Introduction

The recent global concern of environmentalists has been the dehumanizing effect of the pressure inflicted by the exploitative tendencies of humanity on the ecosystems. As trivial as it may sound, these human activities have been attributed to be the major factors behind the evils of climate change, environmental degradation, and ecological distortions currently ravaging today’s world (Alola 2019a; Bekun et al. 2019). It has also been projected that the world is yet to witness the worst effect of climate change on the environment as more lives, more farm output, and more wealth will, and are being wasted at the altar of environmental degradation. Humanity is facing her greatest threat ever and there is urgent need for affirmative actions to curb this doom.

However, environmentalists have pre-occupied themselves in response to the global climate change with researches on population changes, energy use, trade and urbanization, and their effects on the environment (Alola 2019b; Saint Akadiri et al. 2019; Alola and Alola 2018; Sarkodie 2018; Shahbaz and Sinha 2019; Wang and Dong 2019; Nathaniel 2019). Economic growth increases energy demand and drives industrialization which in turn promotes trade. Since CIVETS countries are naturally endowed, the need to facilitate trade promotes natural resource extraction. The extraction and exploitation of natural resources give rise to income increase on one hand while decreasing biocapacity and increasing ecological footprint (EFP) on the other (Panayotou 1993). Just like industrialization, economic growth increases urbanization. Urbanization raises the demand for industrialization and transportation, intensifies fossil fuel consumption, and increases EFP (Ulucak and Khan 2020). Urbanization can improve the purchasing power of urban dwellers which will inform the demand for renewables thereby reducing EFP (Danish and Wang 2019).

Economies are adopting renewable energy (REN) strategies to mitigate CO2 emission effect on the environment (Aliyu et al. 2018; Nathaniel and Iheonu 2019; Moutinho et al. 2018; Sharif et al. 2019), such that the investment in REN was in excess of 214 billion USD as of 2013. As a result, REN consumption has increased from 16% of the total energy consumption in 2007 to 18% in 2016 (World Bank 2017). It is also expected that this percentage will double by the year 2022 owing to the increased concern of economies to adopt cleaner energy and embrace green economy (IEA 2017).

Developing countries mostly focus on how to attain growth through an increase in national output without considering the adverse effects of such growth on the environment. Energy consumption, mainly of non-renewable sources, is observed to be on the increase in these countries, CIVETS inclusive. Non-renewable energy (NREN) is a pollutant. They contribute to greenhouse gas emissions (GHGs) thereby damaging human health, economic activities, and distorting environmental sustainability. This study focused on the CIVETS countries because of their resource endowments, NREN consumption, REN potentials, and vulnerability to climate change. The CIVETS countries are among the highest emitters of carbon dioxide in the world because of their continuous consumption of NREN, mainly fossil fuels. South Africa, for instance, is the 14th highest emitter of CO2 emissions in the world mainly due to increased consumption of coal, a NREN source (Liu et al. 2020a, b). The Living Planet Report (LPR) of 2014 revealed that the EFP in Indonesia fall shut of the world’s average (1.7 gha) biocapacity per person mainly as a result of its continuous consumption of NREN. Turkey emits 500 megatonnes of GHGs each year, more than 1% of the world’s total. The country’s average growth rate was 2.795% for the period 1965 to 2015 and 2.089% for the OECD countries within the same period (World Bank 2018). The average primary energy use per capita growth was 2.627% between 1965 and 2015 in Turkey, whereas for the OECD countries, the same growth rate averaged only 0.585% (Karasoy 2019). As of 2014, total final (energy) consumption (TFC) in Turkey was dominated by fossil fuels: 35.6%, 2.6%, and 12.3% of TFC were from oil, natural gas, and coal, respectively (IEA 2016). In 2016, the energy sector in Turkey contributed the largest share (86.1%) to CO2 emissions (Turkish Statistical Institute 2018). CIVETS, like other developing country, needs sustainable growth and environmental sustainability. The persistent usage of NREN, accompanied by harmful trade and an upward surge in urbanization, will truncate environmental sustainability and inhibit sustainable development in these countries. Hence, the motivation for this study.

