← Cost-effectiveness annex  ·  The Water for Health Fund

Amazi Water Burundi: the community worksheet

Solar boreholes, piped taps, and chlorination for 3.99 million people by 2033, with the under-five mortality pathway the school model deliberately excluded.

This is the portfolio's largest projected credit line and its most complete health case: a community water supply reaching whole households, including under-fives, where the randomized evidence on chlorinated water and child survival applies. Unlike the Rwanda school model, the cost basis here carries every dollar of system construction, so this worksheet shows what the numbers look like when nothing is left out of the denominator.

PRELIMINARY · v1 · AUGUST 2026 · FIRST ISSUANCES EXPECTED 2027 · IN ADVERSARIAL REVIEW
01Headline

All capex in, under-fives counted

$5,307
per DALY averted, raw, on the full $53.4M delivery cost including all 639 systems' construction. Evidence-adjusted: $13,875 (composite 0.383).
$2,450
net donor cost per averted DALY after carbon revenue at the contracted $15 floor. Falls to $1,342 under the Kremer et al. 2023 meta-analytic mortality effect (scenario S1).
54%
of full 2025–2033 cost covered by contracted-floor carbon revenue ($28.8M against $53.4M). Not yet self-financing in-window; the program plan projects carbon self-sufficiency from 2035.
$27.87/t
full delivery cost per credited tonne, 6.6× below the $185 social cost of carbon.
$44.23/t
integrity-adjusted cost per tonne, still 4.2× below the SCC at an integrity ratio of 0.63.
10,067
DALYs averted over the window, raw: 76% from under-five diarrheal mortality, 24% from all-age illness. 17.7M beneficiary-years at $3.01 per person-year.

Read against the Rwanda school worksheet: Burundi's raw $/DALY is lower despite carrying full construction capex, because the community pathway reaches under-fives, where diarrheal disease kills. That is the general shape of the portfolio thesis: the community and household programs are where the health economics concentrate.

02The Worksheet · Series & Cost

Every year, every dollar

Provenance key: Proforma input Evidence-anchored Judgment · flagged Sensor-measured Formula
↓ Download the workbook (.xlsx)
A · Program series, 2025–2033 (Virridy carbon proforma, base scenario)
Series202520262027202820292030203120322033Total
Systems operating (EOY)4499157224297374460548639
New systems installed445558677377868891639
People served274,695618,057980,1631,398,4321,854,1932,334,9092,871,8143,421,2043,989,32517,742,792 ben-yrs
Net credits (tCO₂e)0040,011135,586206,852262,566361,946419,468490,9051,917,334
Carbon revenue at $15 floor ($)00600,1652,033,7903,102,7803,938,4905,429,1906,292,0207,363,57528,760,010
System capex ($, 3%/yr inflation)2,992,0003,852,2004,184,1904,978,4645,587,0266,069,9596,982,8187,359,5657,838,77349,844,995
Chlorine ($80/mo/system)42,24095,040150,720215,040285,120359,040441,600526,080613,4402,728,320
Monitoring ($80/mo, first 18 mo)31,68071,28081,36090,000100,800108,000117,360125,280128,880854,640

All rows trace to the proforma inputs of the base scenario and the program's contract terms: $68,000 full-recovery cost per new system, ~6,000 people per system, credits beginning two years after installation per the crediting methodology, and the contracted $15 per credit floor price. Credit yield basis 0.170 credits per person-year, not yet reconstructed to the 2026 methodology revision (see section 04).

B · Cost basis: one denominator
TOTAL_PROGRAMME_COST, 2025–2033$53,427,955Formula capex + chlorine + monitoring, all years
Beneficiary-years served17,742,792Formula sum of the people row
Cost per beneficiary-year$3.01Formula cost ÷ beneficiary-years

Every figure on this page divides by this one cost; the build fails programmatically otherwise. Donor revenue in the program plan ($4M per year) is deliberately NOT netted out: donor money is the philanthropy whose cost per DALY this page measures. Excluded from the cost: operations beyond chlorine and monitoring (an optimistic omission, flagged in the critique), and all post-2033 years, including the projected carbon-self-sufficient era from 2035 (a conservative omission for donor economics).

