AI-powered mobile money systems are improving transactional efficiency for Nigerian SMEs—boosting logistics knowledge, increasing trust and reducing supply-chain fraud—and their diffusion is reshaping local economic forces in cities such as Lagos and Abuja.

Main Finding

AI-powered mobile money systems (AI-MMS) materially improve SMEs’ supply-chain operations in the Global South (illustrated by Nigeria) by increasing transactional efficiency, enhancing logistics knowledge, reducing fraud and building system trust. However, diffusion is constrained by low awareness, infrastructural and regulatory frictions (bank-led architecture), and sociocultural trust/risk perceptions. The authors extend Rogers’ Diffusion of Innovation (DoI) by incorporating geopolitical and environmental factors (regulation, trust, risk, complexity) to explain AI-MMS adoption dynamics and the emergence of new economic forces following successful diffusion.

Key Points

• Benefits observed
  • Improved transaction speed, transparency and accountability across supply chains.
  • Better bookkeeping automation, expense tracking and predictive cash‑flow/ demand forecasting.
  • Real‑time fraud detection, anomaly detection and credit-scoring enabled by ML/AI, democratizing access to microfinance for SMEs.
  • Operational gains in logistics (route optimisation, inventory visibility) and supplier evaluation.
• Adoption/diffusion dynamics
  • Low awareness/knowledge among many Nigerian SMEs; late diffusion at the “knowledge” stage relative to East African peers.
  • Perceived relative advantage (efficiency, security, cash‑management) motivates adoption once SMEs are aware.
  • Network/externality effects and observable performance drive movement from early adopters to wider uptake.
• Barriers and risks
  • Regulatory architecture: Nigeria’s bank-led MMS model constrains telco-led solutions and reduces system flexibility/cost-effectiveness.
  • Infrastructure limitations (network coverage, POS availability), digital literacy gaps and historical fraud reduce trust.
  • Cost/risk perceptions, fears of hidden fees or system failure, and data governance concerns slow adoption.
• Contextual/technical notes
  • Examples and prior work referenced: M-Pesa (Kenya) as a benchmark; Nigerian AI-MMS examples include Yabx and PayCliq.
  • Technical need for locally‑trained/interpretable models (e.g., SHAP used in fraud work) to improve trust and regulatory oversight.

Data & Methods

• Philosophy & design: Interpretivist epistemology; exploratory multiple case-study design to capture socio-technical dynamics.
• Sample: 40 semi-structured interviews with SME owners/managers and relevant stakeholders across Abuja, Lagos, Kaduna, Kano, Port Harcourt.
  • Sectors sampled: retail/wholesale, food/catering, transport/logistics, medical supplies, manufacturing, agribusiness, FMCG, fintech, public regulators, ICT/consultancy.
  • Interviews lasted ~25–45 minutes; mix of face-to-face and video calls.
• Analysis:
  • Interview guide structured around Rogers’ DoI attributes (relative advantage, complexity, trialability, observability, trust).
  • Coding in NVivo 14 using two-stage approach: template analysis with DoI-based pre-codes, followed by inductive coding to capture context-specific themes (infrastructure, digital literacy, fraud perception).
• Evidence type & limits: Qualitative, purposive sample focused on Nigeria; findings give rich contextual insights but are not causal or broadly generalisable without quantitative follow-up.

Implications for AI Economics

• Transaction costs & productivity
  • AI-MMS can lower transaction and monitoring costs for SMEs, raising firm-level productivity and potentially increasing supply-chain efficiency at industry scale.
  • Predictive analytics and automated bookkeeping may change working-capital needs and the timing of payments.
• Financial markets & access to credit
  • Real‑time credit scoring and alternative data from AI-MMS may expand credit supply to SMEs, alter interest-rate setting and reduce information asymmetries—affecting credit pricing and default risk models.
• Market structure & competition
  • Network effects may create winner-take-most dynamics in payments/fintech; regulatory choices (telco-led vs bank-led) shape market entry, competition, and platform governance.
  • Policy/regulatory frictions can raise barriers to entry and slow diffusion, affecting aggregate adoption and welfare.
• Distributional effects & inequality
  • Uneven diffusion (due to infrastructure, literacy, trust) risks widening gaps between digitally enabled SMEs and those left out; targeted policy can mitigate differential gains.
• Modeling & empirical research recommendations
  • Incorporate institutional/regulatory frictions, trust parameters, network externalities and heterogeneous adoption costs into diffusion and macro‑economic models of fintech adoption.
  • Prioritise causal/quantitative follow-ups: difference-in-differences, panel analyses, or randomized encouragement designs to estimate impacts on productivity, profits, credit access, and employment.
  • Evaluate welfare trade-offs (consumer protection vs innovation), and measure effects on market concentration.
• Governance, interpretability & risk management
  • Interpretable AI and transparent credit/fraud models (e.g., SHAP explanations) are important for regulator trust and SME acceptance.
  • Data governance, KYC/CDD, cybersecurity and clear fee structures are critical policy levers to increase adoption and reduce systemic risk.

Caveat: results are based on qualitative interviews in Nigeria and are exploratory. Quantitative, cross-country and causal work is needed to generalise magnitudes and welfare effects.