Relatively, just a few attempts have been made to examine REN in a framework of population changes and their effects on environment degradation. Some of the studies found that REN reduces environmental degradation, protects the environment, and promotes economic development (Emir and Bekun 2019; Balsalobre-Lorente et al. 2018; Apergis et al. 2018). However, there are studies that discovered no meaningful effect of REN on environmental degradation (Frondel et al. 2010; Marques and Fuinhas 2012) while a few others revealed that REN do not reduce environmental pollution (Al-Mulali et al. 2015a; Pata 2018c).

Most of the studies in the literature on the effect of REN on the environment narrowed the measure of environment degradation to CO2 emissions without considering the effect on the environment of human activities expressed on the available land space for sustained natural resource use. These effects of humanity on the ecosystem, referred to as EFP, reveals human economy dependence on natural capital and it serves as a better measures of environmental degradation than just CO2 emissions (Lin et al. 2015).

The new contributions of this study to the existing literature are in the following ways: (i) there are a lot of studies on the energy-growth-environment nexus on each of the individual CIVETS countries with mixed results. This study is the first to consider the determinants of EFP such as REN, NREN, economic growth, trade, urbanization for CIVETS countries as a group. While most previous studies adopted CO2 emissions (a negative indicator), the present study used EFP (a positive indicator) to measure environmental quality. EFP includes six categories of bio productive land use type (grazing land, forest land, carbon footprint, cropland, built-up land, and ocean). The EFP is arguably the only metrics that compare the resource demand of government, businesses, and individuals against what Earth can renew. Since the CIVETS countries have maintained a fairly stable growth through production over the years, the use of EFP becomes ideal because it measures the environmental consequences of the production of commodities, both goods and services, to promote a required lifestyle (Rashid et al. 2018). (ii) This study is unique in the adopted panel data estimation methods that produce more robust and reliable estimations. We used the Augment Mean Group (AMG) estimator which is consistent with cross-sectional dependence (CSD) and country-specific heterogeneity. The AMG estimator was further complemented with the mean group (MG) and the common correlated effects mean group (CCEMG) estimator that takes care of endogeneity problems.

The remaining parts of this article are grouped in this order: “Review of literature” presents the literature review. “Methodology” addresses the methodology. Results are presented and discussed in “Results and discussion of findings”. “Conclusion and policy directions” concludes with relevant policy directions.

Review of literature

The remarkably increasing economic growth and accelerating process of urbanization have resulted in the increase of energy needs, which is closely related to human activities and thus, lead to environmental degradation (Wang and Dong 2019). Urbanization plays a significant role in influencing the environmental quality via various channels. Industrialization brings along the establishment of new factories and plants providing employment opportunities, leading to people migrating to urban cities from rural areas, and hence increasing the urban population, which ultimately leads to accelerated demand for energy, and subsequently increases CO2 emissions (Sinha et al. 2019). In addition to this particular channel, urbanization also causes environmental degradation in several other ways including transportation, vehicular congestion, international trade, health care, landscape pattern, terrorism, and real income (Saint Akadiri et al. 2019; Alvarado et al. 2018; Farhani and Ozturk 2015; Hanif et al. 2019; Li et al. 2018; Mohamed et al. 2019; Wang et al. 2019). Hence, it is imperative to consider urbanization in the empirical framework while analyzing the ecological footprint across countries.

