Economic growth theories offer critical insights into the long-term drivers of economic progress, each providing a different perspective on the roles of capital, technology, finance and innovation. The subsection reviews the main theories relevant to understanding the finance-growth-energy nexus in the Southern African Development Community (SADC).

Classical Growth Theory emphasizes capital accumulation, labour force expansion and productivity as primary sources of economic growth. While this theory underscores the importance of efficient resource allocation, it lacks the dynamic factors, e.g. technological advancement, that drive modern economic development. For SADC, where capital scarcity and labour productivity constraints are prevalent, classical perspectives are limited in explaining how financial development could spur broader growth.

Neoclassical Growth Theory extends classical thought by incorporating technological progress as an exogenous factor driving long-term growth. According to neoclassical models, continual productivity improvements, typically through innovation and investment in human capital, are crucial for sustainable growth. This aligns with SADC‘s need to improve productivity through investment in education, infrastructure, and technology — a process that robust financial development could help facilitate by providing accessible credit and investment opportunities.

Endogenous Growth Theory, championed by economists such as Lucas and Romer, identifies internal drivers of growth, particularly human capital, knowledge, and innovation. Lucas’s perspective on the finance-growth nexus underscores that well-developed financial systems are essential for accumulating human capital, as they enable investments in education and skill development. In SADC, where human capital is critical for improving productivity and resilience, access to financial services can empower individuals and businesses to contribute meaningfully to economic growth. This theory is particularly relevant as it suggests that targeted financial policies can directly impact growth by fostering human capital and innovation within the region.

Neo-Schumpeterian Theory, an extension of Schumpeter’s work on creative destruction, emphasizes the dynamic role of innovation, finance, industry, and science in economic growth. According to Hanusch and Pyka (2007), financial markets and institutions play a critical role in this process by allocating resources to innovative sectors and facilitating technological advancements. In the SADC context, the Neo-Schumpeterian approach highlights the importance of supporting industries with high growth potential, such as renewable energy, which can address both economic and environmental goals. The theory suggests that proactive financial policies encouraging innovation and industrial upgrading are vital for the SADC countries as they can help diversify their economies and develop energy infrastructure in a way that promotes sustainability.

In summary, these growth theories illustrate the essential roles of financial systems and innovation in driving sustainable economic growth, particularly within structurally constrained regions like SADC. By fostering access to finance and supporting investments in technology and human capital, policymakers can steer the economy to a growth path that is resilient, inclusive, and sustainable.

The question of how financial development and economic growth sway the relationship between energy consumption and other economic variables has been a focal point of inquiry among economists for an extended period. This sustained interest is driven by the significant policy implications of research results, which can determine the choice of strategies to accelerate economic growth, enhance financial development and manage energy consumption. However, empirical studies in this domain have yielded conflicting results, leading to disagreements among economists. This section presents an empirical review of studies that have explored the impacts of financial development and economic growth on energy consumption.

This section is organized into four subsections: (1) causal relationship between financial development, economic growth, and energy consumption; (2) the impact of financial development and economic growth on energy consumption; (3) a summary of the empirical literature; and (4) a conclusion.

The paper uses a sample of annual data for 1980-2023 to estimate a panel data model made up of the sixteen SADC countries. The data was sourced from the World Bank (2023) for all the five variables used in the study: financial development, energy consumption, economic growth, industrialization and urbanization. A sample size of 43 years in the time series was wide enough to allow for stability in the model (Ma and Fu, 2020).

This study modified the model used by Ma and Fu (2020), who examined the relationship between financial development and energy consumption, and included the variables used by Sahoo and Sethi (2020) and by Naeimi, Jahantigh and Varahrami (2023). The technique of econometric analysis is based on the panel model of dynamics as used by Sbia, Shahbaz and Ozturk (2017) with all the elements of cross-sectional and time series employed by Gungor and Simon (2017). The estimated model was formerly specified as follows:

EC_i,t = α + β₀FD_i,t + β₁GDP_i,t + γControl_i,t + μ_i + e_i,t (Eq. 1)

Ultimately, in equation Eq. 1 above EC_i,t represents energy consumption, FD_i,t and GDP_i,t denotes financial development and economic growth, α shows the intercept, β′s represent the coefficient (%) of the conforming independent variables, γControl_i,t denotes the control variables to avoid bias (urbanization and industrialization); μ_i is the unobserved country-specific effect; e_i,t denotes the residual term; and lastly, i and t represent the country (SADC countries) and the period, respectively (Ma and Fu, 2020).

