Every carbon credit rests on a counterfactual. The credit says: this activity happened, therefore that other, higher-emitting activity did not, and here is the difference in tonnes. For the counterfactual to be honest, the baseline against which the project is measured has to be honest as well. That sounds simple. In poor rural communities, it is not simple at all, and the reason is a phenomenon the carbon-accounting world calls suppressed demand.

Suppressed demand is the underpinning concept behind Gold Standard’s safe-drinking-water methodology (ERSDWS v1.0)^[1] and its improved-cookstove methodology (TPDDTEC v4.0)^[2], and it is the reason a safe-water or clean-cooking programme in Madagascar can generate a meaningful volume of high-integrity carbon credits at all. Read this article and the rest of the safe-water carbon logic falls into place.

The problem: crediting emissions that were never emitted

The first-order rule of carbon accounting is that a credit measures emissions avoided. Not emissions that never had a chance to happen. Not emissions from an imagined alternative reality. A clean-cookstove project measures the wood a household actually burns before and after the intervention, credits the difference, and if the difference is small the credits are small.

That works fine in a middle-income country where a household is already consuming its full appetite of cooking energy. The baseline (before the stove) reflects a household cooking as much as it wants to. The project (after the stove) delivers the same cooking output with less fuel. The delta is real, is measurable, and is the credit.

It breaks down in the poorest households in the poorest countries, for the same reason those households are poor: they are already consuming below what they need.

A rural Malagasy household that lives at or below the World Bank poverty line does not boil all its drinking water even when the water is unsafe. It does not because it cannot afford the firewood, cannot afford the time, cannot afford the fuel scarcity of the dry season, and has spent generations rationing every economic activity that consumes anything that costs money. If a survey team walks into that household and asks how many litres of water a day the family boils, the answer might be zero. Perhaps a small pot for the newborn. That is the actual current consumption.

Now bring in the naive baseline. The counterfactual, on this logic, is that the household boils zero litres of water and therefore burns zero firewood to do so. Restore safe drinking water to the community and, on the naive baseline, the emissions reductions from displaced boiling are close to zero. The credit is close to zero. There is no carbon finance available to pay for the intervention.

This is not a hypothetical. This is what happened repeatedly under early CDM safe-water and cookstove projects in the poorest sub-Saharan African settings. The methodology was rigorous in accounting terms, and rigorously produced a null result, because the households were already too poor to have anything the project could be measured to displace.

That result is not just economically inconvenient. It is morally incoherent. It says the poorer the community, the smaller the carbon credit; the further from the safe-water and clean-cooking end-state, the less carbon finance available to move them toward it. It systematically deprives the households the framework was designed to help.

The solution: measure against what should be, not what is

The suppressed demand principle is the fix. It was formalised for CDM by the Executive Board in decision EB 62 (2011)^[3:1], and it has been carried into Gold Standard, Verra and the other major crediting programmes ever since. The principle is precise: where current consumption of a basic service is below a defined minimum decent-living standard, the baseline is set at that standard, not at the pinched actual.

In safe drinking water, the minimum is expressed in litres per person per day. WHO’s basic access standard is 20 litres of safe drinking water per person per day for domestic use. Gold Standard’s ERSDWS methodology takes a defined portion of that as the boil-water baseline: the water that a family would need to boil each day to have safe drinking water at the WHO minimum, if they were doing what a household with adequate resources would do.

In clean cooking, the equivalent minimum is expressed in cooking energy per household per day, calibrated against a decent-living definition of cooked meals per person.

The suppressed-demand baseline says, in effect: the counterfactual is not the pinched present, it is the level of consumption a household would reach if their consumption tracked their need rather than their poverty. Their consumption has been suppressed by poverty; we set the baseline at the un-suppressed level, and we credit against that.

The households do not have to be currently boiling five litres of water a day for the credit to accrue. What matters is that in the absence of the safe-water intervention, they would need to boil at least that much to reach the WHO minimum, and either they would be trapped in the current pinched state (still poor, still with the health consequences) or, if their incomes rose, they would begin to consume closer to the minimum and to burn more biomass to boil more water in the process. Either scenario produces the same conclusion for the counterfactual: without the project, the community is either underserved or on a trajectory to increased biomass combustion. The safe-water project stops both.

Why this is rigorous rather than a loophole

Suppressed demand looks, on first inspection, like a way to inflate credits above measured reality. It is not. It is a strictly-bounded accounting principle with several checks that carbon-market critics sometimes miss.

First, the ceiling is external. The suppressed-demand level is not chosen by the project developer. It is set by an internationally-recognised decent-living threshold: WHO for water, defined kitchen energy tables for cooking. A developer cannot inflate the baseline above that number even if the community would notionally consume more with unlimited income. The credit stops at the international minimum, not at aspirational levels.

