Deliberate design of portfolio-management tools can create governability by making government digital initiatives visible and comparable, changing governance conversations; but evidence that this translates into sustained prioritization, budgets or public-value improvements is not yet established.

Main Finding

Governance of public-sector digital development can be actively specified through the design of portfolio-management artifacts: “governability” is not just an organizational trait but a property produced and continuously shaped by design choices and use. A design-led portfolio solution most clearly increases visibility of initiatives and reshapes governance discussions; effects on prioritization and decision-making are emergent, while longer-term outcomes (strategic alignment, accountability, public value) remain unobserved in this study.

Key Points

• Core concept: Governability — the degree to which digital development initiatives are visible, comparable, steerable, and reflexive.
  • Visibility: making initiatives and their characteristics observable to stakeholders.
  • Comparability: enabling cross-initiative comparison on common dimensions.
  • Steerability: supporting collective prioritization and reallocation of resources.
  • Reflexivity: enabling actors to reflect on governance practices and update them.
• Evidence most strongly supports increases in visibility and changes in governance discourse; changes in prioritization and decision processes are present but still emergent.
• Governance specification is iterative and context-sensitive: design assumptions are repeatedly translated, tested, and revised through use.
• Practical design implication: governance tools should go beyond documenting projects — they must explicitly encode what counts as development, how initiatives map to strategy, and how tradeoffs are surfaced.
• Limitations: single national context, limited organizations, single portfolio solution; long-term public-value outcomes not assessed.

Data & Methods

• Research approach: longitudinal Venture Design Research (design-led, iterative, longitudinal study of artifact development and use).
• Empirical sources:
  • Semi-structured interviews used to surface early governance requirements and assumptions.
  • Developer logs tracking how governance assumptions were implemented and revised in the artifact.
  • Three mini-cases showing how deployment of the solution changed governance conversations and decisions in practice.
• Analytical scope: traced how design decisions translated into governability properties over time; did not use cross-national or large-N quantitative evaluation.

Implications for AI Economics

• Measurement and evaluation
  • Treat governability as an endogenous, design-manipulable variable when modeling public-sector AI/digital programs; measure visibility, comparability, steerability, reflexivity as intermediate outcomes.
  • Economic evaluations of AI interventions should include governance-design features as treatment variables (not just organizational covariates).
• Resource allocation and prioritization
  • Portfolio-specifying tools can change how projects are compared and prioritized, affecting budgetary flows and opportunity costs. Economists should model how information architecture alters selection and sequencing of AI investments.
• Policy design and accountability
  • Design choices that make tradeoffs and strategy links explicit can improve accountability and align incentives. This matters for regulation, procurement rules, and ex-post evaluation of AI initiatives.
• Methodological recommendations for researchers
  • Use mixed-method, longitudinal designs to capture the design–use feedback loop and emergent governance effects; short cross-sectional studies may miss these dynamics.
  • Consider experimental or quasi-experimental variation in tool/design features (e.g., which dimensions are made visible) to identify causal effects on prioritization and outcomes.
  • Track long-run public-value outcomes (service equity, citizen welfare, cost-effectiveness) and distributional impacts, since early governance changes may take time to affect these.
• Heterogeneity and external validity
  • Expect context sensitivity: institutional norms, capacity, and political structures will mediate how design affects governability and economic outcomes. Cross-jurisdictional replication is needed.
• Practical guidance for AI economic policy
  • When funding or scaling public-sector AI, invest in design of governance artifacts that standardize definitions, expose tradeoffs, and link initiatives to strategic metrics — these are low-to-moderate cost interventions that can reshape investment patterns and accountability before larger structural reforms.

Suggested next steps for AI economics research: design experiments that vary the four governability dimensions in portfolio tools, collect quantitative measures of downstream budget and performance impacts, and extend studies across multiple countries and agency types to measure long-term public-value effects.