At the Leaders' Climate Summit in April 2021, the United States, the United Kingdom, and Norway, and nine other companies announced the LEAF Coalition, a public–private partnership to protect tropical forests using forward commitments and price guarantees to pay at least $1 billion for emissions reductions at large jurisdictional (national and state/province) scales (Lubowski, 2021). Jurisdictional crediting systems, with nested accounting of smaller scale project-scale activities, offer the potential to more rapidly scale up emissions reductions as well as to ensure greater environmental integrity compared to stand-alone small projects (Schwartzman et al., 2021; DeFries et al., 2022). However, the design and implementation of such large-scale nested crediting systems raise numerous practical challenges that are only beginning to be considered.

This study focuses on addressing the risk to potential funders, jurisdictions, and nested project implementers (e.g., indigenous communities, private project developers, and landowners) when carbon credits or other result-based payments depend on the uncertain performance of an entire jurisdiction below a crediting baseline. In particular, this study develops a theoretical and simulation model to provide the first quantitative analysis to examine the use of insurance to mitigate this jurisdictional performance risk. The study examines the very particular risk posed within a jurisdictional and nested crediting structure of whether actors implementing activities that are nested within the jurisdiction can, in fact, receive the full market price for reductions or whether this price might be discounted due to jurisdiction-wide underperformance relative to the larger scale crediting baseline.

A conceptual model is developed to examine potential approaches to mitigate this risk based on the theory of individual choice under uncertainty (Mas-Colell et al., 1995; Hayashi, 2021). The model captures the situation in which the jurisdictional authorities are uncertain about the aggregate production that the jurisdiction can achieve due to various factors including the volatility of future agricultural commodity prices, which could increase the opportunity cost of protecting forests. These unexpected developments may lead policymakers and/or the implementing agents to fail to deliver the expected levels of emission reductions.

We calibrate the model to perform simulations to examine how the level of emission reductions might change with and without the use of revenue insurance as well as the willingness to pay to avoid risks of nonperformance. We also examine the option for jurisdictions to access a carbon buffer to hedge against performance risks. The results show that such approaches to insurance have the potential to play an important role in reducing the barriers to public and private investment to scale up emission reductions under jurisdictional scale programs.

There is growing scientific and stakeholder consensus that carbon finance for reducing deforestation should quickly be moving toward accounting at large, jurisdictional scales (e.g., countries or states), with crediting of smaller project-based activities “nested” or integrated within the larger accounting framework (e.g., COICA, 2023). Jurisdictional crediting aligns with the decisions on REDD+ under the UN Framework Convention on Climate Change as affirmed under the Paris Agreement, as well as the standards for REDD+ approved by compliance carbon markets, notably for international civil aviation as well as California.

Crediting systems based on the performance of an entire jurisdiction offer the potential to incentivize governments to pursue policy reforms and other measures that only governments can do. Large-scale accounting over both space and time brings important advantages in terms of environmental integrity, including capturing any leakage or shifts in deforestation within a jurisdiction (Schwartzman et al., 2021). Jurisdictional crediting also creates incentives for law enforcement, recognition of indigenous rights, and other policy measures that only the government can implement. Large-scale accounting also provides greater potential assurance of “additionality” (ensuring that the activity would not have happened anyway) by, for example, avoiding self-selection, which is inherent when smaller projects are free to voluntarily opt-in to participate in the program (van Benthem and Kerr, 2013). Large-scale accounting also supports systemic changes that can better assure “permanence” (ensuring that the climate benefits will not later be reversed) (Schwartzman et al., 2021; DeFries et al., 2022).

The LEAF Coalition represents an important step in terms of demonstrating both private sector and public sector willingness to pay for credits provided via jurisdictional systems as well as interest from various jurisdictions to develop credits. Nevertheless, there are various practical challenges to scaling up the implementation of jurisdictional and nested systems, including how to operate a multi-layer payment for environmental services program (Wunder et al., 2020).

An important feature of REDD+ is that it is result-based such that payments are made for results delivered, ensuring efficient use of climate resources and providing local actors with flexibility for implementation. However, when the achievement of a certain level of forest protection and resulting emissions reductions is uncertain, this creates a delivery risk for a jurisdiction and/or funders or investors supporting the jurisdiction that must make investments up front to later receive payments for certified emission reductions.