The economic growth and energy nexus have been widely studied and extensively investigated empirically (Alshehry and Belloumi 2015; Balcilar et al. 2018; Tiba and Omri 2017) following the very first work of Kraft and Kraft (1978). However, such empirical findings seeking to measure the impact of urbanization on the environment are persuasive but unambiguous due to the implementation of various forms of econometric methods such as correlation analysis (Alam et al. 2015), multivariate and panel cointegration tests (Hatzigeorgiou et al. 2011; Pedroni 1999), regression analysis (Dong et al. 2018; Shafiei and Salim 2014), causality tests (Chang et al. 2015; Mutascu 2016), vector error correction model (Sebri and Ben-Salha 2014; Zhou et al. 2018), and ARDL approach (Dogan et al. 2019; Lau et al. 2018; Rauf et al. 2018; Yang et al. 2018). Hence, such inconclusive empirical evidences cannot assist economic policymakers in planning and developing appropriate plans and programs for long-term economic growth (Ozturk and Acaravci 2010; Payne 2010). Lv and Xu (2019) employed the pooled mean group (PMG) approach to examine the impact of urbanization and trade openness on CO2 emissions and report that urbanization improves environmental quality. Conversely, Destek and Sarkodie (2019) and Wang and Dong (2019) use the augmented mean group (AMG) to investigate the relationship between economic growth, energy consumption, and EFP for 11 newly industrialized countries and 14 Sub-Saharan Africa (SSA) countries and find bidirectional long-run causality among economic growth, non-REN, urbanization, and the EFP. Zafar et al. (2019) conducted a Granger causality analysis among the Asia-Pacific Economic Cooperation (APEC) countries for a period extending from 1990 to 2015 and discovered a unidirectional causal relationship running from REN to economic growth and from economic growth to NREN. Alternatively, both Nathaniel et al. (2019) and Rasoulinezhad and Saboori (2018) employed the FMOLS and DOLS estimation technique for South African data spreading from 1965 to 2014 and for 12 states under the Commonwealth region from 1992 to 2015, respectively, and reported that urbanization and energy use promote environmental quality in the long term. Additionally, Bao and Xu (2019) and Ozcan and Ozturk (2019) employed the bootstrap panel causality test to analyze the cause and effect of REN on the urbanization and economic growth and discovered non-existence of causality indicating that energy-saving policies do not have any negative impact on the growth rates.

In addition to the use of varied econometric analysis, different results also arise due to different variables used and the different sample of countries or cities under study (Azizalrahman 2019). With respect to the variables, Yang et al. (2018) included climate factors based on the socio-economic factors and adopted the extended STIRPAT model to study the Chinese economy during the period 1995–2014, and prove urbanization and economic development as the main drivers of CO2 emissions. On the other hand, Andrés and Padilla (2018) analyzed the structural characteristics of transport activity in terms of transport energy intensity of transport modes and transport volume, as the factors of greenhouse gas emissions for European Union-28 from 1990 to 2014, to highlight on the significant factors and their magnitude of impact. Alvarado et al. (2018) analyzed the real per capita output using 151 countries from 1980 to 2016 and find an inverted U-shaped relationship between the output and CO2 emissions by grouping countries globally on the basis of income levels. While analyzing the long and short run impact of fossil fuel consumption and FDI on CO2 emissions in 15 Asian countries from 1990 to 2013, Hanif et al. (2019) discovered that both fossil fuel consumption and FDI contribute to CO2 emissions deteriorating the environment.

The other strand of the existing literature in this field provides empirical evidence on the relationship between economic growth and CO2 emissions, called as the environmental Kuznets curve (EKC), which states the relationship as non-linear and inverted U-shaped, implying that economic growth increases CO2 emissions initially and then gradually declines it, once it matures (Muhammad et al. 2013). Existing studies validated the existence of EKC including Heidari et al. (2015) and Saboori and Sulaiman (2013) for ASEAN countries; Acaravci and Ozturk (2010), Bento and Moutinho (2016), and Ho and Iyke (2019) for selective European countries; Khoshnevis Yazdi and Ghorchi Beygi (2018) and Sarkodie (2018) for African countries; Dogan et al. (2019) for MINT countries; Hanif (2018c) for East Asia and Pacific countries; and Zhu et al. (2018) for BRIC countries. However, there are also studies that highlight on the non-existence of the U-shaped EKC such as in the case of Turkey where the CO2 emissions increased due to the usage of fuel oil and other traditional energy consumption patterns related to urban development (Katircioğlu and Katircioğlu 2018). Consistently, empirical tests in Russia suggest changes in pollutant emissions due to increased energy usage, concluding no support for EKC hypothesis (Pao 2011). Similarly, EKC hypothesis is not valid for USA as well, as proven by Dogan and Ozturk (2017), where REN mitigates environmental degradation while an increase in NREN contributes to CO2 emissions. Further, EKC literature mostly uses CO2 emissions as a single indicator for environmental degradation, but there exists a gap that environmental degradation cannot be captured by CO2 emissions alone, and hence, there is a need for a broader variable namely, EFP to study the existence of EKC hypothesis (Destek et al. 2018; Dogan et al. 2019). The estimates of EFP rationalize the approach of assessing environmental degradation, since it is a more comprehensive indicator (Rashid et al. 2018; Wang and Dong 2019). This study ameliorates for these deficiencies by using EFP, instead of CO2 emissions, as a broader proxy for environmental degradation.

