Does financial inclusion spur CO₂ emissions? The marginal effects of financial sustainability

Abstract

Purpose

This study aims to examine the relationship between financial inclusion, CO₂ emissions and financial sustainability across 17 African countries.

Design/methodology/approach

Data were sourced from the World Development Indicators for the period 2004-2021. The study performs the principal component analysis, panel fixed effects model and quantile regression estimations to investigate the relationship between financial inclusion, CO₂ emissions and financial sustainability.

Findings

The study finds that an increase in automated teller machine (ATM) penetration rate, savings and credits increases CO₂ emissions. Findings also reveal that financial sustainability reduces financial inclusion, with significant negative effects on the conditional mean of CO₂ emissions and the conditional distribution of CO₂ emissions across quantiles.

Originality/value

This study is beneficial for policymakers, particularly in the age of digitalization and drive for low-carbon emissions, to develop green credits for energy players and investors to take up renewable and green energy projects characterized by high levels of carbon storage and carbon capture. Further, the banking sector’s credits and liquid assets should be used to finance alternative banking energy-related equipment and services, such as solar photovoltaic wireless ATMs, and fewer bank branches.

Citation

Adeneye, Y., Rasheed, S. and Ooi, S.K. (2023), "Does financial inclusion spur CO₂ emissions? The marginal effects of financial sustainability", IIMBG Journal of Sustainable Business and Innovation, Vol. 1 No. 2, pp. 77-97. https://doi.org/10.1108/IJSBI-02-2023-0004

1. Introduction

Contemporary research on CO₂ emissions reports that financial inclusion involving bank account ownership, automated teller machine (ATM) penetration rate, savings and credits drive the total amount of CO₂ emissions (e.g. Le, Le, & Taghizadeh-Hesary, 2020). However, with the drive for low-carbon emissions, the financial system must play a significant role in reducing CO₂ emissions by ensuring financial sustainability through the redistribution of more bank credits and liquid assets to finance renewable and nonrenewable energy initiatives. This study leverages World Development Indicators (WDIs) annual data from African countries to present new evidence to inform the policy implications of financial sustainability on financial inclusion and CO₂ emissions. Understanding the link between the financial industry and environmental pollution in Africa is crucial (Abid, 2016) since most African countries are characterized by poor energy development, increased use of paper notes and the increase in the number of automated teller machines (ATMs) and bank branches in a way to foster financial inclusion. However, such unsustainable paper notes and the carbon emissions from ATMs, bank branches and light vehicles (i.e. that consume more fossil fuels) that are used to transport paper notes from one bank to the other contribute to the increasing rate of CO₂ emissions in Africa (see Supplementary Table S1). For instance, ATMs emit about 3.2 million tons of CO_2, while bank branches emit 383.1 million tons of CO₂ yearly (McCook, 2014). The environmental impact of an average cash transaction is 5.1g CO₂ equivalent with the energy use of ATMs, and currency transportation contributing about a 64% impact on the environment, while the production of coins contributes about 31% impact on the environment (Hanegraaf, Larçin, Jonker, Mandley, & Miedema, 2020). Also, banknotes weighing 93.4 tons emit approximately a volume of 1.6 million tons of CO₂ equivalents (Wettstein, Lieb, & Lieb, 2000).

Indeed, the activities of the financial sector through financial inclusion can exert both direct and indirect effects on environmental pollution and contribute significantly to the total amount of CO₂ emissions. The overall cash in circulation worldwide across countries continues to grow between 5% and 8% annually (Currency Research, 2017). In Spain, for example, 42% of people said that they prefer using cash (Esselink & Hernández, 2017). Interestingly, the existence of Covid-19 may indirectly contribute to CO₂ emissions. The United Nations (2020) confirmed that cash has been used to fight Covid-19 only in the developing world. While one would expect that CO₂ emissions should decrease during Covid-19 due to the stoppage in production across industrial and manufacturing firms, the circulation of cash as an economic palliative may still give rise to CO₂ emissions. In Sweden, the cost of ATM cards and cash amounts to about 0.4% of gross domestic products, causing negative externalities through CO₂ emissions (Bergman, Guibourg, & Segendorf, 2008). The growth in currency in circulation between 2006 and 2016 is over 100% in many African countries when compared to the Eurozone and the UK, where it is −2.5% (As in Figure 1). Also, the number of individuals who visit bank branches and ATMs with their light vehicles that use fossil fuels contributes to CO₂ emissions in G-20 countries (Erdoğan, Taiwo, Altuntaş, & Victor, 2022). Internal combustion vehicles that bank use for cash transportation operations emit large quantities and amounts of CO₂ equivalent (Bank of Finland, 2020). For instance, banking firms can contribute to indirect pollution by lending money to polluting companies or engaging in projects that cause major damage to the environment (Zhang, Yang, & Bi, 2011).

How, then, can we minimize or mitigate the effects of financial inclusion on CO₂ emissions? The most important way is to ensure the sustainability of the financial industry through financial sustainability, which, in turn, can reduce the negative effects of financial inclusion on CO₂ emissions (In Bouma, Jeucken, & Klinkers, 2017). The availability of more bank liquid assets to finance renewable ATMs is important to reduce the carbon emissions that are emitted through ATMs. Equally, the enablement of green credit policy including more online transactions than cash payments, and the granting of more credits from bank deposits to energy policy actors and players to finance renewable energy, would further reduce the environmental effect of the financial industry (Zhang et al., 2011). By extension, banks’ credit policy meets a green standard where more credits are granted to firms that comply with environmental pollution standards, and that credit granted is made through transfer options rather than through cash payments.

Several studies have examined the role played by financial inclusion in African contexts. For example, Zins and Weill (2016) examined the determinants of financial inclusion in Africa and found that education, income and mobile banking drive formal accounts, savings and credit in 37 African countries, using probit estimations. In another study, the use of mobile telephony promotes the likelihood of savings for financial inclusion in Kenya, Malawi, Uganda and Zambia (Ouma, Odongo, & Were, 2017). Gebrehiwot and Makina (2019), using the GMM dynamic panel in 27 African countries, found that financial inclusion positively affects economic growth, while negatively affecting government borrowing. However, whether financial sustainability mitigates the negative impacts of financial inclusion on CO₂ emissions is understudied.

