While empirical evidence regarding the relationship between diversification and banks’ funding costs is limited, the theoretical literature presents mixed predictions. The modern portfolio theory suggests that diversification can reduce banks’ overall risk by lowering the volatility of profits (Ben et al. 2023). This perceived reduction in risk can, in turn, attract more investors and depositors, subsequently leading to lower funding costs for banks. However, the literature on income diversification (e.g., Wang 2023) suggests that diversified firms often face inefficient resource allocation because of agency problems, leading to a discount in their valuation. This decreased valuation, reflecting higher perceived risks and management complexities, can subsequently increase the cost of capital for diversified banks as investors demand higher returns to compensate for the increased risk.

Studies have predominantly focused on the effect of diversification on bank stability. For example, Adem (2022) examines the effect of income diversification on bank stability, providing evidence that diversification of income reduces risk and improves bank stability in Africa. Sissy et al. (2017) analyse the interaction between cross-border banking, revenue diversification and stability across 29 African countries. More recently, Shabir et al. (2024) study the effect of diversification on bank stability in the MENA region. Their findings show that both income and asset diversification improve bank stability. Another strand of literature looks at the impact of diversification on bank performance. Addai et al. (2022), for instance, examine the interaction between income diversification and bank performance in Africa. Similarly, Adesina (2021) analyses the relationship between diversification and bank performance, showing evidence that greater income and asset diversification reduces bank stability, cost efficiency, and profitability.

Furthermore, Githaiga (2021) investigates the moderating effect of income diversification on the relationship between human capital and bank performance in East Africa. Few studies have explored the link between diversification and bank ESG performance. Saif-Alyousfi et al. (2023), for example, analyse whether bank ESG activities affect their income and asset diversification. Their results indicate that environmental and social factors negatively impact bank diversification. A recent study by Kamau and Simo-Kengne (2024) investigates the relationship between legal origin, diversification, and bank lending behaviour in Africa, finding evidence that revenue diversification indirectly mitigates the negative impacts of loan growth in banks with a common-law legal origin. Turning to banks’ funding costs, studies have largely examined the effect of bank solvency risk and funding costs. Aldasoro et al. (2022), for instance, examine the interaction between bank solvency risk and funding cost in Korea.

A related study by Levine et al. (2021) analyses the impact of geographical diversification on the costs of interest-bearing liabilities for banks across U.S. A growing body of literature has explored the effects of ESG performance on banks’ funding costs. For example, Agnese and Giacomini (2023) assess whether ESG factors affect bank bond pricing. Works have also examined the relationship between credit ratings and bank funding costs. For instance, Opoku Mensah et al. (2017) studied the effect of sovereign credit ratings on banks’ funding costs in Africa, finding evidence of lower funding costs for banks with higher sovereign ratings. A related study by Kusi and Opoku-Mensah (2018) analyses the effect of credit information sharing on banks’ funding costs in Africa. They show evidence that the presence of credit information lowers bank funding costs. Most recently, Nguyen et al. (2024) investigated whether banks with more capable managers achieve better funding costs. According to their findings, banks with strong managerial abilities are able to secure funding at reduced costs.

From the reviewed literature, it is clear that no study has specifically explored the impact of diversification on bank funding costs within the African context. Accordingly, this study argues that diversification will likely reduce banks’ cost of funds in Africa for two primary reasons. Firstly, diversification can help mitigate risk, especially in Africa, where banks face higher levels of non-performing loans. Secondly, diversification can provide banks in the region with enhanced access to a wider range of funding sources and earnings, leading to improved liquidity and stability. Consequently, this study proposes the following hypothesis:

Financial regulations designed to curtail non-traditional banking activities have become critical tools in the aftermath of financial crises, particularly after the 2008 global financial crisis (Ben et al. 2023). These regulations were largely implemented in response to the realisation that many large financial institutions had failed because of extensive involvement in non-traditional and often risky financial activities. Literature (e.g., Saif-Alyousfi et al. 2023; Velasco 2022) argues that financial regulation indirectly influences banks’ cost of funds through diversification. For instance, regulatory capital requirements may encourage banks to diversify their income and assets in order to hold lower capital levels. Furthermore, regulations such as interest rate caps could drive banks to explore alternative income and funding sources (Wang & Lin 2021).