Finally, the third strand deals with the relationship between urbanization, EFP, and energy consumption. Baloch et al. (2019a, b) investigated the impact of financial development on EFP and discovered that economic growth, energy consumption, FDI, and urbanization pollute the environment by increasing the EFP. Correspondingly, Wang and Dong (2019) investigated the determinants of environmental degradation using a dataset of 14 Sub-Saharan African countries from 1990 to 2014. They reported that economic growth, NREN, and urbanization exert positive effects on the EFP, while REN plays a negative role in the EFP. Considering EFP as a measure of human demand on earth’s ecosystem, Hassan et al. (2019) opined that natural resources have a positive effect on the EFP, indicating that it deteriorates environmental quality, but also proves the EKC hypothesis. By examining 93 countries, categorized by income, Al-Mulali et al. (2015a, b) reported that energy consumption, urbanization, and trade openness increase environmental damage by exerting positive impact on the EFP of the countries.

Extensive literature is conducted covering either a panel of countries, for instance, for MENA countries (Nathaniel et al. 2020; Saidi et al. 2018); for Sub-Saharan African countries (Boutabba et al. 2018; Hanif 2018a, b, c; Ozturk 2017; Wang and Dong 2019), for N-11 countries (Sinha et al. 2019), for G-20 countries (Destek and Okumus 2019), and for BRIC countries (Zhu et al. 2018). For a single country case, such as for China (Ahmad et al. 2019; Bao and Xu 2019; Chen et al. 2019a, b), for Turkey (Katircioğlu and Katircioğlu 2018; Ozcan et al. 2018), for the US (Alola 2019a), for Europe (Alola et al. 2019; de Souza et al. 2018; Destek et al. 2018; Ho and Iyke 2019), and for Pakistan (Hassan et al. 2019; Luqman et al. 2019).

Methodology

Method

The study proceeds with the CSD test in order to avoid estimator inefficiency and biased estimates. The CSD test is vital especially when variables are nonstationary. It also gives direction as regards the unit root tests to apply. Three different tests (Breusch-Pagan LM, Pesaran scaled LM, and Pesaran CD) are used for this purpose. The null hypothesis of the CSD test is presented in Eq. 1 below:

$$ {H}_0:{\rho}_{ij}= corr\left({\mu}_{it},{\mu}_{jt}\right)=0\forall i\ne j $$
(1)

In the presence of CSD, the conventional unit root tests will give misleading results. Therefore, there is a need to use unit root tests that accounts for CSD. This study applies both first-and-second generation unit root tests. The former does not account for CSD among the countries, while the later does. Levin et al. (2002) test assumes no changes in the autoregressive parameters for all the cross sections, while the PP-Fisher and the Im et al. (2003) unit root tests assume the exact opposite. The LLC (Levin et al. 2002) test is estimated from the equation below:

$$ {\Delta y}_{it}={\phi}_i+{\pi}_i{y}_{i,t-1}+{\sum}_{j=1}^p{P}_i\Delta {y}_{it-j}+{e}_{it} $$
(2)

∆yit is the difference of yit for all ith country. The study further utilized the cross-sectional augmented IPS (CIPS) test of Pesaran (2007) which is robust for CSD. The test equation is given as

$$ {\Delta y}_{it}={\varphi}_i+{\beta}_i{y}_{i,t-1}+{\tau}_i{\overline{y}}_{t-1}+{d}_i\Delta {\overline{y}}_t+{\varepsilon}_{it} $$
(3)