Against this backdrop, this study investigates the impact of financial inclusion on carbon emissions along with the marginal effect of financial sustainability on the relationship between financial inclusion and CO₂ emissions (finance-environment nexus) in 17 African countries over the period 2004-2021. The significant contribution of this study to the understanding of the finance-environment nexus is as follows. Firstly, concerning CO₂ emissions, most studies in the African setting have been linked to economic growth, financial development, sustainable goals, globalization and renewable energy (e.g. Raheem, Tiwari, & Balsalobre-Lorente, 2020; Wang, Mirza, Vasbieva, Abbas, & Xiong, 2020; Zaidi, Zafar, Shahbaz, & Hou, 2019) with the conclusion that CO₂ emissions retard economic growth, and its effects can be reduced through renewable energy and sustainable developmental goals. These studies fail to investigate the link between financial inclusion and CO₂ emissions in African studies and the policy design concerning financial sustainability to reduce the likely negative effects of financial inclusion on CO₂ emissions. Secondly, existing literature (e.g. Li & Wei, 2021) has focused mainly on investigating the environmental impacts of financial development. Specifically, previous studies have often relied on indicators such as domestic credit to the private sector, as a measure of financial development. To bridge the existing gap, this study develops a holistic financial inclusion index by incorporating diverse indicators of financial inclusion to evaluate the finance-CO₂ emission nexus. Thirdly, since SGD-17 and the Paris Agreement 2015 have acknowledged the significant influence of financial sustainability on emission levels, this study examines the combined impact of financial sustainability and financial inclusion on carbon dioxide emissions in the selected African nations. Prior studies have largely ignored the channels by which the financial sector and promoting financial sustainability can collectively contribute to lowering CO₂ emissions in the African context.

The rest of the paper is organized as follows. Section 2 presents a review of related literature that leads to the research gaps and development of research hypotheses. Section 3 provides the data source and methods, while section 4 presents the results and discussions. Section 5 presents the conclusion and policy implications.

2. Literature review

This section presents the literature that forms the development of our hypotheses, and thus the gap for the study.

H1.

Financial inclusion promotes CO₂ emissions.

The studies on the relationship between financial inclusion and CO₂ emissions are beginning to increase. For instance, using Hoechle's (2007) model for Driscoll-Kraay standard errors, Le et al. (2020) investigated the effect of financial inclusion on CO₂ emissions for 31 Asian countries and showed that financial inclusions appear to promote increased CO₂ emissions. Although financial inclusion is an integral part of financial development (Le et al., 2020; World Bank, 2018), most of the studies, except for Le et al. (2020), have linked financial development with CO₂ emissions. As reported by Abbasi and Riaz (2016), financial development promotes, and does not aid, the reduction of CO₂ emissions in Pakistan by adopting ARDL and augmented VAR models. Charfeddine and Kahia (2019) established that financial development enhances CO₂ emissions in 24 countries in the Middle East and North Africa region, using the PVAR model estimator, impulse response function and variance decomposition. The weak financial development in MENA countries does not contribute to environmental quality by reducing CO₂ emissions. Raheem et al. (2020) reported that financial development has a weak effect on CO₂ emissions in G7 countries, using the pooled mean group estimator. Shoaib, Rafique, Nadeem and Huang (2020) conducted a comparative analysis between developed (G8) and developing countries (D8) and found that financial development has significant positive effects on CO₂ emissions in G8 and D8 countries, using the PMG panel ARDL estimator, which is in support of Wang et al. (2020) that study N-11 countries. Using a structural break test, Toda-Yamamoto causality and cointegration models, Shahbaz, Haouas, Sohag and Ozturk (2020) established that financial development positively affected CO₂ emissions and environmental degradation in the United Arab Emirates between 1975 and 2014. In contrast, Zaidi et al. (2019) established that financial development retards CO₂ emissions both in the long run and short run, using Westerlund cointegration, continuously updated fully modified and continuously updated bias-corrected estimation techniques on sample data of Asia Pacific Economic Cooperation. Similarly, Wang et al. (2020) found that financial development reduces CO₂ emissions in the long run in Turkey. Nasir, Huynh and Tram (2019) confirm, using dynamic ordinary least squares and fully modified OLS on a sample of ASEAN-5 economies, that financial development increases CO₂ emissions in the long run. Khan, Peng and Li (2019) further emphasized that financial development positively impacts CO₂ emissions with high pollution spread, using dynamic GMM and seemingly unrelated regression (SUR) on a panel of 193 countries. Guo, Hu and Yu (2019) established that financial development efficiency increases CO₂ emissions, while the financial development scale reduces CO₂ emissions in China, using the STochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) model of environmental pressure. Also, Tian and Li (2022) confirm that financial inclusion increases carbon emissions, using the CS-ARDL and VECM Granger causality method, positing that the availability of financial services boosts funding for more industrial and manufacturing operations, which raises demands for and consumption of fossil fuels.

Similarly, Zaidi et al. (2019) advocate that financial inclusion increases carbon emissions in OECD and BRICS countries through increased economic activity, and access to credit leads to more carbon-intensive financing activities such as transportation or manufacturing. Hussain, Akbar, Gul, Shahzad and Naifar (2023) found that when individuals and businesses have access to affordable financial resources, it can positively affect carbon emissions in the short run only. However, beyond a certain threshold, the accessibility of credit at a reduced cost of capital increases corporate production capacity and energy-intensive household appliances. Another study by Tsimisaraka et al. (2023) advocates that easier access to credit exhibited a greater propensity to purchase a variety of products, such as automobiles and refrigerators, which has contributed to a greater demand for energy derived from fossil fuels, thereby increasing the carbon dioxide (CO₂) emissions. These findings are supported by Jebli and Hakimi (2023), suggesting that companies tend to invest in non-environmentally friendly projects using financial services. Thus, this contributes to an increase in the consumption of energy derived from fossil fuels. Accordingly, the literature on the link between financial development and CO₂ emissions is vast and varies in a contextual sense, an indication that assessing its integral parts such as financial inclusion may enhance our understanding of CO₂ emissions in a particular context.

