Insights · Gross to net
Buffer pool carbon credits: how much of an ARR project goes to the pool
Trees can burn, die in a drought or be cut. So a share of every tree-planting project's credits goes into a shared reserve: the buffer pool. For carbon credits under the VCS Program that share is set by a risk score, and under the current version of the tool it is never below 12%. Here is how the score works, what moves it, and why our screening plans with 15%.
What the buffer pool is
A credit for carbon stored in trees is only worth something if the carbon stays there. A tonne avoided at a power plant cannot come back. A tonne stored in a tree can, the day the tree burns.
The VCS Program handles this risk collectively. Each agriculture, forestry and other land use (AFOLU) project with a reversal risk deposits non-tradable buffer credits into the AFOLU pooled buffer account. In the VCS Standard's own words, buffer credits are cancelled “to cover carbon known or believed to be lost”. As a result, credits already issued are not cancelled when a project loses carbon; the pool absorbs the loss, although after an avoidable reversal the proponent must still make good the deficit. The pool works like insurance, with every project paying a premium in credits.
The premium is taken from the change in carbon stocks only. The VCS Standard sets the number of credits issued as removals, after leakage, minus the buffer, where the buffer is the risk rating multiplied by the stock change. Verra reconciles the pool periodically against verification reports and adjusts the risk criteria so that there are always enough buffer credits to cover losses.
- PoolAFOLU pooled buffer account
- Set byAFOLU Non-Permanence Risk Tool
- Minimum rating (v5.0)12, i.e. 12%
How buffer pool carbon credits are calculated: the risk tool
The current rules are in Verra's Procedure for Applying the AFOLU Non-Permanence Risk Tool, version 5.0, issued on 16 December 2025. The assessment itself is done in the digital tool on the Verra Project Hub, and the validation/verification body checks it.
The tool looks 100 years ahead. Risk factors fall into three categories, each with sub-categories:
| Category | Sub-categories | Category fails above |
|---|---|---|
| Internal risk | Project management, financial viability, opportunity cost, project longevity | 35 |
| External risk | Land and resource tenure, stakeholder engagement, political risk | 20 |
| Natural risk | Historic natural risk (likelihood and significance), projected climate change impact, sea-level rise where coastal | 35 |
Sub-category scores are added up within each category, and the three category totals are added for the overall rating. Mitigation measures can lower a sub-category score, but never below zero. The rating converts straight into a percentage: a rating of 35 means 35% of the stock change goes to the pool. Three limits apply:
- The minimum rating is 12, whatever the calculation gives.
- An overall rating above 60 fails; the project cannot be credited until the risk is reduced.
- Some single answers fail the whole assessment outright, without a score.
What moves the score for a tree-planting project
We will not reproduce the scoring tables here; the procedure is public and the digital tool applies them. What matters for a feasibility decision is which questions carry weight and which can fail a project outright. These are the ones we see most often on ARR sites.
- Fire, drought and pests. Every AFOLU project must assess at least fire, pest and disease outbreaks, extreme weather and geological risks. Each is scored on how often it has happened in the project area over the last 100 years and how much of the carbon stock a single event would destroy. Fire breaks, fuel removal and fire-fighting equipment count as mitigation; so does planting a mix of species against pests. A projected climate change factor can raise the score. Natural risk above 35 fails.
- Land tenure. If due process to find disputes over land and resource rights has not been done, the project fails. It also fails without binding agreements with everyone holding verified rights in the project area, including customary rights holders. Disputes over tenure or over access and use rights each add to the score, as does a history of government expropriation.
- Community. Where people living in or within 20 km of the project area rely on it for food, fuel, fodder, medicines or building materials, stakeholder engagement is scored. Grazing land that is planted is the typical case.
- Money. Financial viability is scored on the payback period and the share of funding secured until breakeven. A payback period of more than 20 years from the assessment fails. Opportunity cost compares the project with the most profitable alternative land use.
- Longevity. A project needs an adaptive management plan with a monitoring plan, and a management, financial and monitoring plan for its whole longevity. For projects that requested registration on or after 1 January 2024, a longevity under 40 years fails. Longer commitments, and legally binding ones, lower the score.