03The Worksheet · Health Chain

Two pathways, one usage number

C · Health chain inputs
InputValueProvenanceBasis
Under-five share of population17.5%Judgment · flaggedCensus lookup pending.
Under-five annual death rate0.011 /child-yrJudgment · flaggedFrom ~52 per 1,000 live births; UN IGME lookup pending. Tested ±30% in S4–S5.
Diarrheal share of under-five deaths9%Judgment · flaggedGBD-order estimate.
Under-five mortality reduction12%Evidence-anchoredGiveWell's conservative in-line chlorination figure. The Kremer et al. 2023 meta-analytic 25% runs as scenario S1.
Years of life lost per under-five death33Judgment · flaggedDiscounting convention pending.
Diarrheal incidence, under-five / over-five0.8 / 0.3 episodes/yrJudgment · flaggedGBD lookups pending.
Diarrhea reduction, chlorination25%Evidence-anchoredWolf et al. 2022 (Lancet), point-of-use chlorination risk ratio ~0.75.
Delivery effectiveness (safe at the tap)90%Sensor-measuredPortfolio monitoring record: 90.3% of samples safe under monitored delivery (n=1,685).
Household usage share70%Judgment · flaggedShare of drinking water actually drawn from the tap; no survey yet. Tested 50–90% in S2–S3.
Effective coverage63%Formula90% × 70%. The same number the credit quantification uses.
YLD per episode0.002215 DALYFormulaDisability weight 0.188 × 4.3 days ÷ 365.
D · Calculation chain
StepValueComputation
Under-five child-years3,105,00017,742,792 beneficiary-years × 17.5%
Under-five diarrheal deaths averted232U5 child-years × 0.011 × 9% × 12% × 63% effective coverage
DALYs from mortality (YLL)7,669deaths × 33 years of life lost. 76% of total health value.
Diarrheal episodes, all ages6,875,000U5 child-years × 0.8 + over-five person-years × 0.3
DALYs from morbidity (YLD)2,398episodes × 25% × 63% × 0.002215. 24% of total.
DALYs averted, total raw10,067
$ / DALY averted, raw$5,307$53,427,955 ÷ 10,067
$ / DALY averted, evidence-adjusted$13,875÷ composite 0.383 → 3,851 adjusted DALYs
xCash (GiveWell / egalitarian / wellbeing / weighted)0.129× / 0.056× / 0.202× / 0.133×DALY moral weights 2.3 / 1.0 / 3.6; GiveDirectly benchmark 0.00335 units per dollar. Ranking stable.
E · Evidence-adjustment pipeline
FactorScoreBasis
Internal validity0.75Strong reduction anchors, but every burden input in the chain is judgment.
External validity0.75In-line chlorination and point-of-use evidence transferred to piped community supply in Burundi.
Publication bias0.80Systematic reviews without an independent funnel-plot confirmation in this model.
Implementation quality0.85Sensor-instrumented delivery plan, but no issuance track record yet: Gold Standard registration expected Q1 2027.
Composite0.383Formula product of the four; reported alongside the raw figure, never silently multiplied in.
04The Worksheet · Carbon

Integrity and coverage, before the first issuance

F · Credit integrity and carbon metrics
Residual baseline-practice factor0.70Judgment · flagged The crediting basis (0.170 credits per person-year) has NOT yet been reconstructed to the 2026 methodology revision, so the residual over-crediting risk is priced larger than Rwanda's 0.78. A measured baseline survey replaces this.
Delivery effectiveness0.90Sensor-measured The same cell the health chain uses.
Leakage / rebound1.00Negligible for water treatment.
Credit integrity ratio0.63Formula 0.70 × 0.90 × 1.00
$ / tonne, credited basis$27.87$53,427,955 ÷ 1,917,334 t · 6.6× below the $185 social cost of carbon
$ / tonne, integrity-adjusted$44.23÷ (1,917,334 × 0.63) · 4.2× below the SCC
Revenue coverage of full cost53.8%$28,760,010 ÷ $53,427,955 at the $15 floor → net philanthropic cost $24,667,945
Net donor $ / DALY averted$2,450the fund's primary metric for a program that is not yet self-financing in-window

