Aid Does Not Fail Only Because There Is Too Little Money
- Ott Sarv
- Jun 3
- 6 min read
In his Foreign Affairs article The End of Foreign Aid Is Not the End of Development, Mark Suzman makes an important argument: the reduction of foreign aid does not have to mean the end of development progress. Aid, in his view, should become narrower, more disciplined, more focused on local capacity, and ultimately designed to make itself unnecessary.
That is the right direction. But it is not the full story.
Aid does not fail only because there is too little money. It also fails because too many programmes are designed to look successful before they are designed to become durable. Development work often becomes a beauty contest. Donors want visible success, positive reporting, strong public narratives, and evidence that their money has produced results. Implementing organisations want to show that they delivered. Governments want to show that they attracted support. Everyone has an incentive to look effective.
That incentive can be dangerous.
When aid becomes a beauty contest, the system rewards appearance before substance. A programme can look impressive in a donor report while leaving very little permanent capability behind. It can count people trained, workshops organised, systems launched, policies drafted, platforms procured, and beneficiaries reached. Yet none of this proves that the national system can continue the work after donor funding, consultants, and external supervision leave.
In digital public infrastructure, there is an additional complication. Aid is often not only aid. It can also function as export support for donor-country enterprises. Technical assistance, pilot projects, procurement support, reference implementations, and advisory programmes can open markets for the donor’s own companies, consultants, platforms, cloud providers, system integrators, and technology standards. The programme may be presented as capacity-building, but it can also create a commercial pathway for external suppliers.
This does not mean every DPI programme is cynical or harmful. Some external support is valuable, especially where it brings scarce expertise, implementation discipline, and tested components. The problem begins when development cooperation quietly becomes market-making. If the architecture, procurement model, standards, hosting arrangements, operating procedures, or support contracts are shaped around external suppliers, the recipient country may gain a system but not sovereign capability. It may become able to operate what was delivered, but not to govern, adapt, replace, audit, or sustain it on its own terms.
This is why DPI support should be treated as a sovereignty decision, not merely as technical assistance. A donor-funded programme can leave behind a functioning platform and still fail to leave behind domestic control. If the country cannot change suppliers, inspect the service logic, correct authoritative records, operate the system during disputes, or reverse harmful outcomes, then the programme has not built autonomy. It has built dependency with better branding.
This is where Suzman’s article is right but incomplete. He argues that aid should invest in local capacity and help countries take over functions that have too often been managed by external institutions. But local capacity must be defined more honestly. It is not only national ownership. It is not only a budget line. It is not only a ministry accepting responsibility for a programme. Local capacity also means qualified people who can operate, supervise, improve, and eventually control the system every day.
That includes trained nurses, teachers, agricultural advisers, procurement officers, system administrators, data managers, supervisors, auditors, cybersecurity specialists, enterprise architects, public officials, and service managers who understand the function they are responsible for. It also means retaining them, paying them, equipping them, and giving them the authority to act. If that workforce is missing, undertrained, overstretched, or constantly replaced, then pushing more money, more targets, or more donor pressure into the system will not produce lasting results.
The same standard must apply to international aid organisations. Weak capacity is not only a problem inside recipient countries. Donor agencies and implementing partners also need people who understand programme design, local context, sequencing, risk, delivery constraints, institutional change, and technology governance. If programme managers do not have that capability, the outcome will not improve simply because the programme has a good theory of change or a polished results framework.
Bad programming does not become good development because it is internationally financed.
This is the uncomfortable part of the aid debate. Many programmes are not weak because the goal is wrong. They are weak because the design is poor, the assumptions are unrealistic, the implementation model is shallow, or the reporting framework rewards activity instead of effect. A programme can meet its KPIs and still fail developmentally. It can satisfy the donor, protect institutional reputations, and produce a clean completion report while the underlying public function remains fragile.
