USAID Consequence Console — Field Guide

1 · What this is — and what it is NOT

The console exists to keep three very different kinds of knowledge honestly separated. The funding cut is real and large and you can verify it live. The human toll is a modeled range with wide uncertainty, contingent on the cuts persisting with no backfill. The geopolitical fallout is interpretive and has many causes. Conflating these three is the single most common way this topic is misreported; the badge system is designed to make conflation impossible.

2 · Reading the three badges

Every number carries one of three badges, on the figure itself (not buried in a footnote):

• MEASURED — hard accounting or observed data (USAspending obligations, observed service/commodity declines). The most reliable; treat as fact.
• MODELED — a published projection with a stated range/CI and a citation. A scenario-dependent estimate, never a count. Always read the range, not the midpoint.
• INTERPRETIVE — qualitative, multi-causal assessment (instability, soft power, "who fills the gap"). Read as "associated with", never "caused by".

The header pill and banner restate this; the data-health strip (bottom right) tells you which feeds are live vs frozen.

3 · The funding collapse MEASURED · LIVE

The headline chart is USAID (agency 072) total obligations by fiscal year, fetched live, in your browser, from the USAspending.gov API — no key, no proxy. It is pure appropriations accounting: ~$38.2B at the FY2023 peak → ~$6.2B in FY2026 (partial, through Q3), about −84%. FY2026 is partial; the console labels it so — do not annualize it. If the live call fails, the console falls back to a frozen snapshot and says so.

4 · Where it dropped — recipient countries MEASURED

Per-country obligations, FY2024 baseline (faint bar) vs FY2026 (red overlay), with the percent drop. Uganda, South Africa, Zimbabwe and others are down 90%+. Two buckets are deliberately excluded: the "United States" line (~$24B, home-coded HQ/administration) and "Switzerland" (~$3B, Global Fund / UN pass-through) — neither is a bilateral recipient, and including them would badly distort the picture.

5 · The cut → consequence cascade

A four-step chain showing how a funding decision propagates. The first two steps (the cut; observed service & supply disruption) are measured. The last two (modeled disease resurgence; cross-border & geopolitical spillover) are modeled/interpretive. Step 3 is the crux and the least certain — the link from reduced coverage to excess incidence is calibrated on 2015–2024 patterns and is out-of-domain for novel pathogens or drug resistance.

6 · Modeled excess deaths — how to read it MODELED

This is the most easily misread panel, so read it carefully:

• Ranges, not points. Each program shows a low–high range with its scenario and citation (e.g. the aggregate Lancet projection ~14M by 2030, 95% UI 8.5–19.7M). Where a source gives no CI (Guttmacher, Gavi) the console says so rather than fake precision.
• Counterfactual = worst case. The figures assume cuts persist with no donor or national backfill. Documented mitigants already bound the realized harm below the worst case — a Feb-2025 PEPFAR lifesaving-services waiver, a rejected PEPFAR rescission, a ~$9B Gavi replenishment, some reinstated food-aid contracts.
• "Avertable benefit" ≠ "cut-attributable deaths." Some entries (e.g. malaria's 13.6M cases) are the full program benefit at risk, not the toll of a partial cut — labeled as such.
• Likely under-stated long-run. One-year models miss onward transmission, lost re-engagement, and eroded trust.

7 · Measured early signals MEASURED

Kept deliberately adjacent to the modeled panel so the difference is visible. These are actually observed declines — South Africa HIV testing −31% and treatment initiation −30%, Mozambique ART initiation −14%, 14% of 56 countries reporting ≤6 months' ARV stock, RUTF (therapeutic food) shortages in 17 countries, 71% of partners cancelling activities. These are service/commodity/workforce declines, not deaths. The measles resurgence card is flagged mixed — it coincides with the cuts but is multi-causal (including U.S. domestic vaccine hesitancy).

★ Beyond lives lost — five added panels MEASURED

Mortality is the headline but the least leading and most-contested signal. These five panels (added from a verified research pass — high correlation, citable data, wireable feeds, low overlap with the panels above) add consequences that move just as strongly and surface earlier.

Made-in-America blowback — the cut came home

Inverts the funding view: the same dollars, seen from the US side. 23,053 confirmed US layoffs across 46 states (USAIDStopWork — a self-reported floor); Food-for-Peace purchases from US farmers fell from ~1.1M MT / ~$2B to a ~211–417k MT USDA remnant; Johns Hopkins alone cut ~2,200 jobs / ~$800M. The economy-wide "~$23B exports & 200k+ jobs at risk" tile is kept MODELED. The one consequence a domestic audience feels directly.

