Safe-water carbon credits are, in some senses, an unusual carbon product. A cookstove credit has an obvious combustion event to displace: the traditional stove is right there, the improved stove replaces it, the counterfactual is visually intuitive. A safe-water credit has a subtler counterfactual: it credits the fuel not burned to boil the contaminated water that a household would otherwise have needed to boil.

Yet the underlying accounting is rigorous, the epidemiology is well-established, and the methodology — Gold Standard’s Emission Reductions from Safe Drinking Water Supply (“ERSDWS”) version 1.0^[1] — is one of the more thoroughly examined instruments in the voluntary carbon market. This article walks through the substantive machinery of a safe-water credit and explains why, correctly implemented, it is one of the higher-integrity carbon products available.

The public-health counterfactual

The counterfactual is anchored in the epidemiology of waterborne disease and the public-health response to it. In settings where drinking-water supplies are microbiologically unsafe — typically informal surface-water sources: rivers, ponds, unprotected wells — the World Health Organization’s recommendation, and the standard household-level intervention, is boiling.^[2]

Boiling water for at least one minute at a rolling boil kills the pathogenic organisms responsible for the great majority of drinking-water related illness, including Vibrio cholerae, most enterotoxigenic Escherichia coli strains, Salmonella, Shigella, and Giardia.^[3] The one-minute-at-boil recommendation is a global standard that appears in national guidance from Madagascar’s Ministry of Public Health, WHO, UNICEF, and USAID guidance to development programmes.^[4]

In settings without electricity or piped gas — which is the setting for rural Madagascar, and for most safe-water carbon projects globally — boiling requires the household to light a fire using biomass fuel (firewood or, in some geographies, charcoal). Each litre of water boiled requires energy input, which requires biomass combustion, which produces CO₂.

The chain from contaminated water to carbon emissions is therefore:

1. Household draws from unsafe source;
2. Public-health-appropriate response is to boil before consumption;
3. Household lights fire with biomass;
4. CO₂ emitted per litre boiled;
5. Aggregate emissions accumulate over the household’s water-consumption timeline.

Break the chain — by restoring reliable access to microbiologically safe water — and the CO₂ is not emitted.

The ERSDWS quantification approach

ERSDWS v1.0 quantifies emissions reductions using the following structure. As with TPDDTEC, the calculation applies conservative discount factors and requires primary data collection.^[5]

Baseline emissions are the CO₂ that would have been emitted through fuel combustion for boiling in the absence of the project. This is calculated from:

• The population served by the water source;
• The fraction of the population that boils water (from baseline surveys);
• The volume of water boiled per capita per day;
• The energy required to bring the volume from ambient to boiling and maintain boil;
• The efficiency of the biomass fire (typically 10–15% for open fires);
• The fraction of biomass that is non-renewable (fNRB) — country-specific;
• The emission factor for the biomass in question.

Project emissions are approximately zero for the boiling activity itself, because the boiling does not occur (the project scenario assumes access to safe water eliminates the need to boil). Project emissions from other sources — for example, energy required to operate a water pump or maintenance transport — are quantified separately and netted from the reduction claim.

Leakage is emissions displaced elsewhere by the project’s activities. In safe-water projects, leakage is typically limited but must be considered — for example, if the reduction in household biomass demand reduces charcoal-market prices such that other households increase consumption, that must be netted out.

Conservative discount factors are applied at multiple points: to boiling rates in the baseline (recognising that not every household boils every litre), to volumes per capita, to project reliability (recognising that safe water access itself may be interrupted), and to the fraction of the population served that is actually served in each accounting period.

The critical role of baseline boiling rates

The single most consequential baseline parameter is the fraction of the population that would boil, in the counterfactual, in the absence of the project.

Empirical evidence on baseline boiling rates in low-income rural settings is mixed. In some geographies, boiling is near-universal because the population is aware of the public-health risk and acts on it. In other geographies — particularly where fuel is expensive relative to income, or where knowledge of waterborne-disease transmission is limited — boiling is intermittent or partial: households may boil the water they know their children will drink but not water they themselves consume, or boil during epidemic periods but not routinely.

ERSDWS v1.0 requires baseline boiling behaviour to be established through direct household surveys in the project area, not through assumption or literature. The surveys must be representative, sampled to methodology-specified sample sizes, and independently verified. Where baseline boiling is measured to be lower than assumed — a common finding in projects that conduct their surveys rigorously — the resulting credit inventory is correspondingly reduced.

This is a substantive integrity feature of the methodology. Over-claim on baseline boiling is the most common failure mode for safe-water project design; ERSDWS v1.0’s insistence on primary measurement structurally guards against it.

