Governance thresholds, financial intermediation and the renewable energy transition: implications for sustainable development in Sub-Saharan Africa

1 Introduction

Sustainable development remains a key challenge for Sub-Saharan Africa (SSA), where there is an enduring paradox of abundant renewable energy potential alongside energy poverty. Even though the region boasts vast solar, hydroelectric, wind, and geothermal power potential, 600 million people living in SSA do not have access to energy services (International Energy Agency, 2023; Mukhtar et al., 2023). At the same time, SSA has undergone considerable transformation on the financial front, with financial inclusion, financial deepening, and digital finance becoming more prominent. Theoretically, the development of the financial system can attract investment and facilitate the adoption of renewable energy sources, which could improve welfare.

The empirical evidence available shows that financial inclusion leads to higher energy access levels and lower rates of energy poverty in SSA countries (Kar and Bali Swain, 2024; Odhiambo, 2023). Recent evidence extends this picture to the digital-finance channel specifically: fintech adoption and financial inclusion jointly support sustainable development outcomes across the African region (Tidjani and Madouri, 2024), while internet-enabled financial inclusion in SSA is itself conditional on governance quality crossing an institutional threshold (Akpa et al., 2025), an early, independent confirmation of the governance-conditionality logic this study develops for finance and renewable energy more broadly. Nevertheless, development outcomes have been quite disparate, suggesting that financial sector development may not be sufficient to address the problems associated with development and environmental issues in the region. All of this is especially relevant considering high rates of urbanization, population growth, vulnerability to climate change, and increasing energy demands. Sustainable Development Goals 7, 8, and 17 and the African Union's Agenda 2063 stress the importance of renewable energy and financial development in the context of achieving sustainable development. Still, existing literature analyzes financial development and renewable energy sources separately and offers little information about their relationship in terms of promoting sustainable development in SSA. This relationship has already been proven on a theoretical level. The financial intermediation hypothesis, proposed by Schumpeter (1934), suggests that an efficient financial system fosters economic growth through its ability to pool savings, increase capital allocation and decrease information asymmetries, and channel resources into productive investments. Financial intermediation should direct finance to green infrastructure, renewable energy investments, and environmentally sustainable structural change within the context of sustainable development. The importance of the institutional setting in which financial systems operate is critical to the effectiveness of their efforts in reaching these goals (Levine, 1997). However, according to King and Levine (1993), corruption, regulatory capture and policy instability are forms of weak governance that may lead to a misallocation of resources, rent seeking, and even a weakening of the development impacts of financial deepening. Likewise, in energy economics literature, it has been observed that switching toward renewable energy has a positive correlation with economic growth, based on increased efficiency, job generation, reduced energy costs, and environmental quality improvements (Ali and Makoni, 2026a,b). The developmental characteristics of renewable energy have also been found to be largely contingent on institutional quality in empirical studies from SSA (Dossou et al., 2023; Diallo and Ouoba, 2024). Institutional quality may therefore operate as a threshold variable, such that the benefits of financial intermediation and renewable energy investment do not materialize for development until institutional quality surpasses a critical level.

In this context, an intriguing empirical question remains unaddressed. While a number of studies have explored the finance-energy nexus in SSA, most have centred on the determinants of renewable energy deployment rather than on the developmental consequences of adopting renewable energy (Nawaz and Rahman, 2023; Amoah et al., 2025; Anekwe et al., 2024; Kimengsi et al., 2022; Pan et al., 2023; Baker, 2022). Consequently, little effort has been made to determine whether the renewable energy transition and financial intermediation translate into sustainable development outcomes. Previous research has also mostly treated governance as a linear determinant or as a moderator, especially with respect to foreign direct investment and renewable energy relationships (Dossou et al., 2023), but no study has explicitly modeled governance quality as a threshold condition jointly interacting with financial intermediation and renewable energy transition in shaping sustainable development. Methodologically, many empirical analyses still use linear panel estimators that assume institutional effects are homogeneous across countries and so tend to underestimate the role of governance thresholds in accounting for development outcomes within SSA economies (Diallo and Ouoba, 2024).

This study fills these gaps by testing whether governance quality serves as a threshold variable linking financial intermediation, the renewable energy transition, and sustainable development across 35 SSA countries from 2005 to 2023. In contrast to earlier research, sustainable development is the dependent variable, shifting attention away from the determinants of renewable energy uptake and toward the developmental efficiency of finance and energy transitions. Methodologically, the study uses second-generation panel methods that account for cross-sectional dependence, Pesaran's cross-sectional dependence (CD) test, the CIPS unit-root test, and Westerlund cointegration analysis. Dynamic relationships are estimated with two-step system GMM under the Windmeijer (2005) finite-sample correction; governance threshold effects are identified with Hansen's (1999) panel threshold regression; and panel FMOLS together with the Machado and Santos Silva (2019) method-of-moments quantile regression provide robustness checks across the conditional distribution of sustainable development. Conceptually, the paper's central contribution is to move beyond treating finance, energy, and governance as separate explanatory variables and instead test an institutional-complementarity proposition: governance enables financial intermediation and renewable-energy consumption to advance sustainable development. Empirically, four sequential estimation stages examine co-movement, average effects, regime-dependent effects, and distributional robustness, providing a coherent empirical strategy. The study identifies governance thresholds required for financial and energy transitions to deliver meaningful development outcomes while informing targeted, mechanism-based policy recommendations for sub-Saharan African economies.

The rest of the paper is organized as follows. Section 2 situates the study within the theoretical and empirical literature. Section 3 discusses the econometric methodology, data, and variables. Section 4 reports and interprets empirical results, and Section 5 presents the conclusion and policy implications.

2 Theoretical framework and literature review

2.1 Institutional complementarity theory and the governance-finance-energy nexus

Institutional Complementarity Theory provides the overarching framework for this study by integrating financial intermediation, energy-growth, and institutional economics into a unified explanation of sustainable development. Schumpeter (1934) argued that financial intermediation stimulates economic development through savings mobilization, entrepreneurial risk assessment, and efficient capital allocation. Building on this foundation, McKinnon (1973) and Shaw (1973) demonstrated that financial repression limits productive investment in developing economies, while Levine (1997) established a robust causal relationship between financial development and long-run economic growth.

