Buffer pools: not fit for purpose

Reconsidering the buffer pool requirement made by registries in favour of insurance to provide more robust protection against reversal risks


Nandini Wilcke

20th Jan 2024



Re-evaluating the risk mechanism that registries were forced to put in place before in-kind insurance was available

A number of carbon credit registries maintain “buffer pools” of carbon credits, designed to protect against carbon reversals – wherein sequestered carbon dioxide is unintentionally re-released into the atmosphere after a credit is issued.

Buffer pools – how they are created, what they cover, how many credits are in them, and how credits are used – are not clearly explained or transparent. What we know is that the buffers are typically calculated using qualitative assessments, lacking the actuarial modelling and customized risk selection approach that CarbonPool brings to the table. As such, buffer pools are likely to inadequate to cover major losses – and thus buyers of carbon credits may be left exposed to reversal risk for many projects. That risk is clearly real, as demonstrated by extreme temperatures and increased wildfire activity around the world. It is also unclear how, and how often, the registries monitor for reversals.

One merit of buffer pools could also be their downfall: because the pools cover multiple projects, they mutualize risk. We expect this to eventually draw the attention of regulators, particularly as the value of carbon removal credits grows.

Buffer pools are, essentially, a form of insurance – but without the any capital behind them which is required to protect customers for unexpectedly bad outcomes. As the carbon market professionalizes, buffer pools are not likely to be an acceptable substitute for carbon insurance. Further, buyers who have staked their reputation on carbon credits from a forest that burns down are likely to face harsh questions if those forests burn down, and are unable to explain how and if the buffer covers the loss of the sequestered carbon.