However, regulatory constraints may lead to a suboptimal allocation of resources by steering banks towards investments that are safer but yield lower returns (Ben et al. 2023). This shift can paradoxically result in reduced income and higher funding costs, as banks may be compelled to offer higher interest rates to attract capital. Prior studies (e.g., Hunjra et al. 2021) demonstrate that capital regulations enhance banks’ capital, enabling them to pursue greater diversification. This diversification can lead to more stable income streams and reduce the volatility of a bank’s overall financial performance, which can lower the risk premium demanded on capital lent to or invested in the bank. Velasco (2022) finds an indirect effect of regulatory capital on bank diversification. Accordingly, this study anticipates that financial regulations will indirectly influence funding costs through diversification. Consequently, the article proposes the following hypothesis:

This study investigates the effects of diversification on banks’ funding costs using a sample of 563 commercial banks drawn from 53 African countries from 2007 to 2021. The selection of these banks and countries was based on the availability of consistent data over the study period. Bank-level data are sourced from the BankFocus database, while data on financial regulations and macroeconomic variables is obtained from the World Bank’s World Development Indicators (WDI) and International Monetary Fund International (IMF) Financial Statistics databases.

This study incorporates both bank-specific and macroeconomic control variables, which have been identified in previous studies (e.g., Andrieș & Sprincean 2023, Levine et al. 2021) to influence bank funding costs. Bank-specific control variables include equity, bank size and return on assets (ROA). According to Saif-Alyousfi et al. (2023), well-capitalised banks are viewed as less risky, making them raise funds at a cheaper cost. Therefore, a negative relationship between equity and banks’ funding costs is expected. Equity is measured as a ratio of equity to total assets. Literature provides mixed predictions on the direction of the relationship between bank size and funding costs. On the one hand, Saif-Alyousfi et al. (2023) show that bigger banks enjoy lower funding costs because they are perceived as ‘too big to fail’.

Conversely, large banks seek more deposits to lend to borrowers, leading to higher interest expenses and potentially increasing their overall funding costs (Kouzez 2023). Thus, the expected sign for the coefficient of bank size is unclear. Bank size is measured as the log of total assets. ROA is a proxy for bank profitability calculated as net income divided by total assets. The coefficient for ROA is expected to be negative, as highly profitable banks may be seen as efficient by investors and thus rewarded with lower funding costs (Andrieș & Sprincean 2023).

For macroeconomic variables, the study includes controls for the policy rate, inflation, gross domestic product (GDP) growth and financial crisis. The monetary policy rate directly affects the interest rates that banks pay to obtain funds (Ćehajić & Košak 2021). Accordingly, the policy rate is expected to positively affect banks’ funding costs. GDP growth improves a country’s economic conditions, simultaneously lowering the riskiness of banks and enhancing the conditions for raising loans (Nguyen et al. 2024). This can result in reduced funding costs.

Therefore, the study expects GDP growth, measured percentage change in a country’s real GDP, to negatively influence banks’ funding costs. Shabir et al. (2024) observe that a rise in inflation rates increases borrowing costs because investors demand higher rates of return on their investments. Thus, inflation is expected to be positively related to the dependent variable. Lastly, crisis is a proxy for the financial crisis, which assumes a dummy value of 1 for the period 2008-2009 and 0 otherwise. Khattak et al. (2023) find that the global financial crisis of 2008 led to increased banks’ funding costs because of the deterioration of the money market and a subsequent increase in financial instruments’ risk premium. Consequently, this study expects a positive coefficient for the crisis variable. Table 1 provides a summary description of the study variables.

The study employs the following baseline model to test hypothesis 1, which examines the effects of diversification on banks’ funding costs, following prior related studies of Agnese and Giacomini (2023) and Wang and Lin (2021): FCikt=α+β1DIVikt+∑j=17∅j Controlsjt  +αi+λt+εikt′[Eqn 5] where FC_ikt denotes funding costs of bank i in country k in year t. DIV indicates bank diversification, which can either be revenue or asset diversification. Controls represent the vector of bank-specific (equity, bank size and ROA) and macroeconomic control variables (policy rate, inflation, GDP growth and financial crisis). In addition, the study includes bank (α_i) and year (λ_t) fixed effects to control for bank-specific and time-varying unobserved heterogeneity. The baseline regression model is estimated using the fixed effects model following the results of the Hausman test and Wang and Lin (2021).

