Do Financial Innovation and Financial Deepening Promote Economic Growth in Sub-Saharan Africa?

Abstract

In this paper, we analyze the impacts of financial innovation and financial deepening on the economic growth of 14 sub-Saharan African (SSA) countries from 1995 to 2023. The autoregressive distributed lag (ARDL) approach and error correction model (ECM) were used to assess short- and long-run effects. The findings indicate that mobile cellular subscriptions and government spending are the main contributors to national economic growth and that money supply has a positive impact. However, the strong negative effect of capital formation on economic growth is contrary to expectations. Conversely, the findings confirm that gross capital formation has a strong positive effect on gross domestic product (GDP) growth in the long run. Bounds testing reveals varying degrees of cointegration across countries. Long-run relationships were confirmed in Senegal, Côte d’Ivoire, Ethiopia, and Zimbabwe, all of which showed evidence of strong cointegration. These findings support policy recommendations aimed at promoting sustainable economic growth in SSA economies through targeted policies that increase domestic credit in the private sector and attract foreign direct investment (FDI).

1. Introduction

Innovation in financial matters, through inventions and technologies that enhance products and services, enables production to become more productive (European Central Bank, 2017). Since innovation can be used to stimulate economic growth and mitigate adverse effects on the environment and society, as described by Chovancová et al. (2024), it can also create new industries, opportunities, and sources of revenue that can enhance economic growth. Therefore, it has been cited as one of the sources of sustainable development.

Financial innovation is also an element of economic development (Noreen, 2024). Schumpeter (1934) suggested that a sound financial system promotes technological advancement by distributing finances between less productive and more productive sectors. Later writers (Revell, 1970; McKinnon, 1973; Lachman, 1974) documented extensive evidence that financial development has a major positive influence on economic development. In addition, Levine (1997) and Mishra (2008) noted that an effective and robust financial system may elevate economic development by enabling economic actors to diversify and expand their portfolios and fulfill their liquidity needs. Moreover, financial innovation is linked to enhanced savings and capital accumulation and, therefore, to higher economic growth. Innovation can also improve the quantity of savings and productivity levels, thereby augmenting the economy (Domeher et al., 2022). In summary, financial innovation has been found to revolutionize and reform financial services, and its beneficial effects on the economy are increasingly gaining relevance. Further drivers of its significant impact on economic growth, particularly in developing nations, include new instruments, technologies, and practices introduced by financial innovation. These advances enable capital formation, more efficient resource allocation, and greater financial inclusion, thereby empowering international trade, productivity, entrepreneurship, and innovation (Egert & Jawadi, 2018).

Africa’s economic growth over the last 25 years has been fragmented. Africa’s macroeconomic situation conceals the variations created by this uneven performance (African Development Bank Group, 2025). Inadequate institutional structures and a lack of financial resources to support the sustainable social and economic transformation of accessible natural capital are major contributors to weak and declining growth in African countries (Achuo, 2023). Furthermore, some researchers argue that Africa’s growth challenges stem from inadequate investment quality, limited project appraisal, and weak governance in supporting industrialization for the transformation and export of natural resource wealth (Soula et al., 2023).

The SSA region has registered low economic growth in recent decades. The annual growth rate of gross domestic product (GDP) per capita between 1961 and 2010 ranged from 0.90 to 2.39, except during 1981–1990 and 1991-2000, when it was −0.96 and −0.32, respectively. The most recent growth rates are −1.71 and 1.72 for 2011–2020 and 2021–2024, respectively (International Monetary Fund (IMF), 2018; World Bank, 2025). Financial systems in most SSA countries grew weakly and were considered vulnerable around the mid-1980s, primarily because of deteriorating macroeconomic performance and, secondarily, because of government interference in the operation of financial institutions, negative real interest rates, and targeted credit policies. These conditions were compounded by structural weaknesses, including the absence of an appropriate legal framework to grant central banks autonomy in pursuing price stability, poor operating practices, the lack of effective tools to control market-oriented monetary practices, and weak, uncompetitive financial market structures. These weaknesses undermined the success of the financial system and its effectiveness in performing financial intermediation functions (Adams, 2021; Briffaut, 1998).

In addition, GDP per capita growth in SSA has fluctuated, with periods of stagnation and growth. Poor performance was observed during the 1980s and early 1990s. This was attributed to the inability of financial sectors to mobilize domestic savings and foreign private capital, as well as to the lack of sound banking institutions and effective financial intermediation (Basu et al., 2000). The COVID-19 pandemic further weakened the economy in 2020, leading to negative per capita GDP growth. As a result, SSA GDP per capita reached 1623 in 2023, representing a decrease of 3.88 relative to 2022 (African Development Bank Group, 2021). In addition, fiscal deficits were expected to reach historic levels of 8.4% of GDP in 2020. In both the short and medium term, the debt margin was projected to increase by ten to fifteen percent. There were also substantial exchange-rate fluctuations, high inflation, and significant disruptions in external financial inflows (African Development Bank Group, 2021).

As such, countries in the SSA region resorted to financial innovation as a means of redress to address increasingly severe economic challenges. The African development challenge can be addressed through financial innovation, as articulated by Beck et al. (2014). Several studies have indicated that financial innovation is more likely to influence the development of SSA economiespositively (Diop et al., 2025; Domeher et al., 2022; Emna et al., 2025). The process of financial innovation forms a foundation for social, political, and cultural realities and norms within the region. On this basis, financial innovation has been introduced in the SSA region to mitigate high-risk challenges in such environments. However, its effects may vary depending on each country’s level of financial development.