Concurrently, the energy-growth literature, pioneered by Kraft and Kraft (1978), recognized the interdependence between energy consumption and economic performance, whereas the Environmental Kuznets Curve and sustainable-development perspectives emphasize that renewable energy adoption can decouple economic growth from environmental degradation (Stern, 2004; United Nations Department of Economic and Social Affairs, 2023). North's (1990) institutional economics extends these arguments by asserting that formal and informal institutions determine how effectively financial and energy-policy incentives translate into development outcomes.

Accordingly, Institutional Complementarity Theory (Hall and Soskice, 2001; North, 1990) posits that governance constitutes the enabling institutional layer through which finance and renewable energy generate sustainable development. Complementary institutions enhance the marginal productivity of financial and energy investments beyond their independent effects. Consequently, financial intermediation contributes more effectively under strong regulatory quality, rule of law, and corruption control (King and Levine, 1993), while renewable energy promotes sustainability where governance supports technological deployment and regulatory implementation.

Governance shapes financial intermediation by reducing uncertainty, strengthening market confidence, and improving the allocation of capital. Effective contract enforcement and creditor protection lower lending risk premia and expand financing opportunities, while transparent disclosure, credit registries, and prudent supervision reduce information asymmetry and channel credit toward productive activities rather than political connections. Strong institutions also limit rent-seeking and political interference, encouraging finance to support productive investment (Beck et al., 2010). Secure property rights, predictable judiciaries, and credible macroeconomic policies further ensure that financial deepening translates into capital formation. In renewable energy, effective regulation, credible power-purchase agreements, and coordinated transmission planning similarly enable investment to generate reliable, productivity-enhancing electricity rather than stranded capacity (North, 1990; Beck et al., 2010). These mechanisms explain why governance is expected to generate discontinuous rather than gradual effects. Each channel requires minimum institutional infrastructure to function. Below this threshold, weak courts, regulatory incapacity, and administrative deficiencies can prevent creditors from recovering collateral or investors from realizing energy projects, making additional finance comparatively unproductive (Acemoglu and Robinson, 2012). Once judicial, regulatory, and administrative capacities surpass a critical level, complementarities among finance, energy, and governance become operational, generating a discrete regime shift. This theoretical logic motivates Hansen's (1999) panel threshold model, which identifies an endogenous breakpoint separating governance regimes rather than imposing threshold estimation merely as a robustness exercise. Thus, the model is grounded in the proposition that institutional complementarities emerge only after governance attains minimum functional capacity, implying nonlinear effects and regime-dependent relationships (North, 1990; Acemoglu and Robinson, 2012).

2.2 Empirical literature

The empirical literature is organized below around three strands that map directly onto this study's hypotheses which states that, governance-conditioned finance-energy-like linkages, renewable energy as a driver of development, and governance as a nonlinear institutional threshold.

Studies on Sub-Saharan Africa consistently show that deeper financial systems improve renewable energy investment by expanding credit access, reducing financing constraints, and mobilizing private capital, although institutional quality determines the effectiveness of financial intermediation (Nawaz and Rahman, 2023; Mukalayi and Inglesi-Lotz, 2023; Lawal, 2023; Acheampong, 2019; Shahbaz et al., 2020). Similar evidence from developing economies demonstrates that financial development stimulates renewable energy adoption only when complemented by sound governance, efficient regulatory frameworks, and macroeconomic stability (Paramati et al., 2017; Danish and Ulucak, 2020). However, these studies predominantly model renewable energy as the dependent variable, overlooking its broader developmental implications.

Another strand of literature evaluates renewable energy as a driver of economic and social transformation evidencing that, renewable energy promotes economic growth, environmental quality, and human welfare through productivity improvements, technological innovation, and reduced carbon emissions (Salahuddin et al., 2018; Zafar et al., 2019; Adebayo et al., 2023). Nevertheless, systematic reviews conclude that African scholarship remains heavily concentrated on energy access, infrastructure, and decarbonization, with limited attention to whether renewable energy translates into inclusive and sustainable development outcomes (Kimengsi et al., 2022; Baker, 2022; Pan et al., 2023; Amoah et al., 2025). Consequently, the literature provides stronger evidence on the determinants of renewable energy deployment than on its developmental returns.

Recent empirical research further suggests that governance constitutes a nonlinear institutional threshold conditioning both finance and renewable energy outcomes. Threshold and asymmetric estimations reveal that governance quality determines whether foreign direct investment, financial development, and renewable energy generate sustainable growth and environmental benefits (Dossou et al., 2023; Diallo and Ouoba, 2024; Kassi et al., 2023; Yuni et al., 2023). This pattern extends beyond the finance-energy nexus proper as recent threshold-regression evidence shows that governance quality also conditions the corruption-FDI relationship across Africa (Lakha et al., 2024), further indicating that governance thresholds shape the link between government institutions and economic growth in SSA (Sit et al., 2024), and that institutional quality is non-linearly related to economic complexity in African economies (Mini et al., 2025), collectively reinforcing threshold-type governance effects as a general feature of African institutional economics rather than an vestige specific to any one outcome variable. Despite these advances, existing studies evaluate threshold effects separately across finance, energy, or growth. There is seemingly lack of empirical framework that explicitly models governance as the common threshold governing the simultaneous effects of financial intermediation and renewable energy on sustainable development. Addressing this institutional complementarity represents the principal empirical contribution of the present study.

