Investigating the Growth Effect of Carbon-Intensive Economic Activities on Economic Growth: Evidence from Angola

Abstract

Despite its immense natural resources, Angola struggles to significantly improve its economy to reduce poverty. Carbon emissions have been increasing over the years, even though the country plans to reduce them by 35% by 2030. This paper attempts to assess the carbon emissions of several sectors (industries, transport, services, and residences) on economic growth, intending to find a balance between environmental protection that requires carbon emissions reduction and economic development that may add to environmental degradation. The study employed time series data on GDP, CO₂, CH₄, and N₂O covering 1971 to 2021 and ARDL and ECM models. This is the first study at the state level in Angola on the relationship between economic development and environmental sustainability considering methane and nitrous oxide emissions. Additionally, the paper assesses the responses of GDP to deviation shock of GDP, CO₂, CH₄, and N₂O by 2032. Phillip Perron and Augmented Dickey-Fuller tests showed that all the data are stationary at the first difference, favoring the application of the ARDL model to explore the short and long-run relationships. The result reveals that methane from agricultural activities and carbon emissions from the building sector and public services contribute to economic growth, whereas carbon emissions from industrial heat systems, non-renewable electricity production, and manufacturing industries harm economic growth. However, no relationship exists between nitrous oxide emissions and economic development. In addition, impulse response function estimates show that appropriate investments can sustain economic development over the years. Therefore, the country should diversify its economy and avoid polluting fuel sources, such as coal. Raising renewable energy’s proportion in the total energy mix can support growth while considering the environmental quality. Investments in skills training, academic projects in renewable energy technologies development, agriculture mechanization, and sustainable job creation are recommended. Additionally, investing in quality seeds adapted to climate realities might help lessen climate change’s adverse effects and promote growth. Manure manufacturing processes must be improved to reduce agriculture and livestock’s methane and nitrous oxide emissions. The country’s leaders are encouraged to promote raw material processing industries while insisting on reducing carbon emissions.

1. Introduction

Economic growth and environmental sustainability issues have gained traction in academic literature. These two themes are all the more interesting because they are related to sustainable development policies. Economic growth is supposed to reduce unemployment, hunger, social inequalities, malnutrition, and poverty. However, according to several researchers, the link between economic development and environmental degradation reports an interdependence, thus threatening the quality of the environment [1].

The EKC is a hypothesized association between various indicators of environmental deterioration and economic growth [2]. That hypothesis stipulates that economic development and the deterioration of the environment are interdependent and positively correlated, indicating that an increase in growth harms the environment while a drop in the economy adds to the quality of the environment. This means that the environmental destruction variables are inverted U-shaped growth functions [2]. The same interpretation has been used to explain the degradation of the quality of the environment, relating greenhouses gas, such as carbon dioxide, methane, and nitrous oxide, with economic development variables, such as GDP [3]. It is, therefore, normal for policymakers to formulate policies balancing economic growth and environmental quality based on studies with GDP, CO₂, CH₄, and N₂O. Since this paper aims to broaden our knowledge of Angola’s environmental performance and economic growth, we will assess the effects of gas emissions on Angola’s economic growth by using data on CO₂, CH₄, N₂O, and GDP. The paper uses the autoregressive distributive lag model, which is efficient for small-sample data, to explore the long- and short-term relationships between the study variables. This is to strengthen the country’s sustainable development policies. Our paper will try to find a fair balance between economic and climate policies so that Angola, which is determined to take up the challenge of significantly reducing poverty, can also, to some extent, formulate policies that will protect the environment in place. Since several studies only focus on the link between carbon dioxide emissions and economic development, our study innovates by incorporating other factors (CH₄, N₂O) not previously used in the Angolan case. Additionally, the paper will help determine which factors might reinforce economic development in the next ten years.

The remainder of the paper is organized as follows. Section Two is about the theoretical background of Angola, while Section Three outlines the literature review about environmental impact and economic development. Section Four details the data and the methods, and Section Five presents the results. The conclusions are provided in Section Six.

2. Theoretical Background on Angola and Review of Economic Development Methods Analysis

2.1. Economic Context in Angola

Angola is a developing Sub-Saharan African country with a population of around 32 million and colossal mineral resource reserves. Despite its natural wealth, the country struggles to eradicate extreme poverty. The malnutrition rate is tricky. A total of 45% of children under the age of 5 have an abnormal height/age ratio, 31% are anorectic, and 1 child in 6 never lives past 5 [4]. The economic crisis in Angola began in 2009 and never got under control again. Between 2012 and 2015, The average GDP rise was 3.7%, sufficient to halt the downward trend in the income per capita. As what has been lost is significant, the poverty rate rises to 54% for multidimensional, 48.9% for poverty intensity, and 42% for monetary [5]. In 2019, the incidence of poverty in Angola was 32.3% at the national level, and it was nearly three times higher in rural areas (54.7%) than in urban areas (17.8%) [6]. Additionally, the COVID-19 crisis came when Angola suffered a negative growth since the oil shock of 2014–2016. The economy contracted by 1.4% annually from 2016 to 2019, exacerbating poverty and social inequalities [7]. Additionally, Doing Business 2020 underlined that Angola is among the 15 worst countries to do business as the country has moved from the 173rd to the 177th position among 190 economies [5]. Several factors seriously hamper the sector’s recovery: the countryside’s isolation because of the road’s network collapse, the prevalence of landmines that have limited arable land, the breakdown of internal trade and distribution networks, the meager availability of domestic credit for agriculture and livestock, and the fragile support from institutions. Therefore, there is a vital interest in any study that may help improve growth.