Second, actual usage is still monitored. Under Gold Standard ERSDWS, project households are surveyed and, on a randomised sample, monitored with third-party sensors and household audits to confirm that the safe-water intervention is actually being used and is displacing the boil-water demand it is credited against. If a household is not drawing water from the restored point (because, say, the point is broken), the credit is not issued for that household. Suppressed demand sets the baseline, but the project scenario still has to be real.

Third, the fNRB parameter caps it further. Only the fraction of the boiled water’s biomass that would have come from non-renewable sources is creditable. In Madagascar’s east coast, the monitored fraction of non-renewable biomass (fNRB) is currently 54% (Cabinet MANO 2025 monitored figure). The other 46% of the biomass a household would have burnt to boil water would have regrown within the accounting period and is not credited. That 54% ceiling on the emissions factor cuts the counterfactual biomass burn nearly in half before any credit is issued.

Fourth, verification is independent. An accredited verification body reviews every step of the suppressed-demand calculation: the baseline choice, the WHO reference standard, the household consumption model, the fNRB, the monitoring evidence. A single failure in the chain and the credits are not issued. This is not a self-declared story.

Between the international ceiling, the monitored fNRB, the actual-usage confirmation and the third-party verification, suppressed demand generates credits that reflect a defined counterfactual under conservative parameters. What it does not do, and this is the substantive change from the naive baseline, is force those parameters to a value that would make the credits vanish in the exact settings the framework exists to reach.

How Gold Standard applies suppressed demand in ERSDWS v1.0

Gold Standard’s Emission Reductions from Safe Drinking Water Supply methodology is the operational framework SaniTap uses on the safe-water strand. Its suppressed-demand logic runs as follows.

Step 1: define the baseline scenario. The baseline is the biomass a household would burn each day to boil their share of the WHO minimum safe-drinking-water need, if they were doing what a household with adequate resources would do. That per-household daily volume is fixed by the methodology and does not vary between projects.

Step 2: apply the emission factor. The volume of boiled water is converted to kilogrammes of firewood-equivalent using published water-boiling energy requirements, and the wood is converted to tonnes of CO₂ using the standard biomass emission factor.

Step 3: apply the fNRB deduction. Only the non-renewable fraction of the wood is credited. Madagascar’s east coast fNRB is monitored at 54% under Cabinet MANO 2025 evidence; only this portion contributes to the baseline emissions figure.

Step 4: net against the project scenario. The project scenario is the biomass the household actually burns for boiling with the safe-water intervention in place, which is close to zero once the community water point is functional and used. Monitored actual usage confirms this.

Step 5: apply the uncertainty and buffer deductions. ERSDWS applies conservative deductions for measurement uncertainty and a buffer pool. The net figure after these deductions is the credit.

The whole chain is verifiable, replicable and audited. Suppressed demand is the first step. Every step after it works to constrain, not to inflate, the resulting credit.

What this looks like in SaniTap’s programme

At current programme scale, SaniTap’s safe-water strand serves approximately 283,000 people across 911 professionally-maintained community water systems. The suppressed-demand baseline, once processed through the ERSDWS chain and the Cabinet MANO 2025 fNRB, delivers on the order of 15,000 tonnes CO₂ per year of avoided emissions from the water strand alone, with the full workbook figure available on the Impact Measures Dashboard.

Every one of those tonnes reflects biomass a Malagasy household would burn to boil water if there were no safe-water intervention and either the household could afford to burn at the WHO minimum level, or the community were, in the coming years, to move toward that consumption level as incomes rose. Either scenario is a counterfactual biomass burn that the safe-water intervention prevents.

That is why the credits carry integrity. The households are receiving safe drinking water, the biomass burn is genuinely displaced against a defined and internationally-recognised baseline, and the emissions figure is deducted at every step for the fraction that would have regrown, for uncertainty, and for a buffer pool. The carbon revenue from those credits funds the professional maintenance of the water systems for fifteen-plus years, so that the safe-water outcome is durable and the emissions reductions accrue year after year.

Why this is central to the ICVCM CCP framework

Suppressed demand is one of the more scrutinised topics in the current integrity-council debate around voluntary carbon credits. The ICVCM has explicitly considered it in its programme-level and methodology-level assessments^[4]. Gold Standard’s ERSDWS has been assessed and confirmed as CCP-aligned. The methodology’s suppressed-demand treatment is not a workaround the ICVCM has overlooked. It is a specific feature the ICVCM has reviewed and endorsed, on the basis that it is bounded, verifiable, and correctly applied.

The reason it is endorsed is that without it, carbon finance is systematically excluded from the poorest communities on the planet. That is exactly the outcome the international framework is designed to prevent. Suppressed demand is not a compromise on integrity, it is what enables integrity-compliant carbon finance to reach the beneficiaries the framework was written for.

Footnotes