Digital Finance, Enterprise Formalization, and Firm Performance in Sub-Saharan Africa: Evidence from Mobile Money Expansion

2.1 Theoretical foundation

The Theory of Transaction Costs holds that high transaction costs make it impossible to conduct formal financial transactions, thereby increasing market entry barriers for small firms. The mobile money, including M-PESA in Kenya, has been indicated to reduce these expenses, improving the performance of micro, small and medium-sized enterprises (Dermish, Kneiding, Leishman, & Mas, 2011; Kim, Wang, Park, & Petalcorin, 2021). Mobile money enhances the security of payments and reduces the costs of transmitting, receiving, and storing money, thereby promoting formalised business practices and greater inclusion of more people in the market (Mensah, Setiadi, & Roessali, 2024; Munyegera & Matsumoto, 2017). Hypothetically, reducing the transaction costs will enhance the productivity and profitability of MSMEs, which may translate into formalisation (Idris, Ebeh, Sule, & Yelwa, 2024). MSMEs tend to be informal due to liquidity constraints, limited access to credit, and unpredictable cash flows. Mobile money provides a source of financial inclusion that is independent of conventional banks (Mohamed & Ahmed, 2022; Senyo, Osabutey, & Kan, 2020). Studies indicate that mobile money widens access to credit and eliminates financial exclusion. Mobile money enables companies to invest in resource development, generating expansion and formalisation by alleviating fiscal constraints (Bongomin, Yourougou, & Munene, 2019; Mhella, 2019).

The Institutional Theory explains why regulations and norms may affect the rate of formalisation. In most African regions, businesses have become informal due to unfavourable institutional conditions and costly regulations (Bongomin & Ntayi, 2020). Mobile money networks facilitate the entry of informal entrepreneurs into the formal financial system, thereby reducing informal economic activity. Digital payments enhance the integrity of MSMEs and reduce the cost of gaining access to formal markets, establishing trust and bridging institutionalities that disincentivise formalisation (Abiona & Koppensteiner, 2018); (Ene, Abba, & Fatokun, 2019).

Female entrepreneurs tend to have less collateral and weaker banking ties (Museba, Ranganai, & Gianfrate, 2021). Mobile money can help overcome these hurdles by making money readily accessible and enabling independent payments. FinTech provides mobile payment incident data to build alternative credit profiles, which benefits historically marginalised women-owned businesses (Bongomin et al., 2019). This technology assists women, particularly those who have to juggle home responsibilities or face safety concerns, by reducing financial barriers and potentially increasing the formalisation of their MSMEs.

The combination of these theories suggests that adopting mobile money can enhance MSME formalisation and performance. Reduced transaction costs lead to improved operational efficiency and market access; decreased liquidity enables greater investment; and adherence to formal standards is enhanced. These impacts can be gender-specific, as female-controlled companies receive the greatest benefits from access to more financial resources. These observations substantiate an excellent hypothesis that promoting the use of mobile money enhances the formalisation of MSMEs and their performance, while accounting for gender-specific limitations.

2.2 Literature review and hypothesis development

A. Mobile money and MSME formalisation

The emergence of mobile money (MM) is a fundamental tool for negotiating institutional barriers that hinder the formalisation of processes in Micro, Small and Medium Enterprises (MSMEs). Misguided bureaucracies, such as lengthy registration processes, inadequate infrastructure, and widespread lawlessness, recurrently hamper the growth of MSMEs in new economies. Such impediments hinder the availability of key financial services and record-keeping modules, thereby frustrating entrepreneurs’ attempts to formalise their organisations. On the other hand, MM significantly lowers the cost of entry, allowing informal firms assured access to transactional services and a consistent accounting infrastructure that can promote voluntary formalisation (S. Gupta & Rhyner, 2022). In empirical research, there is evidence that informal economic activity fell by 4.3 percentage points after the adoption of MM.