To address this research gap, the current study examines the correlation between financial innovation and financial deepening, gross capital formation, government spending, and economic growth in SSA countries. The reviewed literature on the correlation between financial innovation and economic development in emerging economies provides a distinct contribution that has largely been overlooked (Ogbeide & Obadeyi, 2023). Disparities in financial development across developing nations are substantial (Ndako, 2010; Napier, 2014; Qayyum et al., 2025). Most previous research on economic growth in African countries has focused primarily on the roles of natural resources, agriculture, and commodities in growth dynamics, as well as on increased macroeconomic control observed in these countries (Mlachila et al., 2013). Consequently, the exploration of financial innovation represents a relevant area of study. In this context, the present seeks to contribute by examining the economic impact of financial innovation in the SSA region, specifically by analyzing how financial innovation and financial development have affected SSA economies during the period 1995–2023. In particular, this study examines the relationships among domestic credit to the private sector by banks, mobile cellular subscriptions, gross capital formation, government expenditure, and broad money. To better understand the drivers of growth in SSA, the analysis focuses on the influence of these factors. More specifically, the effect of financial innovation on economic growth is examined. SSA performance is assessed in relation to the role of financial innovation in driving growth. Analysis of these variables enables the derivation of relevant policy implications to address economic and social challenges in SSA countries and contributes to the debate on the irrelevance of financial innovation, given the diversity of countries considered.

This study aims to assist policymakers in understanding the economic effects of financial innovation. Moreover, its empirical findings indicate the key directions of financial innovation that should be strengthened to support the evolution of the financial sector and economic development in SSA countries. In addition, this study aims to inform policymakers on how to track financial innovation trends in SSA by identifying challenges faced by users of financial innovation systems and proposing solutions to address them. Furthermore, this paper contributes to the literature by examining the relationship between economic growth, financial development, and financial innovation within the SSA context.

The rest of this study is structured as follows. Section 2 reviews the theoretical and empirical literature. Section 3 outlines the methodology. Section 4 presents and discusses the empirical results. Section 5 concludes this paper and offers policy implications.

2. Literature Review and Theoretical Considerations

2.1. Literature Review

Innovation is commonly defined as the creation, improvement, or replacement of a process, product, or service. The International Organization of Standardization (ISO) (2019) describes innovation as a new or changed item that creates or redirects value, while the Integrated Innovation Operating System (ITONICS) defines innovation as the transformation of an idea into a product or service that solves a specific problem and delivers value to both the organization and the consumer. Innovation can also improve productivity, as the same input can produce a larger output. Owing to increased productivity, imports and exports of goods and services expand, thereby enhancing the economy. Innovation has become a key component of sustainable development by providing tools for the safe and socially responsible development of new products, services, and processes, as outlined by the United Nations (2022). Mauri (1983) defined innovation as a driver of financial development, which he regarded as a precondition for economic growth. Financial development was discussed as a qualitative change in the structure and performance of the financial system. In a similar vein, King and Levine (1993) used financial systems to justify economic expansion. Accordingly, the ideas of innovation and financial systems proposed by Schumpeter (1934) were reinforced by Mauri (1983) and King and Levine (1993), providing a strong argument in support of the role of innovation in economic growth. Innovation may also be described as the emergence of new financial instruments and services, as well as new organizational forms, resulting in more sophisticated and complex financial markets (Mishra, 2008). Tahir et al. (2018) defined financial innovation as the introduction of new financial instruments in financial institutions and markets, as well as new technologies, encompassing process, product, and institutional innovation. Qamruzzaman and Wei (2017) found that financial innovation promotes economic growth by contributing to financial inclusion and the international mobility of goods and services. Financial innovation can further stimulate economic growth through enhanced savings and capital accumulation, as explained by Mishra (2008). Consequently, financial innovation in newly established market economies is welfare-enhancing. Previous studies indicate that financial innovation facilitates efficient payment systems that enable real-time financial transactions and support the safe and effective movement of goods and services.

Many empirical studies have established a relationship between financial innovation and economic growth using cross-sectional or panel data. However, the literature addresses a wide range of macroeconomic variables and perspectives when examining financial innovation and economic growth. This section examines the most significant studies and considers their findings. What remains unclear, however, is the nature of the relationship between financial innovation, financial development, gross capital formation, and government expenditure within the economic growth mechanism. Although this area has been widely explored, the literature provides little empirical evidence on these interrelationships (e.g., Adesete et al. 2020; Bara et al., 2016; Bara & Mudzingiri, 2016; Mwinzi, 2014). Nevertheless, there is broad agreement that financial innovation enhances economic growth by facilitating capital mobilization, financial intermediation, capital formation, and financial system development (Laeven et al., 2015; Nazir et al., 2020). This effect is further strengthened by financial deepening, which expands access to finance for small and medium-sized enterprises (SMEs), a key driver of transformative and sustainable growth and social development in developing countries, particularly in Africa (Moreira, 2016; Sanga & Aziakpono, 2022).

Kirsten (2018) investigated the relationship between economic growth and financial innovation in twenty-five African nations using fixed- and random-effects models. The results obtained using standard ordinary least squares (OLS) estimation were compared across countries to complete the analysis. This was achieved by evaluating three indicators of banking financial innovation, including growth in credit to the private sector, the ratio of broad and narrow money to the economy, and mobile money penetration. Financial innovation was also measured using indicators such as automated teller machines, mobile finance accounts, mobile money agents, and remote banking transactions. The findings indicated that economic growth is positively influenced by mobile financial innovation, particularly in countries with low levels of financial development, while non-mobile-related proxy variables were associated with negative effects.

Similarly, Manu et al. (2020) examined the dynamic nexus between financial development and economic growth using panel data from African countries covering the period from 1980 to 2017. The study employed the panel quantile regression and panel vector autoregression models. The empirical results confirmed a strong correlation between financial development and economic growth. The findings further indicated that per capita GDP growth is Granger-caused by foreign direct investment (FDI). The authors also observed that reforms of the financial system and the acceleration of structural transformation are dynamic strategies for ensuring sustainable economic growth in developing African economies.

Yinusa et al. (2021) examined the influence of financial innovation processes on the economic growth of selected African countries. The study utilized cross-sectional annual panel data for seventeen countries obtained from the World Bank Development Indicators, covering the period from 2004 to 2018. Panel data analysis was conducted using the generalized method of moments (GMM). The findings showed that financial innovation plays a significant role in shaping economic growth in the selected countries. Automated teller machines (ATMs) were identified as one of the most crucial financial innovation processes, with a positive and statistically significant effect on economic growth, whereas growth in the number of bank branches did not have a significant impact. Financial innovation products, including domestic bank credit, were also found to contribute significantly to economic growth in African countries.

