Our starting point is to adopt the conservative view that all NBS-derived credits are likely to be impermanent. We distinguish short-term fluctuations in carbon stock, such as through deciduous leaf fall or the death of individual trees, from the directional release of additional carbon generated by a project, such as through the resumption of deforestation, a major disease outbreak or a change in the fire or climate regime. Impermanence is due to directional loss and can helpfully be conceptualized as the loss of additionality over time. interval, when the counterfactual pixels lose no carbon (as by now they have none to lose), while the project loses its remaining stock. Hence project addibility over this interval (α3; again, simply the difference in the change of the project and the counterfactual carbon stock) is −α1. This is how much previously accrued addibility is lost–and means that in this example all addibility is released over this third interval. The relative permanence of any credit can thus be assessed by considering whether the addibility it was based on is reversed, and when any such release occurs.
To illustrate this point, consider a stylized deforestation-reduction project (Fig. 1; note that the approach is generalizable to other NBS interventions and to different methods for constructing counterfactualals). The project’s addibility is assessed at the end of each of three time intervals by comparing the change in its stock of carbon with the change in stock of a counterfactual set of areas not involved in the intervention but matched to the project site in terms of initial carbon stock, exposure to drivers of deforestation and variables (such as governance) that are likely to predict adoption of conservation actions.
The next stage of the PACT framework links this addibility-based understanding of when impermanence arises with an assessment of the value of impermanent reductions in atmospheric GHG. One view is that if the policy goal is to achieve a time-bound target for limiting temperature increases, any drawdowns of carbon which reverse completely before that target date will not affect temperature at that point and so have limited value (except perhaps in helping the development of more permanent storage technologies)21. We take a different position and consider temporary drawdowns as valuable22. To see this, imagine a health policy motivated by people’s desire to live longer, and with a specific target of increasing the life expectancy of people born after 2050 to 100 years. Interventions that extend the life span of people alive today will not directly help to meet the target. But most of us alive now would benefit from even one extra year of life, so those
Over the first time interval the counterfactual pixels lose half their carbon while the project area loses none. Difference-in-difference analysis thus indicates that the project has generated addibility α1. Over the second interval the counterfactual pixels lose all their remaining carbon while the project ceases to be effective at slowing deforestation and so loses carbon at the same rate. Because changes in carbon stock are the same in the counterfactual and project pixels, no further addibility is generated (α2 = 0) and the overall additionality of the project is unchanged. Impermanence emerges over the final
interventions have social value. Our focus here is on the analogous social value of impermanent reductions in the damages incurred by climate change17,19,20,22,23.
The economic device we use for characterizing that value is the social cost of carbon24 (SCC)–the cumulative long-run cost of the damage caused by releasing one additional tonne of CO2e into the atmosphere, discounted into present-day terms. There are several well-known uncertainties associated with estimating the SCC25 but we use it here as the best-known way of translating future global warming into present-day utility. If the release of one tonne of CO2e has a value equal to the SCC, it follows that one tonne of CO2e permanently withdrawn from (or not emitted to) the atmosphere as a result of an offsetting intervention has an equal but opposite effect, and hence a present value (Vperm) that is identical to the SCC. For an impermanent offset, by comparison, the value of a one tonne drawdown is the SCC of a permanent drawdown minus the present-day cost of the damage caused by the subsequent release of that carbon, estimated from the SCC at the time of the release20. This logic assumes that the project has a small effect on temperature compared to the magnitude of warming from the industrial revolution.