Furthermore, research shows that MM plays a critical role in economic reconstruction in post-conflict environments by establishing formal business setups (Mpofu, 2022),. Mobile financial records also improve access to credit because many companies have shifted towards formal operations by adopting greater financial transparency and credibility (Konté & Tetteh, 2022)&. However, the efficacy of MM is not limited to its inherent advantages; external factors, agent networks and favorable regulatory environments have a strong impact (Asongu et al., 2024). The problem of mobile ecosystem failures stems from a lack of interoperability and inadequate regulatory policies that prevent the scale needed to support large-scale financial inclusion. In turn, uniform regulatory frameworks will enhance operational capabilities and institutional trust, thereby enabling MM systems to operate productively and formalizing MSMEs. Therefore, MM represents an innovative business strategy, as it promotes the development of MSMEs through formalisation and the provision of cheaper financial services; however, its impact is highly dependent on regulatory and market infrastructure.

B. Mobile money and MSME performance

Mobile money (MM) has a significant impact on the performance of micro, small, and medium-sized enterprises (MSMEs) in several ways, including improvements in productivity, market access, financial inclusion, and opportunities for scaling and export. MM also makes MSMEs more effective in their operations, reducing the need for cash and simplifying the payment process, thereby shortening transaction queues and optimizing cash management (Mutiso & Reuben, 2021). MM adoption will enable better sales tracking and more accurate record-keeping, two of the most important dimensions fundamental to MSME growth. MM simplifies cash flow management, thereby enabling entrepreneurs to access formal sources of credit and obtain funds previously unavailable. MM has also been at the forefront of advancing financial inclusion, especially for marginalized businesspersons, such as women and informal sector players. Using fintech-generated credit scores based on MM information, lenders can extend credit to a broader group of business owners (Haq & Dawood, 2023). This democratized credit will help women-owned businesses increase their ability to access the capital needed to grow and expand their businesses (Kim et al., 2021). In inter-border transactions, digital payments reduce friction, particularly when trust in financial transactions is enhanced among partners, which is relevant to international trade (Bongomin & Ntayi, 2020). MSMEs are also enabled to venture into export markets; this is made possible through MM, though indirectly, thereby enhancing their growth and development into larger markets. Formalisation has ceased to be a regulatory prerequisite but has become a strategic asset that provides firms with a high bargaining chip in the international competitive environment. Altogether, the MM process improves operational efficiency, increases inclusiveness, and allows MSMEs to grow by expanding access to credit and markets, which are the fundamental drivers of MSME performance and business success in these environments.

C. Gender-specific constraints and the role of mobile money

The landscape of female entrepreneurs in MSMEs is largely shaped by unique constraints, especially in developing markets. Women also face significant barriers due to the lack of collateral and the absence of banking associations, which also take up a significant part of their free time, as their domestic role absorbs a significant portion of it; thus, access to financial services is a time-consuming exercise (Mndolwa & Alhassan, 2020). Also, prevailing gender ideologies may limit women’s mobility and economic autonomy, thereby hindering their ability to grow businesses (Kamanzi, 2022). Mobile money (MM) has become a viable equaliser. MM allows women to make independent financial choices without the influence of physical banking facilities and ensures their incomes remain intact, since it cannot be misused (Odunga & Kipketer, 2024). Empirical evidence shows that female-owned firms that utilize MM access tend to have higher sales rates, greater levels of formalization, and greater ability to grow their businesses (Asante, Ankrah, Agyei-Holmes, & Prah, 2024); (Warsame & Abdalla, 2024). This indicator not only reflects positive changes in the financial situation but also indicates an ongoing transition to an even more gender-equitable state in the MSME sector. However, MM does not equally enhance gender equality. Gender privileges that affect a continuous state might hinder the implementation and effective utilization of MM services, depriving women of the full benefit of these services. The benefits of MM might still be hidden in places where such norms remain strong, unless supported by institutions and programmes in female business and entrepreneurship. Thus, MM is a quite promising avenue to promote improved financial inclusion and empower women-owned MSMEs, but to achieve meaningful results, a consistent approach is required to address technological adoption and overcome the socio-cultural barriers inherent in the vast majority of communities.