Using two-step system GMM estimation, Bekele and Degu (2021) examined the contribution of financial sector development to economic growth in twenty-five SSA countries in the period between 2010 and 2017. Financial sector development was measured in terms of depth, access, and efficiency, proxied by credit to the private sector as a share of GDP, the number of bank branches per 100,000 adults, and return on assets, respectively. The findings showed a positive and statistically significant impact of financial sector depth, access, and efficiency on economic growth in the selected countries. Domeher et al. (2022) investigated the role of financial inclusion using secondary data for twenty-six SSA countries from 2004 to 2017, employing the GMM estimation technique. The findings indicated that financial inclusion is driven by investment in innovation within the banking industry. Moreover, the relationship between innovation and economic growth was found to be fully mediated by financial inclusion. The study therefore suggests that governments in the subregion should invest in relevant technological infrastructure that the banking sector can leverage to enhance financial inclusion and promote economic development.

Utilizing cross-sectional data from ninety-two countries covering the period 2002–2020, Naeem et al. (2023) analyzed the influence of financial innovation on economic growth in both developing and developed countries. The authors employed a fixed-effects panel regression technique to examine financial innovation dependence. They compared the results and demonstrated that financial innovation may have a negative effect on economic growth, which is undesirable. Accordingly, the study recommends that policymakers should both stimulate beneficial forms of financial innovation and deter those that adversely affect the economy. Jungo et al. (2023) examined the drivers of financial innovation to determine whether it exerts a positive effect on economic development in the presence of corruption in the SSA region. The results showed that competitiveness and crises in the banking sector (financial innovation), the need to diversify deposits and bank loans (financial inclusion), and greater household access to the internet and investment are key drivers of financial innovation in SSA countries. In addition, the study found that corruption negatively moderates the effect of the financial system on economic growth, whereas declines in corruption promote economic growth. This observation highlights the need for policymakers to recognize that financial innovation alone cannot generate substantial economic growth in developing nations without adequate control and curtailment of corruption.

Mlambo (2024) examined whether financial development is correlated with economic growth in low-income countries within the Southern African Development Community (SADC) region. The results of the quantitative panel data analysis indicated a positive relationship. The study further established that the relationship is causal, as economic development is not merely a product of financial development but is also driven by broader financial system development. This finding is consistent with that of Olunuga and Ashoghon (2024), who examined the effects of financial innovation on economic growth and sustainable development in Nigeria using time-series data. The authors applied statistical techniques to analyze secondary data on financial innovation. The results confirmed that financial innovation has a significant influence on economic growth and sustainable development in the Nigerian economy, with financial innovation variables exhibiting positive effects.

Diop et al. (2025) quantified the dynamics of the relationship between banks and stock markets in SSA countries and examined their effects on economic growth. The findings showed that bank growth can be detrimental to stock market expansion. The study further demonstrated that government intervention in stock market development, even at the expense of banks, may be necessary to promote economic growth.

2.2. Theoretical Mechanism of the Financial Innovation Transmission Channel in SSA Countries

From a theoretical perspective, it is a well-established fact that the level of financial deepening impacts economic growth. An increasingly deep financial system, in which information asymmetries are reduced, risks are assessed and managed, and contractual obligations are honored, can positively propel growth by allocating resources efficiently through a self-correcting mechanism. The theory of economic development posits that economic development is spurred by innovation within financial intermediaries (Mishra, 2008). However, several studies have ignored the role of financial innovation in economic growth and instead suggest that the financial system is an endogenous variable (Michalopoulos et al., 2009). Other studies have attempted to relate financial innovation to money demand (Kasekende & Opondo, 2003) and to savings (Ansong et al., 2011; Levine, 1997).

Financial innovation in SSA is predominantly digital and inclusion-oriented, and its growth effects operate through four main channels. First, digital financial services reduce information asymmetry by generating transaction histories, thereby easing credit constraints for households and small firms. Second, mobile savings, transfers, and insurance improve risk sharing and consumption smoothing in high-risk environments (Asongu & De Moor, 2017). Third, digital payments enhance payment efficiency by lowering transaction costs and facilitating market integration, particularly in informal and rural sectors (Asongu, 2015). Fourth, financial innovation mobilizes savings and expands financial inclusion by reducing geographic and cost barriers to formal finance. In this context, mobile phone subscriptions provide a theoretically consistent proxy for financial innovation, as mobile connectivity is a prerequisite for mobile money and branchless banking systems in SSA. Financial innovation thus affects growth in SSA by reducing information asymmetry, improving risk sharing, enhancing payment efficiency, and mobilizing savings through digital inclusion, which justifies the use of mobile phone penetration as a proxy.

3. Methodology

3.1. Analytical Framework

The panel data utilized to identify the indicators of financial development were used in this analysis to determine the predictive role of financial innovation in improving economic performance from a financial perspective. The use of panel data has both merits and limitations compared with cross-sectional and time-series data, as it integrates multiple data dimensions. The main drawback of panel data is the difficulty of comparing values of different units of a particular variable simultaneously. Ibrahim and Bashir (2019) assert that panel data analysis reduces multicollinearity among variables, increases degrees of freedom and test power, improves confidence in the obtained results, and enables the analysis of structural adjustment dynamics, in comparison with time-series and cross-sectional analytical frameworks.

The basic model used in this study is specified as follows:

GDP = f(BCP, MCS, CF, GEX, M2)

(1)

where

GDP = GDP per capita growth (annual %);

BCP = domestic credit to the private sector by banks/GDP;

MCS = mobile cellular subscriptions;

CF = gross capital formation/GDP;

GEX = final expenditure of the general government;

M2 = broad money.

These variables were log-transformed to improve the quality of the analysis. Because the variances are unstable, log transformation was applied to normalize the data and prepare it for regression analysis. This transformation reduces the impact of outliers and nonlinearity and enables easier interpretation of relationships between variables. Logarithmic transformation improves normality and enhances the linearity between dependent and independent variables. It also increases the accuracy and generalizability of the measurements. Ogun (2021) further observed that logarithmic transformation is necessary to ensure the impartiality of point estimates in predicting the dependent variable.

