Do Renewable Energy and the Real Estate Market Promote Environmental Quality in South Africa: Evidence from the Bootstrap ARDL Approach

Abstract

Recent empirical research indicates that South Africa’s present level of wealth and energy, given its fast-expanding population, is unsustainable. Studies in this domain focus on the impact of economic growth and energy use on environmental quality; the role of the real estate market on environmental quality in South Africa is ignored in the emerging literature. The current study aims to deliver a fresh empirical analysis in this context by analyzing the impact of South Africa’s real estate sector expansion and renewable energy sources on carbon emissions. Using the newly developed “bootstrap autoregressive distributed lag (ARDL) approach”, the results of the empirical investigation showed that renewable energy improves South Africa’s environmental quality. The current research also shows that the South African real estate industry has a negative impact on the environment. According to the current research, South African policymakers should create new regulations for the sustainable real estate sector to improve environmental quality by encouraging the usage of and investment in renewable energy.

1. Introduction

Globally energy demand is rising due to economic activities and is one of the most pressing challenges in the twenty-first century because of its significant consequences on global warming [1,2,3,4,5,6,7]. According to global energy balance figures, the world’s energy supply has increased from 6.098 Mtoe in 1973 to 14.282 Mtoe in 2018 (International Energy Agency (IEA), 2020), primarily due to high demand for consumption in economic activities. Because nonrenewable energy sources (fossil fuels) account for 80% of total energy demand [2], the energy sector is the major source of “greenhouse gas (GHG) emissions”, particularly carbon dioxide emissions (CO2), accounting for around two-thirds of overall emissions in the environment [8]. Under the sustainable development goals (SDG-7), environmental degradation is a critical concern to ensure sustainable economic growth (SEG), particularly in developing economies [9,10,11,12,13]. Although many factors have fatal effects on environmental quality, CO₂ is the leading factor in the dangers of temperature, weather, and the ecosystem of the environment [14,15,16]. Therefore, international agreements (Paris Agreement 2015, Kyoto Protocol) have been established by global economies to combat global climate challenges, reduce CO2 emission levels, and obtain sustainable economic growth without compromising ecological quality [17]. Numerous research papers have examined the role of several factors such as FDI (foreign direct investments), “financial development, renewable energy, and the population” on environmental quality [18,19,20,21,22,23,24,25,26,27]. Based on our knowledge, only a handful of limited studies, such as those by Qashou et al. [28] and Wei and Kong [29], have explored the consequence of the real estate market on environmental quality.

Over the past two decades, green construction and sustainability have been persistent themes in the sector. However, the real estate industry is still evolving. Green buildings are considered the norm and mainstream and are expected to provide sustainable features in homes and commercial buildings. According to the National Association of Homebuilders, “homebuyers want—and will pay more for—sustainable features like energy-efficient appliances, windows, and the like, alongside features that ensure better air quality. As younger generations are expected to enter the homebuying market, this socially conscious group is looking out for green features and sustainability that is built in”. On the other hand, several of these concepts should hold true for investors wishing to purchase or enhance rental homes. The greatest renter pool is probably made up of youngsters, and possibly Gen Z, of today who are concerned about reducing their carbon footprint. Therefore, to enrich the existing literature, more studies are needed to validate the findings on the subject referred to. Consequently, the relationship between real estate and environmental quality in the context of South Africa is the primary topic of this research article.

South Africa’s economy is the continent’s most developed, high-tech, diverse, upper-middle-income, and third-biggest economy. The gross domestic product (GDP) of South Africa rebounded in 2021 and increased by 6.4% compared to a pandemic-induced 6.4% decline in 2020. According to Ali et al. [30], South Africa is expected to become the world’s leading producer of economic growth in the near future among the “BRICS economies (Brazil, Russia, India, China, and South Africa)”, due to rapid economic growth.

Energy is the lifeblood of the South African economy, and obtaining, processing, and disseminating energy commodities and services is a major enterprise that creates jobs and value across the economy. As a result, South Africa’s usage of nonrenewable energy sources such as fossil fuels (FF) has grown dramatically. In 2018, coal dominated the energy supply with 65% of the primary energy share; however, crude oil, renewable energy, natural gas, and nuclear energy accounted for 18%, 11%, 3%, and 2%, respectively, of South Africa’s energy supply (South African Energy Sector report, 2021) [31]. In addition to that, Figure 1 shows that fossil fuel energy consumption is the primary energy source in South Africa when compared to renewable energy consumption.

Source: Author

The South African real estate market has improved in recent years due to the increase in economic activity and the growth of urbanization. The real estate market became one of the strongest sectors in South Africa, despite subsequent lockdowns and uncertainties surrounding Coronavirus (COVID-19). In 2020, first-time buyers boosted and fueled the market, taking advantage of rate cuts and low-interest rates (South Africa Residential Real Estate Market, 2021) [32]. The real estate sector plays a critical role in its growth. Despite the economic damage caused by Coronavirus disease (COVID-19), the real estate sector potentially contributes ZAR191 billion, with a direct contribution of ZAR46 billion, to national income. Furthermore, increased spending and lending in the real estate market injects much-needed liquidity into the economy.

Our study advocates that the real estate market may influence carbon emissions via two channels: energy usage and sustainable growth. Household wealth and borrowing rates will benefit from a substantial recovery in the real estate sector, resulting in more investment and sustainable economic development. Many empirical research studies have shown that the real estate sector has a substantial influence on economic growth in this environment [28,33,34,35,36,37]. These studies show that when the real estate market improves, so do household spending and company investment, resulting in better economic growth rates. The considerable effect of the real estate market on economic activity will result in a surge in investment in this sector. As a result, non-renewable energy use increases, contributing to higher emissions.