Agriculture, animal husbandry, and forestry contributed more than 8% of the GDP and 42% of total employment, with women accounting for 70% of the workforce. Agricultural exports accounted for almost 60% of total exports and consisted of coffee (48%), sisal (5%), maize (2%), and many other commodities, including fruits, pulses, legumes, cotton, and cereals. Before independence in 1975, the country was self-sufficient in all major foodstuffs, but by 1990, all those export commodities virtually vanished [4]. The 22-year-old civil war has severely impacted all economic sectors by reducing agricultural activity and productivity and causing a rise in prices and unemployment. The country is currently a net importer of food and spent 2.347 billion USD in 2016, 2.955 billion USD in 2017, 3.283 million USD in 2018, and 8.6 billion USD on food imports in 2019, half of which was spent on essential food basket products, highlighting the fragile situation of food security in the nation where agriculture makes up 6% of GDP [8]. With excellent agricultural conditions, a suitable temperature, and many water resources, Angola has the world’s 16th-largest agricultural potential area. Unfortunately, the lack of necessary investments, modern industrial infrastructures, agricultural equipment, and farmer technical training prevents the nation from being self-sufficient [9]. The TSE invested an average of 1.32 billion USD in agriculture between 2018 and 2019, which is 1.4% of GDP and more than twice the OECD average of 0.6% [10]. Angola’s TSE was equivalent to 29.5% and 27.3% of agriculture GDP in 2018 and 2019, which is greater than South Africa but comparable to Canada and Iceland [10].

2.2. Pollution Sources in Angola

Animal production accounts for 14.5% of human-induced emissions and 37% and 65% of global methane and nitrous oxide emissions [11]. Angola possesses more than 3,160,000 cattle, 1,667,000 goats, 1,108,000 pigs, and 227,000 sheep. Cattle represent around 51% total livestock, while 31% accounts for small ruminants and 18% for pigs [4]. Livestock contributes about 9% of total CO₂ emissions but 37% of CH₄ and 65% of N₂O, and about 18% to the worldwide global warming effect [12]. The livestock sector is among the most significant contributors to GHG emissions, responsible for 18–51% of anthropogenic emissions expressed in CO₂-equivalent [13]. As the world’s livestock numbers and manure emissions increase, so will the emission rate. Other impacts of livestock farming on long-term sustainability are reactive nitrogen losses, of which ammonia emissions from manure are the most worrisome [14]. Figures indicate that agriculture and livestock manure provide 63% and 44% of ammonia emissions, which can adversely affect ecosystem acidification, surface water eutrophication, and human toxicity [14]. Our study is part of considering emissions from all economic sectors.

Angola’s energy emissions nearly quadrupled, increasing 99.45 MtCO₂e (396%) from 1990 to 2014, with fugitive emissions driving most of this increase (84%) [15]. In 2014, the country’s total greenhouse gas emissions amounted to 252.09 MtCO₂e, accounting for 0.52% of global greenhouse gas emissions [16]. The energy sector is the main contributor to greenhouse gas emissions, accounting for 49.4% of emissions, followed by land use change and forestry (37.3%), agriculture (11.7%), garbage (0.9%), and industrial activities (0.6%) [17]. These figures will likely increase as efforts are made to develop the economy. IPCC scenarios projected a mean annual temperature increase between 1.2 and 3.2 °C by the 2060s and between 1.7 to 5.1 °C by the 2090s [18].

2.3. Angola Energy Policies

Concerned with ensuring healthy growth, the authorities have ratified the Paris Agreement and the Kyoto Protocol. Angola is considering unconditionally dwindling GHG emissions by up to 35% by 2030 compared to the BAU scenario (the base year 2005) [19]. Moreover, under a conditional mitigation scenario, the nation is anticipated to reduce emissions by an additional 15% below BAU levels by 2030 [18]. The country plans to cut its emissions trajectory by approximately 50% below the BAU scenario by 2030, at a total cost of more than 14.7 billion USD, by meeting its unconditional and conditional targets [19]. For this, the country plans to invest over 23 billion USD for 2018–2025 with a strong emphasis on promoting universal energy access and increasing the country’s already substantial share of renewable energy [20]. The specific objectives for deploying renewable energy include, among other things, 100 MW of solar power, 500 MW of biomass, and 100 MW of wind power [20].

3. Literature Review on the Methods of Analyzing Economic Development and Environmental Challenges

The vast literature on the link between economic development and environmental degradation supports that the economy’s improvement is likely to accelerate environmental deterioration [11]. Nevertheless, countries plan to reduce unemployment, poverty, and social inequalities and to ensure quality education and environmental preservation. This contrasts with the Kuznet environmental curve hypothesis. Thus, increasing the economy while preserving the environment would amount to reducing GHG emissions while minimizing the effects of this reduction on economic growth. The Paris Agreement recommended a global response to climate change threats and strengthening efforts to eliminate poverty [12,13,14]. Among other things, the Paris Agreement encouraged the maintenance of the global average increase temperature below 2 °C compared to pre-industrial levels. This was done to prevent global warming from attaining 1.5 °C between 2030 and 2052 [21]. Would it, therefore, be preferable for all countries to adopt a commune energy policy? The answer is no. The economists’ conclusions agree that the association between economic development and emissions can vary depending on the study areas, implying several recommendations. It is, therefore, up to each area to conduct its investigations to propose sustainable development policies per the resolutions to which the country is a signatory.

