Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Energy Economics 92 (2020) 104966

Contents lists available at ScienceDirect

Energy Economics

journal homepage: www.elsevier.com/locate/eneeco

The pollution concern in the era of globalization: Do the contribution of


foreign direct investment and trade openness matter?
Sofien Tiba a,⁎, Fateh Belaid b
a
Faculty of Economics and Management, University of Sfax, Tunisia
b
Lille Catholic University, Faculty of Management, Economics & Sciences, UMR 9221-LEM-Lille Économie Management, 60 Boulevard Vouban, Lille 59016, France

a r t i c l e i n f o a b s t r a c t

Article history: This analysis aims to treat the contribution of the foreign direct investment inflows and trade openness to envi-
Received 31 March 2019 ronmental degradation. Drawing on the data for 27 African countries over the period from 1990 to 2013, we de-
Received in revised form 21 September 2020 velop an empirical model based on a set of panel methods. This approach includes Westerlund (2007) panel
Accepted 26 September 2020
cointegration technic, Common Correlated Effects Mean Group (CCE-MG) estimation procedure, and the Granger
Available online 6 October 2020
causality test. In terms of long-term causality, our findings reveal bidirectional long-term causality between CO2/
Keywords:
NOx emissions, GDP, trade openness, and foreign direct investment. Besides, in the short-run, the results recorded
Foreign direct investment a unidirectional causality running from GDP to CO2, and from FDI to CO2. However, bidirectional causality nexus
Trade openness was detected between trade openness and CO2. Additionally, the findings portray bidirectional causality between
CO2 emissions GDP and NOx and between trade openness and NOx, while a unidirectional causality from foreign direct invest-
NOx emissions ment to NOx. Our analysis substantiates the importance of the foreign direct investment inflows and trade open-
Environmental quality ness in mitigating the adverse effect of heavy pollutant activities and resolving the environmental puzzle
(growing without polluting) consistent with the Millennium Development Goals (MGDs) of the UN background.
© 2020 Elsevier B.V. All rights reserved.

1. Introduction nexus. Theoretically, this nexus is known as the “Environmental


Kuznets Curve”.
Global warming and climate change constitute the main concern of The EKC theory implies the existence of a monotonic nexus among
several discourses as a holistic threat, especially, this globalized era the wealth and environmental indicators. This relationship has been
with the demolition of borders and the reduction of spatial heterogene- firstly treated by Grossman and Krueger (1991), where their results
ity between nations. As a response to this worldwide threat, and seen as found an inverted U-shape link among economic growth and income
an emergency, the 1997 Kyoto protocol aimed to reduce greenhouse inequality, which is proved by Kuznets (1955). Nevertheless, to validate
gases that greatly contribute to climate change. The resulting recom- the EKC hypothesis, a quadratic relationship is required. The theoretical
mendation of the protocol suggested the reduction of the pollutant background hidden behind the EKC hypothesis is justified by the transi-
emissions to 5.2% lower than the pollutant emissions level of 1990 for tion process of the economy during the development pathway. Eco-
the spanning time 2008–2012. Further, the carbon dioxide emissions nomic growth and pollution level rise conjointly during the first stages
(CO2) contribute to climate change by around 58.8% (World Bank, of economic development. At this level, the environmental quality has
2014). Besides, the agreement of Paris 2016, which aims to curb the been considered as a luxury good, while the households are more
GHG levels, goes in the same spirit of the Kyoto protocol to fight climate attracted to the necessary goods. Furthermore, in more advanced stages
change and global warming issues. For this reason, each member state of development, and after reaching a specific threshold, the environ-
signed this agreement, should show his intuition, ultimate engagement, ment becomes increasingly claimed. Based on the findings of
and commitment to the global warming threat. In this context, the Dasgupta et al. (2002), at further stages of development, governments
nexus between economic growth and environmental quality is widely should set strict regulations to make a tradeoff between sustained eco-
stressed empirically with conflicting results and without reaching any nomic growth and sustainable ecological quality. Several studies have
unanimous consensus about the existence or not of a significant examined the EKC hypothesis1 by applying diverse econometric tech-
niques (e.g. Ang (2007, 2008); Lee and Lee (2009); Akbostanci et al.
(2009); He and Richard (2010); Iwata et al. (2011); Halicioglu (2009);
⁎ Corresponding author.
1
E-mail addresses: sofienetiba@gmail.com (S. Tiba), fateh.belaid@univ-catholille.fr For an exhaustive theory listing on the EKC hypothesis, please see Tiba and Omri
(F. Belaid). (2016).