2.3 Synthesis and gaps in the literature

Despite the growing interest in digital finance, significant empirical and conceptual gaps remain. Few studies have effectively triangulated mobile money (MM) usage, formalisation processes, and performance outcomes using robust causal inference methods, thereby limiting our understanding of these interrelationships. Additionally, most analyses neglect gender heterogeneity, often treating gender as a mere binary control variable rather than recognising its role as a moderating structural component. Furthermore, the interactions between MM regulation, prevailing gender norms, and digital access have not been comprehensively theorised, leaving a significant gap in the literature. This study aims to fill these gaps by quantifying the influence of MM on the formalisation and performance of micro, small, and medium enterprises (MSMEs), explicitly incorporating gender heterogeneity, and examining the potential mediating mechanisms underlying these dynamics.

3.1 Model specificity

This study examines the relationship between company performance and mobile money penetration (MMP) at the national level, focusing on formalisation, productivity, sales growth, and export participation (EP). The dependent variables (DVs) are derived from the World Bank Enterprise Surveys and include: (i) sales per worker in logarithms as a measure of productivity; (ii) a binary indicator of firm formalization (registered or possessing a tax identification number); (iii) the annual percentage change in sales to capture firm growth; and (iv) a binary indicator of export activity. Variables definition and procy measurement displayed in Table 1.

For robustness, alternative outcomes such as tax filing and bribery incidence were considered. The primary independent variable (IV) of interest is mobile money penetration, calculated at the country-year level as the percentage of individuals who have a mobile money account or use it regularly and lagged by one year. These data were obtained from the World Bank Global Findex database. To address potential endogeneity in MMP, mobile network coverage and regulatory reforms (such as interoperability legislation and agent network rules) were employed. These policy and infrastructural milestones capture exogenous changes in mobile money usage that firms cannot positively influence. Several constraints exist at both the corporate and national levels. Firm-level variables included size, age, ICT use, power interruption, and competitive intensity. Country-level covariates include inflation, GDP per capita (in logarithms), and governance metrics, such as government efficacy and the rule of law. To examine gender heterogeneity, female leadership was introduced as an independent moderator alongside the MMP. Formalisation was further explored as a mediator of the performance-to-MMP pathway. Standard errors are clustered at the nation-by-wave level, and all regressions incorporate fixed effects for sector, country, and survey year to account for unobserved variation. The estimation strategies employ several methods: (i) OLS and linear probability models for core specifications; (ii) logit average marginal effects for binary outcomes; (iii) difference-in-differences designs leveraging the timing of regulatory reform; (iv) 2SLS models utilizing regulatory and coverage instruments; (v) parametric bootstrap methods for mediation analysis; (vi) heterogeneity checks by productivity quantiles, firm size, and gender; and (vii) double machine learning to mitigate bias in high-dimensional settings. Consolidated data can be extracted from (Qamruzzaman, 2026).

3.2 Estimation strategy

Phase 01: Baseline (Core) Models

We begin with baseline regressions of formalisation and performance that control for observable firm and country characteristics and unobserved heterogeneity. For formalisation (an indicator of firm registration status), we estimate a linear probability model with interactions:

For performance, we use an analogous model for log sales per worker:

This includes the same controls and fixed effects, plus Formal. ${\beta }_{\mathit{ict}}$ To capture the formalisation channel. We estimate sales growth and export participation similarly (sales growth by OLS; exporter (binary) by logit with AMEs). All regressions cluster standard errors by country (or an appropriate cluster) to allow for correlated shocks. These core FE estimates capture the direct effect of MMP (and its female interaction) on formalisation and productivity, while flexibly controlling for covariates.