The results of these brief studies reveal some differences. The differences are due to the context in which the study is conducted, the type of estimation techniques, the sample size of the study, the period of study, the kind of theories adopted and the nature of the country (whether developing vs developed countries, the level of per capita income and economic growth and the level of financial structure that exists in the country understudied). Interestingly, one major similarity is that these studies focus on financial development emission linkages. In this paper, we further add to the finance-environment nexus, extending the study of Le et al. (2020), by examining the role that financial inclusion plays in promoting CO₂ emissions across countries in Africa.

H2.

Financial sustainability reduces the impact of financial inclusion on CO₂ emissions.

Several studies have posited that the increasing level of financial constraints is one of the major bottlenecks to taking up renewable and non-renewable energy policies. Supporting this claim, Baulch, Do, and Le (2018) submit that lack of access to capital and lack of credits and loanable funds from government and financial institutions have resulted in a serious barrier in the fight against the reduction of CO₂ emissions in Vietnam. Meanwhile, Le et al. (2020) buttressed that the availability of affordable financial services that can promote the reduction of CO₂ emissions by using alternative fossil fuels and gas such as CO₂ capture and CO₂ storage is relevant in the adoption of environmental sustainability programs. Further, Bayram, Talay, and Feridun (2022) argued that sustainable finance can help to assess climate-related financial risks, which can potentially lead to a reduction in emissions by incentivizing firms to invest in sustainable practices. Moreover, Wan, Pu, and Tavera (2023) claim that financial sustainability through digital finance reduces pollution through innovations, sustainable capital allocation effects and structural adjustments. Similarly, Cao (2023) argued that financial sustainability can be boosted through green finance projects. Another study by Kirikkaleli, Adebayo, and Güngör (2021) submit that sustainability can be accelerated by financing renewable energy technologies and research and development through financial development. Most studies that have linked financial inclusion and financial development having negative effects on CO₂ emissions (e.g. Cao, 2023), imply that sustainable finance, hence financial sustainability, is required to enhance and promote the reduction of CO₂ emissions.

While several studies have suggested different financial strategies and policies for reducing CO₂ emissions using consumption-based carbon tax and other public finance strategies including green finance and digital finance (Bouwer & Aerts, 2006; Wan et al., 2023), two questions are yet to be answered. First, how sustainable is the financing of CO₂ emissions reduction? Second, to what extent can the financing sustainability reduce CO₂ emissions for policy formulation and implementation? Both questions are important because countries differ in terms of the capital market, availability of funds and foreign aid and the level of technological advancement. For instance, most capital markets in Africa are undervalued, and the financial system is less developed to finance CO₂ emissions reduction goals. African countries that are also battling with issues of unemployment, incessant domestic war and insecurity have fewer funds to finance CO₂ emissions. The demand for foreign aid and inflows of energy technologies into Africa has been increasing in the recent decade (Udi, Bekun, & Adedoyin, 2020). Furthermore, better financial sustainability such as energy players’ access to domestic credits to finance CO₂ emissions reduction will enable the government to have long-term solutions to the energy problem and the environmental problems of CO₂ emissions. Government political trust may also be required to promote financial sustainability by ensuring manufacturing and industrialized companies pay suitable tax rates.

To our knowledge, fewer empirical studies have investigated the role of financial sustainability in reducing the positive effect of financial inclusion on CO₂ emissions in Africa. However, studies have identified a few indicators of financial sustainability in financing CO₂ emissions reduction such as bank sector development, that is, the ratio of bank credit to bank deposits (Zafar, Zaidi, Sinha, Gedikli, & Hou, 2019), bank credit to the private sector (Nwani & Omoke, 2020) and liquid assets (Monasterolo & Raberto, 2019). We provide how these financing strategies, grouped as financial sustainability indicators, reduce the positive links between financial inclusion and CO₂ emissions.

3. Data and methods

This study utilizes the panel fixed effects model and quantile regression technique to examine the relationships between financial inclusion, financial sustainability and CO₂ emission across 17 African countries for the period 2004-2021. The description of the variables is presented in Supplementary Table S2. A principal component analysis was performed on the two measures of financial sustainability that support the study of Le, Chuc and Taghizadeh-Hesary (2019). The study controls for GDP, GDP square, energy consumption, trade openness, urbanization and political stability.

3.1 Fixed effect model

Fixed effects estimation is a statistical technique for investigating non-experiment data. It enables the researcher to control individual-specific characteristics that remain constant over time but may be correlated with the predictor variables and exert their influence on the predicted variable (Baltagi and Giles, 1998). These characteristics could encompass unobservable traits or contextual factors that vary across individuals but remain constant over time (Schmidheiny & Basel, 2011). It addresses the potential endogeneity and omitted variable biases that arise in panel data. The fixed effects model provides unbiased estimates of a casual effect than ordinary regression. Therefore, the fixed effects model equation can be written as follows:

(1)lnCO2eit=β0+β1lnAccountsOwnershipit+β2lnATMPenetrationRateit+β3lnSavingsit+β4lnCreditsit+β5lnFSit+β6lnAccountsOwnership*FSit+β7lnATMPenetrationRate*FSit+β8lnSavings*FSit+β9lnCredits*FSit+β10lnGDPit+β11lnGDPsquareit+β12lnEnergyconsit+β13lnTradeopenessit+β14lnUrbanizationit+β15lnPoliticalStabilityit+μit

3.2 Marginal effects model

The interaction term between financial inclusion measures (accounts ownership, ATM penetration rate, savings and credits) and financial sustainability is expected to increase our understanding of the CO₂ emissions equation. At the margin, the effect of financial inclusion measures (accounts ownership, ATM penetration rate, savings and credits) and/or financial sustainability can be calculated by using the interaction model of Brambor and Golder (2006), examining the partial derivative or partial difference of CO₂ emissions concerning each of the indicators of financial inclusion.