Hussain et al. [22] conducted some analyses about the liaison between environmental deterioration and economic development in Pakistan with data on CO₂ emissions, GDP, and energy consumption from 1971 to 2006. They employed the Johansen cointegration, VECM, and Granger causality tests. The findings reported a long-term association between GDP, energy consumption, and CO₂ emissions. Moreover, there is an increasing curve between GDP and CO₂ emissions, rejecting the environmental Kuznet curve relationship. Their results align with Halicioglu [23] in Turkey, He and Sandberg [24] in Canada, and Fodha and Zaghdoud [25] in Tunisia. The authors insisted that a carbon reduction policy would harm economic growth, and if no strict air pollution control is taken, the population will suffer. They recommended network-monitoring policies based on developed countries’ policies.

Similarly, Mohapatra and Giri [26] employed ARDL and VECM on CO₂ emissions, trade openness, energy consumption, urbanization, and gross fixed capital formation data from 1971 to 2012 to show that economic growth adds to energy consumption, which contributes to carbon emissions in the long term in India. In addition, the result supports that urbanization harms environmental quality. Recognizing that the country suffers from an economic deficit and that economic growth and urbanization are deteriorating the environment, the authors had to choose economic development while formulating energy policies that could mitigate the environmental consequences. They recommended that the country invests in clean energy and research to find less-polluting energy sources.

In the words of Ali et al. [27], the EKC hypothesis is applicable in the long term but not in the short term in Pakistan. They reached this conclusion with data from 1980 to 2012 on economic development, CO₂ emissions, and energy consumption. They suggested policymakers consider economic growth impact while formulating energy policy. The main recommendation made is the gradual shift to renewable energy.

On the contrary, Bosah et al. [28] demonstrated that economic development provoked long-term environmental degradation in 15 economies from 1980 to 2017. Moreover, the results indicate that urbanization does not influence CO₂ emissions, while energy consumption enhances them. They reached these conclusions using data on urbanization, GDP, CO₂ emissions, and energy consumption. Technological innovation, population growth control, and public transport improvement were recommended. Their findings support the results of Sheraz et al. [29] in G20 countries and Tong et al. [30] in E7 countries.

Since several studies about economic growth and the environment recommend using clean energy, scholars must esteem renewable energy’s impact on development to formulate appropriate sustainable policies [31,32,33,34]. Even if the ultimate goal of energy policies is to reduce carbon emissions, the fact remains that policymakers manage what energy policy agrees with economic development. Somoye et al. [35] classified renewable energy’s effect on economic growth into three spectrums: positive impact, negative impact, or no impact. This leads to taking precautions while recommending renewable energy adoption in developing countries, as those countries urgently need to promote growth.

According to IRENA, 12.7 million people were employed directly and indirectly by the renewable energy sector in 2021, with 4.3 million jobs coming from photovoltaic solar, 1.3 million from wind power, 2.4 million from hydropower, and 2.4 million from biofuels [36]. This supports the finding of Sari and Akkaya [37] and Proença and Fortes [38], who show that renewable energy adds to employment. Namahoro et al. [39] established that the impact of renewable energy on economic development varies according to regions and income levels in Africa. Their investigation used economic development, energy intensity, carbon emissions, and renewable energy data from over 50 African countries from 1980 to 2018 and Panel CCEMG and PMG. They suggested that investments in renewable energy projects associated with better economic activity management could ensure sustainable development. Additionally, they encouraged country-specific and regional studies with more factors leading to carbon emissions. According to Vural [40], renewable energy and non-renewable’s effect on economic growth energy are very close. This is consistent with Apergis and Payne [41]. This finding means renewable energy could valuably replace non-renewable energy, with the advantage of adding to the environmental quality. This study used renewable and non-renewable energy, capital, and labor from six Sub-Saharan African countries during 1990–2015 and FMOLS. They encouraged more investments in the clean energy sector, tax incentives to encourage the private sector to adopt renewable energy, and skills training.

Maji [42] examined clean energy’s impact on economic development in Nigeria by using the ARDL bounds test, clean energy (alternative and nuclear energy and electric power consumption), and GDP. Their results reveal a mixed impact on the economic development of clean energy. More precisely, alternative and nuclear energy were found to retard economic growth, while combustible renewables and waste favor economic growth. Therefore, particular attention to renewable energy was recommended.

This contrasts with Maji et al. [43], who demonstrated that using renewable energy lowers productivity in 15 West African countries. However, they recommended drawing inspiration from European countries by adopting advanced clean technologies and increasing the proportion of solar, wind, and geothermal energy in the renewable energy mix.

Tsaurai and Ngcobo [44] support that renewable energy use reduces economic development, while education favors reducing the harmful effect of renewable energy on development in BRICS countries. This conclusion was obtained using data on economic growth, renewable energy consumption, education, saving, infrastructural development, trade openness, financial development from 1994 to 2015, and FMOLS. Therefore, they suggested that the countries’ leaders invest more in education.

