307/2118443
applications to model specification in econometrics. The Review of Eco- Guo, L., Qu, Y., & Tseng, M. L. (2017). The interaction effects of environ-
nomic Studies, 47(1), 239. https://doi.org/10.2307/2297111 mental regulation and technological innovation on regional green
Calderón, S., Camilo, A., María, A., Arango, S., Calvin, K., Kober, T., … growth performance. Journal of Cleaner Production, 162, 894–902.
Fisher-vanden, K. (2016). Achieving CO2 reductions in Colombia: https://doi.org/10.1016/j.jclepro.2017.05.210
Effects of carbon taxes and abatement targets. Energy Economics, 56, Halkos, G. E., & Papageorgiou, G. J. (2018). Pollution, environmental taxes
575–586. https://doi.org/10.1016/j.eneco.2015.05.010 and public debt: A game theory setup. Economic Analysis and Policy,
Cheng, C., Ren, X., Wang, Z., & Yan, C. (2019). Heterogeneous impacts of 58, 111–120. https://doi.org/10.1016/j.eap.2018.01.004
renewable energy and environmental patents on CO2 emission - evi- Hansen, B. E. (1999). Threshold effects in non-dynamic panels: Estimation,
dence from the BRIICS. Science of the Total Environment, 668, testing, and inference. Journal of Econometrics, 93(2), 345–368.
1328–1338. https://doi.org/10.1016/j.scitotenv.2019.02.063 https://doi.org/10.1016/S0304-4076(99)00025-1
Copeland, B. R., & Taylor, M. S. (2013). Trade and the environment: Theory Hashmi, R., & Alam, K. (2019). Dynamic relationship among environmental
and evidence. Princeton, NJ: Princeton University Press. regulation, innovation, CO2 emissions, population, and economic
Danish, Baloch, M. A., & Bo, W. (2019). Analyzing the role of governance growth in OECD countries: A panel investigation. Journal of Cleaner
in CO2 emissions mitigation: The BRICS experience. Structural Change Production, 231, 1100–1109. https://doi.org/10.1016/j.jclepro.2019.
and Economic Dynamics, 51, 119–125. https://doi.org/10.1016/j. 05.325
strueco.2019.08.007 He, P., Ya, Q., Chengfeng, L., Yuan, Y., & Xiao, C. (2019). Nexus between
Danish, Ulucak, R., Khan, S. U., Baloch, M. A., & Li, N. (2019). Mitigation environmental tax, economic growth, energy consumption, and carbon
pathways toward sustainable development: Is there any trade-off dioxide emissions: Evidence from China, Finland, and Malaysia based
between environmental regulation and carbon emissions reduction? on a panel-ARDL approach. Emerging Markets Finance and Trade, 1–15.
Sustainable Development, sd.2032. https://doi.org/10.1002/sd.2032 https://doi.org/10.1080/1540496X.2019.1658068
Danish, & Ulucak, R. (2020). How do environmental technologies affect Jiang, Y., Zhou, Z., & Liu, C. (2019). Does economic policy uncertainty mat-
green growth? Evidence from BRICS Economies. Science of The Total ter for carbon emission? Evidence from US sector-level data. Environ-
Environment, 10, 136504, 1–7. https://doi.org/10.1016/j.scitotenv. mental Science and Pollution Research, 26(24), 24380–24394. https://
2020.136504 doi.org/10.1007/s11356-019-05627-8
Danish, Wang, B., & Wang, Z. (2018). Imported technology and CO2 emis- Jordaan, S. M., Romo-Rabago, E., McLeary, R., Reidy, L., Nazari, J., &
sion in China: Collecting evidence through bound testing and VECM Herremans, I. M. (2017). The role of energy technology innovation in
approach. Renewable and Sustainable Energy Reviews, 82, 4204–4214. reducing greenhouse gas emissions: A case study of Canada. Renew-
https://doi.org/10.1016/j.rser.2017.11.002 able and Sustainable Energy Reviews, 78, 1397–1409. https://doi.org/
Danish, & Wang, Z. (2019). Does biomass energy consumption help to 10.1016/j.rser.2017.05.162
control environmental pollution? Evidence from BRICS countries. Sci- Lin, B., & Jia, Z. (2019). How does tax system on energy industries affect
ence of the Total Environment, 670, 1075–1083. https://doi.org/10. energy demand, CO2 emissions, and economy in China? Energy Eco-
1016/j.scitotenv.2019.03.268 nomics, 84, 104496. https://doi.org/10.1016/j.eneco.2019.104496
De Vos, I., & Westerlund, J. (2019). On CCE estimation of factor- Lu, Y., Wang, Y., Zhang, W., Hubacek, K., Bi, F., Zuo, J., … Xue, W. (2019).
augmented models when regressors are not linear in the factors. Eco- Provincial air pollution responsibility and environmental tax of China
nomics Letters, 178, 5–7. https://doi.org/10.1016/j.econlet.2019. based on interregional linkage indicators. Journal of Cleaner Production,
02.001 235, 337–347. https://doi.org/10.1016/j.jclepro.2019.06.293
Devi, S., & Gupta, N. (2019). Effects of inclusion of delay in the imposition Luukkonen, R., Saikkonen, P., & TerÄsvirta, T. (1988). Testing linearity
of environmental tax on the emission of greenhouse gases. Chaos, Soli- against smooth transition autoregressive models. Biometrika, 75(3),
tons and Fractals, 125, 41–53. https://doi.org/10.1016/j.chaos.2019. 491–499. https://doi.org/10.1093/biomet/75.3.491
05.006 Mardones, C., & Baeza, N. (2018). Economic and environmental effects of
Du, G., Liu, S., Lei, N., & Huang, Y. (2018). A test of environmental Kuznets a CO2 tax in Latin American countries. Energy Policy, 114, 262–273.
curve for haze pollution in China: Evidence from the penal data of https://doi.org/10.1016/j.enpol.2017.12.001
27 capital cities. Journal of Cleaner Production, 205, 821–827. https:// Moon, H. R., & Perron, B. (2004). Testing for a unit root in panels with
doi.org/10.1016/j.jclepro.2018.08.330 dynamic factors. Journal of Econometrics, 122(1), 81–126. https://doi.