This is also where face-saving enters the system. Donors rarely want to admit that a programme was poorly designed, that the implementing partner lacked capability, or that the intervention produced little long-term effect. It is easier to point to outputs. It is easier to show a dashboard. It is easier to publish a success story. The KPI culture helps everyone save face because it gives the appearance of accountability without always testing whether the programme changed the system.
The DPI version of this problem is especially visible. A digital platform can be launched. A registry can be demonstrated. A wallet can be piloted. A data exchange layer can be branded. A dashboard can show adoption. These outputs are attractive because they photograph well and report well. But they do not prove that the country has the legal authority, skilled workforce, procurement control, cybersecurity competence, budget, institutional mandate, and vendor independence required to sustain the system.
Nor do they prove that the system can survive contestation. A platform can execute transactions and still fail as public authority if it cannot show who was competent to decide, which record was authoritative, which rule was applied, what evidence was preserved, and how an affected person can obtain correction or reversal. That is why technology in government must be understood as a channel for exercising state power, not merely as a product project.
The more serious question is not whether the activity happened. The question is whether the capability remained.
Can the national authority continue the function when donor funding ends? Is there a qualified workforce in place? Are operating costs covered? Can procurement continue? Are data systems maintained? Can the system be audited? Can the country replace a supplier? Are failures detected and corrected? Are citizens still served when the consultants leave? Does the programme still work after the visibility phase is over?
These are the questions that should determine whether aid has succeeded.
Suzman is right that development institutions should narrow their priorities and focus scarce grant funding on core development investments such as health, education, poverty reduction, and long-term human capability. He is also right that the older architecture of aid has become too broad, especially when compared with the more focused Millennium Development Goals and the much wider Sustainable Development Goals. But narrowing priorities will not be enough if the aid system keeps measuring the wrong things. The future of aid should not be built around doing more with less. It should be built around doing less that lasts.
For DPI, that means funding lawful capability, not only platforms. A serious programme should fund legal origin, institutional mandate, canonical records, governed service logic, evidence-grade execution, rights-preserving interfaces, and operable remedy. If these capabilities are not funded, the project may still deliver software, but it will not deliver a public system that can be trusted under pressure.
That requires a different standard. Every serious aid programme should begin with a succession plan. The plan should identify who will own the function, which professional roles are required, how those roles will be trained and retained, how the function will be financed, what systems must be maintained, what supplier dependencies exist, and what readiness conditions must be met before external support can responsibly decline. Exit should be based on demonstrated capability, not donor fatigue or budget pressure.
This is also where donors need to be more honest about their own role. There is a difference between helping countries build governed public capability and exporting templates, platforms, regulatory models, or vendor ecosystems under the language of cooperation. Europe’s most useful contribution to global DPI, for example, is not the export of a single regulatory template or technical stack. It is the discipline of keeping public functions, institutional authority, trusted digital interaction, shared capability, and remedy distinct under scale. That distinction is essential if development cooperation is to build sovereignty rather than dependency.
Technology, including artificial intelligence, should be treated in the same way. Digital tools can improve agriculture, health logistics, targeting, forecasting, records, and service delivery. But technology does not compensate for weak institutions or an absent workforce. A digital platform only creates public value when qualified people can operate it, maintain it, govern its data, respond to errors, manage suppliers, and embed it into real public service delivery. Otherwise, it becomes another project artefact in a donor presentation.
The future of aid should therefore be neither a defence of the old model nor a celebration of its collapse. The right answer is disciplined transition. Aid should protect essential services now while building the workforce, institutions, systems, and fiscal capability that make external support less necessary over time.
The end of aid should not be declared through budget cuts. It should be proven through functioning public systems. Until a country has the people, institutions, financing, programme capability, supplier independence, and remedy pathways to continue the work, withdrawal is not development success. It is only the relocation of failure.
And until donors are willing to measure long-term institutional effect instead of short-term reputational success, aid will remain too vulnerable to beauty contests, face-saving, export promotion, and programmes that look better than they work.