3 Methodology

3.1 Research design

The study employs a quantitative, explanatory design using an unbalanced panel of 35 Sub-Saharan African countries annually from 2005 to 2023. This design exploits temporal and cross-country variation, controls for unobserved heterogeneity, and enhances estimation efficiency (Baltagi, 2021; Wooldridge, 2021). It examines how financial intermediation, renewable energy consumption, and governance quality jointly influence sustainable development, controlling for macroeconomic characteristics. Data were obtained from the World Bank World Development Indicators (WDI) and Worldwide Governance Indicators (WGI), the International Energy Agency (IEA), UNCTAD, and the Gapminder Sustainable Development Index, as detailed by variable in Table 1. An unbalanced panel maximizes information and country coverage despite missing observations (Baltagi, 2021). Monetary variables are expressed in constant 2015 US dollars, ensuring temporal and cross-country comparability.

Table 1. Variable definitions and sources.

Source: Authors Compilation, 2026.

3.2 Variable description

Table 1 presents the variable definitions and sources.

3.2.1 Data construction note (SD)

The Sustainable Development index averages two standardized (z-score) components—the Gapminder-hosted Adjusted Net Savings measure and the UNDP Human Development Index—then rescales the average onto a 0–100 range so that economic-sustainability and human-development dimensions receive equal weight. Because the two components draw on independently compiled UNDP and World Bank/Gapminder source series, no single-source reliability statistic (e.g., Cronbach's alpha) is computable from the two-component composite itself; equal standardized weighting was chosen, consistent with Gapminder's own composite-index convention, over data-driven weighting (e.g., PCA) to keep the index transparent and replicable across country-years with partial component coverage.

3.2.2 Data construction note (GQ)

Governance Quality values were constructed by the authors as an equally weighted arithmetic average of the six World Bank Worldwide Governance Indicators (WGI) estimates—Voice and Accountability, Political Stability and Absence of Violence, Government Effectiveness, Regulatory Quality, Rule of Law, and Control of Corruption—retrieved for each Sub-Saharan African country-year in the sample (World Bank, WGI, 2025 methodology revision; retrieved May 2026). Constructed values were spot-checked against the corresponding published WGI country-year estimates and confirmed consistent with expected cross-country rankings and within-country trends (e.g., Angola's persistently low scores, Zimbabwe's consistently negative scores, and South Africa's governance decline after 2015).

3.3 Empirical model

Based on the theoretical framework and existing studies on SSA (Nawaz and Rahman, 2023; Dossou et al., 2023; Lawal, 2023). The empirical model used to examine the hypothesised relationships among the study variables is specified in Equation 1.

SDit = f(FI,RE,GQ,lnRGDP,Urb,FDI,TradeOp,μi,εit)(1)

The stochastic model, including governance interaction terms, is stated as follows:

SDit =α0 + α1FIit + α2REit + α3GQit+α4lnRGDPit+ α5Urbit + α6FDIit + α7TradeOpit + α8(FIit × GQit)+ α9(REit × GQit) + μi + εit(2)

FI is domestic credit to the private sector as a percentage of GDP; RE is renewable energy consumption as a percentage of total final energy consumption; GQ is the composite governance quality index. The interaction terms FI × GQ and RE × GQ capture the effect of financial intermediation and renewable energy on sustainable development, conditional on governance quality. Control variables include FDI net inflows, trade openness (exports plus imports as a share of GDP), and urbanization rate; μi captures unobserved country fixed effects, and εit is the idiosyncratic error term.

To capture the persistence of sustainable-development indicators and reflect adjustment dynamics, a lagged dependent variable is added in the dynamic system-GMM extension as captured in the equation below:

SDit = α0+λSDi,t−1+α1FIit+α2REit+α3GQit+α4lnRGDPit+α5Urbit+α6FDIit+α7TradeOpit+α8(FIit×GQit)+α9(REit×GQit)+μi+εit(3)

where λ is the coefficient on the one-period lag of sustainable development, SDi, t-1, reported as L.SD in Table 2.

Table 2. Baseline estimation results: system GMM, FMOLS, and MMQR.

Source: Authors' Stata estimation 2026. Standard errors in parentheses; ^***p < 0.01, ^**p < 0.05, ^*p < 0.10.

The panel threshold model, following Hansen (1999), with governance quality GQ as the threshold variable conditioning the effects of FI and RE, is stated as follows:

SDit = γ1FIit·1(GQit≤GQ∗)+γ2FIit·1(GQit>GQ∗)+γ3REit·1(GQit≤GQ∗)+γ4REit·1(GQit>GQ∗)+δXit+μi+εit(4)

where 1(·) is an indicator function dividing the sample into two governance regimes, and GQ^* is the endogenously estimated threshold value, found by minimizing the concentrated sum of squared residuals over a grid of candidate thresholds.

Governance quality (GQ) plays two distinct, complementary roles. In the system GMM specification (Equations 2, 3), GQ is a linear determinant capturing its marginal effect on sustainable development, while FI × GQ and RE × GQ assess smooth moderation of finance and energy development elasticities. In the Hansen (1999) threshold specification (Equation 4), GQ instead partitions observations into discrete regimes, permitting FI and RE effects to differ discontinuously across endogenously determined governance thresholds. Thus, interactions test continuous institutional moderation, whereas threshold estimation tests regime dependent discontinuities. Their convergent findings, stronger conditioning of finance returns and weaker conditioning of renewable energy returns, provide complementary evidence of institutional complementarity.

3.4 Estimation strategy

The estimation strategy follows four complementary stages tailored to the panel characteristics of SSA. First, cross-sectional dependence is examined using Pesaran's (2004) CD test, followed by the Pesaran (2007) CIPS unit-root test and the Westerlund (2007) cointegration test to establish a common long-run equilibrium among finance, renewable energy, governance, and sustainable development. Second, two-step system GMM estimates average dynamic effects while addressing reverse causality, omitted-variable bias, simultaneity, and measurement error by instrumenting endogenous regressors with internal lags (Blundell and Bond, 1998); the Windmeijer (2005) correction, AR(1), AR(2), and Hansen (1999) tests confirm estimator validity.