The current study addresses a vacuum in the literature by filling it in three strategic ways based on the deficiencies identified in earlier empirical investigations. First, a handful of limited research studies have endeavored to explore the effect of the real estate market on environmental quality, such as those by Qashou et al. [28], who explored how the real estate industry in Turkey affected environmental quality, and Wei and Kong [29], who explored the influence of real estate development on environmental degradation in a “panel of 30 provinces of China” spanning data from 2000 to 2013. It is worth mentioning that none of the researchers have looked at how the real estate sector affects environmental quality in Africa. The primary distinction of the current study is that it evaluates the impact of South Africa’s real estate sector, the region’s biggest polluter, on environmental quality. As an indication of environmental quality, the research utilized emissions calculated in “metric tons per capita”. Additionally, the goal of the current study is to evaluate the impact of the real estate market on the degree of environmental quality using real income, renewable energy, and non-renewable energy for the data collected between 1988 and 2018. Second, the available empirical research has shown a methodological deficit. Several techniques, including “ARDL and Granger causality”, have been used in recent empirical research to examine the influence of renewable energy and economic growth on environmental quality. The “ARDL bootstrap model” is rarely utilized. As a result, the main objective of the current study is to provide fresh empirical evidence by employing a new ARDL cointegration approach. According to McNown et al. [38], the new “ARDL test” is preferred over classic Pesaran [39] “co-integration tests” due to its capability to foresee when resolving power and size constraint concerns, as well as other aspects that traditional co-integration tests do not address. Furthermore, this study uses “Zivot and Andrews [40] (ZA) and Clemente-Montaes-Reyes [41] (CMR)” to assess integration due to its advantage of taking into account the dates of “structural breaks”. Finally, this intends to highlight a number of key policy implications for government and policymakers in South Africa for sustainable development utilizing real estate market indexes, income, and non-renewable and renewable energy channels.

The following is a breakdown of the research structure: Section 2 of this paper delineates the analysis of the empirical literature review and the finding of existing gaps in the literature. Section 3 turns the spotlight on the methods that have been harnessed to accomplish the objectives of this study. Following the conclusion and policy implications in the final part, the empirical data and discussion are provided in Section 4.

2. Literature Review

2.1. Renewable Energy and Environmental Equality

The unequivocal desire to reduce CO₂ emissions has prompted the need for clean sources of energy. Consequently, researchers from the field of energy and environmental economies have given it paramount consideration. The literature on “renewable energy (RE) and CO₂ emissions” is getting much admiration day by day, notably in various developed economies. However, in South Africa, the literature is stranded. Various indicators have been employed to determine “environmental quality”, ranging from CO₂ emissions to “ecological footprint (EF) and material footprint” which depict negative and positive indicators, respectively. The goal of this study is to see how renewable energy and the real estate market affect South Africa’s environmental quality. This necessitates further in-depth research on the issue. Panel data analysis and time series have both been used in studies to undertake research. For panel data analysis, see [42,43,44,45,46,47]. Between 1995 and 2011, Apergis et al. [48] examined the link between renewable energy and CO₂ emissions in Sub-Saharan Africa (SSA). The outcome demonstrated that RE improves environmental quality. Similarly, da Silva et al. [39] in the case of 17 SSA, analyzed the determining factor of RE. The results revealed that “fossil fuel price, population growth, imports, and CO₂ emissions” are the critical hurdles to RE use. Hu et al. [44], for 25 underdeveloped nations from 1996–2012, investigated the “role of trade in commercial service and RE in CO₂ emissions”. The results indicated that RE decreases CO₂ emissions. Furthermore, they went on to say that “trade in commercial services and use of RE can help these countries achieve green growth”. [49]. Bekun et al. [50] used a “Pooled Mean Group” (PMG) in the case of 16 European Union (EU) nations and established that RE is a vital mitigant of CO₂ emissions. They further reiterated that when green energy sources are used, SDGs are attainable. Yazdi and Beygi [51] used the PMG between 1985 and 2015 for 25 African countries. The analysis indicated that RE reduces CO2 emissions, proving the validity of the Environmental Kuznets Curve (EKC) postulate. Additionally, Haldar and Sethi [52] probed the link between RE, GDP, and emissions in underdeveloped nations and affirmed the presence of the EKC postulated. For 25 SSA nations, Zoundi [53] found that RE decreased CO2 emissions. However, in the short and long term, the influence of RE is eclipsed by non-RE. A similar finding was affirmed by Inglesi-Lotz and Dogan [54], who used a “panel evaluation method robust to cross-sectional dependence” for 10 SSA nations from 1980 to 2011 to examine the impact of RE versus non-RE on emissions. Nathaniel and Iheonu [42] examined the influence of ‘renewable and non-renewable energy consumption’ on CO₂ emissions for 19 African nations, using the “Augmented Mean Group (AMG)” for the time period 1990–2014. The outcome indicated that RE decreases emissions insignificantly, while non-RE increases CO₂ emissions. Sahoo and Sethi [55] “explored the link between RE, non-RE, human capital, globalization, and natural resources on ecological footprint from 1990 to 2016, using the Westerlund cointegration method, Mean Group (MG), Augmented Mean Group (AMG) and Dynamic common correlated effects (DCCE) for developing countries”. The “Fully Modified Ordinary Least Square-FMOLS and Dynamic Ordinary Least Square-DOLS approaches” are used to test for robustness, as is the “Dumitrescu and Hurlin causality test”. The discoveries showed that non-RE reduces the quality of the environment while RE improves it. Xue et al. [56] examined the effects of RE use on the environmental footprints of four South Asian “fossil fuel-dependent nations: Bangladesh, India, Pakistan, and Sri Lanka using the panel cross-sectional dependency, slope heterogeneity, and structural break issues from 1990–2016. The results disclose that RE use increases ecological quality while non-RE declines it. Similarly, using the same method and variables in eight Asian nations, Sharma et al. [57] affirm that RE boosts environmental quality.

