Data infrastructure and cloud tools have lowered technical barriers, but governments convert data into public value only when skills, institutions and governance are built alongside technology; the authors propose a five-level national analytics maturity model and a targeted research agenda to close the gap.

Main Finding

National analytics capacity — the enduring ability of a country to collect, integrate, govern, analyse and act on data at scale — is an emergent, systemic property that depends as much on institutions, incentives and distributed human capital as on technology. Recent advances (cloud, open government data, professionalisation of data roles) have lowered some barriers, but persistent fragmentation, skills gaps, governance weaknesses and trust/equity risks mean technology alone will not deliver public-value outcomes. The paper offers a practical five-level maturity model (across six capability dimensions) and a focused research agenda to make analytics capacity measurable, reproducible and policy-actionable.

Key Points

• Definition and framing
  • National analytics capacity = infrastructures + human capital + institutions/governance + ability to act on analytics reliably and repeatedly (a dispositional, chain-like concept).
  • Emphasises the entire chain: collect → integrate → govern → analyse → act. The weakest link constrains value realization.
• Advances documented
  • Data infrastructure and platforms: widespread open government data (OGD) initiatives; cloud and open-source stacks have lowered capital barriers but introduced sovereignty, vendor-dependence and recurring-cost issues.
  • Skills and human capital: professionalisation of data roles and the emergence of data science; emphasis on broad data literacy (distribution of skills across managers, frontline staff and citizens), not just elites.
  • Institutions and governance: rising attention to governance and stewardship (World Bank, UN) though institutional reform lags technical change.
  • Use and value realisation: a growing evidence base and public-sector use cases, but many projects fail at the “last mile” — turning insight into changed decisions.
  • Ethics, privacy and trust: treated as a core capability dimension, not an add‑on; legitimacy and equity shape public-sector deployment differently from private-sector metrics.
• Persistent challenges
  • Fragmented data infrastructures and poor interoperability (semantic and institutional barriers).
  • Acute shortages of skilled personnel and mismatches between imported designs and local contexts.
  • Weak, uneven data governance and low distributed data literacy.
  • Equity, privacy and trust risks that can undermine adoption and public support.

Data & Methods

• Approach: curated state-of-the-art synthesis (not a systematic review) drawing on:
  • Foundational reports (UN “data revolution”, World Bank WDR 2021), PARIS21 and literature on open data, analytics value realisation, and public-sector digital transformation.
  • Cross-disciplinary evidence from information-systems, public administration, technical infrastructure, and sectoral use-cases.
• Analytical framing:
  • Six capability dimensions used to scope the problem: (1) data & infrastructure; (2) skills & human capital; (3) institutions & governance; (4) methods & tooling; (5) use & value realisation; (6) ethics, privacy & trust.
  • Advances organised into four narrative themes (infrastructure & platforms; skills; institutions & governance; public-sector use cases).
• Output tools:
  • A five-level national analytics maturity model spanning the six dimensions (designed to provide a common vocabulary for assessment and target-setting).
• Limitations:
  • Not a formal systematic review; synthesises and curates rather than exhaustively catalogs empirical studies.

Implications for AI Economics

• Productivity and growth
  • Strong national analytics capacity is a precondition for AI-driven productivity gains in public goods and for private-sector complementarities that translate AI into aggregate growth.
  • The distributive nature of capabilities (data literacy + institutional readiness) matters: returns to AI investments depend on complementary investments in governance and human capital.
• Market structure and competition
  • Cloud diffusion lowers entry barriers for computation-intensive AI, enabling smaller governments/firms to experiment, but increases reliance on a few global providers (vendor lock-in, concentration externalities) with implications for market power and national bargaining positions.
  • Public procurement and platform choices can shape domestic AI markets (e.g., favoring local suppliers vs. global cloud incumbents).
• Labor markets and human capital
  • Public–private competition for data/AI talent raises wage premiums and may produce brain drain from public sector; distributing data literacy reduces reliance on scarce specialists and alters the optimal skill-mix.
  • Policy should prioritize pipelines (education, on-the-job training, retention incentives) and consider the macro-labour implications of scaling analytics capacity.
• Data as an economic resource and externality management
  • Data infrastructure, interoperability standards and governance architectures determine whether data creates positive network/externality effects or remains siloed.
  • Privacy-preserving and federated approaches (federated learning, differential privacy) reshape the economics of data reuse, enabling value extraction while internalising privacy/externality costs.
• Public finance and evaluation
  • Measuring the returns on analytics investments is essential for evidence-based public budgeting; without rigorous impact evaluation, spending on analytics risks persistent under‑utilisation.
  • The maturity-model approach enables better targeting of investments (which dimension is the binding constraint) and more accurate cost–benefit assessment.
• Equity, trust and regulatory design
  • Governance and legitimacy constraints in the public sector mean that efficiency gains from AI can be counterbalanced (or negated) by distributional harms and loss of trust; economic assessments of AI deployment must internalise fairness and legitimacy costs.
  • Regulatory design (data stewardship, access rules, liability regimes) will materially influence innovation incentives and the social returns to AI.
• Research priorities relevant to AI economics
  • Quantify aggregate and sectoral returns to national analytics/AI investments, including distributional effects.
  • Model interactions between cloud-provider concentration, national bargaining power and domestic AI capacity.
  • Evaluate the macroeconomic impact of scaling data-literacy and governance reforms on AI adoption rates.
  • Empirically study the welfare trade-offs of privacy-preserving methods versus centralized data aggregation for public-service AI.

If you want, I can (a) extract and summarise the five-level maturity model as presented in the paper (if you can provide the model details), or (b) produce a short policy checklist for economists advising governments on prioritising analytics investments. Which would be more useful?