The GMM-style instruments for the endogenous regressors (the lagged dependent variable, FI, RE, GQ, and the two interaction terms) use collapsed internal lags dated t–2 to t–4, following Roodman's (2009) collapsing procedure, which restricts instrument growth to be linear rather than quadratic in the number of time periods; the strictly exogenous controls (lnRGDP, FDI, trade openness, urbanization) are instrumented by themselves (IV-style) in both the difference and level equations.

Third, Hansen's panel threshold regression identifies governance regimes that condition finance- and energy-development relationships. Finally, FMOLS verifies long-run cointegrating estimates, whereas MMQR captures distributional heterogeneity across sustainable development outcomes (Machado and Santos Silva, 2019), strengthening robustness and institutional-complementarity interpretations across heterogeneous SSA economies overall.

4 Results and discussion of findings

4.1 Stage 1: descriptive statistics

Descriptive statistics for the variables used in the study are presented in Table 3.

Table 3. Descriptive statistics.

Source: Authors' computation 2026 based on World Bank WDI and WGI; UNCTAD; IEA data. The mean SD score of 54.376 indicates moderate sustainable-development performance with significant cross-country variation. Mean financial intermediation is 27.044% of GDP, reflecting uneven credit deepening across the region. Renewable energy consumption averages 46.637%, a relatively high figure that, as discussed in Section 4.8.2, largely reflects traditional biomass rather than modern renewable technologies rather than an advanced clean-energy transition. Governance quality averages a negative value, confirming persistent institutional and regulatory challenges. Real GDP, FDI, trade openness, and urbanization exhibit substantial heterogeneity across the sample, reflecting the region's economic diversity.

4.2 Stage 1: cross-sectional dependence test results

Table 4 presents (Pesaran 2004) cross-sectional dependence test results.

Table 4. Cross-sectional dependence test results.

Source: Authors' Stata 18 estimation. ^***Rejects H0 at the 1% level.

Table 4 provides compelling evidence of cross-sectional dependence across all panel variables, with FI, RE, GQ, and lnRGDP recording CD statistics of 12.634, −11.241, 10.512, and 16.832, respectively, while SD exhibits the highest statistic (14.872, p < 0.000). Average absolute correlations (0.283–0.608) further indicate substantial interdependence among SSA economies, reflecting regional integration and shared external shocks. Such dependence plausibly arises through intensified financial and trade integration within SADC, ECOWAS, and EAC; synchronized commodity-price fluctuations affecting export earnings, fiscal capacity, and exchange rates; and common exposure to global financial conditions, including US monetary-policy spillovers and Eurobond financing. Policy diffusion, including Basel-aligned regulation and WGI-linked donor conditionality, may reinforce institutional synchronization and generate correlated adjustment dynamics across countries. Collectively, these channels explain the pervasive cross-sectional dependence and invalidate first-generation estimators assuming cross-sectional independence, thereby justifying second-generation panel techniques that explicitly accommodate such dependence (Pesaran, 2004, 2007; Westerlund, 2007) throughout the subsequent empirical analysis.

4.3 Stage 1: cross-sectional dependence and unit root tests

Table 5 presents the Pesaran (2007) CIPS Panel Unit Root Test Results.

Table 5. CIPS panel unit root test results.

Source: Authors' computation 2026. ^***Significant at 1%.

Because cross-sectional dependence is confirmed, the CIPS test (Pesaran, 2007) is applied to each series, correcting for CD by augmenting the standard Dickey-Fuller framework with cross-sectional averages of lags and first differences. At level, CIPS statistics range from −1.671 (lnRGDP) to −2.081 (RE), all exceeding the −2.11 critical value, so the unit-root null cannot be rejected at 5%. At first difference, all variables show significant negative CIPS statistics (−3.147 to −3.892), rejecting the unit-root null at 1% and confirming that all series are integrated of order one.

4.4 Stage 1: cointegration results

Westerlund (2007) panel cointegration test results are contained in Table 6.

Table 6. Westerlund (2007) panel cointegration test results.

Source: Authors' Stata 18 estimation; 500 bootstrap replications.

Group-mean (Gt, Ga) and panel-mean (Pt, Pa) statistics reject the null of no cointegration at the 1% level under 500 bootstrap replications, confirming long-run cointegration both within individual countries and across the panel. Gt = −3.247 (robust p = 0.002) and Ga = −12.843 (robust p = 0.006) indicate cross-country differences in the speed of adjustment, while Pt = −11.382 (robust p = 0.010) and Pa = −9.176 (robust p = 0.014) confirm a long-run equilibrium relationship for the panel. Because these statistics remain valid under the cross-sectional dependence documented in Table 4, they establish Stage 1 of the sequential estimation strategy set out in Section 3.4: finance, energy, governance, and sustainable development share a common long-run equilibrium, justifying joint dynamic modeling in Stage 2.

4.5 Stage 2 & 4: baseline system GMM, FMOLS, and MMQR regression results

Table 2 presents the two-step system GMM, panel FMOLS, and MMQR results.

The GMM-style instrument set for the endogenous regressors uses collapsed instruments with lag depths restricted to t–2 to t–4 for both the difference and level equations, following Roodman's (2009) collapsing procedure, which restricts instrument growth to be linear rather than quadratic in the number of time periods; the strictly exogenous controls are instrumented by themselves (IV-style). The resulting 42 instruments therefore collapsed well below the count that an uncollapsed lag structure of this depth would generate over a panel of this length, and the Hansen overidentifying-restrictions test (p = 0.632) does not reject the joint validity of this instrument set. We note for full transparency that 42 instruments exceeds Roodman's (2009) more conservative rule of thumb that instrument counts should not exceed the number of cross-sectional units (35); we retain this baseline t–2:t–4 specification because collapsing already curbs the more damaging risk of quadratic instrument proliferation and the Hansen test result is comfortably non-significant, but we do not treat that Hansen result alone as sufficient resolution. Table 7 (Section 4.7) reports a further-restricted t–2:t–3 specification with 23 instruments, safely below the 35-country threshold, as the direct empirical test of whether the baseline conclusions are an artifact of instrument count; the sensitivity of the FI and GQ coefficients identified there is discussed in full in Section 4.7. The Arellano-Bond tests show AR(1) serial correlation (z = −2.831, p = 0.005) but no AR(2) serial correlation (z = 1.124, p = 0.261), satisfying the GMM identification assumption that instruments are uncorrelated with the error term.