https://doi.org/10.1016/j.eneco.2020.104966
0140-9883/© 2020 Elsevier B.V. All rights reserved.
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Jalil and Mahmud (2009); Fodha and Zaghdoud (2010); Lau et al. investment and trade openness to detect the possible quadratic nexus
(2014); Sahli and Ben Rejeb (2015)). Furthermore, many events have in contradiction of the most of studies which supposed a linear relation-
been undertaken by the United Nations Organization about the emerg- ship. Second, compared to the previous studies, our paper attempts to
ing of the sustainability issue as the ultimate aim which characterized employ the most recent technics of the second-generation estimators
this current millennium. In the same pathway, the highlights of Tiba in panel data that take into consideration the cross-sectional depen-
et al. (2015) pointed out the importance of foreign direct investment in- dence and slope homogeneity observed in the global panel. Specifically,
flows, and trade in environmental sustainability is recognized. Hence, we investigate the quadratic relationship between the environmental
these two factors are considered as a “driving force lies behind the success quality (which proxied by two measures: the CO2 and NOx), foreign di-
story of technologies transfer and providing the additional resources and rect investment, and trade openness. To achieve the main aim, this
capacities to build the structures of development success”. Further, it study employs the Westerlund (2007) panel cointegration methodol-
seems to be crucial to integrate the foreign direct investment inflows ogy to check the presence of long-run relationships. As well as, we use
and trade openness as keys determinants of the environmental sustain- the Cross Correlated Effects (CEE) and the Common Correlated Effects
ability issue. Besides, many scholars' undertaken humble efforts to ex- Mean Group (CCE-MG) estimation procedures, and the Granger causal-
plain the origin of the climate change phenomenon and how it can be ity test for 27 African countries. The empirical analysis covers the period
avoided. Hence, the sustainability theory reveals that this dilemma is 1990–2013.
closely related to the resources or assets portfolio management's way The rest of this article is structured as follows: Section 2 portrays the
of the economy. On the other hand, the re-emergence of the ecological model construction and data. Section 3 records the empirical facts and
dilemma was seen as a stylized fact of outcomes of the Rio Earth Summit discussions. Section 4 concludes the study and provides policy
2012. Accordingly, the sustainability theory underlined the results gen- implications.
erated by polluting activities and high environmental degradation level
for maintaining prosperity expansion. There is a robust causality among 2. Data and modeling construction
these two streams, which appeared as a contradictory but seemingly ro-
bust fact of a significant negative link between economic development 2.1. Data
and ecological sustainability. Other contributions of the economic the-
ory have stressed the crucial roles played by the foreign direct invest- Our study employs annual data related to 27 African countries,
ment inflows and trade openness in terms of mitigating the adverse namely: Algeria, Angola, Benin, Botswana, Cameroon, Congo Demo-
effect of heavy pollutant activities and resolving the environmental puz- cratic Republic, Congo Republic, Cote d'Ivoire, Egypt, Eritrea, Ethiopia,
zle (growing the economy without polluting the environment) consistent Gabon, Ghana, Kenya, Libya, Morocco, Mozambique, Namibia, Nigeria,
with the Millennium Development Goals (MDGs) of the UN Senegal, South Africa, Sudan, Tanzania, Togo, Tunisia, Zambia, and
background. Zimbabwe for the period 1990–2013.2 The data are sourced from the
In this context, the impact of foreign direct investment inflows on World Development Indicators (WDI 2018). The present study uses
the environmental factor has been widely stressed over the last decades. the per capita real GDP in constant 2010 U.S. dollars to measure eco-
Several surveys have been investigated the impact of foreign direct in- nomic growth. A per capita CO2 emission in metric tons and NOx the Ni-
vestment on the pollutant emissions at the individual and global scale, trous oxide emissions in thousand metric tons of CO2 equivalent are
without reaching any anonymous consensus about the existence (posi- used to measure environmental degradation. Trade openness
tive or negative) or absence of an effect. Lee (2009), Zhang (2011), Pao (T) defined as export plus import divided by population to obtain per
and Tsai (2011), Al-mulali (2012), Hitam and Borhan (2012), Kivyiro capita foreign trade. Foreign direct investment net inflows (FDI) as a
and Arminen (2014) Lee and Brahmasrene (2014), and Baek (2016), share of GDP multiplied by GDP and divided by population to obtain
among others, reveal the existence of a negative impact of FDI on envi- per capita FDI. Financial development (FD) is defined as domestic credit
ronmental quality using various techniques. In contrast, Tamazian et al. provided to the private sector as a share of GDP multiplied by GDP and
(2009), Al-mulali and Tang (2013), Shahbaz et al. (2013), Sbia et al. divided by population to obtain per capita financial development. And
(2014), and Zhang and Zhou (2016) treated this issue recorded the energy consumption (EC) is calculated by kg of oil equivalent per capita.
fact that the foreign direct investment leads to decreasing the pollution It's worth highlighting that all the variables are expressed in the loga-
level. However, Perkins and Neumayer (2009), Atici (2012), and Lee rithmic form.
(2013) reveal the absence of any signification effect or relationship be-
tween foreign direct investment and environmental quality. Even more, 2.2. Model specification
several investigations have been treated the impact of trade openness
on the environmental quality at the individual and global scale, without The EKC theory is founded on the research carried by Simon
reaching any unanimous consensus about the existence (positive or Kuznets, Nobel Laureate in Economics in the 1950s. Kuznets indicated
negative) or absence of a significant effect. Baek et al. (2009), an inverted U-shaped linkage among per capita output and income in-
Copeland and Taylor (2003), and Tiba et al. (2015), Tiba and Frikha equality and shed light on the birth of EKC theory. In the early 1990s,
(2018), among others, reveal the existence of a significant positive im- many contributions stated that there is, also, a similar linkage among
pact of trade openness on environmental quality through the use of var- the output level and the environment, and recognized that such anal-
ious techniques. In contrast, Kellenberg (2009), Managi and Kumar ysis is called EKC analysis (e.g. Shafik and Bandyopadhyay, 1992;
(2009), and Boulatoff and Jenkins (2010) have treated this issue, and Panayotou, 1993; Grossman and Krueger, 1995. Since then, many sur-
they recorded the fact that trade openness leads to decreasing the pol- veys have been carried out on various pollution measurements and in
lution level. numerous samples with different economic development paths. The
Even more, the EKC theory supposed a quadratic relationship be- papers that investigated the EKC hypothesis used various environ-
tween the environmental proxy and the output factor. Following mental damage proxies, like carbon dioxide emissions (CO2), nitrous
Apergis and Payne, 2009; Llorca and Meunié, 2009; Lean and Smyth, oxide emissions (N2O), Total Suspended Particulate (TSP), methane
2010; Du et al., 2012; Tiwari et al., 2013; Cho et al., 2014; Tiba et al., emissions (CH4), sulfur dioxide emissions (SO2), and water waste
2015; Tiba and Frikha (2019a, 2019b, 2019c), we include the FDI and (e.g. Apergis and Payne, 2009; Llorca and Meunié, 2009; Lean and
trade openness as determinant variables of the pollution level. Our Smyth, 2010; Du et al., 2012; Tiwari et al., 2013; Cho et al., 2014).
study seeks to contribute to the Environmental Kuznets Curve thesis
by analyzing the contribution of foreign direct investment and trade 2
The selection of variables, sample, and the period in this study are chosen following
to pollution and taking into consideration the square of foreign direct the economic theory and data availability.

2
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Additionally, these papers applied numerous economic indicators,


such as industrial output, per capita income, trade openness, energy
use, foreign direct investments, and financial development. Consistent
with the EKC hypothesis, the majority of the previous papers
employed the output and its square. These two proxies are included
to verify that if an inverted U-shaped link among the environmental
degradation measure and income with its square has existed, thus
the EKC assumption is well verified. Moreover, the majority of the
existing literature confirmed the presence of the EKC assumption in
high-and-middle-income countries.
The extensive conventional EKC literature uses various forms where
findings could be evaluated with the significance of the parameters
computed and the presence or absence of a turning point (Agras and
Chapman, 1999). Following Tiba et al. (2015), we include other funda-
mental aspects that can significantly affect the quality of environments,
such as foreign direct investment inflows and trade openness. More-
over, Tiba et al.'s pointed out that these two aspects are considered as
a “driving force lies behind the success story of technologies transfer and
providing the additional resources and capacities to build the structures
of development success”. Hence, trade openness and foreign direct in-
vestment play a great role in terms of improving environmental quality.
Also, to validate the inverted U-shaped (for the CO2 and NOx) of the EKC
Fig. 1. Modeling strategy.
theory, we use the energy consumption and financial development
(see: Apergis and Payne, 2009; Llorca and Meunié, 2009; Sadorsky,
2009; Lean and Smyth, 2010; Du et al., 2012; Jayanthakumaran et al.,
2012; Shahbaz et al., 2013; Tiwari et al., 2013; Cho et al., 2014; Tiba 2.4. Panel unit root tests
et al., 2015). So the EKC specification can be expressed as follow:
After treating the cross-sectional dependency question, the investi-
gation of the integration properties of the data is a fundamental step
Eit ¼ α 0i þ α 1i GDPit þ α 2i GDP2it þ α 3i FDIit þ α 4i FDI2it þ α 5i Tit in the panel data framework. The econometric literature proposed two
þ α 6i T2it þ α 7i ECit þ α 8i EC2it þ α 9i FDit þ α 10i FD2it þ ξit ð1Þ generations of the unit-roots testing approaches. Before running the re-
gression, it should first investigate the order of integration of the vari-
ables. Hence, these tests tackle as a null hypothesis that all variables
Where i, t, α0i and ξ indicate respectively the sample, the period, the are assumed to contain a unit root against the alternative one.
specific effect, and the error term which is supposed normally distrib- The econometric literature proposed two generations of the unit-
uted. The coefficientsα1i;α2i;α3i;α4i; α5i;α6i;α7i; α8i;α9i and α10i portray roots testing approaches. The first is based on the cross-sectional inde-
the long-run elasticities of the GDP (and it's squared), FDI (and it's pendence hypothesis and includes the contributions of Maddala and
squared), T (and it's squared), EC (and it's squared), and FD (and it's Wu (1999), Choi (2001), and Hadri (2000). Whereas, various unit-
squared) with respect to the environmental proxy, respectively. It root tests have been developed, in response to the need for tests
should be noted that the inclusion of the squared of each variable to which relax the cross-sectional independence assumption and allows
control the possible existence of non-linear relationships. for cross-sectional dependence. The second generation includes the
Eq. (1) suggests that the income level (GDP), his squared (GDP2), contributions of Bai and Ng (2004), Moon and Perron (2004), Smith
foreign direct investment (FDI), his squared (FDI2), trade openness et al. (2004), and Pesaran (2007).
(T), his squared (T2), energy consumption (EC), his squared (EC2), the In this current study, we use various tests, including (i) first-
financial development (FD), and his squared (FD2) that can potentially generation tests: LLC (Levin et al. (2002)), Breitung (2001) and (ii)
determine the level of environmental degradation (e.g. Lee, 2013; second-generation tests: IPS (Im et al. (2003)), Fisher-PP, Fisher-ADF
Shahbaz et al., 2013; Tiba et al., 2015). test, Pesaran (2007), and CADF test (Cross Augmented Dickey-Fuller).
Indeed, our modeling strategy schematically is summarized as Fig. 1
portrayed below.
2.5. Panel cointegration tests

