When ALNAP asked people affected by crisis how aid providers could improve assistance, the third most-common need reported by surveyed aid recipients was livelihoods (10%), behind food (44%) and cash/vouchers (18%). Citing from the Ground Truth Solutions synthesis covering 12 countries between 2023 and 2024, it was found that what humanitarians provided was often out of line with the communities' longer term priorities to be more resilient against future shocks. In fact, it is the most dignified request a person can make of a system built around their need, and it sits in the report beside a harder number: only 53% of aid recipients said the assistance they received met their most important needs at the time.
The State of the Humanitarian System 2026, published today, is the sector's most honest mirror compiling nearly two decades of data, evaluations, operational records, practitioner surveys and, above all, interviews with the people the system exists to serve, assessing how humanitarian action performed through 2022 to 2025. I was one of many contributors to the financial analysis behind this edition, and I write here in a personal capacity, drawing only on the published report. The period it covers ends in what is named a 'generational funding collapse', because international humanitarian assistance fell by around 30%, from US$47.5 billion in 2022 to an estimated US$33.3 billion in 2025. 13 of the 20 largest donors cut their contributions. The UN coordinated appeal was 35% funded.
Within that collapse, one passage from SOHS 2026 keeps pulling me back:
"Prioritisation shifted resources towards immediate survival, but this did not always reflect what people affected by crisis requested. Alongside food assistance, some saw livelihoods, education and self-reliance as integral to dignified survival and wellbeing. Aid recipients identified greater support for independence and resilience as the third most important way providers could improve assistance, yet funding pressures favoured immediate relief. Agriculture remained underfunded in Ethiopia, longer-term approaches were constrained in Yemen, and cuts to livelihoods and resilience programming eroded community resilience in Somalia."
The report records the outcome. What I can add is the mechanism, because in one context I have watched it happen inside a donor's own drafting. Earlier this year I read the European Commission's Humanitarian Implementation Plans for Palestine, the annual decisions that govern DG ECHO's funding there, across three cycles: 2023, 2024 and 2026. These are public documents. Anyone can repeat the exercise, and I would encourage it.
The 2023 decision reads like an answer to the request aid recipients would later rank third. It commits DG ECHO to "advancing on the triple nexus." It describes "a graduation approach, from aid dependency to self-reliance." It funds socio-economic resilience and legal aid in the West Bank explicitly so that communities under pressure can remain in their homes. Then watch the words move. Mentions of the nexus fall from 11 in the 2023 decision to 1 in 2026. Resilience appears 3 times in 2023, 1 in 2024, and never in 2026. Self-reliance follows the same path to zero. Area C, the part of the West Bank where the protective model was built, disappears from the decision text entirely. In their place stands a single organising sentence: lifesaving and first line response will be prioritised, with alignment to recovery pursued "when circumstances allow."
There is a 'needs-based defence' of this shift, and it deserves stating. Gaza's catastrophe forced every donor towards lifesaving, and no serious person disputes that triage. But the defence does not cover what actually vanished. The West Bank is not classified on the famine scale; its crisis is one of protection, livelihoods and staying power, precisely the terrain of the deleted words. And the words that went missing are not surge-phase details. They are the strategic vocabulary: nexus, self-reliance, structural funding, graduation from dependency. A budget cut under pressure is a constraint. The deletion of a vocabulary is a choice, and it is legible in the text before it appears in any funding table.
Why does this matter beyond Palestine? First, because in the West Bank resilience was never a development add-on. Strengthening a community's capacity to remain was the protection strategy against forcible transfer. Deleting the word deletes more than a sector. Second, because the same false economy repeats everywhere the report looks: cutting livelihoods, education and self-reliance does not reduce need, it schedules need, converting this year's savings into next year's relief caseload. And third, because the report finds that prioritisation largely happened without asking anyone: only 13.6% of practitioners said a new needs assessment drove their organisation's response to the cuts. The words disappear first. The budgets follow. The people pay last, and most.
Three things follow:
Donors and agencies could treat the recipients' third-ranked request as a test: does this year's financing decision expand or reduce a community's capacity to need less aid?
Analysts could treat the language of financing decisions as data: which words enter, which vanish, is an early warning indicator in its own right, and one that updates faster than any funding dashboard.
And the sector could name the dynamic honestly. In a recent policy brief with Laura Rahm and James Mayuen, we called it managed abandonment: the rationing of survival. The alternative is not unlimited budgets, which are not coming back. It is shared accountability for what gets rationed, decided with the people affected rather than for them.
The State of the Humanitarian System 2026 is rigorous and uncomfortable in the ways a mirror should be.
Read it here and join the online launch on 15 September and bring the finding that stays with you. Mine is already fixed: the people this system serves asked for the chance to need it less. The least the system can do is stop deleting the words that would make that possible.
09:00 to 10:30 UTC
14:00 to 15:30 UTC