Modelling the Impact of Energy Consumption, Natural Resources, and Urbanization on Ecological Footprint in South Africa: Assessing the Moderating Role of Human Capital

Apart from being the most urbanized and largest emitter of CO2 emissions in Africa, South Africa currently harbours an ecological deficit territory mainly due to its desire to attain more growth with less attention to its natural environment. Since environmental distortions are mainly an outcome of human activities, this study becomes the first to investigate the linkage between natural resource (NR), human capital, energy consumption, and ecological footprint (EF) in South Africa. Findings from the long-run results reveal that energy consumption, natural resource, economic growth, and urbanization add to the EF, while human capital curtails environmental deterioration. The interaction between human capital and urbanization helps in mitigating environmental degradation, which suggests a moderating effect of human capital in urban sustainability. The CCR, DOLS and FMOLS confirm the robustness of the findings. A feedback causality exists between natural resource and economic growth, and between economic growth and EF. Natural resources and urbanization also Granger cause EF. Lastly, policies relating to urban and natural resource sustainability and human capital development are discussed.


INTRODUCTION
The benefits of economic growth are hydra-headed. It can provide the needed infrastructures, reduce poverty, and improve people's living standard. On the flip side, it has its downsides, especially when an economy pays less attention to its natural environment while intensifying its desire for affluence (Meo et al., 2020a(Meo et al., , 2020bAhmed et al., 2019aAhmed et al., , 2019bNathaniel, 2019;Uddin et al., 2019;Omojolaibi and Nathaniel, 2020). In Africa, South Africa is arguably the most developed. This development comes with improved welfare, employment generation, FDI inflow, export expansion; however, keeping paste with this development, South and forest products. It is measured in global hectares of land (gha). Recent studies have adopted the EF to measure anthropogenic activities on our natural environment (Solarin et al., 2021;Nathaniel, 2020;Zameer et al., 2020;Nathaniel et al., 2020aNathaniel et al., , 2020bDanish et al., 2020;Destek and Sinha, 2020;Sharif et al., 2020;Usman et al., 2020;Jin et al., 2020;Altıntaş and Kassouri, 2020;Dogan et al., 2020;Yilanci and Pata 2020;Baz et al., 2020;Omoke et al., 2020). The link between NR and EF is heralded with lots of conflicts. For instance, economic growth which comes with industrialization and urbanization promotes NR extraction, and even its consumption, which in turn can prompt environmental deterioration by increasing the EF (Murshed et al., 2020;. Balsalobre-Lorente et al. (2018) argued that NR could shift a country's attention from non-renewable to renewables. As such, fossil fuel consumption may be minimized. Anthropogenic activities like deforestation, bush burning, mining, etc. destroy the biodiversity, soil, water, and air. All these arguments have opened the door to further investigation into the NR-EF nexus. Studies like (Danish et al., 2020;Zafar et al., 2019) reported that NR does not harm the environment, while Ahmed et al., 2020a;, Hassan et al., 2019 have confirmed that NR increases the EF.
This study is super useful for South Africa where economic advancement has intensified NR extraction, especially coal, due to large energy demand and the desire to earn foreign exchange. Coal accounted for about 72% of energy in South Africa in 2005. This was way above that of China, India, USA, South Korea, and Japan with 63.4%, 38.7%, 23.8%, 23.1%, and 21.1% respectively. South Africa generates 95% of its electricity from coal (Magazzino et al., 2020). Apart from being abundant and cheap, coal is a pollutant and non-renewable. It adds to emissions and truncates environmental sustainability. Unfortunately, coal, a natural resource, is a major source of energy in South Africa, its exploration and consumption have been found to environmentally harmful Udi et al., 2020;Magazzino et al., 2020;. The urbanization rate in South Africa has been on a stable rise. It was 62.4% in 2012 and 65.3% in 2016 (WDI, 2019). South Africa is the biggest economy in sub-Saharan Africa (SSA) with the highest CO 2 emissions of 42.8%, and the 14 th highest emitter in the world (Salahuddin et al., 2019;Ndoricimpa, 2017).
NR exacts an impact on the EF (Zafar et al., 2019). NR like the forest, developed lands, croplands, and grazing lands declines human-caused CO 2 emissions (GFN, 2018). However, NR like oil and coal reduce environmental quality (Ahmadov and van der Borg, 2019). NR is also closely linked with an economies income. As development starts, more energy (more NR) is consumed with little or no attention to the quality of the environment. As development persists, recognition is given to renewables as people increase their demand for NR preservation, clean environment, and energy-efficient commodities. Hence, an improvement in environmental quality. This explains the EKC hypothesis .
Economic growth engineer's industrialization which promotes NR extraction. As NR consumption increases through means like mining, deforestation, and agriculture, the environment could be adversely affected . NR extraction may improve income, but it reduces the biocapacity and increases the EF. Economic growth intensifies NR extraction which drives the EF (Panayotou, 1993;Ozturk and Al-Mulali, 2015). Sustainable management practice is sacrosanct for resource regeneration. The growth of an economy is key to urban population increase. On the other hand, urbanization increases energy and transportation demand, and industrialization which requires more energy (nonrenewable) that increases the EF. As the income urban dwellers improve as a result of urbanization, renewables could form a large chunk of the energy mix, hence the EF decreases (Danish and Wang, 2019). Education and skilled human capital are needed for the sustainable use of NR. Human capital will contribute to societal readiness to adopt energy-efficient and environmental-friendly technologies (Zafar et al., 2019;Ozturk et al., 2016). Human capital has ecological benefit, and when human capital is skilled it can promote growth from different facets, and curtail urban anomaly (Ahmed, 2020b). This is the main reason why we included human capital in our study. The study contributes in several strands to the existing literature. (i) it investigates the association between ecological footprint and natural resource in the presence of urbanization, human capital, energy consumption, and economic growth. This is the first study to explore the relationship between these variables in South Africa's context. (ii) we introduce the interaction term between urbanization and human capital in the model. This will help us identify some new dimensions of urban sustainability, and if human capital moderates the relationship between EF and urbanization. (iii) We test for the EKC for ecological footprint in South Africa, applied the Bayer and Hanck (BH) (2013) cointegration test, and causality test amidst the ARDL estimation technique. These econometric procedures are not only appropriate but produce reliable outputs.