The CIPS statistic is derived from the sample average of Eq. 3 above. Apart from other first-generation cointegration tests, the Westerlund (2007) was used to examine the presence of a cointegrating relationship among the variables. The error correction approach of the test is presented in below as

$$ {\Delta y}_{it}={\delta}_i^{\prime }{d}_t+{\alpha}_i{y}_{it-1}+{\lambda}_i^{\prime }{x}_{it-1}+{\sum}_{j=1}^{pi}{\alpha}_{ij}\Delta {y}_{it-j}+{\sum}_{j=0}^{pi}{\gamma}_{ij}\Delta {x}_{it-j}+{e}_{it} $$
(4)

αi is the error correction parameter. dt = (1, t)′ comprises the deterministic components (constant and trend) with δt = (δi1, δi2)′ representing vector of parameters. Four tests (based on the OLS estimates of αi in Eq. 4) were developed to investigate the existence of cointegration. Two of these four tests are the group mean statistics given as

$$ {G}_{\tau }=\frac{1}{N}\sum \limits_{i=1}^N\frac{\hat{\alpha_i}\ }{SE\left(\hat{\alpha_i}\right)}\ \mathrm{and}\ {G}_{\alpha }=\frac{1}{N}\sum \limits_{i=1}^N\frac{T\ \hat{\alpha_i}\ }{\hat{\alpha_i}(1)} $$

The standard error of \( \hat{\alpha_i} \)is represented by \( SE\Big(\hat{\alpha_i\Big)} \). The semiparametric kernel estimator of \( {\alpha}_i(1)\ is\ \hat{\alpha_i}(1) \). The remaining two tests are the panel mean tests which suggest that the whole panel is cointegrated. They are given as follows:

$$ {P}_{\tau }=\frac{\hat{\alpha_i}\ }{SE\left(\hat{\alpha_i}\right)}\ \mathrm{and}\kern0.5em {P}_{\alpha }=T\hat{\alpha} $$

The study employed the AMG algorithm estimator of Bond and Eberhardt (2013) to examine the effects of each of the variables on the dependent variable. The AMG technique accommodates nonstationary variable(s) (Destek and Sarkodie 2019) and CSD (Solarin and Al-Mulali 2018; Baloch et al. 2019a). We also used the Dumitrescu and Hurlin (2012) Granger non-causality test to ascertain the direction of causality among the variables. The D-H equation is shown in Eq. 5 below.

$$ {y}_{i,t}={\omega}_i+\sum \limits_{i=1}^p{\lambda}_i^{(p)}{y}_{i,t-n}+\sum \limits_{i=1}^p{\vartheta}_i^{(p)}{x}_{i,t-n}+\kern0.5em {\mu}_{i,t} $$
(5)

\( {\lambda}_i^{(p)} \) and \( {\vartheta}_i^{(p)} \) are the autoregressive parameters and the regression coefficient across countries, respectively. x and y are the underlying variables for n cross-section in t time. The two hypotheses associated with the test are

H0 : β1 = 0 and \( {H}_1:\Big\{{}_{\beta_i\ne 0}^{\beta_i=0} \) ∀i = 1,2...N and ∀i = N + 1, N + 2…N.

Data and model specification

The data spans 1990–2014. The time period is consistent with data availability. The data on EFP was derived from the Global Footprint Network (2019), while the remaining data were obtained from the World Bank (2018). The model for the study is given as

$$ {EFP}_{it}={\varphi}_0+{\varphi}_1{REN}_{it}+{\varphi}_2{NREN}_{it}+{\beta}_0{Z}_{it}+{\varepsilon}_{it} $$
(6)

where EFP is the ecological footprint (global hectares per capita), REN is the renewable energy (% of total energy consumption), NREN is the non-renewable energy (kg of oil equivalent), while Zi represents the other control variables. By adding the control variables into Eq. 6 gives Eq. 7.