(2)∂CO2e∂Finclusion=β1:4+β6:9Financialsustainability

MEmean=∂CO2e∂Finclusion=β1:4+β6:9financial sustainability¯it

By applying the method of Brambor and Golder (2006), the standard error can be calculated as follows for each variable:

∂CO2e∂Finclusion=var(β1)+Z2var(β6)+2Zcov(β1β6)

where Z=financial sustainability¯

The t-statistics for marginal effect at the mean value of financial sustainability can be calculated as follows:

t=(β1:4+β6:9financial sustainabilityit¯)σ∂CO2e∂Finclusion

σ∂CO2e∂Finclusion=(β1+β6financial sustainabilityit¯)var(β1)+Z2var(β6)+2Zcov(β1β6)

MEmin=∂CO2e∂Finclusion=β1:4+β6:9min(financial sustainability)it

MEmax=∂CO2e∂Finclusion=β1:4+β6:9max(financial sustainability)it

3.3 Quantile regression model

Unlike the OLS which is based on the conditional mean of the distribution, the quantile regression produces results for all the conditional quantiles of distribution for an outcome variable (Ramdani & Witteloostuijn, 2010). Quantile regression uses a conditional quantile function by splitting the conditional distribution function into segments, and it is appropriate for distribution with heteroskedasticity and the presence of outliers (Koenker & Bassett, 1978). The quantile estimation model is presented as follows:

(3)Quantq(yit|xit)=xitθ( q)

where yit is the dependent variable; CO₂ emissions at quantile q; and xit is the vector of the independent variables.

The application of the fixed effects model, quantile regression and marginal effect analyses provides a comprehensive understanding of the relationship between financial inclusion, sustainability and carbon emission. First, the inclusion of fixed effects accounts for time-invariant unobserved heterogeneity at the individual level (Kansil, 2021). By incorporating individual fixed effects, we can account for any stable individual-level characteristics that may affect the outcome variable (CO₂ emission) but do not vary over time. This helps us to mitigate potential bias arising from unobserved individual heterogeneity, providing more reliable estimates of the relationship between financial inclusion, finance sustainability and carbon emission (Bell, 2015). Second, incorporating quantile regression provides a robust and comprehensive analysis of the relationship between financial inclusion, financial sustainability and carbon emission at different points of the conditional distribution. By estimating the effects at various quantiles (e.g. median, upper and lower quantiles), we can capture the heterogenous effects of financial inclusion and financial sustainability across different emission levels (Syed, Bhowmik, Adedoyin, Alola, & Khalid, 2022). This approach offers a more nuanced understanding of how these factors influence CO₂ emission, accounting for potential non-linear relationships and capturing effects that may be obscured in traditional mean-based regression models (Okada & Samreth, 2017). Third, complementing the fixed and quantile regression models, a marginal effect analysis allows us to examine the incremental impact of financial inclusion and finance sustainability on CO₂ emission associated with a unit change in financial inclusion and finance sustainability. This analysis helps us to evaluate the magnitude and direction of the effects (Okafor, Ede, Chijoke-Mgbame, Ohalehi, & Mgbame, 2021). Through their applications, this study can offer a critical examination of the complex dynamics between financial inclusion, financial sustainability and carbon emission, contributing to the existing body of literature.

4. Results and discussion

4.1 Results

The results of the fixed effects estimations are shown in Table 1. The baseline model (1) shows that account ownership has a negative and significant impact on carbon emissions. This suggests that individuals and entities with account ownership tend to have lower carbon emissions as account owners have better access to financial services, enabling them to invest in energy-efficient technologies or renewable energy sources. while ATM penetration rate, savings and credits have positive and significant coefficients, showing a positive impact on carbon emission. Higher ATM penetration, savings and credit availability may lead to increased economic activity, resulting in higher emissions from production processes and increased competition. The results are in line with Peterson & Ozili (2023). Also, financial institutions are not yet fully aligning their services with environmentally friendly practices, leading to an increase in carbon emissions. In addition, results also show that financial sustainability has significant negative coefficients across the five models, suggesting that sound financial sustainability reduces the total amount of CO₂ emissions. Therefore, financial institutions with strong financial sustainability may have policies that prioritize green lending and environmentally friendly capital investments (Bayram et al., 2022). The interaction role of financial sustainability further reduces CO₂ emissions across the measures of financial inclusion. Therefore, the negative joint effect of higher accounts ownership and financial sustainability suggests that environment-friendly financial practices and services assist in the reduction of carbon emissions by aligning financing with sustainable projects and promoting responsible consumption and investment behaviour (Dou & Li, 2022). Moreover, control variables like GDP, energy consumption, urbanization and trade openness are positively correlated with carbon emissions across the models. Higher GDP is associated with increased economic activity and energy demand. Further rapid industrialization and trade openness drive energy consumption, transportation needs and industrial activities. Also, political stability reduces carbon emissions. This suggests that more stable political environments may be better equipped to implement and enforce environmental regulations leading to reduction in carbon emission (Adebayo, 2022). Table 2 presents the marginal effect of financial sustainability using partial differentiation, while Table 3 presents the quantile regression results to establish the non-linearity effects of financial inclusion and financial sustainability on CO₂ emissions.

As shown in Table 2, the effect of measures of financial inclusion on CO₂ emissions is contingent on the level of financial sustainability. At the mean level of financial sustainability, with a 1% increase in accounts ownership savings and credits, the amount of CO₂ emissions will decrease by 0.01%, 0.08% and 0.11%. Furthermore, at the maximum level of financial sustainability, a 1% increase in accounts ownership, ATM penetration rate, savings and credits, the amount of CO₂ emissions will decrease by approximately 0.12%, 0.38%, 0.12% and 0.20, respectively. This finding supports Bodnar et al. (2018), that increasing public bank credits and long-term loans are successful in financing climate change mitigation goals.

Regarding the findings of quantile regression, this study demonstrates the outcomes at the 10th, 25th, 50th, 75th and 90th quantiles shown in Table 3. The results showed that account ownership appears to lower the amount of CO₂ emissions significantly in both lower and upper quantiles (q = 0.10, 0.25 and 0.90). This suggests that by increasing account ownership, the amount of CO₂ emissions is reduced in both high- and low-emission countries. Conversely, a higher ATM penetration rate appears to increase the total amount of CO₂ emissions significantly at extreme quantiles (q = 0.10 and q = 0.90), while saving positively impacts the emission across both lower and upper quantiles (q = 0.10, q = 0.50, q−0.75 and q = 0.90), though the strength of the relationship is heterogeneous among the quantiles. This implies that financial access (ATM penetration rates and level of saving) among less affluent countries may contribute to higher energy consumption and subsequent CO₂ emissions. Furthermore, at the lower and upper quantiles (q = 0.25, q = 0.50 and q = 0.90, an increase in credit availability is associated with a significant increment in carbon emissions. This implies that among the fewer-emission countries, higher levels of credit access lowers carbon emissions. In addition, financial sustainability negatively impacts carbon emissions, across all the quantiles, reflecting access to sustainable finance services and indicating the adoption of sustainable practices. Furthermore, GDP, energy consumption and urbanization appear to increase CO₂ emissions both at its extreme ends; at the lower quantile and highest quantile and across models, respectively (q = 0.10, q = 0.25, q = 0.50, q− = 0.75 and q = 0.90). In contrast to the aforementioned, political stability, GDP square and trade openness reduce carbon emission across all quantiles. It is worth noting that there is an increase in carbon emissions due to extensive usage of fossil fuel energy, industrialization and the race to a higher standard of living. In contrast, trade openness has a negative impact in lower quantiles (q = 0.10 and 0.25). It might be possible that trade openness and political stability can reduce emissions through technology transfer, improved environmental regulations and a stable political environment.