LITERATURE REVIEW
We divide the literature into two subsections for easier understanding and to make it systematic in a way. The influence of urbanization, as well as, natural resource on the environment is still murky. The outcomes of various researches in this regard have been greeted with lots of discrepancies. Methodology, dataset, region considered, and estimation techniques are the main culprits for these irregularities.

Energy Consumption, Urbanization, Economic Growth and Ecological Footprint
Urbanization has received lots of attention as a potential determinant of EF of recent. Nathaniel (2020) investigated the impact of energy consumption, urbanization and economic growth on the EF in Indonesia from 1971 to 2014. The findings alluded to the fact that both variables add to the EF in Indonesia. Also, economic growth drives the EF. The author called for the use of renewables in Indonesia if the country desires to sustain its growth trajectory. Baloch et al. (2019) examined the effects of urbanization and economic growth on the EF in 59 countries from 1990 to 2016. Similar to Nathaniel (2020), both variables, coupled with energy use increase the EF. Hassan et al. (2019a) conducted a similar study for Pakistan; the results remained consistent with the other reviewed studies in terms of the deteriorating impact of economic growth and energy consumption on the environment. Still for the case of Pakistan, Hassan et al. (2019b) controlled for the influence of biocapacity and human capital in the growth-environment nexus. Economic growth was found not to be harmful as it reduced the EF by 0.60%. Dogan et al. (2019) were the first to conduct a study on the driving factors of EF in MINT countries. Urbanization appears to be the chief cause of environmental deterioration in MINT countries. In contrast to previous studies,

Natural Resources, Human Capital, and Ecological Footprint
There are studies on NR-EF nexus, but only a few studies have added human capital to the nexus. For instance, Ahmed et al. (2020a) tried to establish a link between human capital and EF for G7 countries. Their findings revealed the impact of human capital in mitigating degradation by reducing the EF. However, urbanization exacts an opposite influence on the EF. The findings were indeed revealing. Human capital is developed in G7 countries compared to other emerging economies. An improvement in human capital could just be the missing link to the attainment of environmental preservation. The studies of Ahmed et al. (2020a) is analogous to those of Ahmed et al. (2020b)

Data
This study relied on annual data spanning 1970-2016 for South Africa. The data on EF is mainly responsible for the time period s it ends in 2016. All the variables were converted into natural logarithm following the study of Ahmed et al. (2020b). The models to be estimated are stated below: ψ 0 is the drift constant. u, t, s, r, q, and p are the lag lengths. The long-run multipliers are ψ e , ψ g , ψ n , ψ h , ψ u , ψ i , and ψ z . The white noise and first difference operator are respectively μ t and ∆. Where EF, GR, NR, HC, UB, IN, EN, and GR 2 represent ecological footprint, economic growth, natural resource, human capital, urbanization, interaction term, energy consumption, and the square of GDP respectively. The main focus is on Model 1. We controlled for the interaction term and the square of GDP in Model 2 and 3 respectively. Table 1 for the measurements and sources of the variables.