Phase 02: Mediation (MMP → Formal → Performance)

Next, we assess whether formalisation mediates the MMP effect on performance. We use a causal mediation framework. First, estimate the mediator equation:

Then, for each outcome

Both regressions include the same controls and fixed effects as above. Under the standard linear mediation logic, the indirect effect of MMP on Y through Formal is ${\xi }_{\mathit{ict}}$ . We compute this product and form bootstrap confidence intervals to test its significance (similarly to Preacher and Hayes or Imai et al. techniques. This product-of-coefficients approach generalises the classic Baron–Kenny decomposition. As a robustness check for the binary mediator, we also apply the Karlson–Holm–Breen (KHB) method via generalised SEM, which adjusts for differences in scale between logit and linear mediators. We report the indirect effect with confidence intervals.

Phase 03: Event-study Difference-in-Differences (Policy Timing)

We then exploit the timing of MMP adoption in each country with an event-study specification. Let T _c be the year of adoption of MMP in country c. For example, for formalisation, we estimate:

Phase 04: Instrumental Variables (Endogeneity of MMP)

To address possible endogeneity of MMP adoption, we use country-level instruments. In the first stage, we regress the lagged policy indicator on these instruments and controls:

Here, Reg is an indicator of key regulatory reforms (e.g., mobile-money interoperability or e-money mandates), and Coverage is the population share with 3G/4G network access. These capture exogenous variation in digital finance infrastructure and policy. We include the same fixed effects and time-varying controls. $W_{\mathit{ct}}$ as before, report the first-stage F-statistic to check instrument relevance.

In the second stage, we replace $\mathit{MM}{P}_{c,t-1}$ by its fitted value in the baseline equations (A) and (B), yielding 2SLS estimates of ${\beta }_1$ and $a_1$ . For inference, we supplement the usual t-tests with Anderson–Rubin (AR) robust tests. The AR statistic has an independent distribution from the first-stage coefficients, making it valid even when instruments are weak. In over identified models, we also report Hansen’s J (Sargan) statistic for over identifying restrictions. Standard errors remain clustered as above.

Phase 05: Double/Debiased Machine Learning (Bias Reduction)

Finally, we apply double/debiased machine learning (DML), which was familiarized by Chernozhukov (2018) to control for many covariates and flexibly reduce model-selection bias. We use Neyman-orthogonal moment conditions and cross-fitting: first, we regress $\mathit{MMP}\mathrm{c},\mathrm{t}-1$ on all controls ( $\mathrm{X}\mathit{ict},W_{\mathit{ct}}$ ) using LASSO or random forests, and similarly regress each outcome on the controls. We then form residuals from these regressions in held-out folds and regress the residualized outcome on the residualized MMP. This yields an estimator for ${\beta }_1$ and $a_1$ . That is $N$ -consistent and asymptotically normal under mild conditions. We report DML estimates and standard errors (via cross-fitting) and compare them to the FE-OLS and 2SLS estimates. If the results are similar, this bolsters confidence; large differences could indicate sensitivity to nonlinearities or confounding that DML helped mitigate.

Each phase is grounded in standard causal inference and microeconometric practice. We cite key references to justify our choices: for example, recent work shows the LPM with fixed effects is preferable to logit in panel data with rare outcomes; modern mediation analysis handles nonlinear mediators; the new staggered DiD estimators of Sun–Abraham and Callaway–Sant’Anna address the bias from heterogeneous timing; and the double-ML framework is detailed by Chernozhukov et al. (2018). We will report all coefficient estimates with robust confidence intervals and inference (e.g., bootstrap or heteroskedasticity-robust SEs) to support the validity of our conclusions.

3.3 Justification of estimation techniques

The empirical strategy employs multiple estimation methods to support credible inference. Linear probability models and OLS with fixed effects assess the impact of mobile money penetration on firm formalization and performance, while logit marginal effects validate the binary outcomes. These models control for firm characteristics, macroeconomic conditions, and unobserved heterogeneity through fixed effects, with errors clustered at the country-wave level. An event study difference-in-differences framework leverages staggered reforms in interoperability and agent regulation to identify policy-induced shifts, while controlling for parallel pre-trends and using country-specific linear trends for robustness. Endogeneity concerns are addressed by using regulatory milestones and mobile network coverage as instruments in a two-stage least squares framework. First-stage diagnostics establish validity, while OLS and instrumental-variable comparisons highlight measurement-error attenuation. Mediation analysis examines indirect channels of mobile money’s impact through formalization, using bootstrapped estimates and structural equation modelling for robustness with binary mediators. Double machine learning with cross-fitting extends robustness by partialing out high-dimensional controls using lasso and tree methods. Sensitivity analyses test alternative measures, placebo reforms, and omitted variable bias, providing consistent evidence linking digital finance to firm outcomes.