As mentioned earlier, given the potential indirect influence of financial regulation on diversification, this study employs the two-stage least square (2SLS) method. The 2SLS approach also controls for endogeneity, which is a common concern in the diversification literature (Levine et al. 2021). Such endogeneity may arise because the decision to diversify is at the bank’s discretion; hence the bank-specific factors that influence the revenue diversification strategy may also affect funding costs. Moreover, reverse causality may arise because the decision to diversify a bank’s revenue sources may be influenced by past costs of funds (Velasco 2022). In the first stage, the predicted values of diversification are estimated based on the instrumental variables using the following model: DIVikt=β1Zikt+∑j=17∅j Controlsjt  +αi+λt+εikt′[Eqn 6] where DIV^ikt is the predicted values of diversification and β₁ is the coefficient of the instrumental variable Z_ikt The study uses the financial regulation proxies (a dummy variable for interest rate control and a dummy variable for the adoption of Basel II or Basel III capital) as instruments for revenue and asset diversification. This approach is based on the idea that banks will likely diversify their income and assets to increase earnings when faced with interest rate ceilings. Similarly, the adoption of Basel capital accords, particularly Basel II and Basel III frameworks, has been observed to drive more diversification in the banking industry (Velasco 2022).

In the second stage, the predicted values of diversification are used as exogenous variables to estimate the effect of diversification on banks’ funding costs as follows: FCikt=α+β1DIV^ikt+∑j=17∅j Controlsjt +αi+λt+εikt′[Eqn 7]

The article employs the Sargan test and the Lagrange Multiplier (LM) test to assess the relevance and validity of the 2SLS instrumental variables. Furthermore, the study utilises the Generalized Method of Moments (GMM) estimator to check the robustness of the baseline results. Unlike the first two estimators, the GMM approach accounts for both endogeneity and the persistence of the dependent variable, whereby the previous year’s banks’ funding costs are likely to influence the current year’s banks’ funding costs (Saif-Alyousfi et al. 2023). This study uses the two-step system GMM estimator, which is considered more efficient than the one-step (Khattak et al. 2023). The first step of the two-step GMM is used to generate the instruments for the second-step equation, and it involves estimating the coefficients of the first differenced equation of the endogenous variable. The first step equation can be modelled as follows: DIVikt=∑i=1kρ1DIVikt−1+∑j=17∅j Controlsjt  +αi+λt+μikt[Eqn 8] whereby DIV_ikt is revenue or asset diversification for bank i in country k at year t. DIV_ikt-1 is the lagged values of revenue or asset diversification. α_i and λ_t are bank and year-fixed effects, respectively, while µ_ikt is the error term. The second-step regression of the two-step system GMM involves estimating the effect of diversification on banks’ funding costs by regressing the first-differenced dependent variable on the first differenced predicted values of diversification from the first step, as well as lagged values of the dependent variable and the predicted variable of diversification. The second-step equation can be written as: FCikt=α+γFCikt−1+β1DIV^ikt  +∑j=17∅j Controlsjt+αi+λt+εikt′[Eqn 9] where FC_ikt is banks’ funding and FC_ikt-1 is the lagged value of banks’ funding costs. DIV^ikt is the predicted value of diversification. In performing the GMM estimation, the study uses the 1st lag of the dependent variable and the 1st lag and deeper of revenue diversification as instruments for revenue diversification. To confirm the consistency of the results, tests for second-order autocorrelation and the validity of the instruments are performed using the Arellano-bond test – AR(2) and the Sargan test, respectively. Finally, to account for the time lag in the relationship between diversification and banks’ funding costs, the study includes a lag of the diversification variables in both the baseline and alternative regression models as an additional robustness test.

This study used secondary data and does not contain any studies involving human participants performed by any of the authors.

Table 2 presents the descriptive statistics for the variables under study. The sampled banks exhibit moderate funding costs, with a mean of 16.4%. Furthermore, they display relatively high levels of diversification towards non-interest income, as evidenced by the mean RDIV of 49.3% and ARDIV of 52.8%. The HHI for ADIV indicates significant assets-based diversification, with a mean value of 0.601. Conversely, the low value of AADIV (0.134) indicates a higher concentration in diversification towards interest-earning assets. Lastly, the countries studied show significant variation in financial regulations as shown by the standard deviations of interest rate controls and Basel of 0.496 and 0.50, respectively.