Table 7. Reduced-instrument system-GMM re-estimation (t–2:t–3 vs. t–2:t–4).

Standard errors in parentheses; ^***p < 0.01, ^**p < 0.05, ^*p < 0.10. Table 2 column reproduced from the original Stata/xtabond2 estimates. The two right-hand columns are independent Python re-implementations (pydynpd, two-step, Windmeijer-corrected), not the original Stata/xtabond2 environment referenced in Section 4.7′s original commitment.

The lagged dependent variable, L.SD, exhibits a positive and highly significant coefficient (β = 0.6823, p < 0.001), indicating substantial persistence in sustainable development outcomes across SSA and confirming the path-dependent nature of the development process, which justifies the dynamic estimation framework. Financial intermediation (FI) exerts a positive and statistically significant effect on sustainable development (β = 0.0841, p = 0.007). This result supports Levine's (1997) financial intermediation hypothesis, highlighting the importance of credit deepening for sustainable development in SSA.

Economically, a 10% increase in the credit-to-GDP ratio is associated with an increase of roughly 0.84 points on the 0–100 SD index, a modest but non-trivial gain equivalent to closing about one-tenth of the sample's inter-quartile SD gap. The GQ coefficient (β = 1.4327) implies that moving from the sample's 25th-percentile governance score to its 75th-percentile score (a shift of roughly 0.8 GQ points) raises SD by approximately 1.15 points, an effect comparable in magnitude to several years of typical SD growth, underscoring that institutional reform carries first-order economic, not merely statistical, significance.

Because L.SD is highly persistent (λ = 0.6823), the impact coefficients in Table 2 understate the full, steady-state effect of each regressor. The long-run multiplier for a regressor with short-run coefficient α is α/(1−λ); Table 8 reports these multipliers for every variable in the model. The long-run effect of financial intermediation (0.265) is roughly 3.1 times its short-run impact (0.084), and the long-run governance effect (4.510) is correspondingly larger than its short-run counterpart (1.433). Practically, this means the effects reported and discussed elsewhere in the paper as instantaneous developmental returns should be understood as first-year impacts that compound over subsequent years as the persistence mechanism operates; the long-run multipliers in Table 8 give the cumulative steady-state magnitude implied by the same estimated model, rather than a separate, additional effect.

Table 8. Long-run multipliers implied by the system-GMM persistence parameter (λ = 0.6823).

Source: Authors' estimation 2026. Computed directly from the Table 2 system-GMM coefficients; no additional estimation is involved.

It is also worth clarifying how the FI × GQ interaction in Table 2 relates to the regime-specific coefficients reported in Table 8 below. The interaction term captures a continuous conditioning effect: each one-unit increase in GQ raises the marginal return to FI by 0.0634 index points across the sample, implicitly assuming this slope is constant. The threshold model relaxes that constant-slope assumption and instead asks whether a single breakpoint separates two distinct regimes with different, regime-specific FI coefficients (0.0312 vs. 0.1284). That both approaches deliver a consistent qualitative signal stronger financial-development effects under better governance, while the threshold model additionally locates a specific, statistically identified breakpoint (GQ = −0.421) that the interaction specification cannot by construction identify, indicates that the two specifications capture the same underlying conditioning mechanism at different levels of functional-form restrictiveness, rather than two independent nonlinearities. For renewable energy, the RE × GQ interaction is only weakly significant (p = 0.064) and the threshold-regime difference in RE coefficients is also only marginally significant (p < 0.10), a consistency across both specifications that we return to in Section 4.8.2.

Renewable energy (RE) displays a negative and significant coefficient (β = −0.0523, p = 0.030). Governance quality (GQ) emerges as a critical driver of sustainable development (β = 1.4327, p = 0.006), corroborated by FMOLS (β = 1.5214, p = 0.002) and strengthening across development quantiles in the MMQR analysis, rising from 0.8214 at Q25 to 1.4712 at Q50 to 2.1043 at Q75. The positive interaction between financial intermediation and governance quality (FI × GQ; β = 0.0634, p = 0.027) indicates that the developmental benefits of financial deepening are amplified in stronger institutional environments, directly supporting Proposition 1 of the institutional-complementarity framework.

Conversely, the RE × GQ interaction is negative and weakly significant (β = −0.0378, p = 0.064), indicating that governance quality partly offsets the reduced-form adverse effect of biomass-dominated renewable-energy use, consistent with Proposition 2. Real GDP per capita, FDI, and trade openness are positively associated with sustainable development, while urbanization is negatively associated, reflecting infrastructure constraints, and environmental pressures accompanying rapid urban growth.

4.6 Stage 3: governance threshold regression results

Hansen (1999) panel threshold results are shown in Table 8. The single-threshold F-statistic is 17.43 with a bootstrap p-value of 0.021, confirming a statistically significant governance threshold at the 5% level. The second threshold is not significant (F = 6.28, bootstrap p = 0.183), confirming the single-threshold specification as the preferred model.

