← Cost-effectiveness annex  ·  The Water for Health Fund

MWA Kenya: uptime, not treatment

Sensor-monitored borehole repair across five arid counties, with carbon revenue committed to keeping the pumps running.

This program restores water access during drought; it does not treat water quality. That distinction decides what this page will and will not claim.

PRELIMINARY · v1 · AUGUST 2026 · NOT YET THROUGH ADVERSARIAL REVIEW
01The Role

Why this page publishes no $/DALY

A defensible pathway from borehole uptime to averted DALYs in drought-affected ASAL counties has not yet been designed. Publishing one anyway would be false precision, so v1 declines to.

What the program can honestly claim today: measured service outcomes, climate value, and a financing mechanism that converts credits into pump maintenance. The evidence-adjustment pipeline is not applied because no health effect size is modeled; both facts are stated in the workbook so the omission cannot be reintroduced silently.

214 → 26 days
measured repair time for failed boreholes, before vs under the program (Millennium Water Alliance, gov't partner NDMA Kenya; USAID BHA funded installation)
56% → 91%
measured borehole functionality, against a 90% uptime target, verified by satellite-connected sensors on 200 sites serving 120,000 people today
>70%
of partner-side carbon revenue is committed to ongoing pump maintenance: the credit is the maintenance-financing instrument
02The Worksheet

The model, cell by cell

Provenance key: Proforma input Judgment · flagged Sensor-measured Formula
↓ Download the workbook (.xlsx)
A · Program series, 2025–2033
Series202520262027202820292030203120322033Total
Net credits (tCO₂e)016,23126,01956,19669,84383,49197,13897,13897,138543,194
Buy price to partner ($/cr)9.209.209.309.509.609.709.809.909.90
Sell price, channel ($/cr)151518192025303030
Virridy purchases ($)0149,325241,977533,862670,493809,863951,952961,666961,6665,280,804
Partner upside share, 30% ($)028,24267,910160,159217,910383,223588,656585,742585,7422,617,584
Implied population (credits ÷ 0.103)157,583252,612545,592678,087810,592943,087943,087943,087

Credits and prices are proforma inputs (Voluntary channel 2025–26, CORSIA/LDC from 2027; Mortenson contracted offtake 147,440 credits 2027–33 is the floor case, scenario S1). Yield 0.103 credits per person-year is the Gold Standard SDWS V2.0 bottom-up reconstruction, DHS-anchored. Flagged: the implied population ramp from 158K to 943K people is a proforma assumption; the verified current reach is 120,000 people across 200 sites.

B · Cost basis: one denominator
USAID BHA installation grant$2,000,000installed the boreholes' sensors and repair platform the credits monitor; included so maintenance-only costs are never claimed against full-system benefits
Guaranteed first batch (2026)$210,000MIN(credits, 7,000) × $30, once
Credit purchases, 2025–2033$5,280,804credits × buy price, per year above
TOTAL_PROGRAMME_COST$7,490,804Formula single denominator; the build fails if any figure divides by anything else
Person-years served5,273,728credits ÷ 0.103; cost per person-year $1.42
C · Carbon and financing metrics
$ / tonne, cost basis$13.79$7,490,804 ÷ 543,194 t · 13.4× below the $185 social cost of carbon
Credit integrity ratio0.765baseline residual 0.85 (flagged, V2.0 bottom-up, DHS-anchored) × uptime delivery 0.90 (sensor-measured) × leakage 1.0
$ / tonne, integrity-adjusted$18.0310.3× below the SCC
Partner-side carbon revenue, 2025–33$8,108,389purchases + 30% upside share + first batch
Maintenance financing (70%)$5,675,872the program's committed use of carbon revenue
Maintenance per person-year, steady state$1.140.70 × $15.86 partner revenue per credit × 0.103 credits per person-year; computed at matched population, not the 120,000 current-reach figure
D · Scenarios (one-way; a preliminary model carries no Monte Carlo by design)
ScenarioCredits$/t cost basis$/t integrity-adjMaint $/person-yrReading
Base case543,194$13.79$18.03$1.14expanded-demand base
S1 · contracted-only (Mortenson 147,440)147,440$24.70$32.29$0.71the floor case; still 7.5× below the SCC
S2 · voluntary $10 backstop543,194$13.79$18.03$0.71cost side unchanged; maintenance financing thins
S3 · uptime 0.75543,194$13.79$21.63$0.95integrity falls to 0.64; still 8.6× below the SCC
03Critique

The model's critique of itself

DimensionRatingAssessment
Effect-size qualityFLAGNo health effect size is modeled, deliberately. The service outcomes are measured program data, not modeled effects. Designing a defensible drought-access to health pathway, or formally declining one, is the v2 gate.
Counterfactual robustnessMODERATEThe observed counterfactual is the program's own baseline: 56% functionality and 214-day repairs under government-only maintenance. Carbon-side, the Mortenson offtake is philanthropic demand; a commercial replacement developer is less plausible at this yield.
Moral-weight transparencyFLAGNot applicable in v1: no units of value are computed. Stated so the omission cannot be reintroduced silently.
Attribution and integrityMODERATEInstallation was USAID-funded; the carbon layer finances maintenance. The cost basis therefore includes the $2M grant rather than claiming maintenance-only costs against full-system benefits.
Excluded factorsFLAGExcluded: any health valuation (by design), sensor replacement capex beyond the grant, NDMA co-financing (would lower Virridy-attributable cost), and the flagged population ramp.
Ranking stabilityMODERATEThe climate finding is stable: the worst scenario is $24.70 per tonne, 7.5× below the SCC. The fragile number is maintenance financing per person-year, which halves at the $10 voluntary backstop.