4.2 Robustness test results

To verify the robustness of basic regression results, we analyzed the impact of financial inclusion indicators (account ownership, atm penetration, savings and credits) on carbon emission during the Covid-19 pandemic. The results are presented in Supplementary Table S3. Firstly, across all four models, account ownership, ATM penetration, savings and credits exhibited a positive relationship with carbon emissions. The results in Table S3 demonstrated that account ownership, ATM penetration, savings and credits contribute to increased carbon emissions by 12%, 28%, 19% and 37%, respectively. This could be attributed to the fact that enhanced financial inclusion leads to greater economic activity, resulting in higher energy consumption and subsequently increased carbon emission (Du, Wu, Zhang, Lei, & Saeed, 2022). Secondly, we observed a negative and significant impact of financial sustainability and the Covid-19 dummy variable on carbon emissions across the models. This could be because the Covid-19 pandemic led to a more prolonged and widespread lockdown, resulting in a decrease in economic activity, agricultural, industrial, manufacturing and transportation activities, thus reducing carbon emissions (Ray, Singh, Singh, Acharya, & He, 2022). Also, the negative association with financial sustainability suggests that environmentally sustainable financial practices, such as investments in green technologies or renewable energy sources, may help mitigate carbon emissions. Furthermore, the interaction effect between the Covid-19 period and financial sustainability, as well as each financial indicator, also has a negative and significant relationship with carbon emissions. This indicates that financial activity disruptions and a decrease in money supply made investors hesitant to make new investments and also delayed ongoing capital investment projects due to the Covid-19 pandemic (Anser et al., 2021); this, coupled with a focus on sustainable financial practices, led to a decrease in carbon emission. Moreover, the control variables are consistent with basic results, further validating the robustness of our findings.

4.3 Discussion

Developing economies face a significant challenge in mitigating climate risks, as these nations possess the potential for rapid economic development that may inadvertently harm the environment. In this regard, our results of financial inclusion accelerate carbon emissions due to financial inclusion. This indicates that the conventional policies intended to promote financial inclusion in these nations are not aligned with their commitment to ecological welfare objectives. The results of the positive relationship between financial inclusion and carbon emission can be attributed to the fact that developing nations can be linked to an increase in energy demand from both the production and consumption side. The plausible reason for this is that the availability of ATMs and bank branches enhances access to financial services and involves the construction and operation of physical infrastructure, resulting in higher energy usage. In addition, the energy associated with operating ATMs' electricity usage and maintenance, increased savings and access to credits enables individuals and businesses to pursue personal and economic goals by spending on carbon-intensive goods and services, industrial expansion and infrastructure. Given the heavy reliance on fossil fuels in developing countries, it is anticipated that the rise in energy demand caused by financial inclusion will lead to an increase in carbon dioxide emissions. In the same vein, as the economy experiences growth and its domestic output level increases, it induces a corresponding surge in energy demands, ultimately leading to elevated levels of carbon emission (Liu, Sinha, Destek, Alharthi, & Zafar, 2022).

On the other hand, the finding of a negative nexus between financial sustainability and carbon emission can be explained by encouraging the development and implementation of financing renewable energy projects, such as solar and wind power, which reduce reliance on fossil fuels and consequently reduce carbon dioxide (CO₂) emissions. Besides green financing, it can direct capital toward low-carbon and sustainable initiatives, thereby contributing further to the reduction of greenhouse gas emissions. In addition, financial sustainability can incorporate climate risk assessments, allowing for the identification and mitigation of climate-related investment risks. In addition, it assists businesses in adopting sustainable practices, integrating environmental considerations into strategies and promoting eco-friendly supply chains, which ultimately result in emission reductions throughout the entire value chain. Furthermore, the strengthening of climate change by international and domestic regulatory frameworks like the Paris Agreement (2015) and the African Ministerial Conference on the Environment (AMCE) (1985) has strengthened the call to the financial community to foster financial sustainability. Thereby, there is a pressure on banks to encourage credit financing for renewable energy projects, green financing energy initiatives and promoting sustainable agricultural practices.

The other key findings confirm that economic growth can exacerbate carbon dioxide emissions. This observation is especially pertinent for African nations that are experiencing economic growth and rely primarily on non-renewable energy sources. In these nations, economic activities undertaken are creating a trade-off between pursuing increased economic growth and grappling with intensified environmental challenges. Linking it with the environmental Kuznets curve (EKC) theory suggests that the selected African nations have attained a critical threshold of economic growth, indicating that it is time to adopt and implement policies aimed at reducing CO₂ emissions. Furthermore, another finding of this study confirms that energy consumption and urbanization positively impact carbon emissions in African countries. The scenario can be attributed to the increased combustion of energy (fossil fuels), which contributes to the overall increase in emissions by releasing carbon dioxide into the atmosphere (Peter & Ndubuisi, 2022). Furthermore, urbanization leads to an increase in emissions since urbanization fuels the demand for fossil energy. Besides, it induces waste generation and deforestation due to urban expansion. These findings are in line with Erdoğan et al.’s (2022) study on African countries.

Another key finding derived from this study is regarding political stability and trade openness being linked with the reduction of carbon emissions. Political stability permits the establishment of regulatory frameworks that incentivize the adoption of sustainable practices and renewable energy sources. Therefore, there is a greater likelihood of implementing and maintaining consistent environmental policies and practices in nations with stable political systems, supporting past studies in the African context (e.g. Abid, 2016). Lastly, the scenario in which trade openness reduces can be attributed to the fact that trade openness increases access to advanced technologies and knowledge from more environmentally advanced nations. This is consistent with the trade-related “pollution halo hypothesis” (Cole, Elliott, & Fredriksson, 2006), which posits that trade openness has a positive environmental effect because countries with stricter environmental regulations tend to specialize in cleaner industries and export environmentally friendly goods and services.