The above model was estimated using the autoregressive distributed lag (ARDL) cointegration approach to examine the long-run relationships among the specified variables. The ARDL model employed in this study was initially developed by Pesaran and Shin (1998). A major strength of the ARDL approach is its flexibility, as it can be applied when variables are integrated at different orders. ARDL also performs well in the presence of endogeneity. Moreover, ARDL allows each variable to be treated as either independent or dependent within the equation, incorporates time lags to improve explanatory power, enables causal inference, and identifies policy-relevant variables. On this basis, ARDL was selected as the appropriate framework for modeling economic growth and financial innovation in the present paper.

The unrestricted error correction model (ECM) specification is expressed as follows:

ΔLGDP = β₀ + β₁LBCP_t−1 + β₂LMCS_t−1 + β₃LCF_t−1 + β₄LGEX_t−1 + β₅LM2_t−1 + ∑αiΔGDP_{t−1 +} ∑γiΔLBCP_t−1 + ∑δiΔLMCS_t−1 + ∑μiΔLCF_t−1 + ∑σiΔLGEX_t−1 + ∑θiΔLM2_t−1 + ε_t

(2)

The short-run estimation equation is as follows:

ΔLGDP_t = γ₀ + ∑γ₁LBCP_t−1 + ∑γ₂LMCS_t−1 + ∑γ₃LCF_t−1 + ∑γ₄LGEX_t−1 + ∑γ₅LM2_t−1 + ε_t

(3)

To estimate the factors influencing GDP, we must incorporate short-run dynamics into the long-run model. The long-run estimation equation is expressed as follows:

ΔLGDP_t = β + ∑αiΔLGDP_t−1 + ∑γiΔLBCP_t−1 + ∑δiΔLMCS_t−1 + ∑μiΔLCF_t−1 + ∑σiΔLGEX_t−1 + ∑θiΔLM2_t−1 + φECM_t−1 + ε_t

(4)

Empirical research using panel data models offers several advantages for examining the relationship between economic growth and macroeconomic variables. For example, panel data facilitate the incorporation of cross-sectional country characteristics and capture dynamic interactions between economic growth and macroeconomic variables. Estimation is also more efficient because the number of observations increases and additional degrees of freedom are obtained. Three estimation techniques were employed for panel ARDL estimation: pooled mean group (PMG), mean group (MG), and dynamic fixed effects (DFE). PMG estimates short-run dynamics while constraining long-run parameters, adjustment speed, and error variance. MG imposes fewer restrictions and allows for parameter heterogeneity (Pesaran & Smith, 1995). Both PMG and MG use the Akaike information criterion (AIC) to determine appropriate lag lengths. DFE assumes homogeneous long-run coefficients across countries and constrains both adjustment speeds and short-run coefficients to be identical across panels.

3.2. Variable Definitions and Measurements

The definitions, measurements, and data sources of the variables are provided in

Table 1. Description and measurement of variables.

Variable	Symbol	Expected Sign
Gross domestic product (GDP) per capita growth (annual %)	GDP	N/A
Domestic credit to private sector by banks (% of GDP)	BCP	Positive
Mobile cellular subscriptions (per 100 people) (proxy for financial innovation)	MCS	Positive
Gross capital formation (% of GDP)	CF	Positive
General government final consumption expenditure (% of GDP)	GEX	Positive
Broad money (% of GDP) (proxy for financial deepening)	M2	Positive

. The sample consists of annual data for 14 SSA countries: Benin, Burkina Faso, Cameroon, Chad, Côte d’Ivoire, Ethiopia, Gabon, Ghana, Kenya, Senegal, Sierra Leone, South Africa, Sudan, and Zimbabwe, covering the period from 1995 to 2023. The starting year was chosen because certain explanatory variables, such as mobile cellular subscriptions and household credit provided by banks, are unavailable prior to 1995. The use of panel estimation in this study allows for the control of individual heterogeneity, facilitates the identification of unobservable variables, and provides a more robust basis for the plausibility of the estimations.

3.3. Justification for Selecting 14 Sub-Saharan African Countries

The selection of 14 SSA countries was based on theoretical, empirical, and methodological considerations, as outlined below.

3.3.1. Data Availability and Consistency

This study covers the period 1995–2023, which coincides with the rapid expansion of financial liberalization and mobile-based financial services in SSA. Only 14 countries possessed sufficiently complete and consistent time-series data for the key variables required for panel ARDL estimation. Panel ARDL techniques require balanced or moderately unbalanced panels with relatively long time dimensions, which constrained the country selection (Bashir & Ibrahim, 2022).

3.3.2. Financial Innovation Heterogeneity

The selected SSA countries represent varying levels of financial innovation and financial sector development, ranging from early adopters of mobile financial services to countries with more traditional banking systems. This heterogeneity enables this study to capture cross-country differences in the impact of financial innovation on economic growth while maintaining econometric comparability.

3.3.3. Methodological Suitability for Panel ARDL

Panel ARDL models, particularly the PMG estimator, require a moderate number of cross-sectional units (N) and a large time dimension (T). The selection of 14 countries satisfies these requirements, ensuring reliable estimation of both short-run dynamics and long-run relationships while avoiding bias associated with large-N, small-T panels.

3.3.4. Policy Relevance and Regional Representation

The selected countries represent economies from various sub-regions of SSA (West, East, and Southern Africa), thereby enhancing the policy relevance of the findings. This regional representation allows policymakers to draw broader conclusions regarding the effectiveness of financial innovation and financial deepening strategies across SSA.

3.3.5. Cointegration and Structural Similarities

SSA countries share common structural characteristics, such as weak financial markets, limited access to credit, and reliance on bank-based financial systems, which justify their joint analysis within a panel framework. At the same time, country-specific cointegration results enable the identification of heterogeneous long-run relationships, as observed in Senegal, Côte d’Ivoire, Ethiopia, and Zimbabwe.

3.4. Research Questions

To examine the impact of financial innovation and financial deepening on economic growth in SSA countries, this study aims to answer the following research questions (RQs):

• RQ1: What is the relationship between financial innovation and economic growth in SSA countries over the period 1995–2023?
• RQ2: How does financial deepening affect economic growth in SSA?
• RQ3: Does mobile technology-based financial innovation promote economic growth in SSA?
• RQ4: Does gross capital formation affect economic growth in SSA?
• RQ5: What is the effect of government expenditure on economic growth in SSA?
• RQ6: Do the selected SSA countries exhibit a stable long-run equilibrium relationship between financial innovation, financial development, and economic growth?