Alper and Oguz [45] found no renewable energy effect on economic development in Cyprus, Estonia, Hungary, Poland, and Slovenia. The ARDL model and annual data from 1990 to 2009 were used. This finding supports Bulut and Muratoglu [46] in Turkey. They explain their results by the feeble renewable energy proportion in total energy consumption. However, their main recommendation is to increase clean energy use and promote its production.

The recent development of an asymmetric ARDL cointegration methodology by Shin et al. [47] uses both positive and negative partial sum decompositions to detect both long- and short-run asymmetric effects. The NARDL model devolves into the conventional symmetric ARDL model if the impact of separated components of an explanatory variable is found to be the same [48]. The nonlinear ARDL was used by Katrakilidis and Trachanas [49] to explore the drivers of the housing price dynamic in Greece. Namahoro et al. [50] evaluated the asymmetric nexus of renewable energy and economic development and how agriculture and capital may improve growth using NARDL in Rwanda, with data from 1990–2015. The study outcome encourages prioritizing investments in the agriculture and renewable energy sectors. This extension of the ARDL model was also used by Somoye et al. [35], Bibi and Li [51], and Toumi and Toumi [52] to study the asymmetric effect of renewable energy use on the economy. The main point of their conclusion promotes the use of cleaner technologies while maximizing renewable energy advantages and minimizing its harmful effects.

However, photovoltaic installations, one of the fastest-growing renewable energy sources, have drawbacks [53]. They are dependent on local solar irradiance’s availability and amount, which varies with cloud cover and can increase the voltage in the power grid, affecting individual installations [54]. To overcome the failures of photovoltaic installations, Kut and Pietrucha-Urbanik [53] proposed a new Multiple-Criteria risk assessment method to support the process of managing photovoltaic systems by analyzing the reliability of installation operations, as reliability affects investments and operating costs’ profitability. Following Adachi [55], the improvement of the connection to grid networks in the shielding process and plans for a shift from a niche space to a socio-technical regime in an energy sector structure in an empowering process are the two main factors of renewable energy policies. He also pointed out that delaying the implementation of renewable energy policy can lead to energy policy failure, as is the case in Poland. Germes et al. [56] declared that the success of renewable energy policies is closely tied to skills, competencies, and installations’ places. Forootan et al. [57] reported that machine and deep learning could improve the response methods to energy demand since using the machine and deep learning algorithms has significantly improved models’ accuracy.

This study offers three contributions. Firstly, at the state level, this is the first paper that analyzes the effect of CH₄ and N₂O emissions on Angola’s GDP. Secondly, several papers have used CO₂ emissions as the sole indicator of environmental degradation to test the EKC theory empirically. With its reliance on methane and nitrous oxide emissions, this study adds to the body of research. Thirdly, this paper will assess the responses of GDP to the shock of greenhouse gas emissions by 2032. Indeed, this study’s outcome will help fuel the debate on environmental and development issues.

4. Methodology and Data Used

4.1. Methodology

After the review, the ARDL model was chosen to examine the short and long-run relationship of small-size sample data. The first step in analyzing time series data is determining the lag length and integration order. Properly selecting the lag length may help avoid the parameter’s poor and inefficient estimates or false significance [58]. The second step is the unit root test, which helps specify the appropriate model. Augmented Dickey-Fuller [59] and Phillips Perron [60] were applied for stationarity tests under the null hypothesis of no stationarity. Later, the residual diagnostic tests were used to check our results’ robustness and validity.

Autoregressive distributive lag is one of the most dynamic models in the econometric literature [61]. Bildirici and Ersin [62], Nurgazina et al. [63], and Mirza and Kanwal [64] used it to look into the connection between economic development and environmental quality. The ARDL model was developed by Pesaran and Smith [65] and Pesaran et al. [66]. ARDL can be applied irrespectively of whether the variable is integrated order zero, one, or fractionally cointegrated [67]. Nevertheless, within the ARDL framework, the variable should not be integrated order two or higher because these orders of integration discredit the F-statistics and all critical values [68]. The model crashes when the integration order exceeds one. Regardless of the endogeneity of some regressors, the ARDL technique offers unbiased estimates and correct t-statistics [69]. The ARDL model to study the long-term causal relationships from CO₂, CH₄, and N₂O emissions to GDP can be written as follows:

$$\Delta {\mathrm{GDP}}_{\mathrm{t}}=\mathsf{\mu }+\mathrm{A}+\mathrm{B}+{\mathsf{\epsilon }}_{1,\mathrm{t}}$$

(1)

where:

$$\mathrm{A}=\sum {\mathsf{\beta }}_{\mathrm{i}}\Delta {\mathrm{AME}}_{\mathrm{t}-\mathrm{i}}+\sum {\mathsf{\beta }}_{\mathrm{j}}\Delta {\mathrm{COBR}}_{\mathrm{t}-\mathrm{j}}+\sum {\mathsf{\beta }}_{\mathrm{k}}{\mathrm{COEFT}}_{\mathrm{t}-\mathrm{k}}+\sum {\mathsf{\beta }}_{\mathrm{l}}{\mathrm{COEHE}}_{\mathrm{t}-\mathrm{l}}+\sum {\mathsf{\beta }}_{\mathrm{m}}{\mathrm{COEMC}}_{\mathrm{t}-\mathrm{m}}$$