However, Jebli and Youssef [58] revealed that the use of RE increased emissions for five North African nations. Using 38 nations from the IEA from 1995 to 2015, Khan and Hou [59] applied panel analysis and affirmed that RE reduces the quality of the environment. Farhani et al. [60] studied the long-term impact of GDP and energy use on emissions by means of FMOLS, DOLS, and PMG for five “North African countries (Morocco, Algeria, Tunisia, Libya and Egypt” from 1971–2014. The analysis indicated that an expansion in energy usage creates a decline in environmental quality.

Ben Jebli et al. [43] used the data from 1980–2010 in the case of 24 SSA nations. They found a neutral causal association between RE and CO₂ emissions. Furthermore, an inverse relationship was found, but it was insignificant. Nathaniel and Khan [61], in the case of the “Association of Southeast Asian Nations-ASEAN” nations found that RE usage does not affect the EFs of these nations. However, no RE was noticed driving EFs. Nathaniel et al. [62] studied the effects of RE and non-RE usage on EFs and affirmed that RE has a neutral link with EFs in the “Middle East and North Africa-MENA” nations as a whole. Thus, these studies suggested that promoting RE does not assure a decline in EF. Therefore, it is imperative to explore the association in various countries.

For time-series analysis, see [63]. Using the DOLS, FMOLS, and ARDL methods, Zhang et al. [64] investigated the impact of RE and non-RE use on emissions in Pakistan. The results confirmed that green energy has drastically reduced CO₂ in Pakistan. Moreover, a similar study conducted by Majeed et al. [63] for Pakistan suggested the same findings as that of Zhang et al. [64]. In China, Chen et al. [65], using data from 1980 to 2014, found that RE use lessened the adverse effect of emissions there. Majeed et al. [63] used the “Non-Linear Autoregressive Distributed Lag Model-NARDL” in Pakistan and examined the link among “aggregate and disaggregate energy consumption and environmental quality”. They affirmed that only negative shocks have a considerable effect on EFs. On the contrary, the study of Menyah and Wolde-Rufael [66] for the United States revealed that RE use does not lead to a reduction in CO₂ emissions. However, none of these studies employed the “bootstrap ARDL” to study the link between RE and emissions in South Africa.

2.2. Real Estate Market and Environmental Equality

The literature on the relationship between the real estate market and CO2 emissions is thin; only a few studies have been conducted [29,67,68]. This requires further investigation of the subject. Qashou et al. [28] investigated the influence of the real estate sector on Turkey’s CO₂ emissions from 1988 to 2018, using the “Bootstrap Autoregressive Distributed Lag (BARDL)”. The study revealed that the real estate market reduces environmental quality in Turkey both in the short- and long-run. Wei1 and Kong, [29] “analyzed the effect of real estate investment on CO₂ emissions, in 30 provinces, municipalities, and autonomous regions in China from 2000 to 2013”. The study suggested that increased real estate investment increases CO₂ emissions. Furthermore, they highlighted that real estate investment (“residential house investment in provinces and eastern regions”) is more detrimental to emissions.

However, scholars have examined the link between environmental quality and other markets; for instance, Chen et al. [69] examined the “shadow price of CO₂ of 38 double-digit-growth Chinese industrial sectors from 1980 to 2008” using the directional distance function. The outcome suggested that the fundamental values of the CO₂ shadow price of all sectors tend to increase progressively with the light sector higher than the heavy sector”. Bai [70] also explored the association between industrial development and emissions in China between 1987 and 2008. The findings revealed that the development of “tertiary industry” raises the level of CO₂ emissions in China. Samour et al. [24] assessed the influence of bank development on CO₂ emissions in Turkey using ARDL. A positive association was found between emissions and bank development. Similarly, Tahir et al. [71], Isiksal and Joof [72], and Baloch et al. [73] confirm the same results.

2.3. Economic Growth and Environmental Quality

The link between growth and CO₂ emissions has also been widely studied. Employing the “bootstrap panel unit root tests and cointegration techniques from 1981–2005 in the case of 12 MENA countries”, Arouri et al. [74] explored the link between GDP and CO₂ emissions. The findings showed that CO₂ emissions decreased even though the region showed high economic expansion. Al-mulali et al. [75] used FMOLS between 1980 and 2010 for 18 “South American nations”. The analysis showed an “inverted U-shape” relation between GDP and CO₂, thus validating the “Environmental Kuznets curve hypothesis”. Song et al. [76] use “74 key environmental monitoring cities in China, using the vertical and horizontal grading method to construct comprehensive index of environmental quality” between 2005 and 2017 to examine the link between economic expansion and environmental quality. They concluded that deepening economic growth enhances environmental quality. Nathaniel et al. [77] used the “augmented mean group (AMG) and the common correlated effect mean group (CCEMG) in the case of Latin American and Caribbean countries (LACC) from 1990 to 201results confirm that economic expansion increases emissions. Nathaniel [78] explored the effect of economic complexity on EF and CO₂ emissions using “heterogeneity and cross-sectional dependence in ASEAN countries between 1990–2016”. The results revealed that “economic complexity and economic growth raise CO₂ emissions and EF”. Isiksal et al. [79] in Central Asia and Joof et al. [80] in BRICS countries affirmed similar results. Using the PMG on “Mexico, Indonesia, Nigeria, and Turkey (MINT)” countries between 1975 and 2010, Joof and Isiksal [81] explored the link among human capital, export diversification, economic growth, and CO₂. The analysis suggests that economic growth and the quadratic term of GDP have an inverse effect on emissions in the long term but are not significant in the short run.