$$ {EFP}_{it}={\varphi}_0+{\varphi}_1{REN}_{it}+{\varphi}_2{NREN}_{it}+{\varphi}_3{GDP}_{it}++{\varphi}_4{TRD}_{it}+{\varphi}_5{URB}_{it}+{\varepsilon}_{it} $$
(7)

For uniformity, the per capita form of the variables was used. This was achieved by dividing each of the variables by population.

$$ {efp}_{it}={\varphi}_0+{\varphi}_1{ren}_{it}+{\varphi}_2{nren}_{it}+{\varphi}_3{gdp}_{it}+{\varphi}_4{trd}_{it}+{\varphi}_5{urb}_{it}+{\varepsilon}_{it} $$
(8)

The lower-case letters efp, ren, nren, gdp, trd,and urb are the per capita terms of the variables. The linearized model, after taking the log of each of the variables, is shown in Eq. 9.

$$ {lnefp}_{it}={\varphi}_0+{\varphi}_1\mathit{\ln}{ren}_{it}+{\varphi}_2{lnnren}_{it}+{\varphi}_3{lngdp}_{it}+{\varphi}_4\mathit{\ln}{trd}_{it}+{\varphi}_5\mathit{\ln}{urb}_{it}+{\varepsilon}_{it} $$
(9)

where TRD is the trade (% of GDP), URB is the urbanization (percentage of total population), GDP is the GDP per capita (constant 2010 US$). εi is the error term.

Results and discussion of findings

The results in Table 1 show the smallest and largest mean values of EFP as 18.22 and 19.25 for Indonesia and Colombia, respectively. Considering the mean GDP values, Vietnam is the poorest country (6.774) and Turkey is the richest (9.069). Moreover, on the average, Vietnam has consumed more REN (3.939) than the remaining countries, while Egypt consumed the lowest (1.983). Finally, Vietnam has minimum NREN energy use (5.563), while South Africa has the maximum (7.989).

Table 1 Descriptive statistics
Full size table

Table 2 provides enough evidence that CSD exist among the CIVETS countries. For this reason, the study considers unit root, cointegration, and estimation techniques that account for CSD.

Table 2 Cross-sectional dependence tests
Full size table

Table 3 shows the result of the first-generation, along with the second-generation (CIPS) unit root tests. The four tests (LLC, IPS, and PP-Fisher) affirm that all the variables are I(1). Since these three tests are not robust amidst CSD, the results were complemented with the CIPS tests, to make up for the flaws of the three tests. The findings of the CIPS test revealed that all the variables are I(1). These findings give reason to proceed with the cointegration tests.

Table 3 Unit root tests
Full size table

Table 4 shows the Pedroni cointegration test. The findings suggest that 6 statistics out of 11 statistics presented are significant at 1%. This confirms the presence of cointegration among the variables. The Johansen Fisher cointegration test in Table 5 is also in harmony with the results in Table 4.

Table 4 Panel cointegration test (Pedroni)
Full size table
Table 5 Panel Johansen Fisher cointegration test
Full size table

Since both tests (Johansen Fisher and Pedroni) do not account for CSD, the Westerlund cointegration test provides a better option as the test gives robust estimates in the presence of CSD.

The group mean tests (Gt and Ga) and the panel mean tests (Pt and Pa) in Table 6 support a cointegrating relationship among the variables. See Table 7 for the effects of each of the variables on the dependent variable (EFP).

Table 6 Panel cointegration test (Westerlund)
Full size table
Table 7 AMG estimate result
Full size table

From the results in Table 7, NREN increases environmental degradation by about 57%. As earlier mentioned, NREN is the dominant energy source in these countries. The consumption of NREN energy is responsible for environmental deterioration in these countries. Some of the NREN sources consumed in these countries include coal, natural gas, crude oil, and uranium. NREN sources do not ensure environmental sustainability. These NREN sources cannot be readily replaced by natural means at a quick enough pace to keep up with consumption. This finding is in consonance with earlier studies that concentrated on emerging economies/developing economies. See, for instance, Zhang et al. (2019) and Oganesyan (2017) for BRICS, Muhammad (2019), Gorus and Aydin (2019), Gorus and Aslan (2019), and Al-Mulali and Che Sab (2018) for MENA, Ssali et al. (2019), Nathaniel and Iheonu (2019), and Esso and Keho (2016) for SSA, Sinha et al. (2019) and Sinha et al. (2017) for the Next-11countries, Dogan and Turkekul (2016) and Dogan and Ozturk (2017) for the United States, and Dogan and Seker (2016) for the European Union.