(2)

$$\mathrm{B}={\mathsf{\alpha }}_1{\mathrm{AME}}_{\mathrm{t}-1}+{\mathsf{\alpha }}_2{\mathrm{COBR}}_{\mathrm{t}-1}+{\mathsf{\alpha }}_3{\mathrm{COEFT}}_{\mathrm{t}-1}+{\mathsf{\alpha }}_4{\mathrm{COEHE}}_{\mathrm{t}-1}+{\mathsf{\alpha }}_5{\mathrm{COEMC}}_{\mathrm{t}-1}+{\mathsf{\alpha }}_6{\mathrm{NOEES}}_{\mathrm{t}-1}$$

(3)

ARDL approach to cointegration or ARDL bounds test is determined according to the comparison between the computed F-statistic and the critical value band under the no cointegration hypothesis. No cointegration exists when F-statistics fall well below the critical value’s lower bound. There is cointegration if F-statistics exceeds the critical value’s upper bound. A value of F-statistics inside the critical band value indicates an inconclusive result. Once cointegration is confirmed, error correction can be derived from the ARDL model and (1) become (4):

$$\Delta {\mathrm{GDP}}_{\mathrm{t}}=\mathsf{\mu }+\mathrm{A}+\mathrm{B}+{\mathsf{\gamma }\mathrm{ECT}}_{\mathrm{t}-1}+{\mathsf{\epsilon }}_{1,\mathrm{t}}$$

(4)

where $\mathsf{\gamma }$ represents the coefficient of error correction term, $\Delta$ is the first difference operator, ${\epsilon }_{1,t}$ is the error term. ${\mathsf{\beta }}_{\mathrm{i}}$, ${\mathsf{\beta }}_{\mathrm{j}}$, ${\mathsf{\beta }}_{\mathrm{k}}$, ${\mathsf{\beta }}_{\mathrm{l}}$, ${\mathsf{\beta }}_{\mathrm{m}}$, and ${\mathsf{\beta }}_{\mathrm{m}}$ are the short-term dynamic coefficients, while ${\mathsf{\alpha }}_1$, ${\mathsf{\alpha }}_2$, ${\mathsf{\alpha }}_3$, ${\mathsf{\alpha }}_4$, ${\mathsf{\alpha }}_5$, and ${\mathsf{\alpha }}_6$ are long-term coefficients. $\mathsf{\mu }$ is the drift component. A significant negative value of $\mathsf{\gamma }$ depicts a long-term association with the dependent variable and can be used to determine the adjustment speed to equilibrium [70]. An adjustment speed exceeding 0.5 denotes a quick adjustment to equilibrium, while a value less than 0.5 means a slow adjustment.

4.2. Data Used

Table 1. Data summary.

	AME	COBR	COEFT	COEHE	COEMC	GDP	NOEES
Mean	52.86304	18.45086	39.99152	15.79567	22.43967	2.832000	0.619003
Median	57.20339	20.60606	39.83740	15.79052	19.54545	2.779322	0.546780
Maximum	72.97428	32.45335	71.25000	30.46358	51.89573	3.747771	1.336172
Minimum	17.67796	2.255639	22.25201	7.345972	0.326934	1.655224	0.261097
Std. Dev	14.18966	9.496927	11.04735	4.759229	13.13802	0.655165	0.291017
Skewness	−0.950651	−0.360546	0.614133	0.712397	0.233229	−0.124910	0.852588
Kurtosis	3.015802	1.910852	3.512072	3.448883	2.389191	1.698826	2.739193
Jarque-Bera	7.682295	3.625710	3.763066	4.742008	1.255175	3.730359	6.323239
Obs	51	51	51	51	51	51	51

presents the study variables. Data from 1971 to 2021 on agricultural methane emissions (AME), carbon emissions from residential buildings and commercial and public services (COBR), carbon emissions from heat and electricity (COEHE), carbon emissions from manufacturing industries and construction (COEMC), carbon emissions from transport (COEFT), and gross domestic product (GDP) and Nitrous oxide emissions in the energy sector (NOEES) have been used. Except for the GDP data from International Energy Agency (IEA), the other variables are sourced from Word Bank data. GDP is expressed in billion 2010 USD using exchange rates, AME is expressed in the percentage of total methane emissions, and NOEES is expressed in the percentage of total nitrous oxide from the energy sector, while COEFT, COBR, COEMC, and COEHE are expressed in the percentage of carbon emissions from total fuel combustion. The data spread over a 51-year-old period.

Figure 1 describes the evolution of the study variables over the years. Methane emissions from agriculture decreased over time (Figure 1a). This reduction of methane was accentuated after the war. This could be explained by the decline in agricultural activities since much arable land is infected by landmines. Moreover, a large part of the population has been around agricultural work, causing increased unemployment and decreased agricultural productivity because of the civil war. The carbon emissions from residences, commerces, and public services are experiencing an increasing trend (Figure 1b). From 1971 to 2000, carbon emissions from transport decreased and then increased after 2000 due to the war’s end, the country’s opening to foreigners, and the existence of a modest class (Figure 1c). Indeed, just after the civil war, the country experienced car importations and the construction of road networks. This adds to a rise in energy consumption, increasing carbon emissions from transport. The variations of carbon emissions from heat and electricity are mixed and scattered.(Figure 1d). Carbon emissions from manufacturing industries and construction started increasing in 1971, reached a turning point in 1995, and then decreased (Figure 1e). This is encouraging since the economy is not suffering from that fall. GDP per capita started growing after 1995, even if the GDP decreased for a moment (Figure 1f). It is noted that this growth does not benefit citizens, hence the need to build strong policies. Nitrous oxide emissions from the energy sector are increasing (Figure 1g).

