A carbon offset asks one project to do two jobs: reduce or remove greenhouse gases, then cancel emissions released somewhere else. The first job can be real. The second is where twenty-five years of evidence has broken the template.
That evidence replaces a vague procurement question — “is this a good credit?” — with six tests a sustainability lead, investment committee, or buyer can use.
the receipt and the subtraction claim
what is a carbon credit?
A carbon credit is the issued unit — usually one metric tonne of carbon-dioxide equivalent — sitting in a registry before anyone retires it. It becomes an offset only when a buyer uses it to cancel emissions somewhere else.
what is a carbon offset?
A carbon offset is a credit that a buyer uses to claim that a greenhouse-gas reduction, avoidance, or removal elsewhere balances part of the buyer’s own emissions. The unit is usually denominated as one metric tonne of carbon-dioxide equivalent. The defining feature is not the project type or registry entry; it is the buyer’s subtraction claim.
A contribution can fund the same work without making that subtraction.
| instrument | what the buyer receives | claim it supports | what it does not prove |
|---|---|---|---|
| carbon credit | a quantified, transferable unit | ownership of an issued unit | that the buyer’s emissions disappeared |
| carbon offset | a retired credit applied against emissions | compensation or netting | that gross emissions fell at the buyer |
| contribution | evidence of finance provided to climate work | support beyond the buyer’s own reductions | neutrality or cancellation |
| specific ensurance certificate | a hold tied to one named agent and place | present condition and accountability | an offset elsewhere or a fungible unit of nature |
This distinction matters in board reporting. “We emitted, reduced, and separately contributed” preserves three facts. “We emitted, bought, and became neutral” compresses a physical release and an estimated counterfactual into one net number. The compression is where uncertainty becomes a corporate claim.
six tests the market could not make routine
why do carbon offsets fail?
Romm, Lezak, and Alshamsi’s systematic review of the carbon-offset literature — Are Carbon Offsets Fixable? — treats the recurring problems as connected, not as unrelated quality defects. They write that offset programs “greatly overestimate their probable climate impact often by a factor of five to ten or more.” Barbara Haya’s World Bank presentation on project quality and contributions explains why issuance incentives keep those defects alive.
- Additionality. Haya says a project is not additional when it would proceed without credit revenue, while the project side holds the private information needed to test that counterfactual.
- Leakage. Romm, Lezak, and Alshamsi define leakage as emissions or destructive activity moving outside the project boundary after a reduction is credited inside it.
- Permanence. Romm, Lezak, and Alshamsi show that stored carbon can be released later even though the fossil emissions the offset was meant to balance remain in the atmosphere.
- Double-counting. Romm, Lezak, and Alshamsi identify double-counting when two parties claim the same emissions reduction.
- Over-crediting. Haya finds that uncertainty and aligned incentives can produce more issued credits than the climate benefit a project delivers. A Potsdam summary of Macintosh et al. reports that fewer than 16% of more than 2,300 projects examined achieved the promised reductions.
- Environmental injustice. Romm, Lezak, and Alshamsi name this as a core challenge alongside the mechanism failures — tenure, consent, benefit-sharing, and who captures the revenue. Alshamsi has also pointed to the risk of repeating neocolonial appropriation. Some Indigenous communities run carbon projects on their own terms; that does not erase the failure mode.
Each sentence describes a different break in the same promise: that one purchased unit reliably cancels one emitted unit. Additionality compares reality with a world that did not happen. Leakage follows the activity beyond the map. Permanence tests whether storage outlasts reversal risk. Double-counting follows claims across accounting systems, not just serial numbers inside one registry. Over-crediting compounds errors in all the assumptions beneath issuance.
These are structural tests, not accusations of misconduct. Project developers know their financing and intentions better than buyers. Commodity demand can move extraction outside a boundary. Fire, drought, harvesting, or governance change can reverse biological storage. Host jurisdictions, sponsors, and buyers can report overlapping benefits. Developers benefit from issuance, buyers from inexpensive supply, intermediaries from volume, and auditors work inside the market they assess. Haya calls the result a race-to-the-bottom dynamic, not a failure unique to one registry.