The additionality picture differs from Rwanda's: with no issuances yet, no host-country authorization yet, and construction gated on blended donor and carbon finance, the case that the carbon contract causes the program is currently strong. The corresponding risk moved elsewhere: crediting volume (see S7).

05The Worksheet · Scenarios

What moves the number

G · Scenarios (one-way; a preliminary model carries no Monte Carlo by design)
ScenarioVaried inputDALYs$ / DALY rawxCash GWNet donor $ / DALYReading
Base case10,067$5,3070.129×$2,450as above
S1 · Kremer mortality effect25%18,375$2,9080.236×$1,342the meta-analytic all-cause anchor roughly doubles mortality DALYs
S2 · household usage low50%7,191$7,4300.092×$3,430
S3 · household usage high90%12,944$4,1280.166×$1,906
S4 · under-five death rate −30%0.00777,767$6,8790.100×$3,176
S5 · under-five death rate +30%0.014312,368$4,3200.159×$1,995
S6 · floor price $10$10/t10,067$5,307coverage 35.9% · net $3,403/DALYprice risk is real but bounded
S7 · contracted credits only210,000 t10,067$5,307coverage 5.9% · net $4,994/DALYthe load-bearing carbon risk: crediting volume, not price

Recomputed robustness statement: raw $/DALY spans $2,908 to $7,430 across the single-input health scenarios, and the net donor figure spans $1,342 to $3,430. No scenario brings the health frame near the cash-transfer bar on full cost, and none lifts the climate cost above the SCC. The fragile claim is the carbon ramp itself: if only the currently contracted 210,000 credits materialize, coverage collapses to 6% and the program is donor-funded construction with a carbon topper. Gold Standard registration and the issuance ramp, not the floor price, carry the carbon case.

06The Worksheet · Critique

The model's critique of itself

H · Six questions
DimensionRatingAssessment
Effect-size qualityWEAKThe reduction anchors are strong (GiveWell's conservative chlorination figure, Wolf 2022), but every burden input in the chain (population shares, death rates, incidence, years of life lost) is judgment pending IGME and GBD lookups, and household usage has no survey. Preliminary status is correct.
Counterfactual robustnessMODERATEFew alternative funders operate at this scale in Burundi; funging is low to moderate and unmodeled in v1. Carbon additionality is currently stronger than Rwanda's: no issuances, no authorization yet, construction gated on the blended finance.
Moral-weight transparencySTRONGAll frameworks reported; the mortality-dominant pathway keeps the ranking stable. Net donor cost per DALY (~$2,450 raw) is the decision-relevant number.
Attribution and integrityMODERATEAmazi Water delivers; Virridy develops the carbon and monitoring layer. The model charges full system capex and claims full health benefit, which is internally consistent, but a funder-specific split (carbon investment against donor capex) belongs in v2. The integrity ratio's weak link, the 0.70 baseline residual on an unreconstructed crediting basis, is flagged, not hidden.
Excluded factorsFLAGExcluded: time savings from proximate taps (likely large, conservative), water-quantity and hygiene benefits (conservative), operations beyond chlorine and monitoring (optimistic), and all post-2033 years including the projected self-financing era (conservative for donor economics). The two-year crediting delay also leaves in-window credits understating steady-state yield.
Ranking stabilityMODERATEHealth and climate frames are stable across S1–S7. The sensitive claim is crediting volume: registration and ramp risk, priced in S7, dominates price risk.