5. Results and Discussion

5.1. Variable Lag Selection

The optimal lag length for AME, COBR, COEHE, and NOEES is 1, and the optimal lag length for COEFT, COEMC, and GDP is 2 (

Table 2. Optimal lag length.

Variables	Lags	Lag Selection Criteria
		AIC	SC	HQ
	0	8.2805	8.2805	8.2560
	1 s	26.3218 *	6.1081 *	6.1377 *
AME	2	26.8665	6.2467	6.1730
	3	26.4683	6.2709	6.1727
	4	27.4694	6.3469	6.2242
	0	7.2918	7.3311	7.3066
	1 s	5.3795 *	5.4582 *	5.4091 *
COBR	2	5.4171	5.5352	5.4616
	3	5.4577	5.6152	5.5170
	4	5.4997	5.6965	5.5738
	0	7.1541	7.1935	7.1689
	1	5.9291	6.0078	5.9587
COEFT	2 s	5.9098 *	6.0279 *	5.9542 *
	3	5.9519	6.1094	6.0112
	4	5.9497	6.1465	6.0237
	0	5.9669	6.0062	5.9817
	1 s	5.3339 *	5.4126 *	5.3635 *
COEHE	2	5.3736	5.4917	5.4180
	3	5.4149	5.5724	5.4742
	4	5.4330	5.6298	5.5071
	0	8.0652	8.1046	8.0800
	1	6.0804	6.1592 *	6.1100 *
COEMC	2 s	6.0829	6.2010	6.1274
	3	6.1236	6.2810	6.1828
	4	6.0385 *	6.2354	6.1126
	0	1.9866	2.0260	2.0014
	1	−0.5416	−0.4629	−0.5120
GDP	2 s	−0.9598 *	−0.8417 *	−0.9153 *
	3	−0.9179	−0.7605	−0.8587
	4	−0.8992	−0.7023	−0.8250
	0	0.4657	0.5051	0.4805
	1 s	−0.4443 *	−0.3656 *	−0.4147 *
NOEES	2	−0.4043	−0.2862	−0.3598
	3	−0.3761	−0.2186	−0.3168
	4	−0.3572	−0.1604	−0.2831

Note: ^s represents the selected lag length and * designates the minimum value of lag selection.

).

5.2. Stationarity Tests

Table 3. Unit root tests.

	Variables	ADF-Test		PP-Test
		F-Stat	p-Value	F-Stat	p-Value
At level	AME	1.0296	0.9963	0.0977	0.9625
	COBR	−1.1133	0.7036	−1.1132	0.7036
	COEFT	−2.3791	0.1527	−2.3121	0.1723
	COEHE	−2.4250	0.1415	−3.1083	0.0323 **
	COEMC	−0.9241	0.7718	−0.6681	0.8452
	GDP	−1.7650	0.3931	−1.4004	0.5748
	NOEES	−1.9551	0.3052	−1.8993	0.3300
At first difference	$\Delta$AME	−6.2339	0.000 ***	−7.8658	0.000 ***
	$\Delta$COBR	−4.4112	0.001 ***	−7.6491	0.000 ***
	$\Delta$COEFT	−9.0209	0.000 ***	−8.8025	0.000 ***
	$\Delta$COEHE	−2.8372	0.005 ***	−8.2917	0.000 ***
	$\Delta$COEMC	−2.9930	0.0428 **	−8.5541	0.000 ***
	$\Delta$GDP	−3.7882	0.006 ***	−3.8088	0.005 ***
	$\Delta$NOEES	−7.9752	0.000 ***	−7.9752	0.000 ***

Note: ***, and ** refer to the significance levels at 1%, and 5%.

shows that the variables gained stationarity in the first difference, fulfilling the requirements to apply the ARDL model.

5.3. ARDL Bounds Test

The f-statistic value exceeds the critical band’s upper bound value of 5%, indicating that AME, COBR, COEFT, COEHE, COEMC, GDP, and NOEES share a long-run relationship, according to

Table 4. ARDL bounds tests.

Test Statistic	Value	Signif.	Lower Bounds	Upper Bounds
F-Statistic	4.9728	10%	1.99	2.94
		5%	2.27	3.28
		2.5%	2.55	3.61
		1%	2.88	3.99

Null hypothesis: no levels of relationship.

.

5.4. Short-Run Relationship

In

Table 5. ARDL short-run relationships.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
$\Delta \mathrm{GDP} \left(-1\right)$	0.7015	0.0663	10.5739	0.000 ***
$\Delta \mathrm{AME}$	0.0208	0.0062	3.3325	0.002 ***
$\Delta \mathrm{COBR}$	0.0315	0.0077	4.1110	0.000 ***
$\Delta \mathrm{COBR} \left(-1\right)$	0.0025	0.0076	0.3205	0.7504
$\Delta \mathrm{COBR} \left(-2\right)$	−0.0118	0.0060	−1.9509	0.0585 *
$\Delta \mathrm{COEFT}$	−0.0035	0.0058	−0.6028	0.5502
$\Delta \mathrm{COEFT} \left(-1\right)$	0.0120	0.0048	2.5242	0.0159 **
$\Delta \mathrm{COEHE}$	−0.0133	0.0072	−1.8525	0.0717 *
$\Delta \mathrm{COEMC}$	−0.0250	0.0059	−4.2158	0.000 ***
$\Delta \mathrm{NOEES}$	0.0777	0.1139	0.6821	0.4993
ECT (−1)	−0.2984	0.0434	−6.8637	0.000 ***
C	−0.2957	0.5693	−0.5194	0.6064
R-squared	0.9673
Adj. R-squared	0.9587

Note: ***, **, and * refer to the significance levels at 1%, 5%, and 10%.

