How does a carbon credit work, and why have so many credits not delivered the reductions they claimed?

What a credit is

A carbon credit represents one tonne of carbon dioxide equivalent that was either kept out of the atmosphere or taken out of it. The tonne is counted against a baseline, checked by an independent auditor, and recorded in a registry so it can be sold and then used once.

How a credit is made

1. Baseline. The project estimates what emissions would have been without it. See baseline.
2. Methodology. An approved rulebook sets how the baseline is drawn and how the tonnes are counted. See methodology.
3. Additionality. The project shows the reduction happens because of carbon revenue and would not have happened anyway. See additionality.
4. Validation. An independent validation and verification body checks the project design before crediting starts.
5. Monitoring and verification. The project measures what happened, and the same kind of body checks the results after the fact. See verification.
6. Issuance. The standard issues credits into a registry, one per verified tonne.
7. Retirement. When a buyer uses a credit in a claim, it is removed from circulation for good. See retirement.

Three more checks sit around this chain.

• Leakage. Emissions that move outside the project boundary because of the project are subtracted.
• Permanence. Where stored carbon can be lost, the project sets credits aside in a buffer pool, which is cancelled to cover carbon that is lost.
• Double counting. The same tonne must be issued, claimed and used only once.

Where credits have fallen short

Each failure below is a failure of one link in the chain. The figures come from peer reviewed studies and apply to the projects those studies examined.

Baselines set too high, in forest protection. West and colleagues, in Science in 2023, studied 26 avoided deforestation project sites in six countries. Most had not significantly reduced deforestation, and where they had, the reductions were substantially lower than claimed.

Projects that would have happened anyway, in the Clean Development Mechanism. A 2016 study for the European Commission found that 85 per cent of the projects it covered had a low likelihood that their reductions were additional and not over-estimated. Only 2 per cent had a high likelihood. It included Indian wind projects whose case for carbon revenue depended on the credit price they assumed.

Assumptions too generous, in cookstoves. Gill-Wiehl, Kammen and Haya, in Nature Sustainability in 2024, compared five cookstove methodologies with published research and their own analysis. Their sample of projects was over-credited 9.2 times. Gold Standard’s metered methodology came closest, at 1.5 times.

Across project types. Probst and colleagues, in Nature Communications in 2024, pulled together evidence on 2,346 projects, about one fifth of all credits issued to that date. Less than 16 per cent of the credits issued to those projects were real reductions:

• 11 per cent for cookstoves
• 16 per cent for SF6 destruction
• 25 per cent for avoided deforestation
• 68 per cent for HFC-23 abatement
• No statistically significant reduction for wind power or improved forest management

Carbon lost after it was credited. In California’s forest offset programme, Badgley and colleagues found over-crediting of 29.4 per cent of the credits they analysed. A second study found that wildfires had used up nearly one fifth of the programme’s buffer pool in less than a decade, equal to at least 95 per cent of what was set aside to cover all fire risk for 100 years.

What has changed since

• A common benchmark. The Integrity Council for the Voluntary Carbon Market launched its Core Carbon Principles on 29 March 2023, covering additionality, permanence, robust quantification, independent verification and no double counting. It now rules methodology by methodology.
• Old renewable credits excluded. In August 2024 it found that eight existing renewable energy methodologies failed its additionality requirements. In May 2026 it conditionally approved a new grid renewable methodology from Verra, with a benchmark test, and no past issuances carry the label.
• Forest baselines taken out of developers’ hands. In November 2024 it approved three REDD+ methodologies, including Verra’s VM0048, under which developers no longer set their own baselines from reference areas they choose. Credits under Verra’s older REDD methodologies cannot receive the label.
• Country level accounting. Under the Paris Agreement rules agreed in Glasgow, a country that authorises a credit for another country’s target applies a corresponding adjustment, so the tonne is counted once. The first credits under the United Nations mechanism, approved on 26 February 2026, came to about 40 per cent fewer than the old Clean Development Mechanism rules would have issued. See what Article 6 means for Indian projects.
• India’s own market. The Carbon Credit Trading Scheme gained an offset mechanism in December 2023, and in March 2025 the government approved its procedure and eight methodologies, covering renewable energy, green hydrogen, industrial energy efficiency, landfill methane and mangrove planting. See India’s Carbon Credit Trading Scheme.

What a buyer or developer can check

• The methodology and its version. Whether it carries the Integrity Council’s label, and whether the credits in question do.
• Who drew the baseline. A baseline set by the methodology, or by a jurisdiction, is harder to inflate than one the developer chose.
• What is measured, not modelled. Metered fuel use, surveyed plots and measured soil carbon carry less risk than assumed adoption or assumed growth.
• How losses are covered. The size of the buffer and what happens after a fire or a felling.
• Which claim it supports. A credit authorised with a corresponding adjustment and one without it are different products.

Our restoration designs are measured on fixed plots, with the results published on the scorecard. How carbon on leased land is shared is on the carbon rights page, and the difference between kinds of credit is in avoidance and removal credits.

Standards and Integrity Council decisions change often. Check the current status of a methodology before relying on this page.

Related questions

What does one carbon credit represent?

One metric tonne of carbon dioxide equivalent that has been reduced or removed and verified by an independent validation and verification body. Verra's programme definitions describe its unit this way, and the other major standards use the same one tonne unit.

What share of carbon credits are real reductions?

A 2024 study in Nature Communications by Probst and colleagues reviewed evidence covering 2,346 projects and about one fifth of all credits issued to that date. It found that less than 16 per cent of the credits issued to those projects were real emission reductions. The figure applies to the projects studied, not to every credit.

What is the Integrity Council for the Voluntary Carbon Market?

An independent body that launched the Core Carbon Principles on 29 March 2023, a benchmark for high integrity credits. It assesses methodologies against them. In August 2024 it decided that credits from eight existing renewable energy methodologies would not receive its label, because they failed its additionality requirements.