5. Conclusion and policy implications

Rapid economic expansion in African countries accelerates CO₂ emissions, creating a challenging trade-off between economic benefits and environmental costs. This trade-off is exacerbated by the widespread use of fossil fuels in many African countries, which makes it difficult to accomplish economic expansion without causing environmental degradation. Despite the advocation of the United Nations Framework Convention on Climate change in the reduction of CO₂ emissions, the rising level of financial inclusion and financial development has contributed to increasing CO₂ emissions. Under the given circumstances, it is of paramount importance to identify effective strategies to mitigate the carbon emissions faced by African countries. Against this backdrop, the primary objective of this research was to examine the impact of financial inclusion and carbon dioxide (CO₂) emissions with financial sustainability as a moderator in 17 African economies. Several control variables were introduced into the study to guarantee accurate and dependable results. These control variables included energy consumption, GDP growth, GDP per capita, trade liberalization, urbanization and political stability. The effect of financial inclusion on CO₂ emissions in the African context was analysed using panel fixed effect and quantile regression.

The econometric results suggest that financial inclusion does not significantly contribute to emission reduction in African countries. However, the results reveal that financial sustainability moderates the association between financial inclusion and CO₂ emissions. As evidenced by the findings of this study, financial sustainability does not only reduce CO₂ emissions, but it further reduces the positive effects of account ownership, atm penetration rate and credits on CO₂ emissions. Therefore, financial sustainability measured as bank credit to bank deposits and liquid assets to deposits and short-term funding can mitigate carbon emissions through initiatives like expanding access to green financing, renewable energy projects and factoring in climate risk when determining loan terms reduce the carbon emission. In addition, economic growth, energy consumption and urbanization have a strong positive correlation with CO₂ emissions. On the other hand, it was discovered that political stability and trade openness have a negative relationship with CO₂ emissions.

Based on these important findings, we propose a series of crucial carbon emission mitigation policies. Firstly, to effectively address the environmental challenges associated with promoting financial inclusion, it is essential for African economies to simultaneously promote financial inclusivity and environmental sustainability. These nations must identify the financial risks which are pertinent to the increase in carbon emissions and take proactive steps to align their financial services with sustainable practices. In addition, the implementation of green financial initiatives that provide incentives for investments in renewable energy, energy efficiency and sustainable technologies should be investigated to mitigate the negative environmental effects of promoting financial inclusion. Secondly, increasing the sources of financial sustainability can meet the financing of carbon storage, carbon capture and alternative renewable energy that is cost-efficient and results in the reduction of CO₂ emissions. The government must assess the degree of financial inclusion practices and how each phase and every activity affect CO₂ emission. This will help to plan adequately in the distribution of alternative energy sources into different locations to meet households' and corporates’ energy requirements and needs. For instance, hot and sunny locations may be operated using solar-based ATMs to reduce the CO₂ emissions generated by ATMs. This supports that a solar-based PV power generation system is expected to reduce CO₂ emissions between 69-100 million tons by 2030 (Hosenuzzaman et al., 2015). Thirdly, incorporating climate risk assessments into financial decision-making processes can assist in identifying and managing environmental hazards associated with investments. In addition, capacity-building programs should be implemented to improve financial literacy and environmental awareness, thereby empowering individuals and businesses to make sustainable decisions. Moreover, strict regulatory frameworks should be created and enforced to promote environmentally responsible lending practices and support sustainable initiatives. Fourthly, African economies must swiftly switch to using more renewable energy sources and less fossil fuels to power their economies. To make this change, substantial funds must be invested in R&D to improve renewable electricity generation technology. Fifthly, African countries should encourage ecologically friendly economic growth that does not trade environmental well-being for greater economic growth. Sixthly, urban planning sustainability should be prioritized for mitigating urbanization-led carbon emissions in African countries. Furthermore, a significant aspect of this approach involves that meeting the escalating urban energy demand through the utilization of clean energy resources (solar, wind and hydropower) can diminish their dependence on fossil fuels and effectively mitigate associated carbon emissions.

Lastly, our study has a practical impact on environmental sustainability in Africa through sustainable cash notes. The use of sustainable cotton-fiber paper in the production of banknotes can reduce the negative environmental impacts of cash. Finland, the Netherlands and other members of the Joint Euro Tender (JET) across eight Euro countries are adopting the use of sustainable cotton paper via the application of a protective varnish layer to enhance the durability of banknotes (Bank of Finland, 2020).

Figures

Figure 1

Growth in currency in circulation (i.e. narrow money) between 2006 and 2016

Table 1

Financial inclusion, financial sustainability, and CO₂ emissions

Variables	ln CO2 emissions	ln CO2 emissions	ln CO2 emissions	ln CO2 emissions	ln CO2 emissions
	(1)	(2)	(3)	(4)	(5)
ln accounts ownership	−0.050***	−0.245***
	(0.005)	(0.050)
ln ATM penetration rate	0.200*		0.681***
	(0.120)		(0.088)
ln savings	0.151**			0.2672**
	(0.066)			(0.066)
ln credits	0.221***				0.175***
	(0.045)				(0.057)
ln financial sustainability	−0.434***	−0.070*	−0.126***	−0.246***	−0.091**
	(0.052)	(0.042)	(0.044)	(0.060)	(0.038)
The role of financial sustainability
ln accounts ownership * ln financial sustainability		−0.012***
		(0.027)
ln ATM penetration rate * ln financial sustainability			−0.060*
			(0.035)
ln savings * in financial sustainability				−0.111***
				(0.032)
ln credits * ln financial sustainability					−0.080**
					(0.032)
GDP per capita	0.616***	0.756***	−0.196**	−0.014*	0.768***
	(0.113)	(0.096)	(0.079)	(0.139)	(0.149)
GDP square	−0.457***	−0.199*	−0.247***	−0.593***	−0.519***
	(0.109)	(0.120)	(0.080)	(0.124)	(0.098)
Energy consumption	0.843***	0.629***	0.407***	0.961***	1.096***
	(0.104)	(0.204)	(0.118)	(0.146)	(0.195)
Urbanisation	0.057***	0.026***	0.045***	0.055***	0.019***
	(0.003)	(0.004)	(0.003)	(0.004)	(0.007)
Political stability	−0.564***	−0.900***	−0.641***	−0.683***	−1.011***
	(0.048)	(0.055)	(0.052)	(0.060)	(0.071)
Trade openness	−0.849***	−0.399***	−0.982***	−0.764***	−0.784***
	(0.123)	(0.087)	(0.101)	(0.140)	(0.103)
Constant	5.034***	−4.593***	0.072	−1.125	−5.904***
	(0.729)	(1.403)	(0.766)	(0.821)	(1.325)
Observations	244	244	244	244	244
R-squared	0.942	0.954	0.947	0.882	0.941
Adj R2	0.937	0.949	0.938	0.873	0.931
F-stat	268.846	351.495	256.384	231.768	185.623
Country dummies	Yes	Yes	Yes	Yes	Yes
Year dummies	Yes	Yes	Yes	Yes	Yes

