The dismantling of the US Agency for International Development has produced a polarized debate. Critics portray USAID as a bloated bureaucracy that lost sight of its development mission. Defenders point, correctly, to millions of lives saved, institutions strengthened, and American influence extended around the world.
After spending more than three decades working in international development — beginning in post-conflict reconstruction and ending as one of USAID’s senior fiscal economists — I submit both sides are missing a more useful question: How should we know when a development program has succeeded enough to end?
Consider Jordan. For roughly 25 years, the United States financed successive projects to strengthen budgeting, tax administration, and public financial management. Contractors changed. Projects ended. A 2016-2020 fiscal-reform program even included an “exit strategy” as an explicit component. Yet in 2021, USAID launched another major program pursuing substantially the same objectives. These were not failed projects; several were excellent. That was precisely what made the question difficult: was success another well-executed project, or reaching the point at which another project was no longer necessary? An experience from my consulting career eventually helped me understand why.
In 2001, I mortgaged my recently purchased home to help finance a small consulting firm’s expansion into USAID contracting. I expected the bank to scrutinize the risk. Instead, the loan officer was almost unconcerned.
“USAID contracts are low risk,” he told me. “This is a low-risk business.”
I later realized that he wasn’t evaluating me. He was pricing an industry.
Stable appropriations, multiyear contracts, and recurring procurements had created a remarkably dependable market for professional services. There was nothing improper about this. Contractors behaved rationally, investing in expertise, compliance systems, costly field research, and specialized proposal teams because doing so increased their chances of winning future work. Recipient governments, USAID officials, and congressional appropriators likewise responded to the incentives they faced. The problem was subtler: successful implementation was much easier to define, measure, and reward than successful exit.
USAID became increasingly sophisticated at measuring whether programs spent funds properly, delivered promised outputs, and achieved measurable results. Those safeguards were necessary. But the system was less effective at answering another question: What observable conditions would demonstrate that outside assistance was no longer required? Sometimes success itself makes that question harder.
Zambia’s HIV programs, supported by PEPFAR and other donors, have produced extraordinary humanitarian benefits. Treatment expanded dramatically, lives were saved, and a once-devastating epidemic became far more manageable. But external financing remains deeply embedded in Zambia’s health system, making the system prone to reversals.
That is not evidence that PEPFAR failed. It illustrates something more complicated. When successful externally financed programs become part of ordinary government services, withdrawal becomes increasingly costly — politically and humanly. The question is not whether assistance should suddenly stop, but whether responsibility and financing are progressively transferring to domestic institutions.
Exit must be designed, not assumed. But designing an exit becomes harder when the mission itself keeps expanding.
During my final year at USAID, climate resilience and mitigation had become increasingly prominent across development programming, including in Sub-Saharan Africa. Many of these objectives were worthwhile. But climate joined a growing list of priorities — governance, gender, inclusion, resilience, and others — layered onto traditional development goals.
Each of these goals can be defended on its merits. The difficulty is that an institution that continually acquires new missions becomes less able to determine when any particular mission has been accomplished. A program may make substantial progress toward its original objective while acquiring new objectives along the way. Completion becomes a moving target.
Part of the confusion arises from our use of the single term “foreign aid” to describe three fundamentally different activities.
Humanitarian assistance exists to save lives and should continue wherever human suffering demands it. Strategic assistance advances American interests and may appropriately continue indefinitely. As former USAID Administrator Andrew Natsios has emphasized, continued American engagement should also be judged by US national interests, not simply by conventional development or graduation criteria. The United States need not stop supporting an important partner simply because that country becomes wealthier or more institutionally capable.
Development assistance is different. Its purpose is to help countries build institutions capable of assuming responsibilities themselves. Exit cannot be the only measure of success. But a credible possibility of exit should be built into the institutional design.
Major development programs should begin with explicit “Conditions for Institutional Completion”: measurable criteria describing what domestic capacity would make continued development assistance unnecessary. Long-running missions should periodically justify not simply their budgets but the continued need for the underlying development function. Repeated procurements addressing substantially the same institutional deficiency should trigger independent review. Programs financing recurrent expenditures should include transparent schedules showing how responsibility can progressively shift toward domestic financing.
Procurement incentives should change as well. Contractors are appropriately rewarded for successfully implementing projects. They should also receive credit when they successfully transfer responsibilities to local institutions — even when doing so reduces the need for another contract.
None of this requires abandoning American development assistance. USAID and its partners saved millions of lives, strengthened public institutions, improved agricultural productivity, and helped stabilize countries emerging from conflict. Those accomplishments should inform whatever system comes next. But rebuilding the old architecture without reconsidering its incentives would squander a rare opportunity.
Not every foreign-assistance program should end, and continuation itself does not demonstrate failure. What matters is whether our development institutions are as capable of defining, measuring, and rewarding successful exit as they are of successful continuation.
USAID had a philosophy of self-reliance. What American development assistance still needs is an economics of completion.