Eq. 1 above given by Ma and Fu (2020) is now remodeled into equation Eq. 2 following Shahbaz and Leon (2012). Compared to the linear functional form, the log-linear specification provides better results (Sahoo and Sethi, 2020). Further, all the data in this study were transferred into natural logarithmic ones. The above-estimated model can be written as a log-log model below:

LnEC_i,t = β₀ + υ_i + β₁ · LnFD_i,t +β₂ · LnGDP_i,t + β₃ · Ln(FD · EG)_i,t + β₄ · LnIND_i,t + β₅ · LnURB_i,t + DT + e_i,t (Eq. 2)

Where: Ln — Natural logarithm, EC — Energy consumption, FD — Financial development, EG — Economic growth, IND — Industrialization, URB — Urbanization, (FD · EG) — Interaction variable, i — the country (1,…, N: including all chosen countries), t — time (1,…, T: from 1980-2023), υ_i — random intercepts, β′s = Coefficient, e_i,t — error term and DT — dummy variable.

Energy consumption, denoted as EC, is the total energy consumption measured as kWh per capita (Odhiambo, 2018). kWh per capita is used to capture total energy consumption per capita as the dependent variable (Ma and Fu, 2020). Energy consumption includes indicators such as electricity, gas, oil, and coal consumption (Karakurt and Aykutalp, 2020). Electric power per capita (kWh) is the production of power plants, combined heat and powerless transmission, distribution and transformation losses, and own use by heat and power plants, divided by mid-year population (Strauss, Li and Cui, 2021).

Financial development, denoted as FD, is measured as banks‘ domestic credit to the private sector as a percentage of GDP (Ma and Fu, 2020). There are different measures for financial development, including pension fund assets, mutual fund assets and insurance premiums, life and non-life (Odhiambo, 2018). The study by Sahoo and Sethi (2020) used domestic lending to the private sector to measure financial development. This measure is preferred because it captures the full degree of intermediation in developing countries, where governments borrow from financial markets to build infrastructure for economic development (Ulucak, 2021). Following Odhiambo’s (2018) findings, Karakurt and Aykutalp (2020) expected this variable to increase energy consumption because easy access to credit enables consumers to purchase more energy.

Economic growth denoted as GDP is measured as the GDP per capita growth annual percentage (Ibrahiem, 2020). This shows the rate at which a nation’s GDP grows from year to year (Ma and Fu, 2020); capturing the distribution of income, it enables cross-country comparisons (Lefatsa et al., 2021). Following Denisova’s (2020) findings, the study expected this variable to increase energy consumption because the growing size of the economy results in higher energy use.

Industrialization (control variable), denoted as IND, is measured as value added share (%) of GDP in the industry including construction (Eren, et al., 2019). Industrialization is seen among the drivers of global pollution (Ma and Fu, 2020). More industrialized area demands more energy supply (Sadorsky, 2019). This variable was expected to increase energy consumption, according to Odhiambo (2018).

Urbanization (a control variable), denoted as URB, is measured as the percentage share of urban population in the total population (Sare, 2019). According to Chowdhury, Chowdhury, Chowdhury, Islam, Saidur and Sait (2019), cities have higher energy consumption compared to the countryside. This variable was expected to increase energy consumption (Odhiambo, 2018).

All independent variable coefficients were expected to be positive, except the ambiguous dummy variable (Enoki et al., 2019). The dummy variable was used to gauge any exogenous factors that might affect the tested variables (Sare, 2019). A dummy variable, denoted as DT, was also used to control for the effect of the financial crisis and covid-19 (Cao et al., 2020). There were some years and areas without financial crises (Lefatsa et al., 2021), but covid-19 affected the whole world.

This research followed a study that was conducted in ASEAN + six countries by Hoang (2021) who applied the Bayesian approach via the Metropolis-Hasting and Gibbs samplers as Markov chain Monte-Carlo (MCMC) methods from 1980 to 2016, as recommended in the literature review. The study applied the Bayesian approach via the Metropolis-Hasting and Gibbs samplers as MCMC methods from 1980 to 2023 to estimate the impact of financial development and economic growth on energy consumption for the SADC countries. Finally, the study used Dumitrescu and Hurlin (2012) and Diagnostic tests to check the causality between all the variables and the accuracy of the data and model.