Registers still matter: ownership records, methods, serial numbers, verification files, and retirement make a claim inspectable. They cannot observe a counterfactual or prevent every physical and accounting failure. Retirement removes a serial number, not the buyer’s original emissions.
a diligence memo that separates project value from offset quality
A named standard is not a substitute for an investment view. A serious buyer can respect the registry and still ask for evidence beneath the label.
| test | question for the memo | evidence worth requesting | decision consequence |
|---|---|---|---|
| additionality | Would the activity proceed without this revenue? | financing gap, legal requirements, approvals, other subsidies | Do not subtract if the counterfactual is weak. |
| leakage | Did emissions or extraction move beyond the boundary? | market-area analysis, comparison areas, supply-chain data | Discount or reject unmeasured displacement. |
| permanence | How can storage reverse, and how is it monitored? | reversal plan, monitoring term, liability, replacement | Match the claim to the storage horizon. |
| double-counting | Who else can report this reduction? | title, retirement record, host accounting treatment | Require exclusive accounting for an exclusive claim. |
| over-crediting | How sensitive is issuance to assumptions? | methodology, uncertainty range, conservative deductions | Report uncertainty, not issued quantity as certainty. |
| environmental injustice | Who holds tenure, who consented, who is paid? | FPIC record, tenure, benefit-sharing, grievance mechanism | Do not buy a claim that rests on an unresolved rights file. |
Keep project benefits visible beside this test. Cleaner air, household savings, habitat, employment, or watershed protection do not become worthless because a tonne claim is weak. Assess them on their own evidence, with affected communities’ rights, consent, tenure, and benefit sharing treated as core diligence rather than decorative co-benefits.
This separation lets capital support worthwhile work without buying a stronger climate statement than the evidence can carry.
the honest salvage is contribution, hierarchy, and removal
Haya’s recommendation is direct: shift from offsets to contributions. Under a contribution claim, a company reduces its own emissions and reports external climate finance separately. The money can still reach forests, clean cooking, methane control, restoration, or other useful work; it simply does not purchase permission to subtract estimated benefits from the funder’s footprint.
Contribution language changes the buying brief. Instead of maximizing cheap tonnes, a buyer can select for location, community governance, ecological quality, strategic dependency, or neglected need. Finance can follow valuable work rather than the method that manufactures the largest quantity of interchangeable units. The honest objection is demand: if a contribution buys no subtraction claim, what makes a board approve the spend? Offsets scaled because they let a buyer say something. Contributions move when the money is already owed — procurement, a supply-chain dependency, a license to operate, or a named place the buyer cannot walk away from. That is also the only honest reason to hold a certificate.
The order is the control:
- Measure and disclose gross emissions without netting them away.
- Avoid and reduce emissions within operations and the value chain.
- Use durable, verified removal only for genuinely residual emissions where the claim fits the storage.
- Fund additional climate and nature work as a contribution, reported beside rather than inside the emissions total.
A Potsdam Institute summary of Macintosh, Trencher, Probst and colleagues — Rockström last among eight authors — argues for phasing offsets out of national carbon-pricing schemes and replacing them with tightening price caps. That page is about carbon pricing, not a nature-investment thesis. Durable removals, on that same account, should hold carbon for hundreds to thousands of years and be reserved for genuinely hard-to-abate residual emissions. Their wider infrastructure case belongs in carbon removal infrastructure is a public good.
A defensible corporate position follows the same order: report gross emissions and direct reductions first; use durable removals for residual emissions under a disclosed method; finance further climate and nature work as contributions without treating that finance as permission to continue avoidable harm.
biodiversity inherits the template — and more uncertainty
are biodiversity credits repeating carbon markets?
Biodiversity credits repeat the carbon template when a transferable unit is used to compensate for damage elsewhere, but biodiversity is harder because, as Wauchope et al. (2024) argue, there is no settled “unit of nature” that makes one wetland, species assemblage, or ecological relationship equivalent to another; post 1 defines what a biodiversity credit actually is, and post 5 tests the point where an offset becomes permission to destroy nearby.
Measurement is still necessary for stewardship, finance, and accountability. The narrower lesson is that a measured change is not automatically an interchangeable unit, and a transferable receipt is not a hold on the living system.
a hold on the place is not a better offset
Specific ensurance takes the contribution logic to a named place. A certificate is tied to one agent and the place or mandate that agent represents; proceeds fund present condition and stewardship. It is not sold as carbon neutrality, a biodiversity offset, or a secondary book of interchangeable receipts.
A credit lets a buyer claim a quantified unit elsewhere. A certificate lets a holder stay accountable to the same named place. Price makes stewardship legible to capital, but the price is a bridge to protection, not a claim that the living system is worth only what the instrument records.
Ensurance has live agents, coins, and certificates at small volumes; it is not a liquid replacement for global credit markets. The comparison is architectural: contribution rather than absolution, named responsibility rather than fungible equivalence, and current condition rather than a claim that damage vanished elsewhere.
the decision
Do not ask one purchase to erase a physical emission and finance a worthwhile project at the same time. Cut the emission. Test residual-removal claims against their actual durability. Fund forests and nature because those systems need capital and your business depends on them — then call that funding a contribution.
The carbon market’s evidence is not a victory lap. It is an unusually deep record of what happens when estimates, incentives, and corporate claims are compressed into a transferable unit. Biodiversity finance can use that record now, before a harder-to-measure living system is asked to carry the same promise.
Read next: why carbon removal is public infrastructure, what a biodiversity credit actually is, and why an offset is permission to destroy nearby.