, the short-run analysis puts in evidence of several relationships. AME, COBR, and Lags 1 of GDP and COEFT positively affect GDP, while COEHE, COEMC, and lag 2 of COBR negatively affect GDP. More precisely, a 1% increase in AME, COBR, and lags 1 of GDP and COEFT increases GDP by 2.08%, 3.15%, 70.15%, and 1.20%, respectively, ceteris paribus. This finding supports that investment and assistance funds, agriculture methane emissions, carbon emissions from residential buildings, commercial and public services, and transport add to economic development. This is consistent with the findings of Joo et al. [71] in Chile and Mirza and Kanwal [64] in Pakistan, who found that carbon emissions support economic development. Again, the result indicates that a 1% increase in COEHE, COEMC, and lag 2 of COBR, respectively, reduces GDP by 1.33%, 2.50%, and 1.18%, ceteris paribus, indicating that carbon emissions from heat and electricity, from manufacturing industries and construction, and delayed carbon emissions from residences, commercial, and public services reduce economic growth. This implies that reducing carbon emissions from heat and electricity, manufacturing industries, and, to some extent, from residences, commercial, and public services will not negatively affect GDP. This aligns with the study outcome of Zou and Zhang [72] in 30 provinces of China and Olufemi and Olalekan [73] in Nigeria.

The error correction term has a significant negative coefficient at a 5% level, implying a long-term association and confirming the ARDL bounds tests. The speed adjustment is 29.57%, indicating a slow adjustment rate to equilibrium.

5.5. Long-Run Relationship

The long-run analysis presented in

Table 6. Long-run results.

Variable	Coefficient	Std. Error	t-Statistic	Prob.
AME	0.0695	0.0263	2.6477	0.01 **
COBR	0.0744	0.0237	23.1417	0.00 ***
COEFT	0.0286	0.0137	2.0875	0.04 **
COEHE	−0.0446	0.0251	−1.7776	0.08 *
COEMC	−0.0837	0.0230	−3.6388	0.00 ***
NOOEES	0.2604	0.3868	0.6731	0.50

Note: ***, **, and * refer to the significance levels at 1%, 5%, and 10%.

reveals that AME, COBR, and COEFT increase GDP, while COEHE and COEMC reduce GDP. A 1% increase in AME, COBR, and COEFT raises GDP by 6.95%, 2.37%, and 2.86%, respectively, ceteris paribus. A 1% increase in COEHE and COEMC reduces GDP by 4.46% and 8.37%, respectively, ceteris paribus. However, no long-run relationship ranging from NOEES to GDP was found. These results show that methane from agriculture positively affects economic growth, while carbon emissions have asymmetric effects on economic development in the long run. Nitrous oxide from the energy sector has no significant impact on economic development.

5.6. Residual Diagnostics and Stability Test Results

Table 7. Residual diagnostic tests.

Breusch Godfrey: Serial Correlation		Heteroskedasticity Test: Breusch-Pagan-Godfrey		Jarque–Bera: Normality Test
F-Stat	p-Value	F-Stat	p-Value	F-Stat	p-Value
2.3016	0.1146	0.6265	0.7817	0.0834	0.9592

outlines the ARDL model’s residual diagnostic test results. The serial correlation was examined under Breusch-Godfrey’s test, the heteroskedasticity under Breusch-Pagan-Godfrey’s test, and normality under Jarque-Bera’s test. The result shows that residuals are normally distributed and homogeneous and present no serial correlation issue. Therefore, our model is valid and reliable.

The graphs below present the CUSUM (Figure 2a) and CUSUMQ (Figure 2b) stability test results. Our model is stable as the CUSUM and CUSUMQ graphs (blue line) lie within the 5% significance bounds (red lines).

5.7. Impulse Response Function

Assuming that one of the VAR errors returns to zero in succeeding periods, and that all other errors are equal to zero, the impulse responses function plots the response of each variable’s current and future values to a one-unit rise in the current value of that error [74]. Figure 3 presents the GDP responses to each study variable’s shock from 2022 to 2032. The vertical axes are expressed in units of GDP variable and the horizontal axes vary from 0 to 10, where 2021 marks the origin. The black line represents the variation of GDP because of the other variable shock and red lines represent the bound of 5% significance level.

We ascertain that a standard deviation shock to AME hurts GDP in the short and long terms (Figure 3a). GDP will decrease from 2023 to 2026, hit its steady state value from the 4th to 5th period (2026 to 2027), and then increase till 2032. Similarly, a standard deviation shock to COBR, COEHE, and NOEES hurts GDP (Figure 3b–d). However, with a shock to COBR, GDP will start increasing from the 6th period (2027). Although a standard deviation shock to GDP positively impacts GDP, it may cause, at first, a minor rise from the 1st period to the 2nd (2022 to 2023), a decrease from the 2nd period to the 8th period (2023 to 2029), and then a rise again (Figure 3e). A standard deviation shock to COEMC has an asymmetric effect on GDP, positive from the 1st to the 5th period (2022 to 2026) and negative after 2026 (Figure 3f). GDP will start decreasing just after the 3rd period (2024), enter the negative region from the 5th period (2026), and remain in the negative region. A standard deviation shock from COEFT maintains the GDP in the positive region till the 9th period (2031) and start decreasing just after (Figure 3g).