The score is not set once. Risk can be reassessed at later verifications, again looking 100 years ahead from that point, so a project that resolves a tenure dispute or builds fire defences can bring its rating down, and one whose situation worsens can see it rise. In a grouped project where a risk applies only to part of the area, the area is split into sub-areas with similar risk profiles and assessed separately.
Notice that most of this is about people, rights and money, not trees. A site with excellent growth rates and a contested boundary can end up with a much larger buffer than a slow dryland site with clear title, or fail altogether.
Reversals: what happens when carbon is lost
When an event happens that is likely to count as a loss event, the project proponent must report it under the Registration and Issuance Process. At the next verification the monitoring report restates the loss and calculates net removals for the period including it. The VCS Standard then separates two cases.
- Unavoidable reversals, such as a wildfire the project could not have prevented: the baseline may be reassessed, and buffer credits cover the loss.
- Avoidable reversals, such as clearing the trees: no further credits are issued to the project, or to any other project with the same proponents, until the deficit is remedied. The deficit is the full amount of the reversal.
In both cases the project keeps its boundary, including the burnt or cleared areas, and cannot claim credits for faster regrowth until the loss has been recovered. Permanence must be monitored for at least 40 years. Projects also sign an AFOLU Buffer Deed covering their obligation to compensate the pool in some cases. If a project stops submitting verification reports on time, part of its buffer credits is put on hold on the assumption that the carbon may have been lost.
Why our screening uses 15%, and what that means
A desktop screening cannot score land tenure, community reliance, funding or a fire record from satellite data. So it cannot calculate your risk rating. Instead, the screening applies a planning figure: 15%, the median buffer of 198 registered ARR projects in our registry data. A quarter of those projects sit at 20% or above.
In the anonymised sample study, a dryland planting project in East Africa, the 15% buffer took 10,467 tCO2e out of 69,779 tCO2e of gross removals over 40 years. That line sat between the 10% uncertainty deduction and the leakage estimate in the deduction table, and the report said where the figure came from. The full table is on the gross vs net carbon credits page; leakage, the larger deduction in that study, is covered in leakage in ARR projects.
Read the 15% this way. A project with clear title, secured funding, a long legal commitment and a low fire history may come in near the 12% minimum. A project with disputed land, thin funding or frequent fires can land well above 20%, or fail. The difference between 12% and 25% is 13 percentage points of the stock change, more than an eighth, so it belongs in a financial model as a range, not a single number.
What you can do before validation
The buffer is one of the few deductions you can influence early. Most of the mitigation the tool recognises is about preparation:
- Settle tenure and the right to operate first, with signed agreements, and document how you searched for disputes.
- Write the adaptive management and monitoring plan for the full project longevity, not just the crediting period.
- Build fire breaks and fire response into the budget, and choose species adapted to the site.
- Engage the communities who use the land, and agree benefit sharing before planting.
- Secure funding to breakeven; options contracts do not count as secured funding.
A carbon feasibility study will not give you the validated rating, but it shows where the buffer sits among the other deductions and which risks to work on first. For the remaining unknown in the net figure, the measured baseline, see the dynamic baseline in VM0047.
Frequently asked questions
It is a shared reserve. Under the VCS Program every AFOLU project with a risk of reversal deposits part of its credits into the AFOLU pooled buffer account. The buffer credits cannot be traded. When a project loses carbon, for example in a fire, buffer credits are cancelled to cover the loss, so the credits already sold stay valid.
With the AFOLU Non-Permanence Risk Tool. The project scores internal, external and natural risks, the scores are added up, and the overall rating becomes the percentage: a rating of 35 means 35%. Under the current procedure, version 5.0, the minimum rating is 12, and a project rated above 60 fails.
Because the real figure comes from a risk assessment that a desktop study cannot do. 15% is the median buffer of 198 registered ARR projects in our registry data, and a quarter of them sit at 20% or more. It is a planning assumption, shown as its own line so you can replace it.
They are not yours to sell. The VCS Standard allows projects that show longevity, sustainability and the ability to manage risk to become eligible for release of buffer credits. Plan on the buffer as a cost, and treat any release as upside you cannot count on.
You report the loss event and restate the loss at the next verification. Buffer credits are cancelled to cover it. If the reversal was unavoidable, the baseline may be reassessed. If it was avoidable, for example through poor management, no further credits are issued until the deficit is made good.