The study applied the Bayesian approach via Metropolis-Hasting and Gibbs samples as the MCMC methods to estimate the impact of financial development and economic growth on energy consumption. The study employed panel autoregressive distributive lag to test for integration between the variables and Dumitrescu and Hurlin (2012) and Diagnostic tests to check the causality between all the variables in question and accuracy of the data and model.

To make sure that there is no multicollinearity amongst explanatory variables, the process of estimation starts with preliminary tests that check correlation. The correlation matrix is presented in Table 1 below.

The absence of multicollinearity among explanatory variables, as indicated by the correlation values between 0.0056 and 0.464, suggests that the variables are not highly correlated. With a benchmark threshold of 0.80 (Irfan, 2014), this range confirms no multicollinearity problem. Thus, all variables can be included in the model without concern for inflated variances in estimated coefficients.

This section presents the summary statistics for the primary variables under scrutiny. These statistics are displayed in Table 2, highlighting the mean, standard deviation, minimum, and maximum values, along with the count of observations.

Table 2 shows the Jarque-Bera test results, with probability values exceeding 0.0000, indicating that none of the series in the dataset follows a normal distribution. While non-normality does not invalidate regression analysis, it could affect inference in smaller samples, particularly for estimators that assume normality. However, robust estimators or large-sample properties (central limit theorem) may mitigate this concern.

The unit root tests indicate that the variables, at levels, are non-stationary but become stationary after first differencing. This is observed in the figures, where the series at levels fluctuate without convergence but exhibit stability around zero after first differencing. Formally, the LM Pesaran test confirms that only two variables, IND and GDP, are stationary at levels, while the rest require differencing to achieve stationarity.

Implication: Given that most variables are integrated of order one (I(1)), the dataset is suitable for cointegration analysis to capture long-term relationships. Non-stationary series can lead to spurious results if analyzed directly, so this test supports the use of a model that accommodates cointegration for I (1) series.

4.3.1. Graphical analysis

The study first did a graphical analysis to examine the stationarity status of the variables. Figure 1 shows the graphical analysis of the variables.

All Figures show that the variables were not stationary at levels. The sets labeled (left) show variables at levels. The graphs show that the variables became stationary after being first differenced. This is indicated by the lines hovering around zero in the sets labeled (right). This is an indication that the variables were stationary after first differencing.

Figure 1. 

Energy Consumption. Source: Figure constructed by Eview 9 using statistics from World Bank (2024)

Figure 2. 

Financial Development. Source: Figure constructed by Eview 9 using statistics from World Bank (2024)

Figure 3. 

Economic Growth. Source: Figure constructed by Eview 9 using statistics from World Bank (2024)

Figure 4. 

Industrialisation. Source: Figure constructed by Eview 9 using statistics from World Bank (2024)

Figure 5. 

Urbanisation. Source: Figure constructed by Eview 9 using statistics from World Bank (2024)

The study performed a Pedroni cointegration and the results are displayed in Table 4 below.

The Pedroni Test, which has eight weighted subtests including panel-v, panel-rho, panel PP, and panel ADF statistics, is represented by the first panel in the table. Most of the tests have a p-value that is less than 0.05 and this shows that there was cointegration. This means that the study had to adopt some panel cointegration regressors.

Results show that GDP has a small but positive long run relationship with energy consumption. It means that when GDP is increasing energy consumption also increases. As the SADC countries’ economic growth increases, energy demand rises, too, meaning that if energy is constrained, economic growth pulls back in turn. The results concur with Zhao et al. (2023) who concluded that a long run positive relationship exists between economic growth and energy consumption in 30 Chinese provinces over a period 2000-2019. Also, evidence was drawn from the OECD and European Union Member States between 1980 and 2013, and between 2010 and 2019 by Gozgor et al. (2018); Laszlo (2023), who discovered a long run positive relationship between economic growth and energy consumption. Countries should work on creating new sources of energy and improving energy efficiency. Furthermore, when countries become more productive, they will produce more output and put in more resources, and most importantly, energy will also increase. Conversely, Darrian et al. (2023) discovered a negative and long run relationship between economic growth and energy consumption for Indonesia during the period 1985-2019. Molele (2018); Ahmed and Elfaki (2023) documented a long run negative relationship between economic growth and energy consumption for 15 economies from Asia, the Pacific, and Latin America; and South Africa in 1990-2018 and 1980-2012. Governments of these regions should implement policy measures encouraging efficient use of energy to promote healthy relations in the broader economy.