3.5. Research Hypotheses

Based on prior theoretical and empirical literature, this study proposes the following research hypotheses:

H1. 

Financial innovation has a significant positive impact on economic growth in SSA countries.

H2. 

Financial deepening has a significant positive impact on economic growth in SSA countries.

H3. 

Domestic credit has a significant positive impact on economic growth in SSA countries.

H4. 

Broad money supply has a significant positive impact on economic growth in SSA countries.

H5. 

Capital formation has a significant positive impact on economic growth in SSA countries.

H6. 

Government expenditure has a significant positive impact on economic growth in SSA countries.

4. Results and Discussion

4.1. Preliminary Analysis

The descriptive statistics provide a summary of the central tendencies, dispersion, and distribution of the variables. As shown in

Table 2. Descriptive statistics.

Statistics	GDP	BCP	CF	GEX	M2	MCS
Mean	1851.59	1.084	1.266	1.053	1.474	1.15
Median	1184.77	1.096	1.287	1.072	1.368	1.682
Max	8949.12	1.925	1.779	1.401	4.142	2.268
Min	241.526	−0.006	−0.533	0.311	0.858	−5.551
Std. Dev.	1915.35	0.333	0.228	0.18	0.536	1.173
Skewness	2.064	−0.079	−2.774	−0.6	3.815	−1.627
Kurtosis	5.957	3.63	17.737	3.262	18.7	5.892
Jarque–Bera	436.304	7.149	4184.15	25.555	5040.24	314.439

, the mean GDP per capita growth was 1851.592, and the standard deviation (1915.347) was large, suggesting substantial variation across the 14 countries during the period of analysis. The log-transformed variables (BCP, CF, GEX, M2, and MCS) exhibited varying degrees of dispersion. Notably, M2 had the highest standard deviation (0.536), while GEX had the lowest (0.18). Most variables were non-normally distributed, as indicated by skewness and kurtosis values. This was confirmed by the Jarque–Berra test, with p-values significant at the 0.001 level, indicating the presence of nonlinearity and potential outliers that require careful attention in econometric modeling.

Table 3. Correlation matrix.

Variable	GDP	BCP	CF	GEX	M2	MCS
GDP	1
BCP	0.338	1
CF	0.089	0.286	1
GEX	0.307	0.478	−0.027	1
M2	−0.005	−0.078	−0.07	0.03	1
MCS	0.253	0.334	0.059	0.066	0.134	1

demonstrates the correlations among the variables used in this study. The results illustrate the interrelationships between GDP per capita growth and the explanatory variables. Moderate positive correlations were observed between GDP growth and BCP (0.338), GEX (0.307), and MCS (0.253), suggesting that higher levels of domestic credit, government expenditure, and mobile subscriptions are associated with stronger economic growth. By contrast, M2 and CF exhibited weak negative and weak positive correlations, respectively, indicating low-order relationships. The correlation between BCP and GEX was relatively high (0.478); however, potential multicollinearity concerns were addressed through diagnostic tests and appropriate model specifications.

Table 4. Panel unit root test results.

Variable	LLC	Im, PS	ADF-Fisher	PP-Fisher	Order of Integration
GDP	−8.85425 ***	−8.56748 ***	139.422 ***	140.209 ***	I(1)
MCS	−16.1410 ***	−13.7394 ***	144.757 ***	250.976 ***	I(0)
M2	−15.9650 ***	−14.9009 ***	225.631 ***	243.708 ***	I(1)
GEX	−16.9589 ***	−16.4107 ***	247.859 ***	266.444 ***	I(1)
CF	−1.7126 **	−2.22945 ***	57.2397 ***	45.2211 **	I(0)
BCP	−9.4254 ***	−10.6385 ***	158.800 ***	247.471 ***	I(1)

Notes: ** and *** indicate statistical significance at the 5% and 1% levels, respectively.

reports the stationarity test results, which indicate that GDP, BCP, GEX, and M2 were integrated of order one (I(1)) and therefore required first differencing to achieve stationarity (e.g., D(GDP)). In contrast, MCS and CF were stationary at level I(0). These findings support the application of the ARDL model, as it accommodates variables with mixed orders of integration and allows for the estimation of both short-run and long-run dynamics.

Table 5. Lag length selection criteria.

Lag	LogL	LR	FPE	AIC	SC	HQ
0	−2630.4		8.56	19.174	19.253	19.206
1	−163.808	4807.614	1.80	1.497	2.049 *	1.718 *
2	−126.049	71.949	1.78	1.484	2.51	1.896
3	−83.32	79.553	1.69	1.435	2.934	2.037
4	−31.529 *	94.166	1.51 *	1.32 *	3.293	2.112
5	−11.284	35.926	1.70	1.435	3.881	2.417
6	23.415	60.06	1.72	1.444	4.364	2.616
7	54.25	52.027	1.80	1.482	4.875	2.844
8	88.185	55.776 *	1.84	1.497	5.363	3.049

Notes: * indicates the lag order selected by the criterion. LR: sequential modified LR test statistic (each test at 5% level), FPE: Final prediction error, AIC: Akaike information criterion, SC: Schwarz information criterion, HQ: Hannan-Quinn information criterion.

shows the optimal lag length selection based on several criteria, including LogL, LR, FPE, AIC, SC, and HQ. The SC and HQ criteria selected one lag, whereas LogL, FPE, and AIC selected four lags. Given the dynamic nature of the panel data and the need to balance model complexity with explanatory power, the choice of four lags, as indicated by the AIC, is appropriate for capturing dynamic relationships without overfitting the model.

4.2. Hausman Test Results

In this section, the estimates obtained from the MG, PMG, and DFE models are compared. Based on the consistency and efficiency properties of the estimators, the Hausman test was employed to assess the adequacy of the PMG estimator relative to the MG and DFE estimators, as reported in

Table 6. Hausman test results for PMG versus MG and DFE estimators.

