Aid has always been shaped by donor interests, but development cooperation is now becoming more overtly interest-driven. Can development outcomes survive inside that new framing? In this blog, I highlight three key points and two boldish ideas I covered in my remarks at the recently-held OECD Conference on the Future of Development Cooperation.
In a world of tighter budgets, geopolitical competition, and mounting pressure to demonstrate results, evidence and learning become more important, not less. What changes is the urgency with which we must deploy them.
Scaling the wrong solution faster
Much of the development sector’s attention is focused on scaling what works. But many of the most consequential failures in development are failures of diagnosis – of not correctly identifying the problem before choosing a solution.
Two examples illustrate the point. In Egypt, a major effort to expand access to childcare was based on the assumption that affordability was the primary barrier to women’s participation in the workforce. Evidence showed that the binding constraint was social norms, not cost. In Peru, decades of nutrition policy concentrated heavily on food provision. Evidence demonstrated that significant reductions in child malnutrition came primarily from improvements in basic health services and feeding practices, not food supply.
In both cases, the issue was not an absence of resources or ambition. It was an incorrect diagnosis — and the opportunity costs were enormous.
In an environment where donors and governments are searching for best buys and rapid returns, the pressure to scale visible solutions is intense. But scaling the wrong solution efficiently is still failure. The discipline of problem diagnosis — of investing in understanding the constraint before designing the response — is not a luxury. It is a prerequisite for effectiveness.
The evidence gap is not what we think it is
The issue in many sectors today is not an absence of evidence. A substantial body of rigorous research already exists across health, education, agriculture, social protection, and other domains. The more fundamental challenge is whether governments and organizations are actually structured to learn.
Decision-making in practice happens under real political, institutional, and financial constraints. The goal must be better decisions under those constraints, not technocratic perfection. That requires investing not only in evaluations and research, but in the evidence systems, incentives, embedded learning, and institutional capabilities that allow evidence to actually reach and influence decisions.
The Global Evidence Commitment — a pledge by certain organizations working to strengthen how evidence is used inside institutions and decision processes, rather than simply produced — reflects a growing recognition of this distinction. The challenge is less about generating more evidence and more about embedding learning into the structures through which decisions are actually made.
AI as infrastructure for evidence use
Artificial intelligence may become genuinely transformative in strengthening evidence use — not because it replaces judgment, but because it dramatically lowers the cost of accessing and synthesizing evidence for decision-makers who operate under time and capacity constraints. Development organizations, including 3ie, are already building and evaluating AI tools for evidence synthesis and decision support. The progress is fast, though not yet fully reliable.
The critical condition for AI to work well in this context is human accountability. If the assumptions, trade-offs, and values behind decisions are delegated to algorithmic processes without meaningful oversight, we risk replacing expert bias with algorithmic overconfidence, at scale and with less visibility.
Two proposals
Looking ahead, two concrete proposals deserve serious consideration from the development community.
- Establish an international fund for governments in the Global South to finance evidence-informed decision-making. Such a fund would provide not only financial resources, but would also bolster the skills and institutional infrastructure to use them effectively — including the capacity to access and synthesise existing evidence, interpret it in local context, generate new evidence at scale, and conduct the kind of rigorous problem diagnosis and real-world cost analysis that institutions like the Millennium Challenge Corporation apply as standard practice. The result would be country-owned, evidence-informed decisions made by the governments best placed to make them.
- Track how much ODA is informed by evidence. The international development system currently invests significant effort in measuring how much official development assistance countries provide. Relatively little attention is paid to a more important question: how much of that spending is either informed by existing evidence or designed to generate credible learning about what works? Making this measurable would create a powerful incentive for the quality — not just the volume — of development investment.
Key takeaways
|