The correlation coefficient matrix (available upon request) reveals low correlations between the study variables, with all coefficients below 0.4, suggesting the absence of multicollinearity issues.

Table 3 displays the regression results for testing hypothesis 1, which examines the negative effect of diversification on banks’ funding costs in Africa using the fixed effects model. The results for RDIV and ADIV are presented in columns (1)–(2) and (3)–(4), respectively. According to the results in Table 3, revenue diversification has a consistently negative and significant effect on banks’ funding costs. These results support hypothesis 1 and align with prior related studies of Shabir et al. (2024) and Addai et al. (2022), suggesting that income diversification enhances bank performance, which ultimately leads to reduced risk and, consequently, a lower cost of capital.

Two possible arguments exist for the negative relationship between revenue diversification and banks’ funding costs. Firstly, higher revenue diversification can reduce banks’ risk by minimising income volatility (Shabir et al. 2024; Shahriar et al. 2023). As a result, lenders may view such banks as less risky, leading to a reduced premium in their funding costs (Tran & Nguyen 2023). Secondly, diversification of revenue sources can enhance banks’ profitability and efficiency through economies of scope under the synergy effect as shown by Saif-Alyousfi et al. (2023) and Addai et al. (2022). The increased profitability and efficiency may subsequently reduce banks’ funding costs, as investors and lenders perceive them as low-risk borrowers (Nguyen et al. 2024).

The results in Table 3 show a statistically significant negative relationship between asset diversification and banks’ funding costs, further supporting Hypothesis 1. These findings align with those of Levine et al. (2021), which show that asset diversification reduces funding costs for banks in the U.S. This is possibly because asset diversification reduces risk by adding assets to banks’ portfolios that are imperfectly correlated with existing ones, thereby spreading and minimising potential financial exposure (Ben et al. 2023). Moreover, asset-based diversification enables banks to spread their activities across various industries, potentially reducing the expected costs of financial distress (Shabir et al. 2024). This may result in more stable financial performance for banks, resulting in lower funding costs as the risk to lenders and investors diminishes as shown by Shahriar et al. (2023).

The similarity in results between Africa and the U.S. may be attributed to the shared focus of banks in both African and developed economies on enhancing profitability, managing risks, and reducing funding costs. Accordingly, by diversifying to mitigate risks, banks in both regions align their strategies, resulting in comparable reductions in funding costs (Duho, Onumah & Owodo 2020). Furthermore, the growing integration of African banks into global financial markets exposes them to similar diversification practices, resulting in comparable outcomes on funding costs (Shabir et al. 2024). Looking at the control variables, equity has a positive and significant effect on banks’ funding costs, in contrast to earlier expectations and prior studies by Saif-Alyousfi et al. (2023). This outcome may be expected because an increase in equity reduces banks’ return on capital, leading to higher funding costs (Aldasoro et al. 2022).

The coefficients for ROA are negative and significant, suggesting that higher profitability lowers banks’ funding costs. This result is consistent with Andrieș and Sprincean (2023) finding that profitability lowers funding costs by reducing banks’ risk-taking incentives because of increased charter value. As expected, the policy rate has a significant positive effect on banks’ cost of funds, consistent with earlier findings by Ćehajić and Košak (2021), indicating that an increase in the policy rate leads to higher interest rates for banks’ funding. Lastly, inflation has a positive and significant coefficient, confirming earlier studies by Shabir et al. (2024) that suggest higher inflation rates increase lenders’ and investors’ required rate of return, resulting in higher banks’ funding costs.

Table 4 presents the second-stage estimation results for testing hypothesis 2, which explores the indirect effects of financial regulations on the relationship between diversification and funding costs, employing the 2SLS method with instrumental variables. The Sargan test and the LM test indicate that the instruments generally meet the relevance and validity conditions, thereby confirming the consistency of the findings. The results in columns (1)–(2) and (5)–(6) of Table 4 reveal a significantly negative coefficient for the predicted values of revenue diversification, suggesting that financial regulation indirectly lowers banks’ funding costs. These findings support hypothesis 2 and are consistent with prior studies of Velasco (2022) and Wang and Lin (2021), which provide evidence that financial regulations encourage banks to pursue diversification strategies. This, in turn, can broaden banks’ access to diverse funding sources, ultimately leading to reduced funding costs.