Comparison	Chi-Square Statistic	Degrees of Freedom (df)	p-Value	Decision
PMG vs. MG	347.376	5	<0.001	Reject Null; MG preferred
PMG vs. DFE	72.013	5	<0.001	Reject Null; DFE preferred

,

Table 7. Coefficient differences between MG and PMG estimators.

Variable	MG Coefficient	PMG Coefficient	Var (Diff.)	p-Value
LOG_BCP	−1309.45	4.165	6,071,049.46	0.594
LOG_CF	3479.061	−199.397	7,697,532.00	0.185
LOG_GEX	−4168.82	1117.986	25,055,161.46	0.291
LOG_M2	−813.66	79.556	3,892,839.29	0.651
LOG_MCS	143.664	328.139	40,387.18	0.359

Note: Dependent variable is LOG_GDP.

and

Table 8. Coefficient differences between DFE and PMG estimators.

Variable	DFE Coefficient	PMG Coefficient	Var (Diff.)	p-Value
LOG_BCP	−4.767	4.165	128,596.19	0.98
LOG_CF	1395.314	−199.397	155,439.11	0.001
LOG_GEX	−567.868	1117.986	273,945.99	0.001
LOG_M2	−34.637	79.556	21,060.30	0.431
LOG_MCS	87.455	328.139	2974.13	0.001

Note: Dependent variable is LOG_GDP.

.

The estimates of the PMG, MG, and DFE models were compared using the Hausman test to determine the most appropriate model. The null hypotheses of PMG vs. MG (chi-square = 347.376, p = 0.001) and PMG vs. DFE (chi-square = 72.013, p = 0.001) were rejected, indicating that both the MG and DFE estimators outperformed the PMG estimator. However, the differences between the coefficients of the MG and PMG estimators were not statistically significant (p-values exceeding 0.05), suggesting that the long-run coefficients were broadly similar across these estimators. By contrast, the comparison between the DFE and PMG estimators revealed statistically significant heterogeneity across countries for the variables LOG_CF, LOG_GEX, and LOG_MCS (p < 0.001), indicating that the DFE estimator captures cross- country heterogeneity more effectively.

Given these results, the DFE model was selected to estimate the relationships of interest, as it allows for country-specific fixed effects and does not impose homogeneity of long-run coefficients across countries. Although the MG estimator also provides consistent long-run estimates, it is unsuitable in this context due to the relatively large standard errors, which limit its ability to adequately capture heterogeneity across countries.

4.3. ARDL Estimation

With regard to the implications of the PMG results,

Table 9. Long-run effects in the PMG model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
LOG_BCP	4.165	144.65	0.029	0.977
LOG_CF	−199.397	105.06	−1.898	0.058
LOG_GEX	1117.99	242.87	4.603	<0.001
LOG_M2	79.556	37.206	2.138	0.033
LOG_MCS	328.139	52.838	6.21	<0.001

shows that the PMG model reported positive long-run effects of government expenditure and mobile cellular subscriptions on GDP per capita growth. Specifically, the coefficients for LOG_GEX (1117.99, p < 0.001) and LOG_MCS (328.139, p < 0.001) were positive and statistically significant. These findings are similar to those of Amaghionyeodiwe and Annansingh-Jamieson (2017), who established that mobile technology can generate and accelerate economic growth in these countries. They further argued that the diffusion of new technology has the potential to enhance growth rates.

By contrast, these findings do not align with those of Naceur et al. (2014), who reported that government consumption correlates negatively with financial sector development. They argued that higher government consumption reduces domestic liquidity, thereby constraining financial intermediation and limiting the mobilization of funds needed to finance private sector activity. The present results also differ from those of Maghfiroh and Purwono (2021), who found that increasing government spending may produce adverse effects in developing economies. Specifically, they noted that excessive public spending discourages private consumption, weakens economic incentives in the short term, and harms long-term capital accumulation. Moreover, volatility in government spending was shown to amplify fluctuations in macroeconomic variables.

The results in Table 9 show that LOGM 2 (79.556, p = 0.033) had a positive but relatively weak effect on GDP per capita growth. This suggests that financial deepening contributes to economic growth, although the magnitude of the effect is limited. The results indicate a moderate correlation between economic growth and financial development. By contrast, LOG_BCP was not statistically significant (4.165, p = 0.977), while the coefficient of LOG_CF was negative (−199.397, p = 0.058), contrary to expectations.

According to

Table 10. Short-run effects in the PMG model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
COINTEQ	−0.053	0.028	−1.884	0.06
D(LOG_BCP)	117.312	72.908	1.609	0.108
D(LOG_CF)	267.967	188.053	1.425	0.155
D(LOG_GEX)	138.797	257.241	0.54	0.59
D(LOG_M2)	36.787	177.058	0.208	0.836
D(LOG_MCS)	−33.92	32.335	−1.049	0.295
Constant	57.188	41.543	1.377	0.169

Note: Prefix “D” is first difference, and “LOG” is the logarithm.

, in the short run, the error correction term (COINTEQ = −0.053, p = 0.06) was not statistically significant, indicating a slow adjustment toward long-run equilibrium. None of the differenced variables (D(LOG_BCP), D(LOG_CF), D(LOG_GEX), D(LOG_M2), and D(LOG_MCS)) were significant (p > 0.05), suggesting that short-run dynamics are weak within the PMG framework.

Based on

Table 11. Long-run effects in the MG model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
LOG_BCP	−1309.45	2468.192	−0.531	0.596
LOG_CF	3479.061	2776.431	1.253	0.211
LOG_GEX	−4168.82	5011.402	−0.832	0.406
LOG_M2	−813.66	1973.379	−0.412	0.68
LOG_MCS	143.664	207.796	0.691	0.49

, which reports the MG model results, none of the long-run coefficients were statistically significant (p- values greater than 0.05), and the standard errors were large, indicating substantial cross-country heterogeneity. This reduces the effectiveness of the MG estimator for this analysis, as it does not provide reliable estimates under such conditions.