The leading edge — food insecurity & famine

One rung below the modeled deaths on the same causal chain, but observed in near-real-time: 266 million people in IPC/CH Phase 3+ (crisis or worse), famine confirmed in Gaza and Sudan (first year with two in GRFC history), 35.5M acutely malnourished children. The signature point — the warning system went dark: FEWS NET, the USAID-funded famine early-warning system, went offline in 2025, and the GRFC 2026 recorded the lowest data availability in a decade. Phase populations are MEASURED; sole attribution to USAID is interpretive (conflict, climate and an all-donor shortfall also drive it).

Displacement & migration pressure

Converts the migration story from interpretive to measured: 123.2M forcibly displaced (UNHCR, live keyless API), WFP reach down from 79.9M to 16.7M, and refugee rations halved from Cox's Bazar ($12.50→$6) to Kenya (40%→28% of basket) to Uganda (1.6M→663k reached). The ration-cut → onward-migration link (≈1.9% more migration per +1pp food insecurity) is kept MODELED and multi-causal.

Economic dependency — can the state survive it?

The structural counterpart to the recipient-dollars panel: net development aid (ODA) as a share of each recipient's whole economy (% of GNI). Where it is high, an aid collapse is a macro and fiscal shock to the state itself, not only its clinics — a consequence no death count captures. The most exposed are Central African Republic (27%), Somalia (20%), Burundi (18%), Afghanistan (18%), Mozambique (15%) — several of which are not the biggest dollar recipients, so the panel is genuinely non-redundant (dollars vs share-of-GDP). It fetches live, in-browser, from the World Bank (keyless, CORS-open; 2023 snapshot fallback). The "% of GDP lost from the 2025 cut" projection is a separate MODELED overlay, deliberately not shown as a live figure.

The survivors, permanently diminished — education & human capital

The "survivors" counterpart to the deaths ledger: the far larger group who live but are permanently diminished. 780,000+ children lost their school meal (terminated McGovern-Dole programs — the one MEASURED anchor here), a projected +6 million out of school by end-2026, and the lasting penalty — a child born today in a low/middle-income country may lose ~51% of lifetime earnings to nutrition/learning/skills deficits (World Bank HCI+ 2026), with stunting alone cutting national GDP by up to ~7%. This panel is honestly badged mostly MODELED: most figures are projections, and the out-of-school/aid numbers are all-donor (the US is one of ~3 donors driving ~80% of the cuts). Two anti-double-count rules are stated on the panel: it counts survivors (not the child deaths already in the deaths panel) and the human outcome of school feeding (not the McGovern-Dole dollars already in the funding panel).

8 · Program exposure ranking MODELED

A relative-triage ranking — how deeply each program was cut, weighted by how many depend on it (family planning and climate cut hardest by share; PEPFAR/HIV and malaria highest by absolute exposure). It is for prioritizing attention, not absolute prediction, and it inherits the wide CIs of its source models.

9 · Influence vacuum — the China correction INTERPRETIVE

This panel actively corrects the popular story rather than amplifying it. The best-available evidence is that grant-funded health and humanitarian services mostly stop rather than transfer to another donor: China's grant-equivalent aid is ~$5–8B/yr against the former U.S. ~$63B, and its finance is mostly loans / rescue-lending to governments, not grant health aid. The vacuum is largely unfilled. Russia/Gulf activity is selective and narrow. Everything here is "associated with", never "caused by".

10 · Data sources, honesty & appropriate use

Sources. Funding: USAspending.gov API v2 (live, keyless). Modeled health: Lancet (Rasella et al., 2025), Lancet HIV, PLOS Global Public Health, Topazian/Malaria Atlas Project, Guttmacher, Gavi, IHME. Measured signals: UNAIDS, UNICEF, KFF, aidsmap/Anova, PAHO. Geopolitical: AidData, CGD, Carnegie/ECFR. Each figure carries its citation on the console.

Honest limits. The funding data is current and hard; the death projections are modeled bounds with wide uncertainty and a no-backfill counterfactual; the geopolitical layer is interpretive. The snapshot carries a vintage date; modeled figures update on a stated cadence as new epidemiology arrives, not as a live ticker.