Baseline risk assessment · Amazi Water Burundi

Cookstove overlap and its effect on our Burundi credit baseline

A registered, already-issuing Gold Standard cookstove project (GS 11140, AERA Group / OBEN) sells credits for the same thing our safe-water credits do — reduced woodfuel burning. This page maps where it overlaps Amazi Water’s footprint and forecasts the effect on our credit yield ahead of Q1 2027 registration.

Internal analysis · directional forecast · Aug 2026

Geographic overlap

Effectively certain

Amazi holds a national mandate — all 18 provinces. The cookstove project concentrates in the 7 returnee provinces. Every high-return province is inside our footprint.

Central forecast (2027–33)

~7% credit haircut

≈142k fewer credits, ≈$2.1M at $15 — effective yield 0.170 → 0.157. Plausible band 3%–14% ($0.8M–$3.9M).

Where the real risk sits

Validation, not arithmetic

Gold Standard requires a documented non-overlap / no-double-counting test. Our 0.170 yield is already the portfolio’s flagged integrity residual (0.70).

Where the two projects collide

Reading the map

Every province is inside Amazi’s operating area. Shading shows overlap intensity with concentrated improved-cookstove (ICS) distribution.

High overlap — returnee provincesMakamba, Ruyigi, Rutana, Muyinga, Kirundo, Cankuzo, Rumonge. ~110k returnees settled here; ICS aid packages land here first. ≈ a quarter–third of national population.
Moderate — hub & secondary settlementBujumbura (OBEN stove production, ◯), Bujumbura Rural, Gitega. Distribution staging and urban / return mix.
Amazi baseline — low ICS concentrationBubanza, Bururi, Cibitoke, Karuzi, Kayanza, Muramvya, Mwaro, Ngozi. Amazi present; returnee-ICS presence thin.

OBEN production facility (Bujumbura) — stoves distributed nationally from here.

The two projects, side by side

	Amazi Water (our project)	AERA / OBEN — GS 11140
Credit basis	Safe drinking water — households stop boiling to purify	Improved cookstoves — households cook with less wood
What’s avoided	Biomass burned to boil water	Biomass burned to cook
Methodology	GS Safe Drinking Water Supply	GS TPDDTEC v3.1 (cookstoves)
Geography	All 18 provinces (national mandate, 2021)	Returnee provinces + national OBEN distribution
Scale	Yield 0.170 cr/person/yr · ~1.9M credits 2027–33 · 90% verification	~378k credits/yr · 627k issued · ~236k stoves
Status	Contracted; GS registration Q1 2027; issuances from 2027	Certified & issuing since 2020 (vintages 2020–2025)

The stove project is already registered and issuing and predates our registration — so at validation we are the later project entering an area with an incumbent GS activity claiming woodfuel reductions. The burden to prove non-overlap is on us.

The forecast

We take our proforma credit ramp (2027–33, ~1.92M credits) and apply a baseline haircut h(t) = p·e, where p is the share of Amazi beneficiaries who also run an improved stove and e is the reduction in per-household boiling fuel that an ICS delivers (~25–45%). Three scenarios bracket the uncertainty.

Low case

2.7%

~51k credits · ~$0.8M · yield 0.165

Central — plan to this

7.4%

~142k credits · ~$2.1M · yield 0.157

High case

13.6%

~261k credits · ~$3.9M · yield 0.147

Annual credits at risk, 2027–2033

Green = credits retained (central case) · amber = central at-risk · dashed red = high-case at-risk extent. Values from the Amazi Burundi proforma ramp.

Retained (central) At risk — central (~7%) At risk — high case (~14%)

Scenario	p (ICS share, 2027→33)	e (boiling-fuel cut)	Credits lost	Revenue @ $15	Eff. yield
Low	8% → 12%	25%	51,400	$0.77M	0.165
Central	15% → 24%	35%	141,800	$2.13M	0.157
High	20% → 35%	45%	260,600	$3.91M	0.147

The honest bound. Whether this materializes turns on one methodology choice: the boiling-fuel parameter. If the Safe Drinking Water methodology uses a conservative default specific-fuel value and the DOE accepts documented co-existence, the realized hit clusters near the low end. If the baseline is measured in ICS-heavy catchments, or the DOE requires excluding stove-equipped households, it moves toward the high end. Planning to the central ~7% is the prudent provision.

Why an overlap moves our number

Baseline stove efficiency

Our yield assumes the counterfactual household boils on an inefficient three-stone fire. An ICS household boils the same water on ~30–45% less wood — so its avoided emissions, and its credit, are proportionally smaller.

Double-counting screen

Cooking fuel and boiling fuel are different tonnes, so it isn’t literal double counting. But Gold Standard still requires us to demonstrate end-use separability and that no household’s reductions are claimed twice.

Sampling bias

If our baseline kitchen survey samples returnee provinces without recording ICS presence, it will overstate baseline woodfuel exactly where stove penetration is highest — inflating 0.170 where a DOE looks hardest.

One amplifier: AERA/OBEN also run a broader national household stove programme beyond returnees, so ICS penetration — and baseline erosion — is rising country-wide, not only in the shaded provinces, and grows over the crediting period.

What to do before registration (Q1 2027)

Stratify the baseline survey by stove type. Record actual baseline cooking/boiling technology per catchment; don’t assume three-stone fire. This directly defends 0.170.
Geo-screen against GS 11140. Map planned water systems against the cookstove distribution communes; apply a conservative baseline (or exclude ICS households) where they coincide.
Document end-use separability with survey evidence, satisfying the no-double-counting test up front rather than at DOE review.
Coordinate with AERA/OBEN. A data-sharing or non-overlap letter with the incumbent developer is the cleanest way to close the validation question.
Fold the haircut into the V2.0 yield rebuild. 0.170 is yield-locked and pre-reconstruction; carry ~0.157 (band 0.147–0.165) as the planning value pending survey.