As shown in

Table 12. Short-run effects in the MG model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
COINTEQ	−0.232	0.067	−3.448	<0.001
D(LOG_BCP)	157.318	171.42	0.918	0.359
D(LOG_CF)	170.039	212.974	0.798	0.425
D(LOG_GEX)	216.886	348.515	0.622	0.534
D(LOG_M2)	−74.789	310.138	−0.241	0.81
D(LOG_MCS)	−24.673	64.454	−0.383	0.702
Constant (C)	293.742	357.944	0.821	0.412

, the error correction term (COINTEQ = −0.232, p < 0.001) (Table 12) was highly significant, indicating faster adjustment to equilibrium compared with the PMG model. However, the short-run coefficients were not statistically significant, similar to the PMG results.

In the DFE model, the only significant positive long-run determinant of GDP per capita growth was LOG_CF (1395.314, p < 0.001). This result is consistent with theoretical expectations, as capital formation is a key driver of economic growth. It indicates that capital formation is a major contributor to the economic development of SSA countries. This finding aligns with the theoretical framework and with empirical evidence from Qamruzzaman and Wei (2017), Egert and Jawadi (2018), and Domeher et al. (2022), as well as with the ex post expectations of this study. It also supports the broader literature suggesting that financial innovation promotes economic growth by facilitating capital mobilization, improving financial intermediation, and supporting capital formation and financial system development (Laeven et al., 2015; Nazir et al., 2020). As shown in

Table 13. Long-run effects in the DFE model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
LOG_BCP	−4.767	386.677	−0.012	0.99
LOG_CF	1395.314	408.015	3.42	<0.001
LOG_GEX	−567.868	577.001	−0.984	0.326
LOG_M2	−34.637	149.815	−0.231	0.817
LOG_MCS	87.455	75.934	1.152	0.25

, LOG_BCP (−4.767, p = 0.99), LOG_GEX (−567.868, p = 0.326), LOG_M2 (−34.637, p = 0.817), and LOG_MCS (87.455, p = 0.25) were not statistically significant, indicating weak long-run effects in the DFE framework.

The error correction term in the DFE model (COINTEQ = −0.084, p < 0.001) (

Table 14. Short-run effects in the DFE model.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
COINTEQ	−0.084	0.015	−5.526	0.001
D(LOG_BCP)	137.383	58.43	2.351	0.019
D(LOG_CF)	−41.558	40.066	−1.037	0.300
D(LOG_GEX)	77.883	65.922	1.181	0.238
D(LOG_M2)	40.616	23.254	1.747	0.082
D(LOG_MCS)	−9.336	23.796	−0.392	0.695
Constant (C)	66.457	65.921	1.008	0.314

) was statistically significant, indicating a moderate speed of adjustment toward equilibrium. In the short run, D(LOG_BCP) (137.383, p = 0.019) was statistically significant, suggesting that GDP growth responds positively to increases in domestic credit. This result is consistent with theoretical expectations and prior empirical findings by Qamruzzaman and Wei (2018) and Yinusa et al. (2021), as well as with the a priori expectations of this study. Other short-run coefficients were not statistically significant.

Given the observed relationships between economic growth and the variables of government expenditure, domestic credit to the private sector, and gross capital formation, the following sections compare this study’s findings with those of similar studies, highlighting areas of agreement and disagreement.

4.3.1. Government Expenditure

The findings of this study are consistent with previous research, including Mohamud and Abdulle (2025) and Afonso and Jalles (2015), which found that public investment supports long-term economic growth. However, our findings differ from those of Barro (1991) and Fölster and Henrekson (2001), who reported negative growth effects when government expenditure is inefficient or largely consumption-oriented.

4.3.2. Domestic Bank Credit to the Private Sector

Our results show that domestic bank credit has a positive and statistically significant effect on economic growth. This finding is consistent with Naeem et al. (2023) and Al Mamun et al. (2023). However, it contrasts with Iheonu et al. (2020), who reported a positive but statistically insignificant effect on domestic investment. Overall, empirical evidence on the relationship between finance and growth varies across contexts and model specifications. While some studies identify significant effects, others report weak or insignificant relationships, suggesting that the role of private credit in promoting innovation and growth is not consistent across samples or specifications (Begum & Aziz, 2019).

4.3.3. Gross Capital Formation

Our results suggest that gross capital formation plays an important role in promoting financial development and technological progress by expanding financial markets and creating conditions conducive to financial innovation. This finding is consistent with Alshubiri (2022) and Kong et al. (2020).

4.4. Post-Estimation Diagnostic Tests

The results of the Breusch–Pagan LM test (807.5257, p = 0.00), Pesaran scaled LM test (53.1124, p = 0.00 ), and Pesaran CD test (10.47199, p = 0.00) are reported in

Table 15. Cross-sectional dependence tests.

Test	Statistic	Prob.
Breusch–Pagan LM	807.5257	0.000 ***
Pesaran scaled LM	53.11242	0.000 ***
Pesaran CD	10.47199	0.000 ***

Note: *** indicates a statistical significance level of 0.01.

and

Table 16. Heteroskedasticity test.

Statistic	Value
Likelihood Ratio	1164.87
Restricted LogL	−3456.98
Unrestricted LogL	−2874.55
Probability	6.17 × 10−240

. The results indicate strong cross-sectional dependence, i.e., shocks in one country may affect other countries, likely due to economic linkages within the region or the presence of common external shocks. These results justify the use of cross-sectional panel models. The heteroskedasticity test statistic (1164.87, p < 0.001) indicates substantial heteroskedasticity, indicating that the variance of the residuals is not constant across countries. Although this may affect estimator efficiency, the DFE model partially mitigates this issue by allowing for cross-country heterogeneity.

4.5. Country-Level Analysis

The results of the bounds test presented in

Table 17. Bounds testing for cointegration.