In this section, we propose a conceptual model to describe the jurisdiction's challenge in terms of managing ex-ante investments given the uncertain supply. We then examine potential approaches to mitigate the performance risk based on the theory of individual choice under uncertainty (Mas-Colell et al., 1995; Hayashi, 2021). The model incorporates the option for jurisdictions to access a carbon performance buffer by purchasing insurance, with an adjusted price based on historical performance to hedge against these performance risks.

We then calibrate the model to perform simulations to examine how the level of emission reductions might change with and without the use of the insurance mechanism and the impact on social welfare that jurisdiction would in theory be willing to pay to avoid such risk and the actual payment under an actuarial fair full coverage policy.

We now calibrate the conceptual model to examine (i) the emission reduction level with and without the use of the insurance mechanism and (ii) the impact on economic welfare in terms of a jurisdiction's willingness to pay to avoid the risk of not meeting its emission reduction goal. As per the stylized numerical example mentioned earlier, the minimum threshold requirement Q¯ is 35 million tCO₂. We later consider alternative scenarios with lower levels of this threshold.

International prices for REDD+ credits in the voluntary market were approximately $4 in 2021 (Ecosystem Marketplace, 2022) but have risen sharply. Voluntary market prices for credits are forecast to reach $80—$150 by 2030 (EY, 2022). We examine the global price range between $30 and $90 with $20 increments. The prices offered to domestic agents are set to vary in $1 increments up to the global price.

We consider the case where the jurisdiction chooses a target level of wealth associated with the emission reduction goal, Q_h^*, which is set at the 50th percentile of expected probability such that π_h=0.5. For each level of domestic price P_d, each potential Q^k and its associated probability p_k are derived using the probability density function as shown in Figure 2.

Studies have suggested that risk tolerance R typically ranges from 100 to 150% of net income (McNamee and Celona, 1990; Kirkwood, 2004). While the tolerance might be higher for a jurisdiction than for individuals, for purposes of illustration, we use 125%, the mid-point of this range. Under the global prices ranging between $30 and $90 with $5 increments at 125%, R is approximately $60, $108, $250, and $204 million when the minimum threshold requirement Q¯ as 35, 25, 15, and 5 million tCO₂, respectively. In practice, the risk tolerance level can be adjusted by a jurisdictional program authority to calibrate such analysis to the level of risk that it can bear.

As shown in Figure 4, with a minimum abatement requirement for crediting at 35 million tCO₂, global prices of $30/tCO₂ and $50/tCO₂ are not high enough to induce emission reduction levels (at the 50th percentile) that would be at or higher than the break-even requirement. In addition, with global prices of $70/tCO₂ and $90/tCO₂, break even is possible as long as the domestic price is within a certain range. Relatively, low domestic prices do not provide enough incentive for domestic agents to generate emission reductions above the break-even level. On the other hand, relatively high domestic prices raise the break-even quantity, Q^, to a level that the quantity of emission reductions realized at the domestic level is not enough to meet such a high level of break-even quantity. Figure 5 illustrates this situation with the global price of $70/tCO₂. A domestic price below $20 does not provide enough incentives for domestic agents to generate emission reductions above the break-even level, with at least a 50% probability. To ensure break even with a 50% probability, the jurisdiction needs to set the domestic price in a range from $20 to $38 per ton.

Even within this price range, the jurisdiction still faces uncertainties of not breaking even, since the results only show the emissions reductions achieved or exceeded with a 50% probability. This uncertainty could entail a prohibitive level of risk for committing public funding or obtaining private sector investment (e.g., through a bond issuance) on the back of future carbon market revenues. This is because any emission reduction effort that produces a quantity below the break-even threshold translates to a negative wealth impact on the jurisdiction. Without the option to purchase insurance, therefore, the jurisdiction may decide not to undertake any emission reduction activities at all.

We now simulate the provision of insurance, which could provide confidence to policymakers and potential investors by avoiding the risk of falling below break even. As an example, we consider the goal Q_h, to be the 50% of emissions reduction at each domestic price level, and restrict attention to those levels of the domestic price that are sufficient to ensure break even with a 50% probability. We start with examining scenarios based on a global price of $70/tCO₂ and a minimum abatement requirement for crediting 35 million tCO_2. The expected levels of emission reductions when Q is at or above the 50% goal and when Q falls below this goal are shown in Supplementary Figure S2. The jurisdiction's emission reductions range from 18 to 30 million tCO₂, with an average of 24 million tCO₂ in the case when it is not able to meet its goal. When the jurisdiction can meet its goal, its expected emission reductions increase between 49 and 77 million tCO₂, for prices from $20 to $38 per ton, that is, the jurisdiction's average emission is ~40 million tCO₂ lower when the jurisdiction is not able to meet its goal.