Industrialisation was not seen as having a significant long run relationship with energy consumption and so, presumably, it does not influence energy consumption. These findings concur with Lloyd (2017) who discovered no significant long run relationship between industrialisation and energy consumption in India, China, Indonesia, and Brazil over the period 1960-2014. Abdulkadir and Isik (2020) found no significant relationship between industrialization and Nigeria’s energy consumption between 1985 and 2017. Improving energy efficiency in industry is also difficult because of high complexity of industrial energy systems. On the other hand, Sadorsky (2014) revealed a long run positive relationship between industrialisation and energy consumption in emerging economies for the period 1971-2008. Elfaki et al. (2021) detected a strong positive long run relationship between industrialisation and energy consumption in Indonesia in 1984-2018. Expanding industrialization in the newly industrialised countries is driving the intensive use of energy. Nevertheless, Mentel et al. (2022) discovered a negative long-term relationship between industrialisation and energy consumption over 2000-2018 for Europe and Central Asia. Franck and Galor (2019) also found that industrialisation had negative effect on the use of energy in the long run as the adoption of labour-intensive skilled technology in the early stages of industrialisation at the current human resources level tends to reduce energy consumption and thus become an incentive to further introduce skills-intensive technologies.

Financial development was considered to be having a positive long run relationship with energy consumption, i.e. when financial development improved, energy consumption would grow. Lefatsa et al. (2021) discovered that, in the long run, financial development positively affected energy consumption in South Africa over the period 1980-2018. Thebuho et al. (2022) found a long run positive effect of financial development on energy consumption in 21 SSA countries between 1990 and 2016: well-performing stock markets stimulated growth, thereby increasing investor confidence. Nevertheless, the results obtained by Shahbaz et al. (2019) run contrary to those above. They revealed a long run negative relationship between financial development and energy consumption in India over the period 1960Q1-2015Q4. Ma et al. (2022) detected the same kind of relationship in 67 developing countries between 1995 and 2018. It follows that availability of corporate credit for resolving energy-related issues should be provided for in these countries’ plans.

Urbanisation was seen as having a negative long run relationship with energy consumption: when the former increases, there should be a decrease in the latter. Research discovered negative effects of urbanisation on energy consumption in the Indian and Iranian economies between 1971-2013. Ali (2021) revealed a long run negative relationship between urbanisation and energy consumption in 49 SSA countries over 1980-2014. At the same time, the results obtained by Sheng et al. (2017) show a long run positive relationship between urbanisation and energy consumption in 72 countries during 1995-2012. Zhao and Qamruzzaman (2022) also discovered a long-run positive relationship between urbanisation and energy consumption for Belt and Road countries over the period 2004-2020. Obviously, the urban sprawl and motorisation contributed to energy consumption growth. In either case, the findings suggests that countries must frame their urban policies to create positive externalities.

Running Bayesian inference after estimating a PMG model provides deeper insights into parameter uncertainty, allows for the incorporation of prior knowledge, and enhances the overall robustness and flexibility of the analysis. This approach aligns well with the thesis focus on Bayesian methods, offering a comprehensive framework for interpreting long-term relationships in the panel data. The Bayesian model is the main model of the study; its results are shown in Table 6 below.

These results show that economic growth has a positive effect on electricity consumption. When the economy is growing organizations and individuals consume more energy. This relationship is not very pronounced: even though an increase in economic growth should necessitate higher energy consumption, higher rates of economic growth bring about more productive energy use. These results are in line with Kahouli (2019) and Can and Korkmas (2019) who were the first to observe the nexus between economic growth in 34 OECD countries and Bulgaria over 1990-2016 and found a positive relationship between these variables. The results demonstrated the importance of economic growth for renewable energy consumption but also showed the negative effects of high pollution. It is therefore necessary to expand these countries‘ clean renewable energy projects to curb pollution. At the same time, Ngoc and Tram (2024) discovered a negative effect of economic growth on nuclear energy consumption for 11 Asian countries between 1980 and 2016; Simionescu (2023) saw the same effect in the European Union over 2002-2021: manufacturing activities were reduced resulting in low industrial pollution and environmental degradation.

Research has also shown that financial development positively influences electricity consumption, which means that financial development should not hinder increased energy consumption from the regional perspective but these processes need to be balanced. According to Zhe, Yüksel, Dinçer, Mukhtarov, and Azizov (2021) it holds true for China during in 1990-2015. Also, Ulucak (2021) confirmed a significant and positive relationship between financial development and energy consumption in Pakistan from 1980 to 2017. This suggests that a sound financial system attracts foreign investors and improves economic efficiency. Yet, Ding (2023) and Tariq et al. (2023) found a negative relationship between financial development and energy consumption in the G7 countries over 1990-2020 and in BRI countries over 2000 to 2020 respectively. Financial development may also cause improvements in the environment through funding environmental projects at reduced costs.

