When US foreign assistance was abruptly frozen in January 2025, Yo’ Care South Sudan was providing HIV services to around 10,000 people. Within weeks, nearly 1,000 women had lost access to medication they needed to survive. Another 9,000 people lost the testing, preventative treatment and health services they had relied on for years, and five tonnes of health commodities sat stranded in clinics, while almost the entire workforce was laid off or furloughed. The decision behind all of this was taken thousands of miles away, and its consequences landed in bodies, jobs, medicines, leases and obligations that did not disappear when the money did. 

It is tempting to treat the episode as an aberration produced by an extraordinary political rupture in Washington, and in one sense that is what it was. The scale and speed were unprecedented: by March 2025, more than $76.5 billion of $159 billion in foreign assistance awards had been terminated (oig.usaid.gov). The rupture also exposed a set of relationships that had been in place long before it. 

Only months earlier, USAID had been reporting record progress towards its localisation commitments. In fiscal year 2024, it recorded $2.1 billion in direct funding to local civil society, private-sector and government partners, equivalent to 12.1 per cent of the acquisitions, assistance and government-to-government funding included in its localisation measure. That figure carried a politics of its own. The denominator covered only around 43 per cent of USAID’s total resources. Other calculations using narrower definitions of “local” and a broader funding base produced considerably lower numbers, and USAID’s own definition could include some affiliates of international organisations (publishwhatyoufund.org). The localisation effort was real enough. What the argument over measurement shows is that the apparent scale of change already depended on what an institution chose to count as change, well before anyone reached the question of who held authority. 

When the money stopped, Yo’ Care still had its staff, its responsibilities, its relationships with communities and the expertise it had built over years. What it lacked was enough room to keep those things intact once the institution supporting much of the work changed direction. 

Sight is a strange kind of capacity 

Anticipatory capacity is the capacity of an organisation to continuously develop and apply knowledge acquired through a structured approach to anticipate changing scenarios as needs and context change over time.

Anticipatory capacity is developed, for example, in governance, to equip institutions with the tools to innovate and anticipate challenges or in humanitarian contexts to mitigate the potential impacts of crises before they occur.

Most conversations about anticipatory capacity begin with sight. Can an organisation detect change early enough, pick up the quieter signals before they become crises, understand how a funding shift or an election, or a regulatory change might alter the environment around it? Those capabilities matter, though on their own they make for a strange kind of capacity. 

An organisation may understand perfectly well that its funding is over-concentrated, that political conditions are deteriorating, or that an existing programme has become exposed. It may see the problem earlier than its donor does. If it cannot move money, alter commitments, protect staff or change direction without someone else’s approval, its analysis has travelled a great deal further than its agency. 

So the definition needs to be harder:  

Anticipatory capacity is the ability to preserve and exercise meaningful choices before conditions harden. Imagination sits somewhere in the middle of that sequence; a good deal further from the moment of decision than we usually assume. 

Framed in that way, other things come into view: financial runway, contractual discretion, decision rights, access to unrestricted money, political relationships, and the capacity to absorb a decision that turns out to be wrong. A contract tells us where authority sits at rest, and anticipation tells us whether that authority survives movement. That distinction does more work than most of the language we currently use around “power”. 

The word has become almost compulsory in localisation debates. We shift it, share it, rebalance it, and redistribute it. Yet “power” now stands for so many separate things that it conceals the relationships it is meant to expose. Control over money, agenda-setting authority, contractual discretion, ownership of data, political access, liability, the ability to refuse a condition, the ability to survive a disagreement with the institution paying the bills: these travel independently of each other, and an organisation can acquire several while gaining none of the rest. 

An organisation may lead a programme while somebody else retains the right to terminate it, receive more direct funding without gaining enough unrestricted income to build reserves, or sit at the table with no real capacity to alter the decision that follows. Separating the terms makes the picture less comfortable. Authority moves in some cases. In others responsibility travels much further than authority ever does, and in the worst of them what moves fastest is exposure, with one institution reducing its obligations while another inherits the consequences. 

A relational problem with a geographical name 

Localisation deceives us in a related way, by giving a relational problem a geographical name. The question becomes where resources should sit, how close decisions should be to the people affected, whether leadership should move from international to national institutions. Those are legitimate questions, and they are considerably easier than the one underneath them: what is an institution prepared to stop controlling? 

ODI found that only 3.8 per cent of international humanitarian funding reached local and national actors directly in 2024. Even including indirect flows the figure remained below 10 per cent, and the funding that did move was often smaller, shorter and more tightly earmarked than that received by international counterparts (odi.org). 

Those differences shape what an organisation can do long before a crisis arrives. Whether it can carry staff between grants, build reserves, change course when the world moves faster than the grant agreement. How many months it could survive if its largest funder stopped tomorrow, who holds the severance liability when that happens, who owns the data, how much of the budget can be moved without asking permission. These are unglamorous questions, and they sit much closer to where agency lives than anything in the localisation percentages. 

Yo’ Care shows what happens when that room collapses quickly. Public information does not tell us how many months of unrestricted reserves it held, precisely how concentrated its income was, or what protections sat inside its contracts, and we should not invent what we cannot know. What we can see is the speed at which the range of available choices narrowed once the freeze arrived. 