4.1 Baseline output

The estimates, shown in Table 2, indicate that mobile-money penetration (MMP) has a strong impact on firm behaviour and performance. A ten-percentage-point increase in MMP leads to a one-and-a-half percentage-point increase in the likelihood of formalization (business registration or tax identification), given an average formalization rate of 62 per cent. The advantage of female-led companies is disproportionately higher, with an additional 1.2-percentage-point effect, thus highlighting the impact of digital finance on reducing the reaction to formalization among women. MMP shows a positive response to performance metrics, with modest increases in sales per worker and sales growth. Formalization is closely associated with high performance, with labour productivity and sales growth 3.5 per cent higher and 2.2 percentage points higher, respectively, in formalised firms. They also have a higher likelihood of participating in exports. A 10 percentage point increase in MMP increases the probability of participating in exports by 0.6 per cent, and formalised companies are 0.5 per cent more likely to participate in exports. Despite the historical underperformance of female-led firms in productivity and growth, the positive relationship that exists between MMP and female leadership suggests that mobile money is slightly reducing the disparity. These results show that digital financial infrastructure fosters formalisation, productivity, and market access, thereby advancing SDG5, SDG8, and SDG9, and identifying policy levers to improve the performance of micro-, small-, and medium-sized enterprises (MSMEs).

4.2 Model estimation with Difference-in-Differences

The event-study specification (see Table 3) reveals a noticeable post-policy intervention by the mobile-money regulatory reforms on firms’ behaviour and performance. The 2.2 percentage points of formalization are increased by 2.2 percentage points in two years and by 5 percentage points in three years compared to the pre-policy baseline. The pre-treatment coefficients are negligible, and t=2.5 indicates they are not significant, confirming the parallel-trend assumptions and justifying the causal identification strategy. Productivity is a measure of sales per worker and also indicates a favourable response: it increases by 1.4 per cent over 2 years and 2.6 per cent over 3 years. Similar tendencies are supported by the growth regression: short-term (about 1 percentage point) and long-term (almost 2 percentage points) improvements in sales by firms occur at comparatively high rates among companies after the event. Participation in exports also increases, by 0.5 and 0.9 per cent in the short and medium term, respectively. These are economically substantial, considering that the baseline export share is 12 per cent. The interaction between the female and post-policy indicates that female-led companies will gain disproportionately more from formalisation, with a net effect of 1.8 percentage points. Although productivity and export returns are lower, they are statistically significant for female-led firms, demonstrating that regulatory reforms, along with digital finance, can overcome gender-specific obstacles to entry.