In today’s terms, the damage cost from a release will always be less than the value of the initial drawdown because the rate of increase of the SCC is always less than the discount rate. Formal proof of this is provided in the Supplementary Information, but the intuition is as follows. An emission today results in a relatively constant and eternal small increase in temperature and an associated stream of marginal damages. The SCC is the sum of the discounted value of these marginal damages. An emission next year has an identical stream of marginal damages except that they are discounted by one year less (so the marginal damages have grown in value by the discount rate) and begin one year later (so do not include the cost of damages in the current year). Hence, while it might appear that the SCC increases by the discount rate, because the damages of the current year are now behind us and no longer included, the SCC in fact increases by less than the discount rate. impermanent credits−so, for illustration, this particular schedule pessimistically forecasts that over its second decade the project will lose carbon stock 1.5 times as fast as the counterfactual sites. Because additionality is released at a rate equal to the difference in change in carbon stock in the project and counterfactual sites (demonstrated in Fig. 1), half of the additionality is forecast to be released over this second interval (ˆt1,2, the additionality generated in period 1 which is estimated will be released in period 2; change in project stock −change in counterfactual stock = 1.5 −1.0 = 0.5). During the third interval the project is no longer operational, so the pessimistic forecast is that the project area will now lose carbon twice as fast as the counterfactual sites. Hence the loss of additionality over this interval (ˆt1,3) occurs twice as quickly as before and so, according to this pessimistic schedule, the first decade’s additionality is dissipated entirely by year 25.
Building from the framework of the SCC, if a release schedule of additionality can be estimated, the damage cost (Dtot) from these releases can be subtracted from the value of the initial drawdown to derive the present value of the impermanent offset (Vimp = Vperm −Dtot). We can then calculate the ratio of this value to that of the permanent drawdown of one tonne of CO2e (Vimp/Vperm) to derive the equivalent permanence (EP) of the offset. The inverse of EP (that is 1/EP) can then be used as a multiplier to decide how many present-day impermanent credits need to be purchased to be comparable in welfare terms to geological sequestration.
The ability to set realistic but conservative ex ante release schedules is central to the operation of the PACT framework. If they are too pessimistic, project providers will be deterred, but if they are too optimistic, purchasers will be deterred. In real-world applications, the forecasting of release schedules should be informed by empirical estimates of carbon fluxes over and beyond the lifetimes of comparable projects. Two further considerations are important at this point. First, the derivation of EP should in principle also include the value of the drawdown realized over the assessment interval (the triangle to the left of _ai_ in Fig. 3a); to aid interpretation we have omitted this complexity. Second, one can also make conservative corrections for leakage−the increase in emissions as a result of forgone food, timber or mineral production being displaced to non-project areas26,27. Combining any leakage correction with EP, one can then inform prospective offset buyers of how many impermanent credits constitute a PACT: a bundle of credits which is estimated to have at least the same present value climate benefit as a fully additional, permanent credit.
These ideas can be summarized diagrammatically (Fig. 2, for the same stylized project as Fig. 1). In terms of changes in carbon stock (Fig. 2a), the project successfully stops deforestation over the first time interval so there is net drawdown of carbon, _ai_. However, this additionality is fully released over the third interval (_aj_). In terms of social value (Fig. 2b), the present value of the project (Vimp) is the value of the initial drawdown (Vperm) minus the cost of the damage caused by the release of additionality over interval 3 discounted to its value at the end of interval 1 (Dtot). The EP of the additionality achieved by the project is then the ratio of this impermanent value (Vimp) to that of an equally additional but fully permanent drawdown (Vperm).
Setting out in greater depth how this approach can be operationalized, imagine a simplified, 20-year deforestation-reduction scheme (Fig. 3a; in practice release schedules would be described probabilistically and assessed over shorter time intervals; for a complementary mathematical account see the Supplementary Information). After a decade, ex post comparison of trends in carbon stock in the project and in a set of statistically derived counterfactual sites confirms that the project has generated additionality _ai_. A corresponding carbon credit c1 is issued, with an EP (EP1) based on an ex ante release schedule (Fig. 3b). It is important that this does not overestimate the value of
A third key element in the PACT framework is continued monitoring after a credit has been issued, to allow for ex post correction for the inevitable uncertainty and conservative bias in predicting reversals.