2.3. Cross-section dependency Based on the econometric theory requirements, the following step
after exploring the integration order properties of the data is to investi-
As one of the major important questions in panel data and that gate the existence of a long-run relationship. For this purpose, we
should be treated is the cross-sectional dependence question (e.g. De employed the Westerlund (2007) panel cointegration procedure.
Hoyos and Sarafidis (2006)). Consequently, various tests are proposed Westerlund's cointegration test provides several properties compared
to examine the cross-sectional dependency problem in the panel data to other cointegration methodologies. First, this methodology enables
framework such as the Friedman (1937), Breusch and Pagan (1980), for heterogeneity in the short- and long-term cointegration dynamics
Frees (1995), and Pesaran (2004) tests. Indeed, due to the spatial corre- (Persyn and Westerlund, 2008). Second, this approach has the ability
lations, spillover effects, economic distance, omitted global variables, to overcome the issue of cross-sectional dependence. Lastly, this ap-
and common unobserved shocks, the issue of the cross-section depen- proach proposes an optional bootstrap method that permits various
dency should be treated. The neglecting of the cross-section depen- repetitions of the cointegration tests. Specifically, this methodology as-
dence generates biased estimates and misleading inference. For this sumes that the group-mean tests (Gt and Ga) which admit the alterna-
reason, we employ the cross-section dependency test (CD) of Pesaran tive hypothesis with the existence of at least one long-run relationship,
(2004). The Pesaran's test could be used in a large set of models with even, the panel tests (Pt and Pa) assume as an alternative hypothesis
small/large N and T (e.g. Phillips and Sul, 2003). that the panel is cointegrated (Persyn and Westerlund, 2008).

3
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

2.6. Estimating the long-term relationship the heterogeneity issue in our sample, supporting the nonexistence of
the cross-sectional dependency among the series.
Phillips and Sul (2007) have claimed that the presence of the issues
of cross-sectional dependence, heteroskedasticity, and serial correla- 3.2. Unit root-tests
tion, leads to providing misleading inference and even inconsistent esti-
mators. To overcome these issues, Pesaran (2007) has proposed the To investigate the integration properties of the data, we employed
Common Correlated Effects (CCE), which has been improved later by the first- and second-generation unit root-tests.
Kapetanios et al. (2011) and Chudik et al. (2011). The Pesaran's The highlights of the three-panel unit-roots, which we applied in
(2007) CCE estimator shows a set of properties. Indeed, the CCE estima- this current study, to investigate the order of integration of the series,
tor does not include the estimation of unobserved common factors and are recorded in Table 2. Indeed, the recorded results confirm that all
factor loadings. It enables for unobserved shocks to be correlated with variables have a unit-root at the level rejecting the null hypothesis
exogenous regressors. Besides, this estimator is still consistent even of stationarity at level, which proves that all the variables are inte-
with the presence of correlation in errors, unit-roots in the variables, grated of order one (i.e., all the variables are stationary at the first dif-
and possible cross-sectional dependence of the regressors with the un- ference I (1)).
observed shocks (Chudik et al., 2011). For this reason, we use the Com-
mon Correlated Effects Mean Group (CCEMG) estimator of Pesaran 3.3. The cointegration tests
(2007) to determine the longterm estimators, which assume the
cross-sectional dependence issue. Furthermore, Eberhardt (2012) ad- Table 3 displays the findings of Westerlund's panel cointegration
mitted the robustness of the CCEMG approach with a finite and an infi- testing for the two models (CO2 and NOx were taken as dependent var-
nite number of variables. Additionally, the Pesaran's (2007) estimator is iables). The empirical highlights of Westerlund's test reveal that Group-
robust to non-stationary common variables (Kapetanios et al., 2011). t and Panel-a that Group-t and Panel-t-test statistics are statistically sig-
nificant, and significantly reject the null hypothesis of no cointegration.
2.7. The Granger causality test This confirms the long-run cointegration between the variables for both
models.
The following step after the cointegration issue is to detect the sense
of causality between the variables (Engle and Granger, 1987). For this
3.4. Long-term estimation
purpose, we use the Pooled Mean Group (PMG) estimator of Pesaran
et al. (1997), which assumes the combination of pooling and averaging
The regression findings of the Pesaran (2007) CCE-MG estimations
of coefficients. This approach enables us to avoid the issue of parameter
for the CO2 model are reported in Table 4. The values of the coefficients
heterogeneity. The fundamental contribution of PMG is that it allows
belong to the regressors could be assumed as short-run elasticity. At the
the heterogeneity in short-run coefficients, the speed of adjustment to
1% and 5% level of significance, the majority of the coefficients are statis-
the long-run equilibrium values, and error variances, however, the
tically significant. From Table 4, the estimated coefficients of the CO2
long-term slope coefficients are assumed to be homogeneous across
model specification are 0.296, −0.145, 0.398, −0.368, 0.214, −1.576,
countries. Besides, the PMG estimation approach proves its robustness
0.078, −0.091, 1.719 and − 0.839 for GDP, GDP2, FDI, FDI2, T, T2, FD,
in the presence of outliers and the lag selection issues.
FD2, EC and EC2, respectively. Then, the income factor (his squared)
has a significantly positive (negative) impact on the CO2 emissions at
3. Empirical highlights and discussion
a 1% level of significance. Implying the fact that an inverted U-shaped
is well obtained supporting the EKC hypothesis.3 Regarding the coeffi-
3.1. Cross-section dependency
cient of the foreign direct investment inflows (FDI) where it exerts a
positive influence on the CO2 emissions level at a 5% level of signifi-
The starting point of our empirical analysis is the investigation of the
cance. Hence, the magnitude of 0.398 implies that an increase in the
possible existence of cross-section correlation issues through the use of
FDI level leads to raising the pollution level (CO2 emissions) by around
the Pesaran (2004) test. As illustrated in Table 1, our empirical findings
0.4%. The increase of the CO2 emissions is due to weak environmental
highlight the rejection of the null hypothesis of cross-sectional depen-
regulations background in the African countries that will attract the pol-
dence at the 1% significant level. The results confirm the absence of
lutants investment industries. From this point of view, African countries
have a comparative advantage in polluting industries, thus, the FDI ex-
plained the rise of the pollution level. Furthermore, the FDI squared
Table 1
showed a negative impact on the CO2 emissions indicating that the ex-
The results of cross-sectional dependence test.
istence of a negative relationship. This means that there is evidence of a
Variables Test statistics quadratic nexus between the foreign direct investment inflows and the
Pesaran (2004) CD-test environmental proxy (CO2) in contradiction with the majority of studies
Test statistic P value
which they supposed the linearity of such relationship. Regarding the
trade openness factor, where the magnitude of 0.214 implies that a 1%
lCO2 5.658⁎ 0.0000
rise in trade openness leads to raising CO2 emissions by around
lNOx 5.981⁎ 0.0000
lGDP 33.715⁎ 0.0000 0.214%. This implies that an increase in the openness process generates
lFDI 23.987⁎ 0.0000 an increase in pollutant emissions. Indeed, international trade with its
lT 26.691⁎ 0.0000 negative externalities in terms of importations of the goods that con-
lEC 10.779⁎ 0.0000 sume more energy and contribute heavily to the greenhouse gas due
lFD 16.607⁎ 0.0000
lGDP2 33.715⁎ 0.0000
to the weak environmental regulations in the African countries. Further-
lFDI2 27.369⁎ 0.0000 more, the trade openness squared showed a negative impact on the CO2
lT2 26.691⁎ 0.0000 emissions indicating that the existence of a negative relationship. This
lEC2 11.056⁎ 0.0000
lFD2 17.481⁎ 0.0000 3
An appreciation in the income factor leads to increasing the pollution level in the ear-
Notes: Under the null hypothesis of cross-sectional independence, the Pesaran CD statis- lier stages of development before reaching some specified threshold where the pollution
tics is distributed as a two-tailed normal standard. ⁎. ⁎⁎ are the Statistical significance at level decline confirming an inverted U-shaped. Indeed, in the long-run, the environment
the 1, and 5% level, respectively. has been considered as a luxury good and greatly seeking as confirmed by the EKC theory.