In China, Shahabaz et al. [82] employed the ARDL and found a bidirectional causation between CO₂. Furthermore, they highlighted that GDP increases CO2 emissions. Mirza and Kanwal (2017) applied the “ARDL and VECM-Vector Error Correction Model” in Pakistan between 1971 and 2009. The findings suggested that CO2 and GDP have a feedback causal relationship. Adejumo [83] investigated the relationship between the environment and economic growth in Nigeria using the ARDL from 1970 to 2014. The analysis suggested that rainfall stimulates environmental quality and economic growth. Natural resources, on the other hand, reduce ecological quality while increasing economic expansion but decreasing per capita economic growth. Recently, [30,84,85,86,87,88,89,90,91,92,93] analyzed the linkage between economic growth and environmental quality in different countries. According to these empirical studies, economic growth has a significant impact on environmental quality.

Based on the literature discussed above, we made the following deductions or conclusions: only a few studies examine the link between the real estate market and CO₂ emissions; Qashou et al. [28] explored how Turkey’s real estate industry affected environmental quality, and Wei and Kong [29] explored the influence of real estate development on environmental degradation in a “panel of 30 provinces of China” spanning data from 2000 to 2013. However, none of the researchers have looked at how the real estate sector affects environmental quality in Africa, specifically in South Africa, which is the largest CO₂ emitter in the region.

3. Econometric Methodology

This study mainly aims to demonstrate the impact of the real estate sector, renewable and non-renewable energy utilization, and income on the degree of environmental quality in South Africa. The theoretical linkage among the focused variables is presented as follows:

$${\ln \mathrm{CO}}_{2\mathrm{it}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{InBSD}}_{\mathrm{it}}+{\mathsf{\beta }}_2{\mathrm{InED}}_{\mathrm{it}}+{\mathsf{\beta }}_3{\mathrm{InY}}_{\mathrm{it}}+{\mathsf{\beta }}_4{\mathrm{InREC}}_{\mathrm{it}}+{ \mathsf{\beta }}_5{\mathrm{InNREC}}_{\mathrm{it}}+{ \mathsf{\epsilon }}_{\mathrm{t}},$$

(1)

in Equation (1) above,${ \ln \mathrm{CO}}_{2\mathrm{it}}$ represents South Africa’s carbon emissions, and ${\mathrm{InRES}}_{\mathrm{it}}$ represents the real estate market index. ${\mathrm{InY}}_{\mathrm{it}} \mathrm{is} \mathrm{total}$ GDP in USD (2010 = 100). ${\mathrm{InRE}}_{\mathrm{it}}$ represents the proportion of RE (“renewable energy utilization from the total energy utilization”).${ \mathrm{InNREC}}_{\mathrm{it}}$ represents total non-RE (“non-renewable energy utilization such as coal, petroleum, gas, etc.”). The entire series was transformed into a natural logarithm. Figure 2 shows the time series plot of the series at the natural logarithm, indicating various economic fluctuations and policy changes in South Africa. The data examined in the study ranges from 1988 to 2018.

Table 1. Shows the variables’ descriptions and data sources.

Variable	Variable Description
ln CO2it	CO2 “emissions measured in metric tons per capita”
lnREMt	Housing price index
lnY	“the constant logarithm GDP (2010 = 100) in USD per capita”
lnREt	A portion of total final energy utilization
lnNREt	Non-renewable energy utilization (coal, petroleum, and gas)

Note: The “source of the World-Bank (WB) data” [94].

contains information on the targeted data’s origin and description.

Stationarity and Co-Integration Tests

To evaluate the integration of the targeted variables, Zivot and Andrews [40] and Clemente–Montañes–Reyes [41] assessments have been used. The main advantages of the assessments used were that they took into account the dates of structural failure. There are focused variables that can be used to determine co-integration. The study utilized newly developed bootstrap (ARDL) testing. In this context, the $F{-}_{test}$ and $t{-}_{tests}$ are utilized on lagged facets and dependent and independent facets, respectively, to distinguish between two degenerate cases. The case $no. 1$ happens when the $F{-}_{test}$ on all lagged variables is statistically significant, the $t{-}_{test}$ on the lagged independent variable is statistically significant, but the $t{-}_{test}$ t on the lagged dependent variable is statistically not significant. The second degenerate case $\#2$ occurs when the $F{-}_{test}$ on all lagged focused variables is statistically significant, the $t{-}_{test}$ on the lagged dependent variable is significant, while the $t{-}_{test}$ on a lagged independent variable is statistically not significant. McNown et al. [38] introduced the “bootstrap ARDL technique” to overcome this issue by introducing an additional $t{-}_{test}$ on the degenerate case $\#1$, which in turn provides a robust finding of the integration status of the examined model. A new feature in bootstrap ARDL testing aims to address the stability issue. In this manner, just one chosen variable is accepted by the critical values of the conventional ARDL of the co-integration model as being endogenous. The crucial values in the recently created ARDL tests, however, permit all targeted variables to be endogenous. The level of co-integration among the focused variables will be approved if the values of ($F{\mathrm{statistic}}_{ov}$), ($t{\mathrm{statistic}}_{DV}$), and ($F{\mathrm{statistic}}_{IDV}$) exceed the “Critical Value-CV” as shown by the bootstrap approach. The ARDL test expression is as follows:

$$\begin{array}{cc}\Delta {\mathrm{lnCO}}_{2\mathrm{it}}={\mathsf{\alpha }}_0+\hfill & {\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{r}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnCO}}_{2\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{r}}}{\mathsf{\beta }}_2\Delta \ln RE{M}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{r}}}{\mathsf{\beta }}_3\Delta \ln Y_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{r}}}{\mathsf{\beta }}_4\Delta \ln R{E}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{r}}}{\mathsf{\beta }}_5\Delta \ln NR{E}_{\mathrm{t}-\mathrm{j}}+{\gamma }_1{\mathrm{lnCO}}_{2\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\gamma }_2\ln RE{M}_{\mathrm{t}-1}+{\gamma }_3\ln Y_{\mathrm{t}-1}+{\gamma }_4\ln R{E}_{\mathrm{t}-1}+{\gamma }_5\mathrm{lnN}R{E}_{\mathrm{t}-1}\hfill \\ \hfill & +{ \mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t}}\hfill \end{array}$$

(2)

In Equation (2), ${\mathsf{\epsilon }}_{1\mathrm{t}}$ denotes the error term; ∆ symbolizes the “first difference process”; $\mathsf{\alpha }$ is the constant expression; “${\mathsf{\beta }}_1, {\mathsf{\beta }}_2,{\mathsf{\beta }}_3,{\mathsf{\beta }}_{4,}{\mathsf{\beta }}_5$, and ${\gamma }_1, {\gamma }_2, {\gamma }_3, {\gamma }_4, {\gamma }_5, {\gamma }_6$” denote coefficients in long-term and short-term variables; r means lags order; and ${\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }$ is the ECM, “error correction term” reflecting speed of adjustment among variables.

To reinforce that empirical analysis is performed correctly, the final article employs some diagnostic evaluations, namely, the “Ramsey RESET” test$(X^{-Rt})$; and the “Autoregressive Conditional Heteroskedasticity” (ARCH test$(X^{-At})$, the normality test$(X^{-Nt})$, and the Breush–Pagan–Godfrey heteroskedasticity test$(X^{-Bt})$. Otherwise, the study utilized the “CUSUM and CUSUM-square” assessments to reinforce that the empirical analysis is tested correctly. To determine the short- and long-term causal relations that exist among the series, the “Granger causality” analysis is employed. In this analysis, ($E{C}^{Term}$) defines the variations of short-term level variations from long-run symmetry. The $ECM$ is presented in Equations (3–7):

$$\begin{array}{cc}\Delta {\mathrm{lnCO}}_{2\mathrm{it}}={\mathsf{\alpha }}_0+\hfill & {\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{p}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnCO}}_{2\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_2\Delta {\mathrm{lnREM}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_3\Delta {\mathrm{lnY}}_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_4\Delta {\mathrm{lnRE}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_5\Delta {\mathrm{lnNRE}}_{\mathrm{t}-\mathrm{j}}+{\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t},}\hfill \end{array}$$

(3)

$$\begin{array}{cc}\Delta {\mathrm{lnrREM}}_{\mathrm{it}}={\mathsf{\alpha }}_0+\hfill & {\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{p}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnREM}}_{\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_2\Delta {\mathrm{lnCO}}_{2\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_3\Delta {\mathrm{lnY}}_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_4\Delta {\mathrm{lnRE}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_5\Delta {\mathrm{lnNRE}}_{\mathrm{t}-\mathrm{j}}+{\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t},}\hfill \end{array}$$

(4)

$$\begin{array}{cc}\Delta {\mathrm{lnY}}_{\mathrm{it}}={\mathsf{\alpha }}_0+\hfill & {\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{p}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnY}}_{\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_2\Delta {\mathrm{lnCO}}_{2\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_3\Delta {\mathrm{lnREM}}_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_4\Delta {\mathrm{lnRE}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_5\Delta {\mathrm{lnNRE}}_{\mathrm{t}-\mathrm{j}}+{\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t},}\hfill \end{array}$$

(5)

$$\begin{array}{cc}\Delta {\mathrm{lnrRE}}_{\mathrm{it}}={\mathsf{\alpha }}_0+\hfill & {\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{p}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnRE}}_{\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_2\Delta {\mathrm{lnCO}}_{2\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_3\Delta {\mathrm{lnREM}}_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_4\Delta {\mathrm{lnY}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_5\Delta {\mathrm{lnNRE}}_{\mathrm{t}-\mathrm{j}}+{\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t}},\hfill \end{array}$$

(6)

$$\begin{array}{cc}\Delta {\mathrm{lnrNRE}}_{\mathrm{it}}={\mathsf{\alpha }}_0\hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{p}}}{\mathsf{\beta }}_1\Delta {\mathrm{lnRE}}_{\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_2\Delta {\mathrm{lnCO}}_{2\mathrm{it}-\mathrm{J}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_3\Delta {\mathrm{lnREM}}_{\mathrm{t}-\mathrm{j}}\hfill \\ \hfill & +{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_4\Delta {\mathrm{lnY}}_{\mathrm{t}-\mathrm{j}}+{\displaystyle {\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{q}}}{\mathsf{\beta }}_5\Delta {\mathrm{lnRE}}_{\mathrm{t}-\mathrm{j}}+{\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }+{\mathsf{\epsilon }}_{1\mathrm{t}},\hfill \end{array}$$

(7)

where ∆ means the “first difference” and ${\mathsf{\omega }\mathrm{ECT}}_{\mathrm{t}-1 }$ is the lagged ECT. Causation among series in the short run is assessed using the “Wald test (F) statistics”. Figure 3 depicts the structure of this paper’s econometric methodology.