Returning to our example, suppose the project is reassessed ten years after the first credit issuance, as it draws to a close (Fig. 3c). Imagine that while deforestation in the counterfactual sites has continued, the project has done far better over its second decade than our pessimistic forecast and none of the anticipated deforestation has occurred. In this case, the project will have generated further additionality, denoted a2. However, the new credit issued for this interval, c2, should also include an amount equal to the release previously expected to occur during this interval (ˆ1,2), because its social cost has already been accounted for in the EP value assigned to the first credit (EP1). An anticipated release schedule and new EP value are then developed for this second credit (EP2; Fig. 3d), which might reasonably reflect a slightly more optimistic view of likely post-project releases, given the project’s better than expected performance over the last ten years. second decade’s credit c2 is therefore again calculated as the sum of its observed additionality over that period plus the amount of release of the previous credit that was predicted for this interval. This new credit is assigned its own EP (EP2; Fig. 3f), based on the same anticipated post-project release rate as that in Fig. 3b.
In contrast to the widely used buffer pool approach, this iterative system of tracking and accounting for releases creates an incentive to safeguard already credited carbon, because good post-credit performance increases both the magnitude of future credit issuances and their associated EP values (Supplementary Information). Importantly, however, if already credited carbon is released more rapidly than expected, this too can be corrected through deductions from future credits, and in extremis by withdrawal from a portfolio-wide insurance pool of credits (even after the project ends; Supplementary Fig. 1). However, adopting deliberately conservative release schedules should mean such situations will be uncommon. Conservatism also acts to reduce expectations of non-release placed on future custodians of already credited carbon, helping
An alternative and perhaps more likely outcome over years 10–20 is that carbon stocks do fall in the project area, but at a lower rate than anticipated (Fig. 3e). Additionality over this second interval α2 is less than α1, but because net release has still not happened, this
to alleviate intergenerational equity concerns about dealing with impermanence. schedule derived from an analysis embedded in a representative integrated assessment model28 (Supplementary Fig. 3). Under these assumptions, EP values for these projects’ first round of credits, if issued ex post today, would range from 0.26 to 0.39 (Fig. 4). Combining these EP estimates with headline prices for similar NBS offsets, themselves adjusted for probable overestimation of additionality and underestimation of leakage11–13,27, in turn suggests that PACTs derived from our archetypal projects would cost in the order of US$80–160 (Fig. 4).
Buyers clearly need to make direct comparisons across a diverse array of NBS and other offset classes7. The three-pronged PACT framework enables this by explicitly and transparently expressing the performance of diverse types of projects in a common currency that captures differences in the durability and hence social benefit of the net drawdowns they generate. To illustrate our scheme’s flexibility, consider three archetypal NBS projects (Fig. 4), this time lasting for 40 years and with more plausible—yet still purposely pessimistic—schedules of additionality generation and reversal. To ensure timely corrections for post-credit performance, we suggest the PACT framework would best be deployed over short, iterated assessment intervals (under five years), but for graphical clarity we focus here on a single assessment made a decade into each project.
Significantly, while these calculations indicate that fully offsetting emissions through NBS is substantially more expensive than current market prices suggest, such schemes still appear competitively priced when compared with wholly additional, permanent, geologically sequestered offsets. These reportedly average7 US$140 per tCO2e, but vary widely, with some currently selling at around US$1,000 per tCO2e (https://climeworks.com/subscriptions). This conclusion is insensitive to plausible changes in SCC schedule, release schedule and time horizon, although the cost of NBS-derived PACTs would increase substantially at very low discount rates (<2% per year; see the sensitivity tests in the Supplementary Information and Supplementary Figs. 2 and 4–8). Hence, despite the impermanence of their effects, nature-based interventions, which can also provide important biodiversity and rural livelihood co-benefits, may offer less costly ways of reducing climate damages than some well-known technological solutions.
Estimating the EP values of the credits issued after this first assessment again requires developing conservative release schedules. The first project (Fig. 4a) involves reduced deforestation and, for illustration, a plausible but pessimistic release forecast that previously credited carbon is lost at 10% of the counterfactual rate until the project ends, and at the counterfactual rate after that. Our second project (Fig. 4b) is a fast-growing timber plantation. In this case the release schedule anticipates that 1% of credited carbon is lost each year because of disease, that half of the remainder is lost as a result of wastage at harvesting, and that the wood products generated then last a further 40 years. The final example (Fig. 4c) describes a restored native woodland in a fire-prone biome, where a conservative release schedule reflects a 2% chance of it being lost entirely each year.