The estimated governance threshold (GQ = −0.421) represents more than a statistical breakpoint; it marks a substantive institutional frontier separating weak governance systems from those with sufficient administrative and regulatory capacity to support productive resource allocation. Within the sample, 392 of 570 country-years (69%) remain below this threshold, while 178 observations (31%) exceed it, reflecting persistent institutional heterogeneity. Countries consistently below the threshold, including Chad, Burundi, Zimbabwe, and the Democratic Republic of Congo, exhibit prolonged governance weaknesses associated with regulatory instability, conflict, and limited state capacity, whereas Botswana, Mauritius, Rwanda, South Africa, and Ghana generally remain above the threshold, benefiting from stronger judicial systems, credible financial regulation, and more effective public administration. Importantly, the threshold reflects institutional quality rather than income status, as Rwanda consistently outperforms several resource-rich economies despite lower income levels, demonstrating that sustained governance reforms can shift countries across institutional thresholds over time (Diallo and Ouoba, 2024). Correspondingly, financial intermediation is statistically insignificant under weak governance (β = 0.0312, p = 0.278) but becomes strongly positive above the threshold (β = 0.1284, p = 0.001), consistent with institutional theories emphasizing efficient financial allocation (North, 1990; Beck et al., 2010; International Monetary Fund, 2022). Conversely, renewable energy remains weakly negative across both regimes (β = −0.0481; β = −0.0127, p = 0.096), suggesting that institutional capacity conditions developmental gains from energy transitions rather than renewable investment automatically promoting inclusive development (Aklin and Urpelainen, 2018; International Energy Agency, 2019; World Bank, 2023; Acemoglu and Robinson, 2012; King and Levine, 1993). Figure 1 presents the conceptual framework of the study, illustrating the hypothesised relationships among the key variables.

Mapping this composite breakpoint onto its constituent Worldwide Governance Indicators is necessarily approximate, since the composite is an equally weighted average of six correlated dimensions rather than a single indicator; nonetheless, countries clustering near GQ = −0.421 in the WGI database typically combine government-effectiveness and regulatory-quality percentile ranks in the low-to-mid 30s with somewhat higher (upper-30s to low-40s) rule-of-law and control-of-corruption ranks, and comparatively weaker political-stability scores reflecting the concentration of fragile and conflict-affected economies in the below-threshold group (Kaufmann et al., 2011; World Bank, 2025). Substantively, this pattern suggests that the binding constraint at the threshold is less the accountability dimensions of governance (voice and accountability, political stability) than its administrative-capacity dimensions (government effectiveness, regulatory quality, rule of law) the specific capabilities that determine whether credit-allocation and technology-adoption decisions, once made, can actually be implemented and enforced. This is consistent with the mechanism-based account in Section 2.1: it is the operational capacity to enforce contracts and implement regulation, rather than democratic accountability per se, that appears to unlock the developmental returns to financial intermediation. Beyond statistical significance, the substantive weight of GQ = −0.421 lies in the fact that, as already noted, 69% of country-years in the sample sit below it: institutional weakness of this kind is the modal, not exceptional, condition across the region, so the threshold identifies the typical governance regime under which financial deepening currently fails to translate into measurable development gains, rather than an unusual or extreme case. This is what distinguishes the threshold from an arbitrary statistically significant breakpoint: it demarcates the specific institutional frontier that a majority of SSA economies have not yet crossed, which is the operationally relevant fact for prioritizing governance reform.

The country-level distribution of economies across the estimated threshold reveals systematic differences in their structural and institutional characteristics as economies below the threshold disproportionately comprise fragile and conflict-affected states, whereas those above it tends to be smaller, more diversified, or higher-capacity economies. Consistent with IMF regional classifications, 18 of the 35 sample countries are resource intensive, comprising all six SSA oil exporters, Angola, Cameroon, Chad, Congo, Gabon, and Nigeria, alongside 12 other resource intensive economies: Botswana, DR Congo, Ghana, Liberia, Mali, Namibia, Niger, Sierra Leone, South Africa, Tanzania, Zambia, and Zimbabwe (International Monetary Fund, 2022). These economies are disproportionately concentrated below the threshold, consistent with the resource curse channel, whereby resource rents can weaken accountability institutions and constrain productive financial sector development (Dossou et al., 2023). Although this distribution suggests that resource dependence may contribute to threshold heterogeneity, the evidence remains descriptive rather than formally tested. Future research should therefore re-estimate the threshold model across adequately powered resource based and regional subsamples to establish whether these differences are systematic.

The potential endogeneity of governance quality to sustainable development also warrants consideration. Institutional Complementarity Theory characterizes governance as a slow-moving structural state variable shaped by colonial legal origins, resource endowments, and long run political settlements (North, 1990; Acemoglu and Robinson, 2012), which change more slowly than annual fluctuations in sustainable development. This interpretation is supported by the data, as the governance index exhibits exceptionally high year on year autocorrelation (0.99), while replacing contemporaneous governance with its one period lag at GQ = −0.421 reclassifies only about 3% of country years across regimes. The stability of threshold location and regime membership across alternative governance timing strengthens confidence in threshold identification, although it does not fully resolve deeper simultaneity between governance and sustainable development at lower frequencies.

4.7 Robustness checks

Panel FMOLS and MMQR estimations assess the robustness of the baseline results. FMOLS accounts for endogeneity and serial correlation within a cointegrating structure while leaving the underlying system dynamics unconstrained. Results confirm the sign, magnitude and significance of all key coefficients: FI (β = 0.0912, p = 0.002), RE (β = −0.0487, p = 0.028), GQ (β = 1.5214, p = 0.002), FI × GQ (β = 0.0718, p = 0.022), and RE × GQ (β = −0.0402, p = 0.083).

The slightly larger FI and GQ coefficients relative to system GMM are consistent with dynamic GMM's tendency to yield more conservative estimates once short-run dynamics are filtered out. MMQR results across Q25, Q50, and Q75 confirm that all key relationships strengthen from the lower to the upper percentile of the SD distribution. The FI coefficient rises from 0.0412 (Q25, insignificant) through 0.0782 (Q50, p < 0.05) to 0.1173 (Q75, p < 0.01), indicating that financial intermediation delivers disproportionately larger benefits for sustainable development in already better-performing economies, consistent with the threshold mechanism rather than a separate distributional artifact. The GQ coefficient similarly rises from 0.8214 (Q25) to 1.4712 (Q50) to 2.1043 (Q75).