Country	Obs.	F-Statistic	Decision	t-Statistic	Decision	Cointegration Conclusion
Benin	28	4.55	Near upper bound	−1.803	Not significant	Inconclusive
Burkina Faso	28	1.519	Below I(0)	−2.432	Above I(0)	Weak evidence
Cameroon	23	0.648	Below I(0)	−0.994	Not significant	No cointegration
Chad	27	1.597	Below I(0)	−1.691	Not significant	No cointegration
Côte d’Ivoire	28	9.647	Above I(1)	−2.515	Borderline	Yes
Ethiopia	25	7.749	Above I(1)	−0.383	Not significant	Yes (F only)
Gabon	28	1.393	Below I(0)	−1.009	Not significant	No cointegration
Ghana	24	1.995	Below I(0)	2.308	Invalid (positive)	No cointegration
Kenya	28	0.831	Below I(0)	0.676	Invalid (positive)	No cointegration
Senegal	28	5.264	Above I(1)	−5.045	Below I(1)	Strong cointegration
Sierra Leone	26	2.795	Between bounds	−1.526	Not significant	Inconclusive
South Africa	28	3.683	Between bounds	−0.161	Not significant	Inconclusive
Sudan	27	2.476	Below I(0)	−1.192	Not significant	No cointegration
Zimbabwe	25	9.048	Above I(1)	−3.207	Below I(0)	Yes (F only)

indicate heterogeneous cointegration outcomes across countries. Evidence of long-run relationships was found for Senegal (F = 5.264, t = −5.045), Côte d’Ivoire (F = 9.647), Ethiopia (F = 7.749), and Zimbabwe (F = 9.048), whose variables were cointegrated.

The remaining countries exhibited no cointegration or inconclusive results, indicating that long-run relationships were not uniform across the panel. This heterogeneity supports the use of the DFE model with country-specific effects.

As shown in

Table 18. Error correction model results.

Country	ECM Statistic	Coefficient	Std. Error	t-Statistic	p-Value
Benin	COINTEQ	−0.003	0.017	−0.153	0.880
Burkina Faso	COINTEQ	−0.008	0.011	−0.734	0.471
Cameroon	COINTEQ	−0.002	0.010	−0.219	0.829
Chad	COINTEQ	−0.178	0.066	−2.691	0.014
Côte d’Ivoire	COINTEQ	0.032	0.044	0.714	0.483
Ethiopia	COINTEQ	−0.035	0.006	−5.570	0.000
Gabon	COINTEQ	−0.084	0.046	−1.827	0.082
Ghana	COINTEQ	−0.028	0.030	−0.939	0.361
Kenya	COINTEQ	0.047	0.031	1.498	0.149
Senegal	COINTEQ	0.008	0.039	0.207	0.838
Sierra Leone	COINTEQ	−0.266	0.074	−3.582	0.002
South Africa	COINTEQ	−0.106	0.081	−1.304	0.206
Sudan	COINTEQ	0.099	0.034	2.877	0.009
Zimbabwe	COINTEQ	−0.213	0.038	−5.624	0.000

, the ECM results captured short-run country-specific dynamics. Chad (−0.178, p = 0.014), Ethiopia (−0.035, p = 0.001), Sierra Leone (−0.266, p = 0.002), and Zimbabwe (−0.213, p = 0.001) exhibited statistically significant error correction terms (COINTEQ), indicating adjustment toward long-run equilibrium. Short-run coefficients varied widely across countries:

• Chad: D(LOG_M2) (−819.912, p < 0.001) showed a strong negative effect, indicating that increases in broad money reduce short-run GDP growth, possibly due to inflationary pressures. This interpretation is consistent with Bajrami et al. (2025), who documented similar inflationary effects of money supply on economic growth in Western Balkan countries.
• Ethiopia: D(LOG_GEX) (−41.594, p = 0.057) and D(LOG_MCS) (−25.068, p = 0.053) were marginally significant, indicating negative short-run effects.
• Ghana: D(LOG_MCS) (−205.820, p = 0.012) was negative and significant, suggesting that mobile subscriptions do not generate immediate economic growth.
• Zimbabwe: D(LOG_BCP) (259.672, p = 0.001) exhibited a strong positive effect, indicating that access to credit plays an important role in short-run economic growth.

5. Conclusions

The panel ARDL and ECM analyses indicate generally weak relationships between the macroeconomic variables and GDP per capita growth in the sample of fourteen SSA countries. The DFE model, selected based on the Hausman test, shows that domestic credit is a significant short-run determinant, while gross capital formation is a key long-run driver of economic growth. This study therefore concludes that financial innovation plays a significant role in influencing economic growth in SSA countries.

The country-specific ECM and bounds testing results reveal heterogeneity in cointegration relationships as well as in short-run dynamics, indicating the need for country-specific policy responses. Diagnostic tests also confirm the presence of cross-sectional dependence and heteroskedasticity, which further justifies the selection of the DFE model. These findings provide useful insights for policymakers seeking to promote sustainable economic growth in African economies. The long-run results from the DFE model emphasize the importance of gross capital formation as a major contributor to GDP growth, highlighting the role of investment in physical capital in economic development.

The limited effects of BCP, GEX, M2, and MCS suggest that these variables are context-dependent and vary across countries. The PMG results indicate that government spending and mobile cellular subscriptions positively influence long-run economic growth, implying that government spending and technological adoption are important drivers of growth. Consequently, governments in the SSA region should strengthen efforts to promote the further development of mobile technologies as a tool for enhancing regional economic growth.

Short-run analysis from the DFE model and country-specific ECM results, particularly for countries such as Zimbabwe, shows that financial development plays an important role in stimulating short-term economic growth. The negative short-run effects of M2 and MCS observed in some countries may reflect challenges such as inflationary pressures arising from monetary expansion and the delayed economic benefits of mobile technology adoption.

Policymakers should introduce additional incentives to attract foreign direct investment and improve the business environment. Expanding access to bank credit for the private sector is also essential for short-run growth, especially in countries such as Zimbabwe. The effective allocation of government expenditure remains important, as the PMG results suggest that public spending can exert a positive influence on growth when efficiently managed. Regarding monetary policy, policymakers should adopt prudent money supply management strategies to mitigate inflationary pressures, given that negative monetary effects appear to be short-lived but potentially disruptive if implemented poorly.

Although mobile subscriptions generate long-run economic benefits, these gains should be supported through policies that foster technological progress in business activities and financial systems. Governments should also consider reducing structural and regulatory barriers that may hinder technological diffusion. Future research should focus on country- or region-specific analyses, such as East, West, Central, and North Africa, rather than treating SSA as a homogeneous group, while accounting for differences in income levels, technological development, and institutional capacity.

Future studies should also examine the role of financial innovation in economic growth, with particular emphasis on the application of artificial intelligence in banking and financial services. Country-specific case studies would provide deeper insights into how financial innovation interacts with national economic structures and growth dynamics.