We now examine the wealth impact with and without the option to purchase performance insurance when a jurisdiction fails to meet its goal. In regards to the price of performance insurance, results from the simulation suggest, with Q¯ Q at 35 million tCO₂, the actuarially fair full insurance ranges between $28 and $42/tCO₂ for a global price between $70 and $90/tCO₂. By construction, under an actuarially fair full insurance policy, the jurisdiction's expected wealth remains the same as without the insurance policy after deducting the price paid for the insurance premium. However, the key here is that without the insurance, the jurisdiction may otherwise decide not to undertake any emission reduction activities at all given the excessive level of risk. This is because any emission reduction effort that produces a quantity below the break-even threshold translates to a negative wealth impact on the jurisdiction.

Figure 6 shows the negative impact on wealth, with and without the option to purchase performance insurance when the jurisdiction fails to meet its reduction goal Q_h (set at the 50th percentile of the probability distribution). In the case when the global price of $70//tCO₂ and  Q¯ at 35 million tCO₂, without an option to purchase insurance, this translates to a potential loss of $199 million if the jurisdiction does not meet its goals and a potential loss of $338 million if the jurisdiction decided not to undertake any emission reduction activities. In the case when the global price of $90//tCO₂ and Q¯ at 35 million tCO₂, without an option to purchase insurance, this translates to a potential loss of $320 million if the jurisdiction does not meet its goals and a potential loss of $952 million if the jurisdiction decided not to undertake any emission reduction activities.

As discussed earlier, in addition to making it possible for the jurisdiction to meet or go beyond the break-even emission reduction level, the provision of insurance could create an economic surplus when the jurisdiction's willingness to pay is lower than the actual payment to mitigate such risk. We now examine economic surplus. We also examine the effect on the economic surplus of setting the minimum abatement requirement for crediting Q¯ at 5, 15, and 25 million tCO₂.

Figure 7 shows the expected economic surplus for global prices of $70 and $90/tCO₂ and the threshold requirement at 35 million tCO₂. For an international price of $70, the economic average of the economic surplus is $81 million. For an international price of $90, the expected economic surplus jumps to $572 million, given that there is a jump not only in the price but also in the number of profitable outcomes. This is consistent with the expectation that as global price increases, with domestic price remaining constant, the quantity of emission reductions that could be covered by the international market increases. Figure 7 also illustrates the effect of changing the minimum abatement requirement for crediting. As discussed earlier, at the global price of $70/tCO₂ and a minimum threshold requirement of 25 million tCO₂, the jurisdiction will participate when the domestic price ranges from $20 to $38 per metric ton. When the minimum threshold requirement falls to 25, 15, and 5 million tCO₂, the price range expands to $11 to $51, $5 to $60, and $2 to $67 per metric ton, respectively. The results also indicate that as the minimum crediting requirement decreases, at a given global price and with access to insurance, the positive impact on consumer surplus to emission reductions increases. This is because there is a wider range of domestic prices such that the jurisdiction can meet break even as the minimum abatement requirement decreases.

Earlier, we noted that a jurisdiction that has established a record of meeting its threshold and break-even level could be eligible to access the insurance at a discounted price based on an assigned performance risk score. The better a jurisdiction's performance in meeting threshold and break-even quantity, the lower its performance score. As the score decreases, economic surplus increases. In our illustrative example, with a global price of $70 and $90 t/CO₂ and a minimum requirement of 35 million metric tons, a one-point decrease in performance risk score increases the average economic surplus by about $3 million.

Similar to reducing the minimum abatement requirement for crediting, a physical “buffer” of emissions reductions that could be tapped and sold to international buyers in the case of underperformance would provide another approach to insuring the risk of failing to meet a break-even level of reductions. This buffer could potentially be built up by jurisdiction over time, providing a form of self-insurance. Figure 8 shows the potential of such a buffer to unlock emissions reductions at the 50th percentile of supply for different levels of international prices, based on the different levels of domestic prices that the jurisdiction would be willing to offer that is consistent with break even, as shown in Figure 3. The buffer level is varied at 5 million ton intervals. No buffer is needed to allow reductions consistent with break even at prices of $70 and $90, but emissions reductions are not feasible at lower prices without a buffer. However, a buffer of 0–5, 5–10, 10–15, and 15–20 million tons, respectively, enables emissions reductions consistent with break even at international prices of $50, $40, $30, and $20/t CO₂. Progressively greater buffers allow for greater reductions at each price level. Notably, for most of the price ranges and buffer sizes, each million ton of buffer enables more than 1 million ton of emissions reductions (a purely one-to-one linear relationship with the buffer is illustrated by the 45-degree line in black).