4. Empirical Results and Discussion

In

Table 2. The findings of the ZA unit root assessment.

	Level		First Differences
Variables	$\mathbf{t}{.}^{\mathit{S}\mathit{T}\mathit{A}\mathit{T}}$	${\mathrm{DSB}}^{ 1}$	Variables	$\mathbf{t}{.}^{\mathit{S}\mathit{T}\mathit{A}\mathit{T}}$	${\mathrm{DSB}}^{ 1}$
ln CO2it	−2.003	2003	${\ln \mathrm{CO}}_{2\mathrm{t}}$	−9.131 **	2011
lnREMt	−2.215	1991	${\mathrm{lnREM}}_{\mathrm{t}}$	−9.951 **	2008
lnY	−3.001	2008	$\mathrm{lnY}$	8.134 **	2008
lnREt	−2.663	2013	${\mathrm{lnRE}}_{\mathrm{t}}$	−8.851 **	2016
lnNREt	−1.776	2002	${\mathrm{lnNRE}}_{\mathrm{t}}$	−7.465 **	2012

Note: “** means the significance of the variables at the 5% level. DSB ¹ means dates of structural break”.

and

Table 3. The CMR unit root test assessment.

		Level			First Differences
Variables	$\mathbf{t}{.}^{\mathit{S}\mathit{T}\mathit{A}\mathit{T}}$	${\mathrm{DSB}}^{ 1}$	${\mathrm{DSB}}^{ 2}$	Variables	$\mathbf{t}{-}^{\mathit{S}\mathit{T}\mathit{A}\mathit{T}}$	${\mathrm{DSB}}^{ 1}$	${\mathbf{DSB}}^{ 2}$
ln CO2it	−1.445	1996	1999	${\ln \mathrm{CO}}_{2\mathrm{t}}$	−8.101 **	2003	2005
lnREMt	−2.321	1996	2003	${\mathrm{lnREM}}_{\mathrm{t}}$	−8.322 **	2008	2016
lnY	−1.002	2008	2009	$\mathrm{lnY}$	−7.112 **	2009	2013
lnREt	−1.996	1994	2003	${\mathrm{lnRE}}_{\mathrm{t}}$	−9.363 **	1996	2015
lnNREt	−0.451	1994	2001	${\mathrm{lnNRE}}_{\mathrm{t}}$	9.886 **	2007	2009

Note: “** means the significance of the variables at the 5% level. DSB ¹ and BSD ² mean dates of structural break”.

, the findings illustrate that CO2, REM, Y, REC, and NRE variables are integrated and stationary at the first order ($\Delta )$. In this way, the study findings provide empirical evidence that study facets are “integrated at the I(1) level”. Hence, Equation (1) revealed the co-integration model. The findings of the “bootstrap ARDL model” are reported in

Table 4. The bootstrap ARDL co-integration approach.

Bootstrap ARDL Findings Diagnostic Test Findings
	$F.statisti{c}_{-OV}$	$t.statisti{c}_{-DV}$	$F.statisti{c}_{-IDV}$	${\mathit{x}}^{-\mathit{R}\mathit{t}}$	${\mathit{x}}^{-\mathit{A}\mathit{t}}$	${\mathit{x}}^{-\mathit{N}\mathit{t}}$	${\mathit{x}}^{-\mathit{B}\mathit{t}}$
	6.012 **	−5.651 **	6.551 **	$1.521{ }^{FS}$	$1.411{ }^{FS}$	$1.312{ }^{FS}$	$1.811{ }^{FS}$
Bootstrap-based table CV 5%	2.798	−3.864	4.475	$0.56{ }^{PV}$	0.6$5{ }^{PV}$	0.5$6{ }^{PV}$	0.8$5{ }^{PV}$

Note: ** “denote significance level at 5%; p-value is represented by PV; $F{\mathrm{statistic}}_{-ov}$ is the F-test on. All the lagged level variables; $t{\mathrm{statistic}}_{-DV}$ is the t-test on the lagged dependent variables; and $F{\mathrm{statistic}}_{-IDV}$ is the F-test on the lagged independent variable”.

. The statistical outcomes confirm that the values of the three co-integration assessments (the “$F.{\mathrm{statistic}}_{-OV}$, $t.{\mathrm{statistic}}_{-DV}$, and $F{\mathrm{statistic}}_{-IDV}$”) are higher than the “bootstrap ARDL CVs”. These outcomes reinforce that the co-integration among the CO2, REM, Y, RE, and NRE is valid.

The outcomes of the ARDL technique in the short- and long-runs are shown in

Table 5. ARDL test results.

