MacKenzie Scott's giving, in QALYs | Max Ghenis

How it works

Each Monte Carlo draw takes the giving — each year's gifts inflated to 2026 dollars ($26.39 billion nominal ≈ $30.3 billion, so the dollars and the cost-effectiveness evidence share one price level) — allocates it across the archetypes (a Dirichlet whose centers come from Scott's own gift database — dollar amounts are disclosed for about two-thirds of the money, and each organization's dollars are split across its reported focus areas and mapped to the archetypes — health dollars split by each organization's reported service locations, with the non-US share priced at global-health anchors; the undisclosed remainder is imputed from her announced year totals, scaled by each recipient's pre-gift IRS 990 revenue), assigns each a cost-per-QALY, and multiplies by two independent discounts:

The gift-size data yields one measured regularity along the way: across the 1,313 disclosed gift–revenue pairs, gift size scales with the recipient's pre-gift revenue to the power 0.41 (R² 0.37) — a 10× larger organization receives about 2.5× more money, not 10× more. That fitted elasticity, not proportionality, weights the imputation of the undisclosed gifts.

Geography enters only through health. Across all causes, an estimated 20% of the disclosed dollars go to organizations reporting service locations outside the US — environment, education, and funding intermediaries carry the largest non-health shares — but only the health-and-development slice (about 5% of the ledger) is priced differently for it, because global health is the one category with a published cost-per-QALY evidence base. Education or climate dollars keep their US-study anchors wherever they're delivered; if geography changes their effectiveness, the model doesn't capture it.

• Causal credibility — how well the effect is identified, drawn from the study's design tier: a lottery RCT (income → mortality) is trusted; an associational SNAP correlation is shrunk hard; an assumption-only bucket (arts, civic) goes to near zero. This is the evidence stance slider.
• Realization — the fraction of the studied effect a marginal unrestricted grant actually delivers.

The model also prices the same dollars at the global-health frontier. GiveWell's current impact estimates put the 2022–2024 program averages at ~$4,000 per life saved (Malaria Consortium) to ~$5,500 (AMF nets). A child death averted at ~age 1 is ~25 discounted QALYs under this model's own conventions (~65 remaining years, 3% discount, utility ~0.87), so those endpoints — inflated to 2026 dollars — become roughly $175–$241 per QALY-equivalent; I model the benchmark as loguniform $150–$260, handicapped with the same realization and credibility as Scott's portfolio (and rescaled at other discount rates) so the comparison is like-for-like. At the best-guess defaults, the frontier delivers roughly 500× more health per marginal dollar.

That multiple is a marginal comparison — the next dollar, not the whole portfolio. Frontier-priced opportunities are scarce: GiveWell directed $397 million in all of 2024 and moves its cost-effectiveness bar with the money it expects to raise — funding down to ~6× cash when flush, back up to 10× when projections fell. Malaria control, the deepest frontier bucket, absorbed $3.9 billion in 2024 against a $9.3 billion target, while ~610,000 people died. And the implied frontier counterfactual — ~105 million QALYs at the default settings — would mean averting ~4.2 million child deaths, most of a full year of the world's under-5 deaths; no amount of money buys that at bed-net prices. Redeploying the full $30 billion would ride up the marginal-cost curve from ~500× toward the floor: direct cash, the one option with effectively unbounded capacity, which GiveWell now scores at 3–4× its own historic benchmark: in health-only terms, somewhere under 20–40×.

What this doesn't capture

A QALY is a health metric. Most of Scott's giving targets economic mobility, education, and equity, whose value is largely non-health — income, opportunity, rights, wellbeing. The model therefore understates her total social impact; it answers one specific question. The largest dollar buckets (equity & justice at ~22%, education at ~18%) contribute little health precisely because no credible study ties those grants to QALYs, not because the giving lacks value.

Key sources

• Sommers (2017), AJHE 3(3) — Medicaid cost per life saved $327k–$867k in 2007 dollars, CPI-U inflated here to $529k–$1.40M; mortality effect corroborated by Miller, Johnson & Wherry (2021), QJE.
• Bailey & Goodman-Bacon (2015), AER — community health centers at ~$54k per life-year (2012 dollars), ~$79k in 2026 dollars, converted to $/QALY with the model's utility draw.
• Holtgrave et al. (2013), AIDS and Behavior — Housing & Health intervention $62,493/QALY (HIV-positive unstably-housed cohort, stated in 2005 dollars; ~$107k in 2026 dollars); NASEM (2018) for the mixed-evidence assessment of supportive housing.
• Collaborative-care depression reviews (Community Guide, 2008 dollars: $17k–$39k; van Steenbergen-Weijenburg 2010: $21.5k–$49.5k, unnormalized) — roughly $26k–$77k in 2026 dollars.
• Cesarini et al. (2016), QJE — lottery evidence of a ~null causal income→mortality effect.
• HHS ASPE (2026), Table 3 — value per QALY, $726k central at a 3% discount rate in constant 2025 dollars (~$756k in May-2026 dollars), used for the benefit/cost ratio.
• GiveWell current impact estimates — 2022–2024 lives-saved program averages converted here to $150–$260 per QALY-equivalent.
• Yield Giving gift database — 2,711 gifts with dollar amounts disclosed for about two-thirds of the $26.39B; org-reported focus areas drive the allocation split (area-to-archetype mapping); undisclosed amounts imputed from year totals in proportion to pre-gift IRS 990 revenue, elasticity fit on the disclosed gifts.

Full annotated bibliography, parameter file, and the tested Python package: github.com/MaxGhenis/mackenzie-scott-qaly.