Results show that industrial development has a positive effect on electricity consumption. This suggests that, as industries develop, they consume more electricity. Hence there is a positive relationship between industrial development and electricity consumption. That is what we observe in the SADC region which is experiencing a huge growth in the industrial sector that brings about increased demand for energy. According to Simionescu (2023), growing industrial activity leads to greater use of advanced machinery compared to traditional agriculture and basic manufacturing, and this, too is always linked with expanding energy consumption. Simonescu’s findings are in line with those by Lefatsa et al. (2021) who proved the existence of a positive link between industrial development and energy use in South Africa for the period 1980-2023. Also, according to Sadorsky’s (2014) findings, industrialization is a major factor contributing to higher levels of energy consumption caused by the need to fuel machinery and increase automation of production processes. However, Li and Lin (2015) obtained the opposite results: their analysis showed that industrialization reduced energy consumption in 73 countries over the period 1971-2010. Zhang, Qamruzzaman, Karim, and Jahan (2021) also discovered a negative relationship between industrialization and energy consumption in 30 China’s provinces and cities in 2008-2017. Energy poverty inhibits the optimization of the energy consumption structure and thus prevents improvement in energy efficiency of the construction industry.

Results show that urbanization has a positive effect on electricity consumption because individuals and organizations in cities need more energy. Rapid growth of the urban population creates huge demand for the construction of housing and commercial real estate, roads, bridges and other infrastructure, thus increasing energy use. These results are consistent with the findings of Liu (2009), and Shahbaz and Lean (2012) who posited that urbanization significantly increased energy consumption. Empirical evidence shows that production methods and processes can cause urbanization and hence growing energy consumption as they become more and more energy intensive (Lefatsa, 2021). This, however, contradicts the conclusions made by Sadorsky (2014) who found that energy consumption may be reduced by urbanization, albeit not significantly, in a small sample of 18 developing economies, including South Africa. Furthermore, Sekantsi and Motlokoa (2015) detected no significant correlation between these variables in Lesotho over the period 1972-2011. The growing demand for energy threatens the energy supply, leading to a peak power deficit being met through exports.

Based on the test results, the study concludes that there is a bi-directional Granger causality between economic growth and energy consumption. These findings concur with Chen, Saud, Bano, and Haseeb (2022) and Iqbal, Tang and Rasool (2022) who made a research into the BRICS economies in 1990-2019 and 2000-2018 and discovered a bidirectional causality effect running between economic growth and energy consumption. They maintain that further openness to international markets contributes to the use of advanced energy-efficient technologies and production methods. At the same time, a research by Aydin (2023) and Cheriyambadan et al. (2023) into the economy of Asian countries in 1980-2018 and 1975-2021 detected no causality effect between economic growth and energy consumption. The neutrality hypothesis explains this by the uneven and insufficient exploitation of renewable energy sources.

The results show a unidirectional causality running from financial development to energy consumption, i.e. financial development causes greater energy consumption. Adebayo and Ağa (2022) in MINT countries from 1990Q1-2019Q4 and Zhang (2023) in China between 2010-2020 obtained similar results, discovering a unidirectional causality running from financial development to energy consumption. It signals the need to address socioeconomic problems and energy deficits to create sustainable environment and encourage new economic activity.

The results presented in the table do not indicate any causality between the variables in question. These findings concur with the results obtained by Shahbaz and Lean (2012) in the United Arab Emirates over the period 1975-2011 and by Zhang and Broadstock (2016) in China between 1963 and 2010, who both failed to find a causality effect between industrialization and energy consumption even though energy is the cornerstone of the modern industrial economy: affordable and reliable energy as a co-requisite for improved industrial productivity and competitiveness is a crucial element in the process of economic diversification. Their conclusions contradict the findings of Gungor and Simon (2017) who discovered a bidirectional causality effect between industrialization and energy consumption in South Africa during the period 1970-2014. In a similar way, using data from Bangladesh over the period 1975-2018, Rahman and Alam (2021) detected bidirectional causality between industrialization and energy consumption. The efficient availability of energy increases industrial productivity.