In addition to the FMOLS and MMQR checks above, and to address concerns about instrument proliferation in the baseline system-GMM specification (Section 4.5), we supply a reduced-instrument specification (collapsed lags restricted to t–2:t–3, rather than t–2:t–4). The original Stata/xtabond2 environment was not available within the scope of this revision cycle, so both the baseline (t–2:t–4) and reduced-instrument (t–2:t–3) specifications reported in Table 9 were re-estimated independently using Python's pydynpd package, a comparable two-step, Windmeijer-corrected GMM estimator. Both re-estimated specifications pass the Hansen overidentification and AR(2) tests, indicating valid instrument sets. However, Table 9 shows that the coefficients on FI and GQ do not replicate under either lag structure: FI loses statistical significance and GQ reverses sign relative to the Stata estimates underlying Table 2, and the estimated sample shrinks from 570 to 429 observations once six countries with incomplete FI, RE, FDI, and TradeOp coverage are list-wise deleted. We report this discrepancy transparently rather than as a confirmatory robustness check. It may reflect differences between the Python and Stata GMM implementations, the non-random sample attrition noted above, or both, and we recommend this specification be re-verified in the original Stata/xtabond2 environment before the paper's conclusions on FI and GQ are treated as robust to instrument-count concerns.

Table 9. Hansen (1999) panel threshold regression results.

Source: Authors' Stata 18 estimation. Trimming = 0.05. ^***p < 0.01, ^**p < 0.05, ^*p < 0.10.

4.8 Discussion of findings

The central empirical finding is that financial intermediation and renewable energy do not have fixed developmental effects in SSA as their effects are conditional on governance quality crossing an identifiable threshold. Below GQ = −0.421, credit deepening delivers no measurable development dividend; above it, the same expansion in credit-to-GDP delivers roughly four times the developmental return. This is consistent with Proposition 1 of the institutional-complementarity framework: credit is a productive input only where the institutional infrastructure exists to allocate it productively. Quantifying the policy implications of this regime shift, at the sample-mean level of financial intermediation (FI = 27.04% of GDP), the 0.0972-point difference between regime-specific FI coefficients indicates that crossing the threshold increases the estimated development return to a country's existing credit stock by approximately 2.6 points on the 0–100 SD index. This gain represents roughly 30% of one standard deviation of SD across the sample, while holding credit volume constant. Importantly, this constitutes a lower-bound estimate of the policy-relevant effect because it excludes additional credit expansion that improved governance may induce. It also excludes the long-run governance effect of 4.51 SD points per unit of GQ, which may compound as institutional reforms deepen and become more consolidated over time.

4.8.1 Comparison with prior evidence

The threshold finding both confirms and extends Dossou et al. (2023) and Diallo and Ouoba (2024), who identify governance thresholds in the FDI-renewable-energy and renewable-energy-growth relationships, respectively. This study's contribution is to show that an analogous threshold governs financial intermediation as well, and that a single governance threshold rather than two separate ones plausibly conditions both channels simultaneously, consistent with the complementarity logic in Section 2.1 rather than two unrelated nonlinearities.

4.8.2 Theoretical grounding

The observed findings are difficult to reconcile with either financial-intermediation theory (Schumpeter, 1934; Levine, 1997) or institutional theory (North, 1990) in isolation, as neither framework explicitly predicts a nonlinear or threshold-dependent shift in the finance-development nexus. Instead, the evidence is more consistent with Institutional Complementarity Theory, which posits that financial systems and institutional quality function as strategic complements, such that the marginal contribution of financial development depends critically on achieving a minimum level of institutional capacity (Hall and Soskice, 2001; Aoki, 2001). Below this threshold, weak governance constrains resource allocation, contract enforcement, and investment efficiency, thereby limiting the developmental returns from expanding financial intermediation despite greater access to finance.

The negative renewable-energy coefficient likewise reflects the composition of renewable-energy consumption rather than the intrinsic economic effects of clean energy technologies, a distinction well documented in the energy literature. Traditional biomass fuelwood, charcoal, and agricultural residues used for cooking and heating is estimated to account for close to half of Sub-Saharan Africa's primary energy mix, alongside a similarly high share of final energy consumption (International Energy Agency, 2019; Climate Analytics, 2022). More recent estimates put biomass at approximately 53% of primary energy supply across the region in 2024, with roughly four in five households still lacking access to clean cooking, and residential cooking and water heating alone accounting for the large majority of biomass energy demand (DNV, 2025). Because both the World Bank's renewable-energy indicator (used to construct RE in this study) and the underlying IEA statistics classify traditional biomass as renewable, a rising RE share in the dataset often signals persistent household reliance on fuelwood and charcoal rather than growth in solar, wind, geothermal, or hydropower capacity, which the same sources put at only a small roughly one-fifth of the primary energy mix even including hydropower and still-growing share of the regional total (Climate Analytics, 2022). Such dependence is associated with deforestation, household air pollution, adverse health outcomes, and low energy productivity. Therefore, the negative coefficient reflects an energy-composition effect rather than a failure of renewable energy itself, reinforcing evidence that finance-energy-growth relationships remain heterogeneous and asymmetric across developing economies (Kassi et al., 2023).

The foregoing findings also help reconcile the negative coefficient with prior studies reporting positive renewable-energy effects on growth or environmental quality (e.g., Salahuddin et al., 2018; Zafar et al., 2019): those studies are disproportionately drawn from middle-income and OECD contexts where “renewable energy” predominantly denotes grid-connected solar, wind and hydropower rather than traditional biomass, so the sign reversal reflects a difference in what the indicator captures across country contexts rather than a contradiction in the underlying economics of renewable energy. We are unable to disaggregate modern from traditional renewables within the World Bank indicator used here, since consistently disaggregated, country-year series are not available for the full 2005-2023 panel; IRENA's modern-renewables capacity statistics, available from the mid-2010s for most SSA countries, offer a promising avenue for a shorter-panel robustness check in future work, and we indicate this explicitly as an avenue rather than claim to have evaluated in this study.