Unit root
Unlike most previous studies that adopted the conventional unit root tests like the Dickey and Fuller (1981) and Phillips and Perron (1988), we applied three tests that do not only improved on the weaknesses of the aforementioned tests, but also account for breaks. One of these tests is the Zivot and Andrews (ZA) (1992). The ZA test suggested the three models are shown in Eq. 4, 5 and 6. ' DT t and DU t are the trend shift and the dummy variable

Cointegration tests
In addition to the bounds test, the BH was applied. The benefits of the BH test are numerous. Cointegration tests (like Johansen and Juselius, 1990;Banerjee et al., 1998;Boswijk, 1995;Johansen 1991;Engle and Granger, 1987) have their weaknesses. For instance, inconsistency in one-step, in relation to the Engle and Granger (1987) test, which is a two-step test, could be transferred to the next step. For the Johansen and Juselius (1990) and Johansen (1991) tests to be efficient, the sample must be large, and all variables must be I(1). The BH test adopts a combined approach. Hence, overcoming the weakness of the previous test (Ahmed et al., 2019a).
The Fisher form of the BH test is given as: ρ JOH , ρ BDM , ρ EG , and ρ BO are the test probabilities of individual cointegration tests.

ARDL technique
The ARDL technique of Pesaran et al. (2001) was preferred because it is not biased to small sample size (Keho, 2019). It can simultaneously correct for endogeneity and serial correlation inasmuch as the lag length is efficiently modified. Variables with different optimum lag lengths and integration are accommodated, except for I(2) variables . The Toda and Yamamoto (TY) (1995) was used to check for the direction of causality among the variables. Apart from being the modified version of the Wald test, it is way superior and provides robust estimates than the conventional Granger causality test, irrespective of variables order of integration.

RESULTS AND DISCUSSION
From Table 2, GR 2 has the highest average. Also, NR remains the most volatile of the variables. All the variables are positively skewed except EF. The variables are platykurtic; indicating less or few extreme outliers. The normality of the variables is confirmed based on their probability values, but for HC.
Unit root tests are needed to guide the aid the direction of the analysis and techniques to be applied in this study. The unit root results are presented in Tables 3 and 4. Here the DF-GLS, Ng and Perrron (2001), and the Zivot and Andrews (ZA) (1992) was applied. The first two tests are in harmony. They confirmed a mixed   level of integration, but the later (ZA) affirmed stationarity at I(1). Structural breaks exist in the series of EF, GR, NR, HC, and UB in the year 2003, 2004, 1994, 1987, 2001, etc. These breaks are due to economic and political reforms in the country.
For robustness, two cointegration tests were applied for the three models (Tables 5 and 6). The outcome of the tests is in the affirmative. They both confirmed that a long-run relationship is evident in the models.
Since the F-stat from the three models is greater than the critical values at 5% levels, we can't deny the existence of a long-run relationship in Table 5. The BH test requires the values of both EG-JOH and EG-JOH-BO-BDM to be greater than the 5% critical values. Since that is the case in Table 6, then cointegration exists. Table 7 reports the ARDL results for the three models. The long-run results are the same in terms of the impact of each of the variables on the EF. Growth, NR, energy consumption, and urbanization add to environmental degradation, while human capital promotes environmental quality. South Africa, like other BRICS countries, have witnessed a fairly stable growth over the years and is arguably the most developed in the continent. However, one important factor that has contributed to the country's growth is energy consumption Magazzino et al., 2020). Unfortunately, the country consumes more of non-renewable energy sources. These energy sources are pollutants and their impact on the environment could be far-reaching. This is consistent with the findings of Danish et al. (2020) for BRICS; Ahmed et al. (2020a) and Liu et al. (2020) for G7 countries. NR can either promote or deteriorate the environment depending on the resource and how it is being explored. In South Africa, coal is one of the most explored resources. Coal is finite, it is non-renewable and hence pollutes the environment. Studies have confirmed coal consumption to be particularly harmful in South Africa (Udi et al. 2020;. This finding complements the studies of Hassan et al. (2019) for Pakistan, Ahmed et al. (2020b)  This finding is evidence that the consumption and exploration of NR have not been sustainable in South Africa. The country has exacted enormous pressure on its NR to meet its energy demands. South Africa generates 95% of its electricity from coal. Some researchers are of the opinion that NR abundance should reduce a country's reliance on energy import, with a gradual shift to lesspolluting sources which will abate environmental degradation. But, the case is, however, different for South Africa where coal is a major energy source. Just like economic growth and energy consumption, urbanization also exacts a negative impact on the environment. This confirms the earlier findings of Sarkodie and Adams (2018), Salahuddin et al. (2019), andNathaniel et al. (2019) for South Africa. This finding is intuitive and appealing because urbanization triggers economic growth and other social  Source: Authors computation. ***and **represent 0.01% and 0.05% significance levels respectively  Source: Authors' computations. Note: ***, ** and * represent statistical significance at the 1%, 5%, and10% levels of significance respectively. t-statistics are in parentheses activities that promote energy consumption. Urbanization in South Africa is linked to industrialization that is characterized by massive energy use which stimulates environmental deterioration. This is, however, not the only way urbanization contributes to environmental degradation in South Africa, it could also contribute through enormous waste generation, and increasing demand for water, food, transportation, and other resources.
Human capital reduces the EF. This reason is justified as human capital has played a key role in South Africa's drive for economic prosperity. An educated human capital will crave for a better environment, which will propel an efficient use and conservation of NR, energysaving, and the adoption of environmental friendly technology at the collective and individual level. Human capital is also key for financial development, and could serve as a pathway through which NR can trigger growth (Zallé, 2019). This outcome is in consonance with those of (Zafar et al., 2019;Ahmed and Wang, 2019).
Interestingly, the interaction term exhibits a significant and negative coefficient inferring the moderating role of human capital in mitigating environmental deterioration in South Africa. Urbanization increases the EF, but its interaction with human capital reduces it. This provides a great insight that human capital is germane for urban sustainability. We further observed the existence of the EKC for ecological footprint in South Africa evident from the positive/negative coefficient of GDP/GDP 2 in Table 7, Model 3. This is a confirmation that pollution reduces as growth persists in South Africa. The short-run outcomes are consistent with those of the long-run but for the inconsistent impact of human capital on the EF. The short-run evidence affirmed that human capital is   not potent enough to reduce the EF in Model 1 and 3. This calls for human capital development for a better environment in South Africa. Table 8 confirmed the robustness of our findings in Table 7. The three results (FMOLS, DOLS, and CCR) are in harmony with all the long-run findings; therefore similar discussion/interpretation applies.
The causality test in Table 9 showed different directions of causality.
A feedback causality between NR and GR, and GR and EF revealed that Natural resource plays an important role in South Africa's growth trajectory and also contributes to pollution. As such, there is a need to ensure its efficient exploration and consumption. A unidirectional causality from EN to EF, EN to UB, and from EN to GR re-emphasizes urban sustainability, sustainable exploration, and the need for clean energy consumption in South Africa.
The CUSUM and CUSUMsq plots for each of the models confirmed that the models are stable and can be used for forecast.
Cusum and Cusumsq Plots for Model 3