We suggest that more important than the direction of these preliminary findings, though, is the ability of the PACT framing to integrate real concerns about credit reversals into assessments of NBS (and indeed those of technology-based offsets at risk of reversal29). This facilitates project comparability and, by increasing accountability, has the potential to promote buyer confidence. This may in turn boost sales of NBS offsets to existing and new customers, although the higher cost of
Each of these schedules describes the anticipated complete release of the carbon credited after the first decade and is used to derive an associated EP value assuming a 3% per year discount rate and an SCC
PACTs compared with unadjusted NBS credits may discourage those buyers who are satisfied with low-integrity offsets. If demand for robust credits does grow, this should help lift the price paid for them, thereby encouraging more NBS projects to enter the carbon offset market—a critical policy goal.
In addition, tailoring and revising the estimation of EP according to the recent performance of a project (and others like it) should incentivize project providers to adopt actions likely to increase permanence— such as improving land tenure and reducing opportunity costs borne by local communities, for instance by boosting farm yields on already cleared land. If successful, these actions could generate additional benefits by enhancing project additionality, reducing risks of leakage of forgone production and hence emissions elsewhere27, and improving local livelihoods. Moreover, by being explicitly geared towards frequent low-cost analysis of remotely derived data, the PACT framework offers the twin prospects of greater accountability for offset buyers and reduced transaction costs of project proponents, as well as aligning directly with calls for digital monitoring, reporting and verification in carbon markets30. Continued monitoring would also enable separate ongoing accounting of the physical climate impacts of projects (essential for tracking progress towards temperature-based goals21). Crucially, such monitoring–if linked, as we propose, with ex post repayment for lower-than-anticipated releases–incentivizes project stakeholders to continue to safeguard already credited carbon into the future.
The increasing availability of near-time remote-sensing data will be key in continuously updating the information provided to offset purchasers about what they are buying. Procedures for estimating NBS additionality will need regular revision as counterfactual estimation techniques improve, socioeconomic drivers change and new national and sectoral commitments to stopping deforestation are made. Some NBS (and, indeed, technology-based schemes) will also become less additional if their costs fall so that they become financially viable without offset payments31. Methods for estimating permanence will need updating as our ability to forecast release schedules improves and as threats to emissions drawdowns change15. Techniques for estimating leakage will require further work, especially as trade expands such that carbon-emitting production, forgone as a result of project activities, becomes increasingly likely to be displaced far away from intervention sites26,27. The dynamic accounting central to the PACT framework means that it is readily capable of accommodating such new procedures and information.
Investors face trade-offs in deciding which offsets to buy. Well-designed NBS projects present singular opportunities for benefitting biodiversity and rural livelihoods5. Moreover, while NBS schemes may be more vulnerable to impermanence than some other offset classes, they can and do mitigate the social costs of climate change considerably. Our new generalizable and scalable formulation suggests how this contribution can be valued, enabling the direct comparison of nature-based and technological offset options for progressing towards net zero.
All data are available in the main text or the supplementary materials. For more information on PACT see www.cambridgepact.org.
The code for producing carbon release schedules and calculating EP is available on request.
This work was developed with support from the Royal Society, ESRC, NERC, Grantham Research Institute on Climate Change and the Environment, Frank Jackson Trust, Dragon Capital and the Tezos Foundation. We thank B. Balmford, G. Cerullo, A. Eyres, H. Hannable, S. Jaffer, E. Quigley, E.-P. Rau and C. Wheeler for their help and ideas.
All the authors conceived the initial idea. A.B., S.K., F.V., B.G. and T.S. developed the method. T.S. created the figures. A.B., S.K. and T.S. wrote the manuscript and all co-authors revised it.
A.B. is a trustee of the World Land Trust, a non-governmental organization that supports forest-based carbon projects. The Cambridge Centre for Carbon Credits (4C) has no commercial interest in carbon credits.
Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41558-023-01815-0.
Correspondence should be addressed to Andrew Balmford.
Peer review information Nature Climate Change thanks Per Kristian Rørstad, Susan Cook-Patton and Lucas Joppa for their contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© Springer Nature Limited 2023