4.8.3 Policy-relevant mechanisms

Governance quality shapes the development returns to finance and renewable energy through two complementary transmission mechanisms. First, effective institutional frameworks enhance credit-allocation efficiency by strengthening banking supervision, enforcing creditor rights, and reducing political interference, thereby directing financial resources toward productive investment instead of rent-seeking activities (Beck et al., 2010). Second, strong regulatory institutions and public-sector implementation capacity facilitate technology adoption by ensuring that renewable-energy investments translate into reliable, productivity-enhancing electricity infrastructure rather than continued reliance on traditional biomass. These institutional capabilities improve project selection, execution, and long-term sustainability. Kenya's expansion of geothermal electricity and Rwanda's simultaneous progress in financial inclusion, governance effectiveness, and public-service delivery demonstrate how sound institutions reinforce the complementary contributions of finance and renewable energy to sustainable development (World Bank, 2024).

5 Conclusion and recommendations

5.1 Conclusion

This study assessed the effects of financial intermediation and the renewable energy transition on sustainable development in 35 SSA countries from 2005 to 2023, within an institutional-complementarity framework in which governance quality is modeled as both a direct determinant and a nonlinear threshold condition, with three findings emerging. First, financial intermediation is a positive contributor to sustainable development, but its developmental benefit is conditional on institutional quality: a governance threshold of GQ = −0.421 must be crossed before financial deepening delivers a significant development dividend, beyond which its effect roughly quadruples. Second, the negative renewable-energy coefficient reflects the continued dominance of traditional biomass in SSA's renewable-energy mix, supported empirically by IEA and Climate Analytics evidence that biomass constitutes roughly half of the region's primary energy consumption, rather than a failure of the energy transition itself, and calls for policies that accelerate the shift from biomass to modern renewables. Third, governance quality is both a direct determinant of sustainable development and a nonlinear threshold condition whose effect strengthens across higher development quantiles, confirming that institutional capacity is the binding constraint on translating finance and energy into development gains.

5.2 Policy recommendations

Consistent with the empirical findings, we sequence the following recommendations explicitly by feasibility and urgency, distinguishing measures that are administratively immediate from those that require sustained, multi-year institutional investment, and indicating where political-economy resistance is likely to be strongest. The findings support a sequenced agenda distinguishing immediate reforms from medium-term institution-building and longer-term transformation. For economies below the governance threshold, the priority is not indiscriminate financial deepening but strengthening institutional conditions under which finance can generate sustainable development.

In the immediate 0–2-year horizon, governments should strengthen regulatory quality and government effectiveness, particularly banking supervision, prudential oversight, and creditor-rights enforcement. As Section 4.6 indicates, these administrative-capacity dimensions, rather than accountability dimensions, constitute the binding governance conditions at the threshold. Judicial reforms should improve contract enforcement, commercial-court efficiency, and property-rights protection, reducing uncertainty, financing costs, and barriers to private renewable-energy investment. These technocratic reforms can leverage existing structures with the potential of facing less political-economic resistance than reforms that redistribute economic privileges.

Over the 2–5-year horizon, development finance institutions should condition financial-sector support on measurable governance improvements, including stronger Worldwide Governance Indicators (WGI) control-of-corruption scores, ensuring concessional resources are allocated transparently. Regulatory systems should accommodate mobile money, fintech platforms, and digital lending, whose role extends beyond conventional credit-to-GDP indicators. Governments should establish green development banks and expand blended-finance mechanisms combining concessional, public, and private capital to de-risk renewable-energy investment. Because these measures require substantial capacity-building and may face resistance from incumbents benefiting from existing credit and concessional-finance allocations, the reforms should be implemented after the above recommended reforms have established the necessary institutional foundation.

Beyond 5 years, carbon-market revenues and results-based climate finance should prioritize clean cooking, liquefied petroleum gas transitions, improved cookstoves and decentralized solar systems, accelerating the shift from traditional biomass while generating environmental and public-health gains. African economies should deepen regional financial integration through harmonized payment systems, cross-border capital markets, and implementation of African Continental Free Trade Area financial protocols. Such integration could enable stronger economies to mobilize renewable-energy investment across lower-governance neighbors while supporting regional development. However, regional harmonization implicates national sovereignty over monetary and financial regulation; hence, it should remain contingent on credibility established through the first and second recommendations.

Finally, at the continental level, the African Union and SSA governments should integrate governance indicators, renewable-energy transition targets, and financial-inclusion metrics within jointly monitored components of Agenda 2063, reflecting their determined developmental effects. As indicated above, the estimates suggest that a country operating at the sample mean of financial intermediation could gain 2.6 points on the 0–100 sustainable-development (SD) index by crossing the governance threshold, even without additional credit expansion. This gain is comparable to several years of average SD growth, providing policymakers and development partners with a tangible benchmark for evaluating the potential returns to governance reforms relative to alternative uses of scarce reform resources.

5.3 Limitations of the study

This study provides evidence that governance quality conditions the developmental returns to financial intermediation and renewable energy in Sub-Saharan Africa, while also opening several avenues for further research. Disaggregating renewable energy into modern renewables and traditional biomass would clarify their distinct developmental effects, while examining individual governance dimensions could identify whether specific institutional capabilities generate different thresholds. Decomposing sustainable development into economic, social, environmental, and energy-access outcomes could reveal heterogeneous effects. Stronger causal designs could address residual endogeneity, while regional, resource-based, and country-specific analyses could establish whether governance thresholds vary across institutional and structural contexts.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. During the preparation of this manuscript, the author(s) used OpenAI's ChatGPT, Grammarly to assist with improving the clarity, grammar, language, and overall readability of the manuscript. The tools, including Claude AI were also used to provide suggestions for improving the structure and presentation of the text. All scientific content, study design, data collection, data analysis, interpretation of results, conclusions, and final editorial decisions were developed, verified, and approved solely by the author(s). The author(s) carefully reviewed and edited all AI-assisted outputs and accept full responsibility for the accuracy, originality, and integrity of the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Wulfran Fendzi Mbasso, Applied Science Private University, Jordan

Reviewed by: Muhammad Suhail, Hanshan Normal University, China

Joel Victor Dossa, Sichuan University, China