4
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Table 2
Panel unit root tests results.

Test Breitung LLC IPS Fisher-ADF Fisher-PP

Level Δ Level Δ Level Δ Level Δ Level Δ

lCO2 −0.057 −6.709⁎ −0.629 −7.587⁎ −0.958 −11.853⁎ 68.449⁎⁎⁎ 240.403⁎ 142.314⁎ 684.735⁎
(0.4773) (0.0000) (0.2645) (0.0000) (0.1690) (0.0000) (0.0892) (0.0000) (0.0000) (0.0000)
lNOx −2.345⁎ −11.732⁎ 2.2981 −11.451⁎ 4.055 (1.0000) −10.147⁎ 18.458 202.553⁎ 18.2976 375.463⁎
(0.0095) (0.0000) (0.9892) (0.0000) (0.0000) (1.0000) (0.0000) (1.0000) (0.0000)
lGDP 3.042 (0.9988) −2.869⁎ −4.966⁎ −65.897⁎ −1.650⁎ −5.249⁎ 286.541⁎ 410.218⁎ 65.0259 251.512⁎
(0.0021) (0.0000) (0.0000) (0.0.0494) (0.0000) (0.0000) (0.0000) (0.1447) (0.0000)
lFDI −3.816⁎ −11.670⁎ −9.711⁎ −15.453⁎ −6.061⁎ −18.671⁎ 163.843⁎ 385.919⁎ 200.450⁎ 1650.48⁎
(0.0001) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
lT −1.059 −5.967⁎ −0.444 −11.825⁎ 1.293 (0.9021) −11.691⁎ 39.908 239.792 38.228 438.497
(0.1447) (0.0000) (0.3282) (0.0000) (0.0000) (0.9237) (0.0000) (0.9486) (0.0000)
lEC 0.992 (0.8395) −1.148⁎ −6.972⁎ −38.977⁎ −5.345⁎ −17.121⁎ 414.565 688.074 90.457 466.781
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0014) (0.0000)
lFD 2.693 (0.9965) −2.850 −2.933⁎ −10.326⁎ 0.030 (0.5122) −8.907⁎ 67.712⁎⁎⁎ 185.701⁎ 43.526 293.777⁎
(0.0022) (0.0017) (0.0000) (0.0000) (0.0994) (0.0000) (0.8452) (0.0000)
lGDP2 2.661 (0.9961) −3.088⁎ −4.966⁎ −55.119⁎ −1.650⁎⁎ −16.754⁎ 286.541⁎ 388.566⁎ 65.025 275.334⁎
(0.0010) (0.0000) (0.0000) (0.0494) (0.0000) (0.0000) (0.0000) (0.1447) (0.0000)
LFDI2 −11.881⁎ −9.009⁎ −3.263⁎ −11.053⁎ −7.052⁎ −15.607⁎ 364.341⁎ 299.482⁎ 431.097⁎ 2145.84⁎
(0.0000) (0.0000) (0.0005) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
LT2 0.461 (0.6780) −3.051⁎ −1.252 −10.325⁎ −1.211 −9.310⁎ 68.726⁎⁎⁎ 189.874⁎ 77.411⁎⁎ 455.019⁎
(0.0011) (0.1052) (0.0000) (0.1129) (0.0000) (0.0856) (0.0000) (0.0200) (0.0000)
LEC2 1.032 (0.8491) −3.044⁎ −35.082⁎ −29.347⁎ −11.002⁎ −13.872⁎ 458.946⁎ 189.366⁎ 114.996⁎ 382.109⁎
(0.0079) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
LFD2 2.281 (0.9887) −5.076⁎ −6.763⁎ −9.693⁎ −2.342⁎ −8.461⁎ 120.787⁎ 171.453⁎ 63.165 350.617⁎
(0.0000) (0.0000) (0.0000) (0.0096) (0.0000) (0.0000) (0.0000) (0.1841) (0.0000)

Note: Δ portrays the first difference term. ⁎, ⁎⁎, ⁎⁎⁎ are the Statistical significance at the 1, 5 and 10% level, respectively (P-values are presented in parentheses).

Table 3
Werterland cointegration test highlights.

Dependent variable: CO2 Dependent variable: Nox

Statistic Value Z-value P-value Statistic Value Z-value P-value

Gt −5.897⁎ −20.197 0.000 Gt −3.117⁎ −2.567 0.005


Ga −37.118⁎ −19.317 0.000 Ga −32.963⁎ −11.339 0.000
Pt −20.309⁎ −9.774 0.000 Pt −14.395⁎ −1.910 0.028
Pa −14.857 ⁎ −5.862 0.000 Pa −17.458⁎ −3.830 0.000

Notes: ⁎ indicates the test statistics are significant at 1% level. Following Westerlund (2007) and Persyn and Westerlund (2008), the maximum lag length is selected
according to 4(T/100)2/9.

means that there is evidence of a quadratic nexus between the trade rise in energy consumption leads to raising CO2 emissions by around
openness and the environmental proxy (CO2) in contradiction with 1.72%. This implies that a rise in the pollutions level is justified by the in-
the majority of studies in which they supposed the linearity of such a re- crease in energy consumption. Due to the great consumption of energy
lationship. Specifically, the imports of goods that seek more energy con- by the industrial activity that exerts heavy pressure on the environment.
sumption exert heavy pressure on environmental quality. Indeed, the Also, the EC squared revealed a negative effect on the CO2 emissions in-
pressure on the environmental quality caused by such goods justified dicating that the existence of a negative relationship. This means that
the additional demand for energy consumption (e.g. cars' fuel, electric- there is evidence of a quadratic nexus between the energy consumption
ity, and coal) leads to increasing the pollution level, and, therefore, the and the environmental proxy (CO2) in contradiction with the majority
pressure on the environmental quality. Moreover, the magnitude of of studies in which they supposed the linearity of such a relationship.
0.078 related to FD implies that a 1% rise in financial development Therefore the Environmental Kuznets Curve hypothesis is well verified
leads to an increase in CO2 emissions approximately by 0.08%. This im- for the CO2 emissions, however, the positive impact of foreign direct in-
plies that an increase in the financial development level generates an in- vestments and trade openness offers a snapshot that the African coun-
crease in the pollution level and the pressure on environmental quality. tries have polluting industries with a heavy share of greenhouse gas
This finding is justified by the increase of the total credit to the private emissions (e.g. Grossman and Krueger, 1995).
sector, which generates an increase in the demand for energy consump- The Pesaran (2007) CCE-MG estimated coefficients belong to the
tion, and thus the emissions level improves. Moreover, the additional NOx model, displayed in Table 5, are assumed as short-run elasticity.
volume of credits generates additional demand for luxury goods that The income factor (his squared) has a significantly positive (negative)
seek more energy use (e.g. cars, sophisticated technologies, etc.), and impact on the NOx emissions at a 1% level of significance, implying
thus the increase of the environmental pressures. Besides, the FD that an inverted U-shaped is well obtained, supporting the EKC
squared recorded a negative effect on the CO2 emissions indicating hypothesis4 for the NOx. Regarding the coefficient of the foreign direct
that the existence of a negative relationship. This means that there is ev-
idence of a quadratic nexus between the financial development and the 4
The increase of the income factor leads to increasing the pollution level in the earlier
environmental proxy (CO2) in contradiction with the majority of studies stages of development before reaching some specified threshold where the pollution level
in which they supposed the linearity of such a relationship. Regarding decline confirming an inverted U-shaped. Indeed, in the long-run, the environment was
the energy use aspect, where the magnitude of 1.719 means that a 1% considered as a luxury good and greatly seeking as confirmed by the EKC theory.