As indicated earlier, this article employs the GMM estimator to verify the robustness of the baseline results. Table 5 presents the GMM results. The insignificant AR(2) test values indicate the absence of second-order autocorrelation. Similarly, the Sagan test provides evidence that the lagged values of revenue and assets diversification are valid instruments across most of the estimation models. The lagged dependent variables are highly significant, confirming the persistence of funding costs, hence, supporting the use of the dynamic model. Overall, the GMM results in Table 5 largely align with the findings from the baseline models, confirming that diversification reduces banks’ funding costs. Lastly, the lagged coefficients for revenue and asset diversification (available upon request), consistently remain negative and significant, further confirming that diversification lowers banks’ funding costs in Africa.

Studies (e.g., Opoku Mensah et al. 2017; Tran, Nguyen & Hoang 2024) suggest that the relationship between diversification, financial regulations, and funding costs can vary significantly across banks of varying sizes. Large banks typically have greater resources and market access, enabling them to diversify risks across a broader range of assets and efficiently navigate complex financial regulations (Shabir et al. 2024). This can enhance the benefits of diversification in lowering funding costs. In contrast, smaller banks often have limited resources and face relatively higher compliance costs, which can reduce the effectiveness of diversification and increase the sensitivity of funding costs to regulatory pressures (Andrieș & Sprincean 2023).

Accordingly, to investigate the possibility that large banks benefit from cheaper funding costs, this study further examines the effect of diversification on banks’ funding costs and the indirect role of financial regulations across different levels of bank size by re-estimating equations (1)–(9) for small and large banks. In this article, banks are classified as small if their total assets fall below the 50th percentile of the total assets of banks operating in country j, and as large if their total assets are at or above the 50th percentile following Andrieș and Sprincean (2023). The results are presented in Table 6. For large banks, the coefficient for revenue diversification remains significantly negative as shown in columns (1)–(2). However, this significant effect disappears for small banks as indicated in columns (7)–(8). These findings support Huynh and Dang (2022), indicating that large banks are more stable, which can significantly lower their funding costs.

Columns (3)–(4) and (9)–(10) show no significant difference in the indirect effect of financial regulation on banks’ funding costs for small and large banks contrary to the results shown in Table 4. A possible argument for this outcome is that banks, regardless of size, are subject to a uniform regulatory framework. Thus, regulations may not provide differential funding cost advantages based on diversification (Velasco 2022). Furthermore, large banks possess a significant advantage in managing operating leverage associated with shifts towards non-interest income, attributed to their access to resources as shown by Shabir et al. (2024). Surprisingly, the results in Table 5 indicate no notable difference between asset diversification and banks’ funding costs for the two subsamples. This is possibly because of limited investment options in Africa that may make banks, regardless of their size, to invest in similar asset classes, resulting in a minimal difference in the impact of asset diversification on banks’ funding costs (Sissy et al. 2017).

Lastly, Table 7 displays the results of re-estimating equations (5)–(9) for a sub-sample of local and international banks in the context of revenue and asset diversification. The results reveal that a shift towards non-interest income activities has an insignificant effect on funding costs for local banks, but significantly and negatively influences funding costs for international banks, as shown in columns (1)–(2) and (7)–(8), respectively. These findings are in line with Wang (2023), indicating that banks with broader geographical reach gain from economies of scale and enhanced financial access. Regarding the effects of asset diversification on funding costs, the findings from Panels A and B of Table 7 do not exhibit significant differences between the two groups.

Furthermore, the results in columns (3)–(4) and (9)–(10) of Table 7 show a more significant indirect inverse effect of financial regulation on funding costs for international banks compared to local banks, suggesting that the indirect effect of financial regulation on funding costs varies with banks’ geographical coverage. This variation is likely because investors perceive international banks as less risky, given their operation under multiple regulatory environments, which can significantly reduce their funding costs (Poshakwale et al. 2020). Furthermore, banks with higher geographical coverage can take advantage of regulatory arbitrage to access cheaper funding sources (Avdjiev, Aysun & Tseng 2022).