USAID WITHDRAWAL — Consequence Console

USAID (agency 072) total obligations by fiscal year — agency financial reports from USAspending.gov when reachable. Latest fiscal-year totals can be partial; completed and partial windows are shown separately, without an unmatched annual cut percentage.

Embedded recipient-country obligations, vintage 19 June 2026: full FY2024 versus partial FY2026 through Q3. These are unequal windows, not a matched annual cut estimate. This country allocation table is not refreshed by the agency-budget feed. Bars use the FY2024 scale; overlay is partial FY2026. Home-coded buckets (United States HQ, Switzerland = Global Fund/UN pass-through) are excluded.

Net development aid (ODA) as a share of each recipient's whole economy (% of GNI). Where this is high, an aid collapse is a macro and fiscal shock to the state itself — a consequence no death count captures. Distinct from the recipient panel above (which ranks USAID dollars): this ranks aid against GDP, surfacing the most structurally exposed economies — several of which are not the biggest dollar recipients. The US was the largest bilateral donor, so its withdrawal removes a large slice.

The same dollars, seen from the US side. Foreign-aid cuts also hit American farmers, implementers and universities — the consequence a domestic audience feels directly. Inverts the funding panel above (dollars leaving) to show where they originated and landed.

How a funding cut propagates. The left two steps are measured; the right two are modeled with wide confidence intervals — the service-disruption→resurgence link is the crux and the least certain.

Published projections, each with its range and citation. Counterfactual: assumes cuts persist with no donor/national backfill — a published scenario, not a realized count. Not a running death ticker.

One rung below the modeled deaths above, on the same causal chain — but observed in near-real-time, so it shows catastrophe building months before mortality registers. The single best beyond-deaths signal on this console.

The mortality panels count children who die; this counts the far larger group who survive but are permanently diminished — the toddler stunted before age two, the girl pulled from school when the meal that drew her there vanishes. The slow-burn consequence, and an economic-sovereignty story: lost human capital compounds into lost lifetime earnings and GDP. Read as stakes, not a realized count.

Auto-written from the panels, for a non-specialist. Labels what is measured vs modeled vs interpretive.

What is actually observed on the ground (service, commodity, workforce declines) — distinct from the modeled death projections. These are real declines, not deaths.

Converts the “instability / migration” story from the interpretive panel below into measured quantities: ration depth in camps and displacement stocks, bridged to migration by one published elasticity.

Relative exposure = how deeply each program was cut × how many depend on it. For prioritization, not absolute prediction; inherits the wide CIs of the source models.

Multi-causal and qualitative — read as "associated with", never "caused by". The best-evidenced finding is the opposite of the popular story.

Live USAspending vs baked snapshot vs frozen narrative sources, with vintages. A single feed failure never blanks the console.