5
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Table 4 Table 5
Results of the CCE-MG estimator. Results of the CCE-MG estimator.

The dependent variable is CO2 The dependent variable is NOx

Variables Coef. P-value Variables Coef. P-value

GDP 0.296⁎ (0.0005) GDP 0.606⁎ (0.0004)


GDP2 −0.145⁎ (0.0008) GDP2 −0.254⁎ (0.0020)
FDI 0.398⁎⁎ (0.0175) FDI 0.013⁎⁎⁎ (0.0961)
FDI2 −0.368⁎ (0.0049) FDI2 −0.025⁎⁎ (0.0490)
T 0.214⁎ (0.0000) T 0.232⁎ (0.0022)
T2 −1.576⁎ (0.0090) T2 −0.476⁎⁎ (0.0171)
FD 0.078⁎ (0.0000) FD 0.025⁎⁎⁎ (0.0920)
FD2 −0.091⁎⁎ (0.0201) FD2 −0.007⁎⁎⁎ (0.0840)
EC 1.719⁎ (0.0000) EC 0.194⁎ (0.0000)
EC2 −0.839⁎⁎ (0.0300) EC2 −0.200⁎⁎ (0.0170)
Wald chi2 (10) 11.22 Wald chi2 (10) 13.56
Prob > chi 2 0.0109 Prob > chi 2 0.0187

Note: Root Mean Squared Error (sigma): 0.0483 and ⁎, ⁎⁎, and ⁎⁎⁎indicate statistical signif- Note: Root Mean Squared Error (sigma): 0.0757 and ⁎, ⁎⁎, and ⁎⁎⁎indicate statistical signif-
icance at the 1%, 5%, and 10% level, respectively. icance at the 1%, 5%, and 10% level, respectively.

investment inflows (FDI) where it exerts a positive influence on the NOx 3.5. Short-run and long-run causality analysis
emissions level at a 5% level of significance. Hence, the magnitude of
0.013 implies that an increase in the FDI level leads to raising the pollu- The findings of the Granger short- and long-term causality test
tion level (NOx emissions) by around 0.013%. The increase of the NOx are recorded in Table 6. Our findings are interpreted essentially for
emissions due to the African countries characterized by weak environ- the relationships between CO2/NOx and foreign direct investment
mental regulations background will attract the pollutants investments and trade openness. In terms of long-term causality, for both
industries. From this perspective, African countries have a comparative models, the ECT's coefficients are negative and statistically signif-
advantage in polluting industries, thus, the FDI explained the rise of the icant. This implies that there is a bidirectional long-term causality
pollution level. Furthermore, the FDI squared showed a negative impact between CO2/NOx emissions, GDP, trade openness, and foreign di-
on NOx emissions, indicating that the existence of a negative relation- rect investment. These facts are in line with the results of Tiba
ship. This means that there is evidence of a quadratic nexus between et al. (2015), Amri et al. (2018), Tiba and Frikha (2019c), and
the foreign direct investment inflows and the environmental proxy Belaïd and Zrelli (2019).
(NOx) in contradiction with the majority of studies which they sup- Regarding the short-run Granger causality analysis, we pointed
posed the linearity of such relationship. Regarding the openness factor, out unidirectional causality running from GDP to CO 2 , and from
where the magnitude of 0.232 implies that a 1% rise in trade openness FDI to CO 2. However, bidirectional causality nexus was detected
leads to raising NOx emissions by around 0.23%. This implies that inter- between trade openness and CO2. Besides, our empirical facts re-
national trade through foreign imported goods contributes heavily to vealed short-run bidirectional causality between GDP and NO x
greenhouse gas due to the weak environmental regulations in African and between trade openness and NO x . However, we revealed a
countries. Furthermore, the square of trade openness showed a signifi- unidirectional short-run causality from foreign direct investment
cant negative impact on the NOx emissions, supporting a quadratic to NO x . These results are in line with the findings of Belaïd and
nexus between the trade openness and the environmental proxy Youssef (2017) and Tiba and Frikha (2018). Fig. 2 below reflects
(NOx). Moreover, the magnitude of 0.025 related to FD implies that a schematically the short-run and long-run causality interactions
1% rise in financial development leads to an increase in NOx emissions between the variables.
approximately by 0.025%. This implies that an increase in the total credit The results of causality state the importance of understanding the
to the private sector5 generates an increase in the demand for energy controversial interaction between foreign direct investments, trade,
consumption and, thus, the raising of the emissions level. Besides, the and environmental proxy. Indeed, the increase of the environmental
FD squared recorded a significant negative effect on the NOx emissions, damages is due to the polluting quality of the foreign investment
confirming the existence of a quadratic relationship between financial coupled with the weak environmental regulations background in
development and the environmental proxy (NOx). Also, the magnitude these economies, therefore, these elements will attract the pollutant in-
of 0.194 means that a 1% increase in the energy demand, leads to an in- dustries. Then, African countries have a comparative advantage in pol-
crease in the NOx emissions by 0.194%. The squared of energy consump- luting industries.
tion has a significant negative effect on NOx emissions, indicating that Also, the increase in foreign trade generates an increase in pol-
the existence of quadratic nexus between energy consumption and lutant emissions. Indeed, imported goods below the international
the environmental proxy (NOx). standards contribute heavily to greenhouse gas due to the weak
Therefore the Environmental Kuznets Curve hypothesis is well veri- environmental regulations in the African countries. Additionally,
fied for the NOx emissions, however, the positive impact of foreign di- this means that the importation of goods that seek more energy
rect investments and trade openness offers a snapshot that the African consumption exerts heavy pressure on environmental quality.
countries have a comparative advantage in polluting industries with a Hence, the pressure on the environmental quality caused by such
heavy share of greenhouse gas emissions (e.g. Grossman and Krueger, goods justified by the additional demand for energy consumption
1995). It implies that a rise in foreign direct investments and trade leads to increasing the pollution level, and, therefore, the pressure
openness will enhance the pressure of pollution due to a comparative on environmental quality. Therefore, the African industrial sector
advantage in polluting industries under weaker environmental regula- exerts great pressure on the environment through pollutant emis-
tions (e.g. Jayanthakumaran et al., 2012). These findings are in accor- sions. In fact, because of the comparative advantage in polluting
dance with the highlights of Tiba et al. (2015). industries conjointly with weaker ecological regulations, the ris-
ing of trade openness generates an increase in environmental
5
The additional volume of credits generates additional demand for luxury goods that
damages. From this perspective, foreign trade through its positive
seek more energy use and thus the increase of the environmental pressures. externalities in terms of technology transfer, international

6
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Table 6
Granger causality test highlights.