A more complicated comparison 

The Titi Foundation, another South Sudanese organisation, offers a more complicated comparison. After UK aid cuts in 2021 and 2022, Titi had already been forced to confront its dependence on funding channelled through UN agencies and international NGOs, and its executive director, Gloria Soma, described the lesson plainly. The organisation changed its funding base. By February 2025 she estimated that around 70 per cent of its budget came from philanthropic foundations, and Titi had also begun experimenting with earned income from its technical expertise (thenewhumanitarian.org). When the USAID freeze arrived, Titi suspended one programme while the rest of the organisation remained standing. 

For a few weeks it looked like a clean story about diversification working, and then it stopped looking like one. By September, Soma was reporting that Titi had lost $3.5 million in UNICEF-channelled funding after US support disappeared, and that nearly half of South Sudan’s national aid organisations had closed since March (cbc.ca). 

The later losses make the earlier lesson more precise. Diversification gave Titi more room when the first shock hit, buying it time, alternatives and a wider range of moves, and it left the organisation fully exposed to a contraction large enough to tear through the entire system. Anticipatory capacity is therefore a matter of degree: how much room remains, for how long, and against what scale of disruption. 

The comparison should also not be moralised, because prudence explains very little of it. Organisations working in HIV treatment, humanitarian relief, advocacy, or livelihoods face different funding markets. Philanthropy reaches them on very different terms, and earned income is no answer at all to the financing of essential health services. The structure around an organisation shapes the choices it is able to create. 

Localisation, decolonisation, decoloniality

This is also why localisation2, decolonisation and decoloniality should not be treated as interchangeable terms. Localisation is largely concerned with where resources, responsibility and authority sit. Decolonisation confronts the structures created through colonial domination and asks what must be dismantled. Decoloniality goes somewhere harder, into the assumptions that survive after formal structures change: the ideas about whose knowledge counts, what competence looks like, which risks deserve attention and whose interpretation of reality is considered credible. 

The difference matters for anticipatory work, because a process can be entirely localised and still leave its underlying rules untouched. A national organisation can lead the exercise while uncertainty is still interpreted through categories inherited from somewhere else. Community knowledge can be actively sought and then translated into the language of indicators, risk matrices and donor priorities before anyone is willing to act on it. 

Reversing the hierarchy answers none of this, and local knowledge carries no guarantee of being emancipatory. Local institutions contain their own hierarchies, and class, caste, ethnicity, gender, geography and political access determine whose knowledge travels everywhere. Moving a decision across a border leaves those fractures intact. 

The demanding question is whether another interpretation can change a consequential decision, which sets a far higher bar than participation and leads back to anticipation. Seeing first means surprisingly little when someone else still decides whether the thing you saw is real enough to act upon. 

Imposed, and designed 

Donors are constrained too. Parliamentary appropriations, audit rules, procurement systems and counterterrorism obligations shape what public agencies are able to fund and how, and some rigidity is imposed rather than chosen. The rest is designed. ODI’s research on humanitarian overheads found that many local and national actors receive only 4 to 7 per cent in indirect cost recovery, if they receive anything at all, and that flexibility available higher up the funding chain often disappears before the money reaches local sub-grantees. Restrictions on overheads then make it harder for those organisations to build precisely the reserves they are later told they should have had when the crisis comes (odi.org). 

Some constraints are legal, some are political, and some are institutional habits dressed up as inevitability. Anticipatory analysis earns its keep by telling them apart. 

Optionality in a market that is contracting 

The difficulty is that we are now asking these questions in a market where everyone has fewer options. ALNAP’s State of the Humanitarian System 2026 estimates that international humanitarian assistance fell from a peak of $47.5 billion in 2022 to $33.3 billion in 2025. Telling hundreds of organisations to diversify in that environment quickly becomes absurd. Diversify towards whom? Reserves buy time, and they cannot manufacture funders that no longer exist. 

There is a cruel irony in the timing. USAID had been one of the strongest institutional champions of locally led development, and its collapse exposed how dependent parts of that agenda remained on the continued choices of the institution promoting it (devex.com). 

The same research on overheads points to a different kind of optionality, one that sits outside any single organisation. Where local and national actors have advocated collectively, and in some cases declined contracts that arrived without adequate cost recovery, donors and intermediaries have moved. Default overhead rates have risen. Terms presented as fixed turned out to be negotiable once they were being negotiated with more than one organisation at a time (odi.org). 

That changes the unit of analysis. An organisation bargaining alone in a contracting market has almost nothing to refuse with, and the smaller it is, the truer that becomes, while a group declining the same terms changes the position of the institution offering them. Coordination expands neither the pool nor the room within it. It changes who takes part in deciding how scarcity is passed down. 

The obvious objection is that refusal requires something worth refusing, and that leverage of this kind is easiest to build when funding is stable. That is fair, and it is exactly why the question belongs to anticipation. Collective positions are slow to construct. They get built in the interval between recognising a dependency and being overtaken by it, which is the interval this piece has been describing throughout. 

Where the argument lands 

So the argument lands some distance from prediction, and some distance from resilience. Certain shocks are too large for any single organisation to absorb, certain dependencies cannot be diversified away, and certain choices are impossible to preserve alone. What remains available is the work of identifying where room to act is disappearing while enough of it survives to defend, widen or reconstruct. 

Organisations across this sector saw 2025 coming in outline, and many had been describing their own dependence for years. What they arrived at, when the moment came, was a decision that had already been taken somewhere else. 

The useful question for anyone funding, designing or evaluating anticipatory work is where in the sequence an organisation’s room to act disappears, and whether anything in the current architecture is built to widen it. Very little of it is.