4.3 Instrumental variables (IV), mediation, and heterogeneity results

The baseline results (see Table 4) are checked using instrumental-variable (IV) estimation, thereby alleviating endogeneity in mobile-money penetration. Using tools based on interoperability laws, agent-network legislation, and mobile network coverage, the second-stage coefficients are positive and significant. A 10% point rise in the fitted penetration increases formalization by 2.5 percentage points, productivity by 2.2 percentage points, sales growth by 1.6 percentage points, and the probability of being an exporter by 0.7 percentage points. Specification F -statistics (first stage) are greater than 19, and this does not give concern about weak instruments. Anderson, Rubin, and Hansen-J tests fail to reject the null of instrument validity. The IV coefficients are higher than those from ordinary least squares, indicating that the attenuation bias in the baseline models is due to measurement error in the survey indicators. These findings suggest a strong local average treatment effect for firms affected by regulatory changes and immediate coverage adoption. The analysis of mediation suggests that an element of the performance increase is realized by formalization. The direct impact of mobile money on productivity is 1.1 per cent, and the indirect impact is 0.8 per cent mediated by formalization, thus suggesting that 42 per cent of the overall impact is mediated. The same ratios apply to sales growth (46%) and exporter probability (40%). The photographic evidence of significant statistical significance in both direct and indirect pathways, as interpreted by bootstrapped confidence intervals, supports the theoretical proposal that digital payments support registration, which in turn opens markets and credit to the poor. Heterogeneity tests witness the significance of distribution patterns. There is a stronger outcome from formalisation and exportation for women-led firms, and the mean marginal effect relative to male-led firms is almost twice as large; these effects are significant at customary levels. Splits by size show a greater impact on small and medium enterprises, but less on micro and large enterprises, indicating scale effects in achieving gains from digital finance. The productivity quantile regressions reveal larger changes in the upper part of the distribution: the coefficient increases by 0.9 percentiles at the 25th percentile, up to 2.2 percentiles at the 75th percentile. Such trends indicate that mobile-money penetration not only increases average performance but also raises the performance of larger producers.

The difference-in-differences staggered adoption findings (see Table 5) provide strong evidence that interoperability and agent-network rules associated with mobile-money regulatory reforms lead to a perceptible formalization of firms over time. All of these estimation results indicate consistent average treatment effects with the ATT between 0.018 and 0.021 among the Sun-Abraham and Callaway -Sant-Anna estimators. These effects imply that regulatory reforms increase the likelihood of firm formalization by about 1.8 to 2.1 percentage points compared to untreated firms in the same time period. Divergence among the estimators strengthens internal validity. When treatment timing is staggered, the methods reduce the bias inherent in the traditional two-way fixed-effects model.

The pre-trend tests enhance the credibility of the identifying assumptions. The pre-reform leads joint significance test results for both estimators yield p-values far beyond the customary thresholds (0.36, 0.41). This suggests that there was no statistically significant difference in the trends of formalization of treated and untreated firms in the run-up to the regulatory changes. The absence of such anticipatory effects or divergent dynamics before the reforms underlies the conclusion that the post-reform benefits can be explained by policy interventions rather than differences in pathways.

Dynamic treatment effects show one explanation of how the impact changes over time. The reform’s treatment effect is 0.014 a year after the reform, indicating an initial, but slight, improvement. In year three, this effect is 0.019, and in year four, it is 0.025. It indicates that the changes in the reform do not lead to immediate expansion of full-scale adoption and compliance; rather, formalization occurs progressively as mobile-money systems are integrated into local financial systems, agent networks grow, and interoperability removes the hassle of conducting digital transactions. The gradual growth is strongly consistent with the digital-finance literature, which highlights a diffusion lag in technology adoption and changing behavior among small firms.

These results support the hypothesis that regulatory reforms are catalytic institutional interventions that reduce barriers to entry into formal financial systems. On such reforms, by improving the reliability, accessibility, and reach of mobile-money platforms, they reduce the transaction costs of registration, tax compliance, and participation in formal markets. Its long-lasting and cumulative effects, even at t +3, further suggest that policy-improved mobile-money infrastructure is complementary to gradual firm learning, network effects, and market adaptation. The event-study and staggered DiD results substantiate the premise that the mobile-money regulatory changes have a meaningful impact on enterprise formalization, which contributes to the overall goals of financial inclusion and structural change in developing economies.

4.4 Robustness check with double machine learning (DML) methodology

High-dimensional control variables are included (see Table 6). The double-machine-learning (DML) method produces the same estimates while also accounting for HDI control variables. By partialing out and leveraging LASSO, the coefficient for mobile-money penetration (MMP) increases to 0.017, with a corresponding decline in standard errors, thereby providing evidence of noise attenuation. An analogous coefficient of 0.016 that a random forest specification produces proves that nonlinearities do not modify the core. Cross-fitting operations ensure that estimates are not biased by overfitting. The beneficial effect of MMP on productivity is strong even when the model is assigned other alternative specifications, confirming that the presence of small-scale confounders does not condition the relation. These findings highlight the importance of MMP as a determinant of formalisation, especially when other latent confounders are present. The DML framework also shows that empirical findings are robust across alternative model choices, proving the policy irrelevance of digital finance. The fact that OLS, LASSO, and tree-based methods converge indicates that the baseline fixed-effects specification effectively captures salient confounding variables.