Dependent variable Sources of causation (independent variable) Long-run

Short-run

Model ΔCO2 ΔGDP (ΔGDP2) ΔFDI (ΔFDI2) ΔT (ΔT2) ΔFD (ΔFD2) ΔEC (ΔEC2) ECT

ΔCO2 # 0.017⁎ (0.0097) −1.182 (0.2418) 1.170⁎⁎ (0.0248) 2.201⁎(0.0378) −0.1306⁎ (0.0082) −0.053⁎⁎(0.0139)
ΔGDP (ΔGDP2) 0.207 (0.1731) # −1.206 (0.2145) 1.487 (0.1152) 2.973 (0.1755) −2.718⁎⁎ (0.0381) −0.002⁎(0.0036)
ΔFDI (ΔFDI2) −0.191⁎ (0.0062) −0.454⁎ (0.0016) # 2.092⁎ (0.0041) −0.380⁎⁎ (0.0063) −0.074⁎ (0.0013) −0.230⁎⁎ (0.0345)
ΔT (ΔT2) 0.673⁎⁎⁎(0.0639) 3.410⁎⁎ (0.0166) 2.295 (0.4347) # 2.004⁎⁎⁎ (0.0647) 1.319⁎⁎ (0.0140) −0.023⁎ (0.0092)
ΔFD (ΔFD2) 0.987⁎⁎ (0.0441) 1.572⁎⁎ (0.0115) 0.144 (0.2857) 1.120⁎⁎ (0.0293) # −0.121⁎ (0.0097) −0.721⁎⁎(0.0141)
ΔEC (ΔEC2) 0.629(0.1488) −2.948⁎⁎ (0.0388) 0.577 (0.9637) −2.080⁎⁎⁎ (0.0990) 1.488 (0.1584) # −0.0003⁎(0.0030)
Model ΔNOx ΔGDP (ΔGDP2) ΔFDI (ΔFDI2) ΔT (ΔT2) ΔFD (ΔFD2) ΔEC (ΔEC2) ECT
ΔNOx # −0.345⁎⁎ (0.0115) 0.219 (0.5989) −0.182⁎⁎⁎ (0.0946) −1.206 (0.1152) 0.168 (0.1020) −0.001⁎(0.0012)
ΔGDP (ΔGDP2) 0.248⁎⁎(0.0429) # 2.932 (0.5738) 0.597⁎⁎⁎ (0.0906) 0.592 (0.1104) 0.520⁎⁎ (0.0977) −0.0003⁎ (0.0005)
ΔFDI (ΔFDI2) 0.200⁎ (0.0038) −4.343⁎ (0.0009) # −9.763⁎ (0.0081) −0.913⁎⁎ (0.0099) 1.081⁎ (0.0088) −0.071⁎⁎ (0.0126)
ΔT (ΔT2) −0.369⁎⁎(0.0213) 0.894⁎ (0.0054) −6.252 (0.2854) # 0.886⁎⁎⁎ (0.0549) −1.179⁎⁎ (0.0486) −0.004⁎ (0.0013)
ΔFD (ΔFD2) 0.139⁎ (0.0196) 1.380⁎ (0.0050) 1.241 (0.2619) −0.506⁎⁎ (0.0413) # 3.853⁎⁎ (0.0446) 1.434 (0.5954)
ΔEC (ΔEC2) −0.060⁎⁎(0.0196) −1.834⁎⁎ (0.0320) −1.095 (0.2620) −1.022⁎⁎ (0.0414) −0.863⁎⁎⁎ (0.0504) # −1.103⁎ (0.0028)

Note: P-values are listed in parentheses and ⁎, ⁎⁎, and ⁎⁎⁎indicate statistical significance at the 1%, 5%, and 10% level, respectively.

Fig. 2. The Short-run and long-run causal interactions results.

expertise, improving innovation and R&D, and the attraction of 4. Concluding remarks
the foreign resources to build the required capacities (e.g. finan-
cial, technical, human). Indeed, this later will help the African The third millennium is characterized by the increasing conscience
economies to enhance productivity generating additional jobs, de- about the challenging threat of climate change, precisely, with the
mand, and revenue. These highlights are similar to the findings of Kyoto protocol 1997, which this protocol includes the environment as
Costantini and Liberati (2014) and Tiba et al. (2015), Tiba and a relevant factor that determines the sustainability of development con-
Frikha (2019a, 2019b, 2019c), and Tiba (2019a, 2019b, 2019c). sistent with the fifth generation of human rights. However, the growth
From this outlook, our survey shares the same intuition of the find- path exerts heavy pressure on the environment through several trans-
ings of Tiba et al. (2015) where they supposed that the foreign direct in- mission channels, which generates conflicting results and goals at
vestments inflows (FDI) and trade openness seem to be the driving odds to environmental aims. With the launching of a large scale event
forces behind the environmental sustainability through the transfer of such as the agreement of Paris-2016 under the United Nations back-
green technologies and providing the necessary expertise and capacities ground, following the climate change agreement of Kyoto1997, each
(financial and human) towards the sustainable development into the member state of this agreement should renew his intuition, ultimate
UN millennium development challenge. engagement, and strong commitment for fighting the global warming
threat. Besides, we added two important relevant variables that contrib-
ute very well to the climate change phenomenon, we contribute to the
3.6. Robustness test EKC theory by emphasizing on the trade openness and foreign direct in-
vestment inflows as determinants of pollution level. We analyze the sig-
We perform a robustness check-in to assess the sensitivity of the re- nificant contribution of the Trade openness and Foreign Direct
sults to model specification. Accordingly, ss an alternative for the Investment inflows as a major explication to the climate change threat,
Pesaran (2007) CCE-MG model, we develop a Dynamic Ordinary Least which they are missed in most of the cases, by validating the EKC
Squares (DOLS) model to estimate the long-run relationships for our hypothesis for foreign direct investment and trade openness.
variables. Without any doubt, foreign direct investment inflows and trade
The results are displayed in Tables A1 and A2 in the appendix. We openness play keys roles in environmental sustainability through their
note that the results of the DOLS models are consistent with the results positive externalities. Even more, the EKC has been widely stressed
of the earlier analysis, and support the findings of the CCE-MG model. yet without shedding more light on the key role of the foreign

