An Empirical Approach to Quantifying Forest Carbon Risks

Wildfires are everywhere these days, and their impacts are getting worse. What used to be a rare disaster feels commonplace. On a recent vacation in southern California, for example, my family and I drove by multiple wildfires. And on a call last week, a research collaborator checked whether a mandatory evacuation zone for a nearby wildfire in Colorado had expanded to include her home.

You might think that forest carbon crediting mechanisms would have long grappled with forest carbon losses—known as reversals, such as from wildfire, drought, or disease—and that they incorporate the latest science into their efforts. I wish I could say you were right.

While many mechanisms require forest projects to set aside a fraction of their credits in a common “buffer pool” to cover the risk of future reversals, the way they quantify reversal risk leaves a lot to be desired. Academic research finds that reversal risk parameters used in carbon crediting mechanisms are published without a traceable account of data or methods and that actual reversal risks are much higher than what the mechanisms assume (see, e.g., Badgley et al. (2022), Haya et al. (2023), and Anderegg et al. (2025)).

In a new Nature article led by Tsinghua University Professor Chao Wu, my colleagues and I set out to quantify how large carbon crediting buffer pools would need to be cover 100 years of forest carbon reversal risks across the United States (where forest projects are eligible to earn credits in California’s carbon market). Unfortunately, we find that natural reversal risks are about four to eight times larger than California’s rules assume.

The good news is that this kind of work shows how standard-setters can ground reversal risk analysis in empirical data on historical forest disturbances and project these risks forward under a changing climate. The bad news is that, at least so far, few carbon crediting mechanisms have been willing to adopt empirical approaches, let alone disclose the technical basis for their own reversal risk numbers.

One significant and important exception is the Paris Agreement Crediting Mechanism (PACM), on whose Methodological Expert Panel I serve. For a quick overview of the PACM and its expert panels, see my recent post; as always, these comments reflect my personal views only.

At its June meeting, the Methodological Expert Panel released several important documents, including a recommendation to the PACM Supervisory Body to adopt a new methodology for crediting clean cooking projects that reduce fuelwood or charcoal consumption. Consistent with the PACM standard on addressing non-permanence and reversals, the clean cooking methodology applies a reversal risk assessment and requires projects to contribute some of the carbon credits they earn to a mechanism-wide buffer pool.

The Panel also released a draft reversal risk assessment tool for public comment. The draft tool provides default parameters for reversal risks that vary by geography and are based on a global extension of the newly published Nature results, as documented in Appendix 2 of the draft tool and in a new preprint from Wu et al. A call for public input is open through July 24.

The promise of the Paris Agreement Crediting Mechanism is that it will identify best practices and raise the bar for quality in carbon markets. Adopting an empirical approach to forest carbon reversal risks would be a key step forward, both in terms of the PACM’s overall environmental integrity as well as the incentives it would send to market participants to price risks appropriately—rather than pass them on to the atmosphere.

Danny Cullenward

Shleifer Senior Fellow

Danny Cullenward is a Shleifer Senior Fellow at the Kleinman Center. He is an economist and lawyer focused on the scientific integrity of climate policy.