Why fNRB matters as much here as for cookstoves

The same country-specific fraction of non-renewable biomass (“fNRB”) that anchors the cookstove methodology anchors ERSDWS v1.0.^[6] The reasoning is identical: not every kilogram of biomass burned contributes equally to net atmospheric CO₂, and the country-level fNRB captures the counterfactual sustainability of the biomass harvest.

Madagascar’s fNRB is above 0.9 — one of the highest globally — reflecting the extraordinary rate of tree-cover loss the country has experienced since 2000.^[7] For a safe-water project in Madagascar, this means that a very high fraction of the biomass combustion avoided by not boiling translates into net avoided CO₂ emissions. The public-health benefit is the same regardless of country context; the carbon-integrity translation is uniquely favourable in high-fNRB geographies.

Monitoring: usage and reliability

A rehabilitated water point is not, by itself, a safe-water intervention. A functional water point being used continuously by the population it was designed to serve is a safe-water intervention. The distance between these two states is significant, and ERSDWS v1.0 requires the second, not the first, to be evidenced.

Monitoring under the methodology includes:

• Water-point functionality checks at defined intervals, evidencing that the water point remains operational;
• Usage monitoring through community-level indicators (queue length, throughput surveys, community water-guardian reporting);
• Water quality monitoring at defined intervals, evidencing that the water actually delivered meets safety standards;
• User-side behavioural monitoring, through periodic surveys, evidencing that households are using the safe water source rather than continuing to draw from unsafe sources.

For SaniTap’s project in Madagascar, monitoring includes an additional layer beyond methodology minimum: third-party sensor monitoring on a randomised sample of installations, providing independently verified usage data. This is not required by ERSDWS v1.0; it is an integrity choice that materially strengthens the audit-defensibility of the resulting credits, particularly for buyers whose diligence exceeds programme minimums.

What is not counted

The methodology is conservative by construction, and there are several categories of substantial real-world benefit from safe-water restoration that are not included in the carbon claim:

• Reduced deforestation from charcoal supply chains where the reduced-demand household would have obtained biomass from a charcoal market rather than direct collection. The upstream deforestation avoided is a real benefit but is not quantified in the credit;
• Time savings for women and girls, who bear the majority of the water-collection and boiling burden in most rural settings, freeing time for other economically or socially productive activity;
• Reduced health-system costs from lower waterborne-disease incidence — reduced clinic visits, reduced pharmaceutical demand, reduced lost work days;
• Reduced child mortality — diarrhoeal disease being a leading cause of under-five mortality in Madagascar and comparable settings.^[8]

These outcomes are documented in the project’s SDG Impact reporting and contribute to the broader sustainable-development narrative around the credit, but the carbon claim itself rests only on the avoided combustion.

Common failure modes

Three failure modes distinguish poorly-designed safe-water projects from well-designed ones:

1. Assumed rather than measured baseline boiling rates. A project that relies on regional averages or aspirational assumptions rather than direct household surveys produces a credit inventory that will not survive rigorous audit.
2. Water-point restoration without maintenance commitment. If the water point falls back into disrepair within the credit-issuance timeline, the reduction is not delivered and the credit claim collapses. Long-term maintenance commitment is essential; without carbon-finance-backed maintenance, most historical water-point restoration programmes have not delivered sustained functionality.
3. Insufficient user-side monitoring. A functional water point that the intended population is not actually using does not deliver the reduction. Baseline behavioural surveys plus follow-up monitoring are required, not optional.

For a buyer, the diligence question is what evidence exists on each of these three failure modes for the specific project under consideration.

Where SaniTap sits

SaniTap’s safe-water project in Madagascar operates under Gold Standard ERSDWS v1.0, with several features specifically addressed to the failure modes above:

• Baseline boiling rates measured through direct household surveys in the project area, with statistically defensible sampling and independent verification.
• Long-term maintenance commitment funded through the sustained credit-revenue stream, with local maintenance capacity built through the SaniTap–MadAvance partnership.^[9]
• Third-party sensor monitoring on a randomised sample of installations, providing independently verified usage data beyond the methodology minimum.
• 725 water points rehabilitated as of the most recent reporting period, serving communities across the project geography, with detailed condition, geolocation, and failure-mode records for each installation.

For buyers evaluating safe-water credits at the rigour level appropriate to CORSIA Phase 2 or ICVCM CCP-aligned integrity requirements, the project’s design documents, monitoring reports, and verification statements are available on request via our commercial team.

Further reading

• Gold Standard, ERSDWS methodology (v1.0), the primary source.^[10]
• WHO/UNICEF, Progress on Household Drinking Water, Sanitation and Hygiene 2000–2022, current edition.^[11]
• Rosa, G. & Clasen, T. (2010), Household drinking water treatment methodologies and health impact.^[12]