6. Conclusions

This paper analyzes the effects of CO₂, N₂O, and CH₄ emissions on economic development in Angola. ARDL and ECM were employed on data on agricultural methane emissions, GDP per capita, carbon emissions from residences, commerces, and public services, carbon emissions from heat and electricity, carbon emissions from manufacturing industries and construction, carbon emissions from transport, and nitrous oxide emissions in the energy sector, covering the period 1971–2021. The results show that agricultural methane and carbon emissions from residences, commerces, and public services contribute to economic development. Carbon emissions from heat and electricity, manufacturing industries, and construction harm economic growth in the short and long term. The results demonstrate that nitrous oxide emissions from the energy sector do not influence economic growth. The generalized findings are that carbon emissions have asymmetric effects on economic growth, methane emissions from agriculture add to growth, and nitrous oxide emissions from the energy sector do not affect growth. These results lead to several policy implications, the most important of which follows.

Agricultural methane, carbon from residences, commerces, and public services emissions contribute to economic development. Reducing emissions from agricultural methane and residences and commerces and public services emissions could somewhat deteriorate the Angolan economy because commerce and public services support employment, reduce inequalities, and may help the authorities in their poverty reduction policies. Therefore, environmental policy protection might harm economic growth if appropriate measures are taken. The country should start diversifying its economies away from heavily polluting fuel sources, such as coal. Increasing the proportion of renewable energy in the total energy mix may support growth while considering the environmental quality. The uptake of renewable consumption is very slow, with traditional hydropower playing a significant role in the generation mix. Renewable energy sources, such as wind and solar, are gradually being deployed. For renewable energy to be fully adopted, it is imperative that the country produces it and umpires energy prices. Better manure management may help to reduce methane from agriculture and livestock. Additionally, livestock food diets with cerebral palsy and balanced fiber would optimize the release of N₂O and CH₄. Since agricultural methane emissions are experiencing a gradual decrease, the government’s efforts should be encouraged.

Carbon emissions from heat, electricity, manufacturing industries, and construction harm economic growth in the short and long term. This finding means that reducing carbon emissions from heat, electricity, manufacturing industries, and construction will not severely affect economic development. It is equally conspicuous from the study that Angola is experiencing the worst consequences of climate change, given that the greenhouse gas emission effects are significant in the long- and short-run analyses. This leads to the reality that the country will continue to bear the brunt of climate change. Policymakers should formulate economic policies encouraging manufacturing industries to move towards clean energy. The country might reward industries complying with its energy policies. Implementing command and control instruments will go a long way to protect the environment and bring about sustainable development. Taxes can be put on heavily polluting generation sources or agents that pollute the economies to fund projects or handouts to people experiencing poverty.

The lag in GDP supports economic development in the short term. This finding implies that sound investment may sustain economic development. The country should invest in training skills, build infrastructures, promote entrepreneurship, and modernize agriculture. The country’s economic development is low and inequitable. Therefore, there is a need to move economic growth to a path of inclusive growth by formulating macroeconomic and fiscal policy that will balance this disequilibrium; that is, the amalgamation of economic and environmental policies will attain sustainable development.

The impulse response function test ascertains that a shock to GDP will significantly benefit economic growth. This means that huge investments should accompany economic policies. For economic development to occur, leaders should promote job creation and reinforce existing jobs by valuing salaries and subsidizing agricultural intrants and tractors. Investing in the raw materials processing industries might add to the country’s economy. Additionally, the authorities might invest in quality seeds adapted to climatic realities. This could increase agricultural productivity, lessen climate change’s adverse effects, and promote employment and economic growth.

As the effects of CO₂, CH₄, and N₂O are significant, this implies that the consequences of these emissions will increase over the years if no effective measures are taken. As a result, Angola will be increasingly affected by gas emissions. This is a motivating cause for the transition to renewable energies and to promote solar photovoltaics, the most easily deployable renewable energy in Angola and Sub-Saharan Africa. However, to meet the increasing demand for energy in a country with relatively high poverty rates, the authorities need to establish an energy policy that facilitates access to energy for the most disadvantaged citizens and enables them to carry out their energy-dependent economic and social activities. There will be a need for a sufficient energy supply, quality equipment, skills, and competencies (for installations and maintenance), and a reasonable energy price. This will push authorities to make more efforts to optimize the production and use of renewable energy and might be a model for other countries in Sub-Saharan Africa.

Future research should consider economic development’s potential effects, climate adaptation, and mitigation in the Sub-Saharan Africa region. Studying pathways to optimize the production and installation of photovoltaic solar energy in Sub-Saharan Africa would be interesting. This might reduce the risk of failures in photovoltaic installations and help address the rising demand for energy in remote areas of the country. Testing machine learning methods in optimizing the operational performance of photovoltaic installations would be significant in managing renewable energy installation. The limitation was due to the absence of data for certain parameter periods.