The results in Table 10 point to a unidirectional causality that is running from urbanization to energy consumption, i.e. urbanization is causing higher energy consumption. They are in line with Khan, Amin and Rahman (2018) who found a unidirectional causality running from urbanization to energy consumption in Bangladesh in 1980-2015. Also, Wang, Li, and Tarasov (2021) obtained similar results for China between 2000 and 2017. As cities grow and more people move into them, the demand for energy increases.

The results from the Bayesian analysis were subjected to a number of tests to see if the model was correct.

Figure 6 shows a diagnostic graph that contains a trace plot, a histogram and density plots for the MCMC sample, and a correlogram. The trace plot has a stationary pattern, which shows that the model was correct. The histogram does not have any unusual features, such as multiple modes. The kernel density plots for the full sample, the first half of the chain, and the last half of the chain all look similar and do not show any strange features such as different densities for the first and last half of the chain.

Figure 7 shows a diagnostic graph that contains a trace plot, a histogram and density plots for the MCMC sample, and a correlogram. The trace plot has a stationary pattern, which shows that the model was correct. The histogram does not have any unusual features such as multiple modes. The kernel density plots for the full sample, the first half of the chain, and the last half of the chain all look similar and do not show any strange features such as different densities for the first and last half of the chain.

Figure 8 shows a diagnostic graph that contains a trace plot, a histogram and density plots for the MCMC sample, and a correlogram. The trace plot has a stationary pattern, which shows that the model was correct. The histogram does not have any unusual features such as multiple modes. The kernel density plots for the full sample, the first half of the chain, and the last half of the chain all look similar and do not show any strange features such as different densities for the first and last half of the chain.

Figure 9 shows a diagnostic graph that contains a trace plot, a histogram and density plots for the MCMC sample, and a correlogram. The trace plot has a stationary pattern, which shows that the model was correct. The histogram does not have any unusual features such as multiple modes. The kernel density plots for the full sample, the first half of the chain, and the last half of the chain all look similar and do not show any strange features such as different densities for the first and last half of the chain.

Figure 6. 

Convergence of Energy Consumption and Economic Growth. Source: Figure constructed by Eviews 9 using statistics from the World Bank (2024)

Figure 7. 

Convergence of Energy Consumption and Financial Development. Source: Figure constructed by Eviews 9 using statistics from the World Bank (2024)

Figure 8. 

Convergence of Energy Consumption and Urbanisation. Source: Figure constructed by Eviews 9 using statistics from the World Bank (2024)

Figure 9. 

Convergence of Energy Consumption and Industrialisation. Source: Authors own computation, Eviews 9 using statistics from the World Bank (2024)

This research underscores the importance of aligning financial and energy policies to support sustainable economic development in the SADC countries. The key recommendations include:

Incentivizing Renewable Energy Investments: Encouraging renewable energy adoption can help reduce dependency on traditional energy sources, contributing to energy security and environmental sustainability.

Expanding Financial Access: Improving credit availability for small-to-medium enterprises and industries can facilitate capital flows toward energy-efficient and innovative technologies.

Strengthening Energy Infrastructure: Upgrading infrastructure to meet rising energy demand from industrial and urban growth is essential for economic resilience.

Implementing these strategies will enable the SADC countries to unlock economic potential, improve quality of life and position the region competitively on a global scale by integrating economic, environmental and social priorities.

This study has several limitations worth noting. First, it relies on data spanning from 1980 to 2023, and inconsistencies in reporting standards and data availability across the SADC countries may impact the accuracy of the findings. Additionally, the key variables like financial development and energy consumption are measured using broad indicators, potentially obscuring sector-specific differences, such as those between industrial and residential energy use. Although Bayesian inference and the PMG model provide robust insights, they may not fully capture non-linear dynamics or structural breaks, especially during significant events like economic crises. Moreover, the study does not account for external shocks, such as global energy price fluctuations or geopolitical factors, which may influence energy consumption and economic growth in the region.

Future research could address these limitations by incorporating disaggregated sectoral data to better understand how financial development impacts different industries. Including environmental variables, such as carbon emissions and renewable energy adoption, would provide a more comprehensive view of sustainable growth. Using non-linear or threshold models could capture complex interactions and reveal structural shifts over time. Comparative studies with other regional blocs, such as ECOWAS or ASEAN, could further contextualize the SADC region’s energy-economy relationship, highlighting unique dynamics and exploring the applicability of similar policy interventions across regions. Finally, examining the effects of specific policy changes or external shocks on these relationships could improve our understanding and inform more targeted policy recommendations for sustainable development in the SADC region.