Findings further revealed that economic growth is not particularly harmful to the environment in these countries, rather it adds to environmental quality. However, the impact of economic growth on environmental quality is not significant. The insignificant influence of economic growth on environmental quality is consistent with the studies of Nathaniel et al. (2020) for MENA countries and Khan et al. (2020) for Pakistan, but contradicts those of Ulucak et al. (2020) for BRICS countries, Liu et al. (2020a, b) and Ahmed et al. (2020) for G7 countries.

On the other hand, the consumption of REN mitigates environmental degradation (though insignificantly). This points to the meager investment and inadequate usage of REN in these countries. This finding suggests that environmental quality could be enhanced if these countries invest and promote the usage of REN. The CIVETS countries, over the years, have concentrated more on how to enhance economic growth without paying adequate attention to the quality of their environment. These do not in any way suggest that these countries have not invested in REN, but the investment in REN is still meager, and its consumption remains inadequate. On the flipside, urbanization contributes about 0.20% to environmental degradation in the CIVETS countries. This is indeed revealing. Urbanization increases energy demand. Since NREN is mostly consumed by these countries, the tendency for energy consumption to add to environmental degradation abounds. This is the tale of the CIVETS countries. Policies are indeed needed to curb this upward surge in urbanization, since the demand for energy is not expected to decline any time soon.

Unlike energy consumption and urbanization, trade exacts no harmful impact on the environment. A 1% increase in trade is expected to increase environmental quality by 0.10%. This is an indication that the CIVETS countries are not involved in trade activities that deteriorate environmental quality. This outcome is consistent with the findings of Sinha et al. (2019) for both BRICS and N-11 countries.

For country-specific cases, NREN promotes environmental degradation in all the countries, but the situation is even more severe in Turkey. Studies like Pata (2018a), Karasoy (2019), and Pata (2018b) have confirmed the horrendous effects of NREN on environmental quality in Turkey. On the other hand, Román et al. (2018), Kurniawan et al. (2018), Shahbaz et al. (2019) and Al-Mulali et al. (2015a, b), Ibrahiem (2016), and Rafindadi and Usman (2019) have discovered a similar results for Colombia, Indonesia, Vietnam, Egypt, and South Africa, respectively. This confirms the negative impact of NREN on environmental quality and also justifies the ubiquitous calls for the consumption of REN in emerging economies like the CIVETS countries. Economic growth improves environmental quality in Colombia, South Africa, and Turkey, but not in Indonesia, Vietnam, and Egypt. This affirms that Colombia, South Africa, and Turkey are not growing at the expense of their environment, while growth is particularly harmful in the remaining countries. Earlier studies like Kurniawan and Managi (2018) and Alam et al. (2016) discovered that economic growth is not environmentally friendly in Indonesia.

Interestingly, REN consumption reduces environmental degradation insignificantly in Colombia, South Africa, Turkey, Indonesia, and Egypt. Pata (2018c) had earlier reported that REN was not at a desirable level to reduce environmental degradation in Turkey. This goes to show that investment in REN among these countries is still infinitesimal. REN investment in CIVETS countries has not yielded the desire impact, which is to reduce environmental deterioration. There is a dire need for the CIVETS countries to increase their investments in ‘clean’ energy sources in order to achieve sustainable growth amidst low emissions (Chen et al. 2019a; Ghazali and Ali 2019; Baloch et al. 2019a; Ma et al. 2019; Chen et al. 2019b; Baloch et al. 2019b).