Fouquau, J., Hurlin, C., & Rabaud, I. (2008). The Feldstein-Horioka puzzle: org/10.1016/J.JECONOM.2003.10.020
A panel smooth transition regression approach. Economic Modelling, 25 Niu, T., Yao, X., Shao, S., Li, D., & Wang, W. (2018). Environmental tax
(2), 284–299. https://doi.org/10.1016/j.econmod.2007.06.008 shocks and carbon emissions: An estimated DSGE model. Structural
Ganda, F. (2019). The impact of innovation and technology investments Change and Economic Dynamics, 47, 9–17. https://doi.org/10.1016/j.
on carbon emissions in selected organisation for economic co- strueco.2018.06.005
operation and development countries. Journal of Cleaner Production, Pesaran, M. H. (2004). General diagnostic tests for cross
217, 469–483. https://doi.org/10.1016/j.jclepro.2019.01.235 section dependence in panels (No. 1233).
ULUCAK ET AL. 9

Pesaran, M. H., Ullah, A., & Yamagata, T. (2008). A bias-adjusted LM test Wang, H., & Wei, W. (2019). Coordinating technological progress and
of error cross-section independence. The Econometrics Journal, 11(1), environmental regulation in CO2 mitigation: The optimal levels for
105–127. https://doi.org/10.1111/j.1368-423X.2007.00227.x OECD countries & emerging economies. Energy Economics, 104510.
Sarkodie, S. A., & Strezov, V. (2019). A review on environmental Kuznets https://doi.org/10.1016/j.eneco.2019.104510
curve hypothesis using bibliometric and meta-analysis. Science of the Wang, J., Wang, K., Shi, X., & Wei, Y. M. (2019). Spatial heterogeneity and driv-
Total Environment, 649, 128–145. https://doi.org/10.1016/j.scitotenv. ing forces of environmental productivity growth in China: Would it help to
2018.08.276 switch pollutant discharge fees to environmental taxes? Journal of Cleaner
Shi, H., Qiao, Y., Shao, X., & Wang, P. (2019). The effect of pollutant char- Production, 223, 36–44. https://doi.org/10.1016/j.jclepro.2019.03.045
ges on economic and environmental performances: Evidence from Wang, Y., Yan, W., & Komonpipat, S. (2019). How does the capacity utili-
Shandong Province in China. Journal of Cleaner Production, 232, zation of thermal power generation affect pollutant emissions? Evi-
250–256. https://doi.org/10.1016/j.jclepro.2019.05.272 dence from the panel data of China's provinces. Energy Policy, 132,
Tiba, S. (2019). Modelling the nexus between resources abundance and 440–451. https://doi.org/10.1016/j.enpol.2019.06.010
economic growth: An overview from the PSTR model. Resources Policy, Yuan, X., Zhang, M., Wang, Q., Wang, Y., & Zuo, J. (2017). Evolution analy-
64, 101503. https://doi.org/10.1016/j.resourpol.2019.101503 sis of environmental standards: Effectiveness on air pollutant emis-
Ulucak, R., Koçak, E. (2018). Rebound effect for energy consumption: The sions reduction. Journal of Cleaner Production, 149, 511–520. https://
case of Turkey;1–10. Retrieved from https://amsterdam2018. doi.org/10.1016/j.jclepro.2017.02.127
econworld.org/papers/Ulucak_Kocak_Rebound.pdf Zhang, J., & Lin Lawell, C.-Y. C. (2017). The macroeconomic rebound effect
Ulucak, R., Yücel, A. G., & Koçak, E. (2019). The process of sustainability: in China. Energy Economics, 67, 202–212. https://doi.org/10.1016/J.
From past to present. In B. Ozcan & _I. Ozturk (Eds.), Environmental ENECO.2017.08.020
Kuznets Curve (EKC) London: Academic Press Elsevier. (pp. 37–53). Zhang, Z., Hu, W., & Chang, T. (2019). Non-linear effects of P2P lending
https://doi.org/10.1016/B978-0-12-816797-7.00005-9 on bank loans in a panel smooth transition regression model. Interna-
Van Der Werf, E., & Di Maria, C. (2012). Imperfect environmental policy tional Review of Economics and Finance, 59, 468–473. https://doi.org/
and polluting emissions: The green paradox and beyond. International 10.1016/j.iref.2018.10.010
Review of Environmental and Resource Economics, 6, 153–194. https://
doi.org/10.1561/101.00000050
Vatn, A. (2015). Markets in environmental governance: From theory to
practice. Ecological Economics, 117, 225–233. https://doi.org/10. How to cite this article: Ulucak R, Danish, Kassouri Y. An
1016/j.ecolecon.2014.07.017
assessment of the environmental sustainability corridor:
Wang, C., Zhang, L., Zhou, P., Chang, Y., Zhou, D., Pang, M., & Yin, H.
(2019). Assessing the environmental externalities for biomass- and Investigating the non-linear effects of environmental taxation
coal-fired electricity generation in China: A supply chain perspective. on CO2 emissions. Sustainable Development. 2020;1–9.
Journal of Environmental Management, 246, 758–767. https://doi.org/ https://doi.org/10.1002/sd.2057
10.1016/j.jenvman.2019.06.047