The consolidated robustness analysis (see Table 7) provides a valid assessment of the fold in the expected effect of mobile-money penetration on the firm’s formalization and performance. The results show excellent convergence across alternative specifications, measurement techniques, estimators, and falsification tests, thus supporting the empirical basis of the research study. The robustness checks for the first category, that is, alternative measures, indicate that both account ownership and active use measures yield highly stable estimated coefficients in determining mobile-money penetration. The active-use condition has somewhat larger effects, consistent with theoretical predictions: when firms are actively involved in digital payments, they derive more benefits than when merely holding accounts. The underlying mechanisms of formalization are supported using alternative dependent variables, including tax filing and the incidence of bribes. A higher rate of tax filings would mean they are more integrated into formal systems, and a decline in the rate of bribes would imply that, within digital transactions, firms can skip informal intermediaries, who usually require bribes to streamline processes. The number of exports also increases, indicating that access to markets enabled by formalization through digital finance reduces friction.

Specification robustness checks are additional tests of the validity of the baseline. The different growth paths can be predicted using country-specific trends, but the core estimate would remain virtually the same. The drop-one-country test is used to verify that no single country has a disproportionate effect on the outcomes, thereby demonstrating the wide consistency across regions. The policy-implementation year also yields an exclusionary donut estimation that almost matches the outcome, thus dispelling fears of transitional noise or policy foreseeability. Placebo tests are a strict evaluation of identification. The treatment effects of artificial policy dates are generated in no important way and show that the model does not mechanically attribute treatment effects to arbitrary shocks. Comparably, unrelated outcomes of phone ownership would show no association with mobile-money regulatory changes and would contribute to the discussion by providing evidence that the reported effects are due to the adoption of digital finance rather than other socioeconomic factors.

The omitted-variable bias analysis that uses the Oster method finds that unobserved confounding would have to be at least twice as strong as observed covariate imbalance to confound effect estimates to zero, which is often quite high in applied microeconomic data. In addition, the corrected coefficient is almost identical to the fixed-effects estimate, implying little residual bias. Correction of measurement errors increases the weight of inference. The 2SLS coefficient is larger than the OLS coefficient, as expected, due to attenuation from classical measurement error in survey-based measures of digital adoption. The fact that the attenuation ratio is 0.68 indicates that OLS is underestimating the true effect size, and that regulatory reforms and network-coverage instruments should be employed. When standard shocks such as GDP per capita, inflation, and governance indices are included, the results remain the same. This means the relationship between mobile money and firm outcomes is not confounded by macroeconomic cycles or institutional changes.

Machine-learning robustness tests (DML) will have another validation point. These estimators address model selection and nonlinear confounding by partialing out high-dimensional controls selected via Lasso and random forests via cross-fitting. The ensuing coefficients are consistent with the traditional fixed-effects estimates, implying that the fundamental findings do not rely on restrictive functional-form predications. Treatment effects are almost equal to those of baseline regressions with staggered estimates of adoption based on Sun.-Abraham methods and Callaway-Sant. Methods. This shows that policy reforms, rather than the adoption of endogenous technology, lead to the observed differences in formalization. Lastly, the mediation analysis shows that 40-46 percent of the total effect flows through formalization, implying a substantive pathway through which digital finance enhances firms’ performance. Collectively, this evidence of rigor provides an interesting and sensible story: the impact of mobile-money penetration on formalization and firm performance is statistically robust and economically significant and able to withstand a host of potential alternative empirical difficulties.