Note(s): The data used in estimating the results from columns (1)-(5) were extracted from the World Development Indicators. In brackets, t values are reported. *, ** and *** are significant at 10%, 5% and 1% level, respectively

Source(s): Analysis provided by authors

Table 2

Marginal effects (using partial differentiation) of financial sustainability using the interaction model of Brambor and Golder (2006) ME=[∂CO2emissions∂financialinclusion=β1+β4financial sustainabilityit]

The measures of financial inclusion
	Accounts ownership	ATM penetration rate	Savings	Credits
MEminimum	−0.002	0.060	−0.055	−0.029
	(0.563)	(0.704) (0.077)	(0.505)
MEmean	−0.006***	−0.162	−0.089***	−0.118***
	(0.056)	(0.384)	(0.08) (0.002)
MEmaximum	−0.012***	−0.385***	−0.121***	−0.206***
	(0.004)	(0.080)	(0.019)	(0.001)

Note(s): The data used in estimating the results from columns (1)-(4) were extracted from the World Development Indicators

***, ** and * denote significant at the 1, 5 and 10% levels, respectively. Values in the parentheses (.) indicate the standard errors. ME = marginal effect

Source(s): Analysis provided by author

Table 3

Quantile regression of CO₂ emissions, financial inclusion and financial sustainability [Qq(yit|xit)=xitθ(q)]

Variables	0.10 (1)	0.25 (2)	0.50 (3)	0.75 (4)	0.90 (5)
Accounts ownership	−0.052***	−0.046***	−0.053***	−0.057***	−0.042***
	(−8.16)	(−5.55)	(−9.40)	(−7.85)	(−6.43)
ATM penetration	0.281*	0.155	0.174	0.189	0.0625*
	(2.04)	(0.86)	(1.42)	(1.21)	(0.44)
Savings	0.173*	0.155	0.152*	0.419***	0.218**
	(2.23)	(1.53)	(2.21)	(4.76)	(2.73)
Credits	−0.003	0.204**	0.280***	0.095	0.199***
	(−0.07)	(2.90)	(5.89)	(1.56)	(3.59)
Financial sustainability	−0.452***	−0.472***	−0.447***	−0.357***	−0.341***
	(−7.29)	(−5.81)	(−8.13)	(−5.06)	(−5.32)
GDP per capita	0.935***	0.663***	0.636***	0.588***	0.606***
	(7.07)	(3.83)	(5.42)	(3.91)	(4.43)
GDPsqr	−0.341*	−0.624***	−0.562***	−0.290	−0.569***
	(−2.57)	(−3.60)	(−4.78)	(−1.92)	(−4.16)
Energy consumption	0.631***	0.756***	0.720***	0.826***	0.748***
	(4.61)	(4.22)	(5.94)	(5.31)	(5.29)
Urbanization	0.066***	0.064***	0.055***	0.038***	0.032***
	(15.83)	(11.70)	(14.78)	(7.95)	(7.32)
Political stability	−0.525***	−0.602***	−0.584***	−0.578***	−0.648***
	(−8.40)	(−7.38)	(−10.55)	(−8.15)	(−10.04)
Trade openness	−0.644***	−0.795***	−0.757***	−0.797***	−0.691***
	(−4.89)	(−4.62)	(−6.49)	(−5.33)	(−5.08)
Constant	−6.355***	−5.246***	−4.346***	−4.574***	−2.850**
	(−6.09)	(−3.85)	(−4.70)	(−3.86)	(−2.64)
Observations	244	244	244	244	244

Note(s): The data used in estimating the results from columns (1)-(5) were extracted from the World Development Indicators

In brackets, t values are reported. *, ** and *** are significant at 10%, 5% and 1% levels, respectively. The equation is explained as follows. The elements of the parameter vector θ give the marginal effect of the corresponding explanatory variable. The notation θ(q) highlights the presence of a potentially different parameter vector at each respective quantile q of the distribution

Source(s): Analysis provided by author

Table S1

Currency-CO₂ emissions impact (mean 2004-2014)