7
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

investments and openness aspects. The motivation of our study is to in- Table A2 (continued)
vestigate the role of FDI and trade openness in shaping the EKC theory. Dependent variable: LNOX
To achieve this goal, our study employed the Westerlund (2007) panel
Variable Coefficient Prob.
cointegration methodology to check the presence of long-run relation-
ships as well as we used the Cross Correlated Effects (CEE) and the Com- LT2 −9.130⁎⁎⁎ 0.0622
LT −0.393⁎ 0.0038
mon Correlated Effects Mean Group (CCE-MG) estimation procedures,
LGDP 1.102⁎⁎ 0.0407
and the short- and long-term Granger causality, for 27 African countries LGDP2 −6.809⁎⁎⁎ 0.0801
over the period from 1990 to 2013. C 8.686⁎ 0.0000
Our findings reveal bidirectional long-term causality between CO2/ R-squared 0.698
NOx emissions, GDP, trade openness, and foreign direct investment. Be- Adjusted R-squared 0.688
F-statistic 231.102
sides, in the short-run, the results recorded a unidirectional causality Prob(F-statistic) 0.0000
running from GDP to CO2, and from FDI to CO2. However, bidirectional
Note: *, **, and *** indicate statistical significance at the 1%, 5%, and 10% level, respectively.
causality nexus was detected between trade openness and CO2. Addi-
tionally, the findings portray bidirectional causality between GDP and
NOx and between trade openness and NOx, while a unidirectional causal- Appendix B. Supplementary data
ity from foreign direct investment to NOx. Our highlights pointe out the
validation of the EKC hypothesis for the CO2 and NOx emissions, hence, Supplementary data to this article can be found online at https://doi.
we recorded a quadratic relationship between the FDI, trade openness, org/10.1016/j.eneco.2020.104966.
and the environmental damage proxies. Indeed, we shed light on the im-
portance of the foreign direct investment inflows and trade openness in References
mitigating the adverse effect of heavy pollutant activities and resolving
Agras, J., Chapman, D., 1999. A dynamic approach to the environmental Kuznets curve hy-
the environmental puzzle “growing the economy without polluting the en- pothesis. Ecol. Econ. 28, 267–277.
vironment” consistent with the MDGs of the UN background. Also, their Akbostanci, E., Türüt-AsIk, S., Tunç, G.I., 2009. The relationship between income and envi-
positive externalities of the FDI and trade provide the necessary capaci- ronment in Turkey: is there an environmental Kuznets curve? Energy Policy 37,
861–867.
ties and additional resources (e.g. capital inflows, human expertise, effi- Al-mulali, U., 2012. Factors affecting CO2 emission in the Middle East: a panel data anal-
cient partnerships with national, international organizations, and ysis. Energy 44 (1), 564–569.
international civil society, financial assistance, etc.). Thus, the quality of Al-mulali, U., Tang, F.C., 2013. Investigating the validity of pollution haven hypothesis in
the gulf cooperation council (GCC) countries. Energy Policy 60, 813–819.
foreign investment and trade openness plays a key role in terms of sus- Amri, F., Bélaïd, F., Roubaud, D., 2018. Does technological innovation improve environ-
taining the growth of the economy and providing a sustainable environ- mental sustainability in developing countries? Some evidence from Tunisia.
ment through the diffusion and the adoption of sustainable goods and J. Energy Develop. 44 (1/2), 41–60.
Ang, J.B., 2007. CO2 emissions, energy consumption, and output in France. Energy Policy
sustainable practices. Hence, the integration of the FDI and trade open-
35, 4772–4778.
ness into the core of ambitious development strategies seems to be cru- Ang, J.B., 2008. Economic development, pollutant emissions and energy consumption in
cial to achieving the challenge of growth without pollution. Malaysia. J. Policy Model 30, 271–278.
Apergis, N., Payne, J.E., 2009. CO2 emissions, energy usage, and output in Central America.
Energy Policy 37, 3282–3286.
Appendix A. Appendix Atici, C., 2012. Carbon emissions, trade openness, and the Japan–ASEAN interaction: a
group-wise examination. J. Jpn. Int. Econ. 26 (1), 167–178.
Baek, J., 2016. A new look at the FDI–income–energy–environment nexus: dynamic panel
Our empirical robustness check results, which we employ the Panel data analysis of ASEAN. Energy Policy 91, 22–27.
Least-squares, confirm our results in terms of signs, significance, and the Baek, J., Cho, Y., Koo, W.W., 2009. The environmental consequences of globalization: a
analysis remains the same for both models. country specific time-series analysis. Ecol. Econ. 68, 2255–2264.
Bai, J., Ng, S., 2004. A PANIC attack on unit roots and cointegration. Econometrica 72,
Table A1 1127–1177.
Belaïd, F., Youssef, M., 2017. Environmental degradation, renewable and non-renewable
Panel least squares estimation results for LCO2 equation.
electricity consumption, and economic growth: assessing the evidence from
Dependent variable: LCO2 Algeria. Energy Policy 102, 277–287.
Belaïd, F., Zrelli, M.H., 2019. Renewable and non-renewable electricity consumption, envi-
Variable Coefficient Prob. ronmental degradation and economic development: evidence from Mediterranean
countries. Energy Policy 133, 110929.
LFDI 5.534⁎⁎ 0.0357
Boulatoff, C., Jenkins, M., 2010. Long-term nexus between openness, income and environ-
LFDI2 −6.347⁎ 0.0068
mental quality. Int. Adv. Econ. Res. 16, 410–418.
LEC 9.785⁎ 0.0000 Breitung, J., 2001. The local power of some unit root tests for panel data. In: Baltagi, B.H.,
LEC2 −2.409⁎ 0.0002 Fomby, T.B., Hill, R.C. (Eds.), Nonstationary Panels, Panel Cointegration, and Dynamic
LT2 −6.129⁎ 0.0051 Panels (Advances in Econometrics, Vol.15). Emerald Group Publishing Limited,
LT 0.243⁎ 0.0040 United Kingdom, pp. 161–177.
LGDP 1.306⁎ 0.0000 Breusch, T.S., Pagan, A.R., 1980. The Lagrange multiplier test and its applications to model
LGDP2 −1.068⁎⁎ 0.0244 specification in econometrics. Rev. Econ. Stud. 47, 239–253.
C −9.458⁎ 0.0000 Cho, C.H., Chu, Y.P., Yang, H.Y., 2014. An environment Kuznets curve for GHG emissions: a
R-squared 0.805 panel cointegration analysis. Energy Sour. Part B: Econ. Plan. Policy 9, 120–129.
Adjusted R-squared 0.802 Choi, I., 2001. Unit root tests for panel data. J. Int. Money Financ. 20, 249–272.
F-statistic 290.812 Chudik, A., Pesaran, M.H., Tosetti, E., 2011. Weak and strong cross-section dependence
Prob(F-statistic) 0.0000 and estimation of large panels. Econ. J. 14, C45–C90.
Copeland, B., Taylor, M.S., 2003. Trade and the Environment: Theory and Evidence.
Note: *, **, and *** indicate statistical significance at the 1%, 5%, and 10% level, respectively. Princeton University Press, Princeton, NJ.
Costantini, V., Liberati, P., 2014. Technology transfer, institutions and development.
Table A2 Technol. Forecast. Soc. Chang. 88, 26–48.
Panel least squares estimation results for LNOX equation. Dasgupta, S., Laplante, B., Wang, H., Wheeler, D., 2002. Confronting the environment
Kuznets curve. J. Econ. Perspect. 16, 147–168.
Dependent variable: LNOX De Hoyos, R.E., Sarafidis, V., 2006. Testing for cross-sectional dependence in panel-data
models. Stata J. 6, 482–496.
Variable Coefficient Prob.
Du, L., Wei, Chu, Cai, S., 2012. Economic development and carbon dioxide emissions in
LFDI 5.174⁎ 0.0002 China: provincial panel data analysis. China Econ. Rev. 23, 371–384.
LFDI2 −5.827⁎ 0.0053 Eberhardt, M., 2012. Estimating panel time-series models with heterogeneous slopes.
LEC 1.525⁎⁎⁎ 0.0720 Stata J. 12, 61–71.
LEC2 −6.600⁎ 0.0022 Engle, R.F., Granger, C.W.J., 1987. Co-integration and error correction: representation, es-
timation, and testing. Econometrica 55, 251–276.