The relationship between a buffer and the feasible emission reductions consistent with break even varies according to different percentiles of the supply distribution (Supplementary Figure S3). A buffer of 10–15 million tons is needed to ensure break even if it is necessary to have a 95% confidence level of supply. The required buffer size falls to 5–10 million tons at the 50th (median) and 75th percentiles, at 0–5 million tons at the 25th percentile, and no buffer at the 5th percentile of supply. This shows how greater reductions are feasible if the jurisdiction can take on more risk of not breaking even, or similarly if the jurisdiction can increase its confidence in the feasibility of different levels of supply.

There is increasing consensus around the environmental and social benefits of moving from small to larger scale emissions reductions under carbon market crediting systems for REDD+ (COICA, 2023). However, the mobilization of large-scale finance will be necessary to implement these approaches. This study demonstrates the critical role of performance uncertainty in constraining the potential for large-scale mitigation supply when payments are contingent on emissions reductions exceeding a threshold level and when there is a need to break even on payments to domestic actors. This would be the case of a jurisdiction that is budget constrained, rather than able to finance emissions reductions with its own domestic resources to meet climate and/or other environmental and social goals. This would also be the case for private investors or other private actors investing with the requirement of a commercial return.

One approach would entail lowering the performance requirements, but this would have tradeoffs with environmental integrity, also potentially limiting demand. Higher international prices could also compensate for the risk but could greatly limit the demand. Our numerical simulations likely overstate the level of prices that will be needed as it is based on cost curves estimated from the price sensitivity of historic land-use decisions rather than the range of government policies (e.g., law enforcement, designation of protected areas) that have demonstrated potential to reduce emissions at low budgetary costs. We also consider the case where the jurisdiction only seeks to cover its budgetary costs without considering the economic benefits to local actors. Different jurisdictions may also have higher levels of risk tolerance. Nevertheless, our analysis suggests that uncertainty over performance could require significantly higher prices to assure break even. Performance uncertainty is, thus, an important concern for REDD+ and other forms of large-scale emissions reductions. Even if uncertainties can be characterized, the risk of failing short on performance is likely to be a powerful hindrance to public and private investors that are seeking to finance high-quality and high-integrity emissions reductions at scale.

We illustrate how insurance has the potential to play an important role in addressing this risk and mitigating the barriers to large-scale public and private investments. We show that actuarially fair insurance would allow jurisdictions to deliver large-scale emissions reductions despite performance uncertainty. Based on our illustrative simulations, given sufficiently high carbon prices, jurisdictions would be willing to pay for insurance and could develop a significant welfare surplus from the purchase of insurance to mitigate uncertainties over performance. As jurisdictions can develop a track record of performance, insurance costs could fall, providing further welfare gains.

Major insurance companies have already identified promoting natural climate solutions, including forest protection, as a top priority for new insurance products as well as direct investments (e.g., AXA, 2022). However, it may be difficult for private companies to provide this insurance given unfamiliarity with the likely probabilities and potential concerns over the moral hazard, such that jurisdictions receiving insurance will have lower incentives to manage risks, particularly if insurance is not conditional on meeting a particular level of performance. A central feature of REDD+ is that it is performance-based. A concern could be that access to insurance might encourage jurisdictions to underperform. Given the high cost of insurance, this is less likely to occur if future insurance rates are conditioned on past performance, creating an incentive for jurisdictions to establish a successful track record.

The challenges faced by private insurers suggest a potential role for philanthropy and potential public and multilateral donors and development finance institutions to work with jurisdictions and private insurers to jump-start the mechanism. Organizations such as the World Bank, Green Climate Fund (GCF), and Development Finance Corporation, together with philanthropic and public donors could potentially act as reinsurers to facilitate the creation of insurance businesses for REDD+ and reduce insurance costs with a consequential increase of private investment in jurisdictional REDD+. They might also offer insurance directly to jurisdictions on a concessional basis, based on a clear definition, monitoring, and verification of the investments to be made by the jurisdiction. Philanthropic and public donors also have an important role to play in supporting jurisdictions in capacity development and other investments that reduce risks of non-performance.