Country	CO2 from automated teller machines (per 100,000 adults)	CO2 emissions from bank branches (per 100,000 adults)	CO2 from ATMs as a percentage of total CO2 emissions (per 100,000 adults)	CO2 from bank branches as a percentage of total CO2 emissions (per 100,000 adults)
Algeria	6.148	3234.430	0.005	2.800
Angola	11.927	3864.937	0.040	13.367
Benin	2.185	1594.446	0.039	33.840
Botswana	33.239	5061.952	0.743	113.583
Burkina Faso	1.425	1162.699	0.055	60.060
Burundi	0.693	1415.006	0.245	616.607
Cabo Verde	46.466	17235.572	9.236	3482.817
Cameroon	2.264	889.699	0.037	15.591
Central African Republic	0.586	423.697	0.206	156.033
Chad	0.465	366.389	0.079	68.512
Comoros	3.741	937.660	2.577	661.369
Congo, Dem. Rep	0.422	356.893	0.014	17.133
Congo, Rep	2.603	1194.068	0.118	60.877
Cote d'Ivoire	3.000	1955.785	0.033	24.641
Djibouti	3.560	2324.126	0.657	457.520
Egypt, Arab Rep	9.805	2751.327	0.005	1.418
Equatorial Guinea	3.705	2073.537	0.072	41.518
Eswatini	28.955	4202.985	2.651	390.681
Ethiopia	0.164	722.355	0.002	10.782
Gabon	9.023	2784.874	0.192	58.926
Ghana	4.495	3019.457	0.040	30.359
Guinea	0.965	833.868	0.041	36.122
Guinea-Bissau	1.572	935.772	0.611	379.938
Kenya	8.492	2628.639	0.071	23.064
Lesotho	9.069	1818.612	0.399	81.270
Liberia	1.258	1453.830	0.163	183.517
Libya	4.769	6868.312	0.009	13.000
Madagascar	1.662	903.854	0.075	42.318
Malawi	3.590	1306.796	0.328	121.929
Mali	2.144	2327.346	0.198	230.126
Mauritania	2.742	2225.614	0.104	95.510
Mauritius	52.808	13014.646	1.404	347.199
Morocco	24.072	11108.599	0.044	20.363
Mozambique	6.839	1902.427	0.226	67.122
Namibia	41.897	7806.109	1.371	283.683
Niger	0.599	571.801	0.035	48.466
Nigeria	11.076	3535.537	0.012	3.734
Rwanda	2.659	2385.504	0.364	361.528
Sao Tome and Principe	10.105	11468.321	9.100	11117.126
Senegal	2.862	2398.539	0.035	37.380
Seychelles	60.697	29017.738	12.043	5626.192
Sierra Leone	0.221	1030.261	0.029	141.190
Somalia	0.000	0.000	0.000	0.000
South Africa	60.865	5410.695	0.013	1.152
Sudan	3.052	1745.772	0.020	12.390
Tanzania	3.692	1067.223	0.043	14.066
Togo	1.825	2165.718	0.071	102.281
Tunisia	22.799	9727.890	0.088	38.146
Uganda	4.116	1354.044	0.115	38.698
Zambia	7.177	2385.996	0.237	87.331
Zimbabwe	6.785	4327.591	0.080	46.906

Note(s): We computed the average CO₂ emissions from automated teller machines and bank branches using the methodology of McCook (2014). McCook (2014) found that 600,000 bank branches emit 383.1 million tons of CO₂ emissions per year. This amounts to an average of 638.5 tons emitted per branch in a year. We then multiply 638.5 by the number of bank branches in each of the countries annually. About 3.2 million CO₂ emissions are emitted from 2,394,700 ATMs in the world annually (McCook, 2014). By calculation, each ATM emits 1.336 tons of CO₂ emissions per year. We adopt the same process as in bank branches

Source(s): Analysis provided by author

Table S2

Description of variables

Variables	Measurement	Source
CO2 emission	The total carbon dioxide emission from the consumption of energy measured in million metric tons	European Union Publication on CO2 and GHG emissions, WDI
Financial inclusion	Accounts ownership (automated teller machines per 100,000 adults)	WDI
	ATM penetration Rate (bank branches per 100,000 adults)	WDI
	Savings (deposit accounts with commercial banks per 1000 adults)	WDI
	Credits (borrowers from commercial banks per 1000 adults)	WDI
Financial sustainability	Bank credit to bank deposits (%)	WDI
	Liquid assets to deposits and short-term funding (%)	WDI
GDP	GDP per capita (constant 2015 US$)	WDI
GDP square	Square of GDP per capita (constant 2015 US$)
Energy consumption	Energy use (kg of oil equivalent per capita)	WDI
Urban population	The population is measured in millions of persons to indicate urbanization	WDI
Trade openness	(Exports + imports (BoP, current US$)/GDP	WDI
Political stability	Perceptions of political instability, politically motivated violence, including terrorism	WDI

Note(s): This table presents variables used in this paper and sources of raw data. *WDI is World Development Indicator

Source(s): Table provided by author

Table S3

Financial inclusion, Covid-19 and CO₂ emissions

Variables	CO2 emissions (1)	CO2 emissions (2)	CO2 emissions (3)	CO2 emissions (4)
Accounts ownership	0.116*
	(.065)
ATM penetration		.0276**
		(0.109)
Savings			0.189***
			(0.000)
Credit				0.366***
				(0.038)
Financial sustainability	−0.203***	−0.177***	−0.096**	−0.268***
	(0.074)	(0.052)	(0.041)	(0.035)
Covid. dummy	−0.230*	−0.271**	−0.1453*	−0.297***
	(0.132)	(0.113)	(0.086)	(0.098)
The role of Covid-19
Account ownership* Covid*Sustainability	−0.063*
	(0.037)
ATM penetration* Covid*sustainability		−0.069**
		(0.033)
Savings* Covid* sustainability			−0.025**
			(0.010)
Credits* Covid* sustainability				−0.036**
				(0.040)
GDPPC	0.065	0.563***	0.734***	0.270***
	(0.161)	(0.161)	(0.175)	(0.133)
GDP square	−0.664***	−1.595***	−1.787***	−1.360***
	(0.144)	(0.209)	(0.166)	(0.108)
Energy consumption	0.968***	2.091***	1.068***	0.011
	(0.217)	(0.234)	(0.169)	(0.167)
Urbanization	0.036***	0.026***	−0.004	0.036***
	(0.005)	(0.005)	(0.006)	(0.003)
Trade openness	−0.863***	−0.803***	−0.632***	−0.602***
FDI	0.256**
	(0.038)
Political stability	−0.404***	−0.629***	−0.536***	−0.713***
	(0.086)	(0.072)	(0.043)	(0.052)
Constant	−4.885***	−10.70***	−4.132***	4.446***
	(1.246)	(2.118)	(1.397)	(1.172)
Observations	244	244	244	244
R-squared	0.900	0.946	0.970	0.952
Adj R2	0.891	0.941	0.967	0.948
F-stat	372.421	301.481	505.280	559.232
Country dummies	Yes	Yes	Yes	Yes
Year dummies	Yes	Yes	Yes	Yes

Note(s): The data used in estimating the results from columns (1)-(4) were extracted from the World Development Indicators. In brackets, t values are reported. *, ** and *** are significant at 10%, 5%, and 1% level, respectively

Source(s): Analysis provided by author

Funding: This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.