6.1 Conclusion

The discussion has shown that mobile money penetration significantly affects the formalisation and performance of companies in sub-Saharan Africa. Across varying empirical specifications, the level of penetration is higher, which augurs the possibility of formal registration, higher productivity, faster sales growth, and increased export participation. These findings highlight the disruptive opportunities of digital financial services not only to reconfigure firm behaviour, reduce dependence on unstructured systems of electronic cash, and gain access to formal markets, but also to reduce dependence on unstructured systems of electronic cash and gain access to formal markets. The first notable aspect of these results is that they are gender-heterogeneous: funding from female-led companies is more likely to be formalised, indicating that mobile money reduces the structural barriers that have traditionally limited women’s access to finance and market entry. Thus, the evidence demonstrates the importance of digital financial services in minimising gender differences in entrepreneurship and providing women with an inclusive economic experience. The mediation analyses confirm that formalization plays a central role in directing the performance benefits; a significant portion of the benefits accruing from mobile money is mediated by firm registration; hence, legal and institutional incorporation enhances the productivity and growth impact of the innovation. These relationships are causal, based on event-study and instrumental-variable designs, and are highly robust to comprehensive robustness tests. Altogether, these findings highlight that digital finance is not only a technological innovation but also an institutional mechanism that facilitates the sustainable development of enterprises. This paper can be seen to add to the existing body of literature on mobile-money adoption and its link to better firm performance by linking it to formalisation. This understanding is especially noteworthy to policymakers, regulators, and development actors aiming to utilise digital finance to achieve inclusive growth and define it in relation to the Sustainable Development Goals.

6.1 Policy suggestions

The findings provide policy-makers and regulators in sub-Saharan Africa with practical information. To begin with, the fact that mobile-money penetration increases firm formalisation and performance suggests that governments should focus on regulatory changes related to enhancing access to digital financial services. The emergence of interoperability laws, the growth of agent networks, and consumer security systems have become key levers for strengthening market penetration. Such reforms enhance the benefits of micro, small, and medium-sized enterprises (MSMEs) by reducing fragmentation and increasing trust. Second, the fact that the gains in women-led firms are disproportionately greater underscores the need to incorporate gender-sensitive strategies into the financial inclusion strategies of developing nations. Policymakers should encourage the inclusion of female agents, the adoption of digital identity options to meet the needs of women, and the expansion of financial literacy programmes specifically tailored to the needs of female entrepreneurs. Development partners and regulators may also foster gender inclusivity by introducing gender-disaggregated reporting by mobile-money providers to develop evidence-based interventions to overcome chronic barriers. Third, the mediating role of formalisation underscores the importance of linking digital financial inclusion to broader institutional reforms. Governments need to simplify registration processes, make tax compliance easier for small firms, and offer incentives for the legal use of digital platforms by firms. A further way of optimising the institutional benefits of digital finance is to combine mobile payments with government-to-government payments, including procurement and licensing. Fourth, there is a positive relationship between mobile money and export participation, indicating that digital financial ecosystems should be incorporated into trade facilitation strategies. Assistance to MSME participation in international markets can be achieved by policymakers supporting cross-border interoperability, harmonising digital payment standards, and increasing regulatory collaboration across regions. In this way, digital finance can trigger trade diversification and competitive expansion.

6.2 Future research and limitations

This paper is based on firm-level survey data, which includes regulatory indicators and may be affected by reporting issues and measurement errors. Although endogeneity issues are addressed using instrumental variables and machine-learning techniques, they would be better addressed using transaction-based mobile-money data, which would enable superior measurement of its use. Outcomes based on surveys, such as growth in sales and reductions in bribery, could misrepresent practice. The case study of sub-Saharan Africa, where mobile money payments have been widely accepted, is discussed. Future studies should investigate the impact in other emerging markets, including South Asia and Latin America, where digital financial systems are growing under different parameters than in Africa. Comparative research would assess the extent to which regulations and institutional capacity affect mobile-money reforms. The next line of study should examine the extent to which continued mobile money use influences firm survival, innovation, and employment performance. The direction of the industry and rural-urban differences will show the most prominent area of digital finance influence on the economy. The inclusion of household and community data is used to identify spillover effects and to associate firm performance with inclusive development.