While insurance could play an important role to unlock large-scale reductions, the creation of a performance buffer offers the nonlinear potential to unlock supply. Such buffer could be accumulated at the level of a jurisdiction over time (or across different actors within a jurisdiction) to enable self-insurance or could potentially be funded initially by philanthropic or public donors. A question will be the willingness of the buyers to pay for buffer tons to “top up” jurisdictional performance to ensure the break-even level of reductions is met. There is also potential for a pooled buffer across jurisdictions if buyers or other funders would be willing to accept reductions that are not tied to any particular location. Furthermore, the greater the flexibility of the buyers, the greater the potential to develop a buffer with a greater set of emissions reductions from different time periods and potentially different types of emission reduction activities.

Insurance and a performance buffer could not only be used as stand-alone strategies but also have potential synergies for a mixed strategy to reduce risk and deliver high-quality emissions reductions. Insurance could help enable emissions reductions that support the creation of a buffer over time. Furthermore, the creation of a buffer at the level of a jurisdiction could in turn make a jurisdiction more insurable. Potential insurance providers themselves may also build up a buffer as the basis for their willingness to provide insurance or act as re-insurers.

There is a growing awareness of the importance of credit quality and the importance of large-scale crediting approaches for protecting forests. Buyers of REDD+ emissions reductions are looking for emissions reductions that would satisfy strict jurisdictional crediting criteria such as the ART-TREES standards used by the LEAF Coalition. Higher quality standards may make emission reductions more expensive, but protection and restoration of tropical forests remain the most cost-efficient mechanism to reduce emissions, with a benefit-cost ratio of >5 (Fuss et al., 2021). In addition, REDD+ can provide a large volume of emissions reduction in a timely manner when it is needed most to narrow the emissions. Nevertheless, scaling up REDD+ requires both private and public investments.

We develop a conceptual model and numerical simulations that demonstrate the potential power of REDD+ performance insurance and buffer creation strategy as important instruments to scale up REDD+ supply under jurisdictional crediting programs. Our analysis demonstrates how uncertainty over future performance can significantly increase the cost and decrease potential investment when there is a performance threshold that must be achieved before crediting begins. We, then, show how insurance mechanisms are a solution that could reduce risks and unlock environmental and economic benefits.

We considered the case where the jurisdictional government is the primary actor and can obtain a form of actuarially-fair revenue insurance to mitigate the risk. We also examine the case of a performance buffer to mitigate risk. In practice, a combination of instruments could be used. There may also be important uncertainties about the performance of government authorities, rather than about more decentralized local agents. Performance insurance and buffers could thus be applied at different levels of performance from national, subnational, and actor-specific levels. A surety bond is one widely used approach in construction and public works projects where contractors must post a bond that provides insurance to investors or other clients against the risk of default (Jing et al., 2020; Giacomelli and Passalacqua, 2021). In the case of jurisdictional REDD+ and other climate mitigation, such a mechanism could be appropriate at the level of different local agents, project developers, or government authorities depending on the nature of the performance risk and who is closest to managing it.

The insurance and simulation model could be extended to consider multiple periods of crediting, including the potential resetting of the baseline over time. Additional research could also focus on empirical estimation of potential performance risks to support estimates of actuarially fair insurance rates. Research could also inform how risks can be mitigated through different program designs for jurisdictional REDD+ programs, as well as how these could be diversified at the international level. This could include spatial designs for reducing risks of reversals (McCallister et al., 2022), payments for landowners indexed to agricultural commodity prices (Engel et al., 2021), as well as financial approaches to address the underlying pressures on forests [e.g., via a revolving fund to support the intensification of cattle running on already deforested land instead of additional deforestation (Golub et al., 2021)].

High-integrity carbon markets and other forms of result-based finance will be needed to protect tropical forests and achieve climate mitigation at the scales and speed required. However, given a broad set of uncertainties at this early stage of market implementation, result-based finance approaches that require ambitious performance entail the risk that could significantly diminish government and private investors' appetite to commit capital at sufficient scales to unlock the required mitigation. Once these programs are underway, this should simplify further rounds of insurance by providing more data as well as emissions reductions that could potentially start building a buffer that could support future rounds of insurance. Private insurers and philanthropic and public donors have a critical role to play in kickstarting and building an insurance market that can de-risk these programs and unlock natural climate solutions and other forms of urgent climate and nature protection at scale.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

KKC, AG, and RL jointly designed research, developed conceptual model, conducted simulations, and wrote the manuscript. All authors contributed to the article and approved the submitted version.