Variable	Coefficient	$\mathbf{t}{.}^{\mathit{S}\mathit{T}\mathit{A}\mathit{T}}$	$\mathit{p}{.}^{\mathit{v}\mathit{a}\mathit{l}\mathit{u}\mathit{e}}$
${\mathrm{lnREM}}_{\mathrm{t}}$	0.113 **	2.384	0.023
$\mathrm{lnY}$	2.221 ***	2.999	0.000
${\mathrm{lnRE}}_{\mathrm{t}}$	−0.252 ***	−4.741	0.000
${\mathrm{lnNRE}}_{\mathrm{t}}$	1.992 ***	6.551	0.000
${\mathrm{lnREM}}_{\mathrm{t}}$	−0.051 ***	2.451	0.016
$\mathrm{lnY}$	1.131 ***	4.461	0.000
${\mathrm{lnRE}}_{\mathrm{t}}$	−0. 114 ***	−6.771	0.000
${\mathrm{lnNRE}}_{\mathrm{t}}$	1.445 ***	3.553	0.000
${\mathrm{ECT}}_{\mathrm{t}-1}$	−0.251 ***	−5.331	0.000

** and *** denote the significance level of 10%, 5% and 1%, respectively.

. The results demonstrate that, at a 1% level of significance, the real estate market is positively and substantially linked with carbon emissions, both in the short- and long-term. In South Africa, a 1% increase in the real estate market is accompanied by a 0.113% reduction in carbon emissions over the medium term, but over the long run, it is detrimental to the environment. These outcomes are in line with Qashou et al. [28] and Wei and Kong [29]. In view of the results of this study, it is suggested that the stakeholders in South Africa, particularly the government, environmentalists, and policymakers consider green sources to build green houses that have a beneficial effect on the environment. Furthermore, investors must be encouraged to invest in real estate market projects that reduce environmental impacts while providing a standard living environment to the next generation.

Additionally, the data demonstrate a positive and significant relationship between real income and CO₂ emissions across both short- and long-time periods, with a significance level of one in South Africa, where carbon emissions significantly rise by 2.221% and 1.131% over the course of the medium- and long-terms, respectively, for every 1% growth in real income. These results concur with those of [76]. The results of this do not validate the “Environmental Kuznets Curve (EKC) hypothesis” which presents an inverted U-shaped association between environment and income. The results suggest that South Africa must use clean energy sources to boost its economy without affecting the environment in the long- or short-run.

The results show that NREC is responsible for worsening the environmental quality of the study nation. This revealed that in the medium- and long-run, a 1% increase in NREC resulted in 1.992 and 1.445% increases in carbon emissions in South Africa, respectively. Accordingly, a 1% increase in RE reduced CO₂ over the medium and long term by 0.252 and 0.44%, respectively. This outcome is in line with the inferences of [56,69]. According to the results, utilizing renewable energy sources is more climate-friendly as compared to non-renewable sources. Therefore, the South African authorities, particularly the government, must encourage clean and green projects such as solar systems, hydropower stations, and wind energy. Moreover, fossil fuel use must be taxed to discourage and minimize its usage. The industry must be funded and encouraged to shift the energy system away from fossil fuels and toward renewable energy sources.

According to ${\mathrm{ECT}}_{\mathrm{t}-1}$ in the results, the production functions change to a long-run equilibrium route, with a 5.1% speed adjustment from the short- to the long-run. The stability of the research model was verified using diagnostic tests, “CUSUM, and CUSUM square tests”. Table 4 displays the results of the diagnostic tests. The “Breush–Pagan–Godfrey $(X^{-Bt})$ heteroscedasticity test” results show that there is no autocorrelation, and the examined model is homoscedastic. Furthermore, the normality test $(X^{-Nt})$ ensures that the study model has a normal distribution, whereas the “Ramsey RESET test $(X^{-Rt})$” ensures that the model is stable and correct. Figure 4 and Figure 5 show the “CUSUM and CUSUM square tests”. According to Figure 4 and Figure 5, the model has been correctly built “when the green line falls inside the brown and green lines”.

Using a cointegrated testing series, the VECM is employed in this work to assess the causal association. In order to promote green energy in “South Africa” for the long-term growth of energy, we may make some reasonable recommendations by examining the causal link between the components under examination.

Table 6. Results of the Granger causality testing approach.

	$\Delta lnC{O}_{2t}$	$\Delta lnRE{M}_t$	$\Delta ln{Y}_t$	$\Delta lnR{E}_t$	$\Delta lnNR{E}_t$	ECT
$\Delta {\mathrm{lnCO}}_{2\mathrm{t}}$	-	8.332 **	8.521 **	7.770 **	8.001 **	−0.431 **
$\Delta {\mathrm{lnREM}}_{\mathrm{t}}$	0.441	-	1.31	1.351	0.995	−0.112
$\Delta {\mathrm{lnY}}_{\mathrm{t}}$	0.801	8.410 **	-	1.741	0.765	−0.125
$\Delta {\mathrm{lnRE}}_{\mathrm{t}}$	0.441	6.031 *	6.113 *	-	2.756	−0.952
$\Delta {\mathrm{lnNRE}}_{\mathrm{t}}$	0.661	7.225 *	7.112 **	1.851	-	−0.113

Note: * and ** indicate significance levels at between 5 and 10%.

shows that in the long-run, the South African real estate market, real income, usage of renewable and non-renewable energy, and carbon emissions are all causally related. Furthermore, the results suggest that in the short-term there is a one-way causal relationship between the examined variables (REM, Y, RE, NRE, and CO2). Additionally, the findings reveal a one-way causal relationship between the real estate market, REC, income, and NRE. These estimates show that the South African real estate industry has a significant impact on carbon emissions via NREC and real income channels. However, this result indicates that the real estate sector can have a significant influence on emissions. As a result, the current study proposes that authorities in South Africa should leverage real estate expansion to encourage renewable energy investment. The conclusions of the present study are critical for South Africa in terms of altering its energy mix by including more renewable energy sources.