8
S. Tiba and F. Belaid Energy Economics 92 (2020) 104966

Fodha, M., Zaghdoud, O., 2010. Economic growth and environmental degradation in Persyn, D., Westerlund, J., 2008. Error-correction-based cointegration tests for panel data.
Tunisia: an empirical analysis of the environmental Kuznets curve. Energy Policy Stata J. 8, 232–241.
38, 1150–1156. Pesaran, M.H., 2004. General Diagnostic Tests for Cross Section Dependence in Panels
Frees, E.W., 1995. Assessing cross-sectional correlation in panel data. J. Econ. 69 (2), Working. Paper 435 (CESifo).
393–414. Pesaran, M.H., 2007. A simple panel unit root test in the presence of cross-section depen-
Friedman, M., 1937. The use of ranks to avoid the assumption of normality implicit in the dence. J. Appl. Econ. 22, 265–312.
analysis of variance. J. Am. Stat. Assoc. 32 (200), 675–701. Pesaran, M.H., Smith, R., Shin, Y., 1997. Pooled Mean Group Estimation of Dynamic Het-
Grossman, G.M., Krueger, A.B., 1991. Environmental impacts of a north American free erogeneous Panels. DAE Working Paper Amalgamated Series. University of
trade agreement. National Bureau of Economic Research Working Paper 3914. Edinburgh.
NBER, Cambridge MA. Phillips, P.C.B., Sul, D., 2003. Dynamic panel estimation and homogeneity testing under
Grossman, G.M., Krueger, A.B., 1995. Economic growth and the environment. Q. J. Econ. cross section dependence. Econ. J. 6, 217–259.
110, 353–378. Phillips, P.C.B., Sul, D., 2007. Bias in dynamic panel estimation with fixed effects, incidental
Hadri, K., 2000. Testing for stationarity in heterogeneous panel data. Econ. J. 3, 148–161. trends and cross section dependence. J. Econ. 137, 162–188.
Halicioglu, F., 2009. An econometric study of CO2 emissions, energy consumption, income Sadorsky, P., 2009. Renewable energy consumption, CO2 emissions and oil prices in
and foreign trade in Turkey. Energy Policy 37, 1156–1164. theG7 countries. Energy Econ. 31, 456–462.
He, J., Richard, P., 2010. Environmental Kuznets curve for CO2 in Canada. Ecol. Econ. 69, Sahli, I., Ben Rejeb, J., 2015. The environmental Kuznets curve and corruption in the Mena
1083–1093. region. Procedia Soc. Behav. Sci. 195, 1648–1657.
Hitam, M.B., Borhan, H.B., 2012. FDI, growth and the environment: impact on quality of Sbia, R., Shahbaz, M., Hamdi, H., 2014. A contribution of foreign direct investment, clean
life in Malaysia. Procedia-Soc. Behav. Sci. 50, 333–342. energy, trade openness, carbon emissions and economic growth to energy demand
Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels. in UAE. Econ. Model. 36, 191–197.
J. Econ. 115, 53–74. Shafik, N., Bandyopadhyay, S., 1992. Economic growth and environmental quality: Time-
Iwata, H., Okada, K., Samreth, S., 2011. A note on the environmental Kuznets curve for series and cross-country evidence. World Bank Working Papers, WPS 904,
CO2: a pooled mean group approach. Appl. Energy 88, 1986–1996. Washington, Vol. 52, p. 13.
Jalil, A., Mahmud, S.F., 2009. Environment Kuznets curve for CO2 emissions: A co- Shahbaz, M., Ozturk, I., Afza, T., Ali, A., 2013. Revisiting the environmental Kuznets curve
integration analysis for China. Energy Policy 37, 5167–5172. in a global economy. Renew. Sust. Energ. Rev. 25, 494–502.
Jayanthakumaran, K., Verma, R., Liu, Y., 2012. CO2 emissions, energy consumption, trade Smith, L.V., Leybourne, S., Kim, T.H., Newbold, P., 2004. More powerful panel data unit
and income: a comparative analysis of China and India. Energy Policy 42, 450–460. root tests with an application to mean reversion in real exchange rates. J. Appl.
Kapetanios, G., Pesaran, M.H., Yamagata, T., 2011. Panels with non-stationary multifactor Econ. 19, 147–170.
error structures. J. Econ. 160, 326–348. Tamazian, A., Chousa, J.P., Vadlamannati, K.C., 2009. Does higher economic andfinancial
Kellenberg, D.K., 2009. An empirical investigation of the pollution haven effect with stra- development lead to environmental degradation: evidence from BRICcountries. En-
tegic environment and trade policy. J. Int. Econ. 78, 242–255. ergy Policy 37 (1), 246–253.
Kivyiro, P., Arminen, H., 2014. Carbon dioxide emissions, energy consumption, economic Tiba, S., 2019a. Exploring the nexus between oil availability and economic growth: in-
growth, and foreign direct investment: causality analysis for Sub-Saharan Africa. En- sights from non-linear model. Environ. Model. Assess. 24 (6), 691–702.
ergy 74, 595–606. Tiba, S., 2019b. Modeling the nexus between resources abundance and economic growth:
Kuznets, S., 1955. Economic growth and income inequality. Am. Econ. Rev. 45, 1–28. an overview from the PSTR model. Res. Policy 64, 101503.
Lau, L.S., Choong, C.K., Eng, Y.K., 2014. Carbon dioxide emission, institutional quality, and Tiba, S., 2019c. A non-linear assessment of the urbanization and climate change nexus:
economic growth: empirical evidence in Malaysia. Renew. Energy 68, 276–281. the African context. Environ. Sci. Pollut. Res. 26 (31), 32311–32321.
Lean, H.H., Smyth, R., 2010. CO2 emissions, electricity consumption and output in ASEAN. Tiba, S., Frikha, M., 2018. Income, trade openness and energy interactions: evidence from
Appl. Energy 87, 1858–1864. simultaneous equation modeling. Energy 147, 799–811.
Lee, C.G., 2009. Foreign direct investment, pollution and economic growth: evidence from Tiba, S., Frikha, M., 2019a. ECK and macroeconomics aspects of well-being: a critical vision
Malaysia. Appl. Econ. 41, 1709–1716. for a sustainable future. J. Knowl. Econ. 11 (3), 1171–1197.
Lee, J.W., 2013. The contribution of foreign direct investment to clean energy use, carbon Tiba, S., Frikha, M., 2019b. Sustainability concern in the agenda of African countries: evi-
emissions and economic growth. Energy Policy 55, 483–489. dence from simultaneous equations models. J. Knowl. Econ. 11 (3), 1270–1294.
Lee, J.W., Brahmasrene, T., 2014. ICT, CO2 emissions and economic growth: evidence from Tiba, S., Frikha, M., 2019c. The controversy of the resource curse and the environment in
a panel of ASEAN. Glob. Econ. Rev. 43 (2), 93–109. the SDGs background: the African context. Res. Policy 62, 437–452.
Lee, C.C., Lee, J.D., 2009. Income and CO2 emissions: evidence from panel unit root and Tiba, S., Omri, A., 2016. Literature survey on the relationships between energy, environ-
cointegration tests. Energy Policy 37, 413–423. ment and economic growth. Renew. Sust. Energ. Rev. 69, 1129–1146.
Levin, A., Lin, C.F., Chu, C.S., 2002. Unit root tests in panel data: asymptotic and finite sam- Tiba, S., Omri, A., Frikha, M., 2015. The four-way linkages between renewable energy, en-
ple properties. J. Econ. 108, 1–24. vironmental quality, trade and economic growth: a comparative analysis between
Llorca, M., Meunié, A., 2009. SO2 emissions and the environmental Kuznets curve: the high and middle-income countries. Energy Syst. 7, 103–144.
case of Chinese provinces. J. Chin. Econ. Bus. Stud. 7, 1–16. Tiwari, A.K., Shahbaz, M., Hye, Q.M.A., 2013. The environmental Kuznets curve and the
Maddala, G.S., Wu, S., 1999. A comparative study of unit root tests with panel data and a role of coal consumption in India: cointegration and causality analysis in an open
new simple test. Oxf. Bull. Econ. Stat. 61, 631–652. economy. Renew. Sust. Energ. Rev. 18, 519–527.
Managi, S., Kumar, S., 2009. Trade-induced technological change: analyzing economic and Westerlund, J., 2007. Testing for error correction in panel data. Oxf. Bull. Econ. Stat. 69,
environmental outcomes. Econ. Model. 26, 721–732. 709–748.
Moon, H.R., Perron, B., 2004. Testing for a unit root in panels with dynamic factors. J. Econ. World Bank, 2014. World Development Indicators. Available at:. http://www.worldbank.
122, 81–126. org/data/on line data bases/on linedatabases.html.
Panayotou, T., 1993. Empirical Tests and Policy Analysis of Environmental Degradation at Zhang, Y.J., 2011. The impact of financial growth on carbon emissions: an empirical anal-
Different Stages of Economic Development. Working Paper WP238. Technology and ysis in China. Energy Policy 39 (4), 2197–2203.
Employment Programme, International Labour Office, Geneva. Zhang, C., Zhou, X., 2016. Does foreign direct investment lead to lower CO2 emissions?
Pao, H.T., Tsai, C.M., 2011. Multivariate Granger causality between CO2 emissions, energy Evidence from a regional analysis in China. Renew. Sust. Energ. Rev. 58 (2016),
consumption, FDI (foreign direct investment) and GDP (gross domestic product): ev- 943–951.
idence from a panel of BRIC (Brazil, Russian Federation, India, and China) countries.
Energy 36, 685–693.
Perkins, R., Neumayer, E., 2009. Transnational linkages and the spillover of environment-
efficiency into developing countries. Glob. Environ. Chang. 19, 375–383.

You might also like