Urbanization happens to be a significant problem in most of the counties; more severe in Indonesia, but less severe in Colombia. A 1% increase in urbanization reduces environmental quality by 0.50% in Egypt, 0.71% in Indonesia, 0.56% in South Africa, 0.33% in Turkey, and 0.20% in Vietnam. Indonesia, for instance, has experienced an unprecedented increase in urbanization over the years. More than one half of Indonesians settles in the urban areas (UNDP 2017). It has also been projected that coal (NREN) will be the major source of energy to meet the energy demand of the urban population in Indonesia mainly due to its large reserve (Kurniawan and Managi 2018). The same problem (urbanization) is witnessed in Vietnam where basic facilities and economic activities are concentrated in Ho Chi Minh and Hanoi City. Fan et al. (2019) discovered the same result for Vietnam, while Al-Mulali et al. (2015a) noted that urbanization, energy consumption, industrialization, and natural resource consumption are culpable for the increasing environmental pressure in Vietnam. Also, South Africa is the most urbanized country in SSA (Salahuddin et al. 2019). Recent studies like Salahuddin et al. (2019) and Sarkodie and Adams (2018) discovered that urbanization deteriorates the environment in South Africa. Trade increases environmental quality significantly in Egypt, Indonesia, and Turkey, insignificantly in Colombia and South Africa, but adds to environmental quality in Vietnam. This suggests that the Vietnamese government has been involved in harmful trade with its trading partners. This is in line with the findings of Shahbaz et al. (2019), Jha and Mani (2006), and Al-Mulali et al. (2015a) for Vietnam.

Since effect does not necessarily mean causation, the D-H Granger non-causality test, in Table 8, presents the direction of causality among the various variables. From the findings, EFP drives GDP, energy use, and trade, while urbanization drives REN and EFP. This points to the fact that urbanization is really a serious problem in these countries. This is a peculiar problem in emerging economies. At the initial stage of development, social amenities and other facilities that make life meaningful are mostly concentrated in the urban centers.

Table 8 D-H Granger non-causality test
Full size table

Interestingly, GDP drives energy use in the CIVETS countries. This result confirmed the growth-led energy consumption hypothesis in these countries. This complements the findings of Chen and Fang (2018), Kirikkaleli et al. (2018), and Nyasha et al. (2018) and calls for caution as regards the energy sources consumed in these countries. For growth to be sustainable, the energy source should be one that does not pollute the environment. A feedback causality was further discovered between trade and economic growth. This is an indication that trade is core among the variables that promote economic growth and advancement in these countries. The MG and the CCEMG estimator tests confirmed the robustness of the results in Table 8. See Table 9.

Table 9 Check for robustness
Full size table

Controlling for CSD and endogeneity, results from the MG and CCEMG estimations reveal that NREN and trade exhibit statistically significant relationships at 1% and 5% levels, respectively. The coefficients of all the five variables are similar to those of the AMG in sign and magnitude. Hence, a similar interpretation holds.

Conclusion and policy directions

This study investigated the effect of renewable energy, urbanization, economic expansion, and trade on the ecological footprint in CIVETS countries by applying the augmented mean group estimator. With the evidence of CSD, the second generation unit root and cointegration tests were applied. Findings confirmed long run interaction among the variables. REN (though not significant), economic expansion, and trade reduce environmental degradation, while NREN and urbanization deteriorate the quality of the environment. Urbanization arises from discrepancies in development factors. These factors include household income, basic amenities, infrastructural provisions, etc. The inadequate provision of these factors in the rural areas will encourage urbanization. Therefore, to address this issue, the provision of basic infrastructures in the rural areas in the CIVETS countries will help mitigate the upward surge in urbanization and all the anomaly it creates. The consumption of NREN was the major contributor to environmental degradation. Policymakers in these countries should invest and encourage the consumption of REN (like solar, tide, wind, hydropower, geothermal, etc.). These REN sources are different from fossil fuels, because they are clean and low in emissions. An improved investment in environmentally friendly technologies will not only curtail emissions, but also enhance growth.

This study was conducted on a panel of CIVETS countries. The study was limited by data availability which informed the exclusion of some determinants of ecological footprint. The future research should be conducted on the European Union and BRICS countries and compare the findings to obtain more robust inferences which will promote policies that will aid the attainment of the SDGs by 2030.