5. Conclusions and Policy Implications

The existing status of energy and wealth in South Africa, despite a fast-rising population, appears to be untenable, according to current empirical literature. The influence of economic expansion and energy usage on environmental quality is the focus of most studies in this field; the influence of the REM (real estate market) on environmental quality in South Africa is mostly neglected in the existing literature. In this light, the current study attempts to give a unique empirical analysis of the impact of real estate sector development and external debt on carbon emissions in South Africa by utilizing the newly developed “bootstrap autoregressive distributed lag (BARDL)” testing from 1988 to 2018. Empirical evidence suggests that renewable energy improves environmental quality in South Africa in both the long- and short-term. Furthermore, the results demonstrate that both immediately and over time, the real estate market has a detrimental effect on environmental quality. Similarly, real income and NRE are significantly positive and strongly correlated with CO₂ emissions across both short- and long-time periods, at a 1% level of significance. In the short-term, the “Granger causality test” demonstrated a unidirectional causal relationship between the variables investigated (REM, Y, RE, NRE, and CO2). Additionally, the findings reveal a one-way causal relationship between the real estate market, REC, income, and NRE.

Having been a fossil fuel-rich nation, South Africa is largely dependent on non-RE energy, considering its catastrophic impacts on the environment. Previous and current findings, including this study, have affirmed that RE curtails air pollution (CO₂ emissions). In order to protect its natural environment from depletion and enhance energy generation, it is incumbent on the South African authorities to aggressively invest in renewable energies such as solar, biogas, wave power, geothermal, and tidal power. This will ensure a decline in CO₂ emissions, thereby leading to a sustainable environment and “green growth” [95]. It does not imply that South Africa is unaware of the benefits of RE and has not initiated programs to stimulate and invest in RE. In actuality, over the years, South Africa has heavily financed RE projects. For example, on the commercial use of biomass, the Minister of Finance in his budget speech of 2002 announced a reduction in the fuel levy on diesel produced from biomass by 30%. In addition, PetroSA has produced an environmentally friendly product called eco-diesel, which is already available around Cape Town [96]. Regardless of these promising investments, “the result is simply a tiny speck of good that suggests that movements towards green energy might just be a wild goose chase after all”. This indicates that formal policies are not sufficient for achieving clean energy. The culprits could be weak institutions [78].

The policy recommendations are, firstly, that as a fossil-dependent country, in order to protect its natural environment from depletion and improve energy generation, it is incumbent on the South African authorities to aggressively invest in renewable energies such as solar, biogas, wave power, geothermal, and tidal power. However, considering the infrastructural and technological constraints, improving the share of RE in the aggregate energy profile might not be easily attained by South Africa. Thus, in addition to promoting RE consumption, conventional ‘fossil fuels’ should be gradually replenished with comparatively fewer toxic substitutes. These can be considered "transitional fuels" until the dynamics limiting South Africa’s RE sector growth are overcome. Secondly, to initiate an epitomic transition from “carbon-intensive industries and energies to a service-oriented economy” thereby leading to changes in economic structure as a result and assisting in alleviating the problems of climate change. Thirdly, it is necessary to introduce technological innovations (carbon capture and storage, etc.) in the production and power sectors to uphold energy efficacy. Finally, economic reform by the government to initiate and uphold policies to “decarbonize the economy” and offer a favorable environment that stimulates clean energy. In order to achieve the “Sustainable Development Goals” South Africa should strengthen its institutions.

Furthermore, real estate was found to increase the level of emissions in South Africa. Real estate is perceived to impact the environment through various channels, notable among which are energy consumption, deforestation, and economic growth. Real estate development may trigger the cutting of trees or the destruction of forest resources, which can adversely affect the quality of the environment. Similarly, expansion of the real estate market may increase household consumption of non-renewable energy, thus decreasing air quality. Additionally, real estate augments investment growth, thereby stimulating economic activities; economic expansion raises energy use and eventually increases CO₂ emissions.

In this regard, our analysis establishes numerous policy consequences for South African authorities. Primarily, authorities should initiate, promote, and guide real estate firms to invest in R&D to devise new expertise, methods, and strategies so as to lower environmental degradation in construction. Secondly, persuade, subsidize, and give incentives for firms to use ecologically friendly building materials. Third, promote the use of renewable energies (wind energy, solar energy, etc.). Fourthly, authorities should initiate taxonomy regulations to guide the investment decisions of real estate firms toward a sustainable environment. This can be achieved by creating awareness among stakeholders in the real estate industry, building and strengthening the law enforcement institutions, and developing the necessary capacity to enable them to assess the environmental impacts arising from the activities of this sector. Finally, authorities should initiate a policy coordination among all the facets of the economy, especially with the financial sector, which provides capital to the real estate industry, because low or zero-carbon emissions cannot be achieved in the absence of finance. In conclusion, South African firms should strive to develop “low-carbon buildings” by importing suitable technology that supports low emissions. Additionally, the building standards should factor in minimum pollution requirements.

However, the limitation of the current work is the unavailability of assessed data after 2018 for some selected valuables. Future empirical research should be devoted to assessing the effect of real estate markets on the level of ecological quality using other panel and time series methods. Areas for future research may include using a panel technique as well as dividing South Africa into different provinces or regions to explore the most impacted regions.