CONCLUSION
This study investigated the linkage between natural resource, energy consumption, urbanization, human capital, and EF in South Africa. DF-GLS and Ng-Perron tests alluded to a mixed level of integration, while the BH test affirmed that cointegration abounds in the series. The ARDL, complemented with the FMOLS, DOLS, and CCR results confirmed that economic growth, energy consumption, natural resource, and urbanization increase the EF, while human capital reduces it in the long-run. Results further suggest that the EKC for EF exists for South Africa, and the interaction of urbanization and human capital can ensure urban sustainability in the country. A feedback causality exists between natural resource and economic growth, and between economic growth and EF. Natural resources and urbanization also Granger causes EF. These results have policy implications and calls for reasonable policy directions.
Since natural resource drives growth, and coal consumption is a major natural resource used for electricity generation and other activities, and natural resource increases the EF; the need arises for policymakers to adjust the country's energy portfolio and make a paradigm shift to clean energy sources. This might not be an easy sail considering the country's financial wellbeing. However, creating awareness, providing the household with palliatives (subsidies, tax and interest rate holidays, etc.), encouraging firms to embark on cleaner production while taxing the dirtier ones could be a good step in the right direction. The country also needs to gradually shift attention to other natural resources that are less of a pollutant and ensure its sustainable exploration, without altering the growth process. These will ease the difficulties associated with attaining SDG 7 (Affordable and Clean Energy).
Development issues like inequality, lack of amenities, and low income are the major cause of urban explosion. The concentration of infrastructures in cities like Johannesburg, Pretoria, Durban, and Cape Town, and a dearth of such infrastructures in Ntabankulu, Nkangala, and uMgungundlovu will only encourage urbanization as people will prefer to live in cities where they can easily make earns meet with fewer efforts. The need for smart cities cannot be overemphasized. Smart cities enhance the performance of urban services like transportation and energy to attain sustainability and innovation. All these will put the country in the pathway of achieving the SDG 11 (Sustainable Cities and Communities). Most importantly, human capital needs to be developed in South Africa for sustainable communities, cities and energy. Once human capital is developed, the demand for renewables will increase, and the sustainability of the environment will be a priority.