Rohini Pande in Science Magazine: Fixing forest carbon credits

This article first appeared in Science on January 11, 2024

The world’s forests store an estimated 861 billion tonnes of carbon – equivalent to approximately 100 years' worth of anthropogenic emissions at current rates – and, with good husbandry, could potentially store up to 226 billion tonnes more. Clearly, preserving and repairing forests is a vital element of mitigating climate breakdown.

However, forests worldwide are under threat, particularly in low- and middle-income countries; for example, 17% of Amazon forests have been destroyed, with another 17% degraded, and 11% of all carbon emissions come from deforestation. In an ongoing climate breakdown, preserving forests as a carbon sink likely has higher overall benefits than clearing them, selling the wood for charcoal, and using the space to grow both food and cash crops. But the benefits of preservation accrue slowly to all of humanity, whereas the benefits of deforestation all accrue more rapidly to local landowners, businesses, residents, and governments. Even where country-level authorities attempt to preserve forests, the financial and political incentives to clear land can often defeat attempts at regulation by a low-capacity state.

One answer is to transfer some of the global benefits of preservation to forest communities, paying them to preserve the forest that they would otherwise have cleared and thus preserving the forest while compensating communities for foregone income.

We know that, at least in some environments, paying communities not to clear forests can be effective; for example, Jayachandran et al. (2017) conducted a randomized trial of payments to landowners in Uganda and found that these payments significantly reduced forest degradation.

However, attempts to formalize this process internationally through a carbon market structured by the United Nations (UN), an initiative known as REDD+ (reducing emissions from deforestation and forest degradation), have been relatively slow and inadequate owing to both concerns about measuring and monitoring of carbon sink preservation – and so about the actual value of carbon credits – and inadequate international finance.

In the absence of formal transfers between countries on anywhere near the scale required and of a functional international market for carbon, a voluntary carbon market has emerged in which companies and other institutions buy carbon credits to, in theory, offset the emissions generated by their own activities. More than $36 billion has been invested in this market since 2012, and buyers include over 36% of Fortune 500 companies.

This market suffered a severe correction in 2023 when researchers examined voluntary carbon market forest preservation schemes across several countries and found that they have not meaningfully reduced deforestation. If a scheme does not preserve forests relative to a baseline, it does not prevent carbon emissions relative to that baseline – it does not provide "additionality" – and the voluntary carbon credits (VCCs) that were sold on the basis of the expected success of preventing emissions become worthless.

So, what went wrong?

Several parties are involved in the creation and trading of a VCC: the developer of a mitigation project or activity aimed at reducing or removing GHG emissions from the atmosphere; local stakeholders who, in theory, benefit from VCC payments in exchange for either ceasing economic activities that might damage the forest, engaging in active stewardship, or both; a crediting program that uses its own methodology to calculate the potential carbon benefits of mitigation projects and issues VCCs for the projects; an independent third-party auditor that verifies and validates the calculation of carbon credits from the mitigation project or activity; and last, the buyers who buy these VCCs to trade or to offset their own emissions.

In January 2023, it was reported that over 90% of the offsets certified by Verra, one of the largest offset crediting programs, were essentially worthless. In October 2023, the New Yorker published a story describing the failure of one carbon offset venture certified by Verra in Kariba, Zimbabwe. Two fundamental problems were identified: First, the projects had not prevented deforestation and carbon emissions anywhere near equivalent to the value of credits that had been sold; and second, in the Kariba project at least, it was not clear how much of the share of credit sales receipts promised to local communities had been delivered to them.

What went wrong? Well, first, there was no single agreed-upon method to estimate potential emissions reductions from certified projects, leaving the certifier with substantial leeway in what they chose. Second, they were paid according to the number of credits they certified, so there was an incentive to choose the methods that gave the highest number of credits rather than the most credible. Third, the project paid "independent" third-party auditors directly. When there is a competitive market for auditors, there is also an incentive for those auditors to collude with those being audited to give them the results they want. This phenomenon is not restricted to the auditing of VCCs. In research conducted in Gujarat, India, co-authors and I found that auditors conducting emission audits of polluting firms, whom those firms also paid, tended to find that firms emitted pollution at levels that clustered just below the legal limit.

Despite the recent setbacks that have been encountered, VCC markets, particularly those centered around preserving forests, are still considered a key policy in working to prevent climate breakdown. This year, Blue Carbon, a United Arab Emirates–based broker with no experience in carbon trading, is negotiating deals to manage forests across one-tenth of Liberia’s land; one-fifth of Zimbabwe’s land; and vast tracts of land in Kenya, Zambia, and Tanzania, with the intention of trading carbon credits from these areas. At COP28 in Dubai, the US government, along with the Bezos Earth Fund and the Rockefeller Foundation, announced the "Energy Transition Accelerator," which aims to use what they term "high-integrity" carbon crediting.

Meanwhile, the UN has been negotiating – although failed to agree on, at least at COP28 – rules for including VCCs in formal international carbon trading mechanisms through Article 6.4 of the Paris Agreement. The UK’s Climate Change Committee has set out guidance on developing functional voluntary carbon markets, and in the United States, the Commodity Futures Trading Commission has been developing its own rules for trading derivatives based on VCCs.

The rules in development all have a similar focus. They prioritize transparency in methodology, measurements, and finances; additionality, ensuring that projects genuinely lead to a reduction in emissions, or an increase in sequestration, that otherwise would not have taken place without the project; and robust quantification of emissions reduction and sequestration.

However, these rules don’t tell us how to implement forest carbon credit projects that abide by them. So how can we implement carbon credit projects that are actually as credible as these regulatory institutions insist they should be and, more importantly, that genuinely help to prevent climate breakdown and improve the welfare of forest communities in low-income countries?

First, we can remove perverse financial incentives, ensuring that certifiers and auditors are rewarded for accuracy, not for providing their clients with the most optimistic version of reality that they can argue for. We know that this is possible; in our paper on pollution audits in Gujarat, we conducted a randomized trial of a new system in which auditors for firms were chosen randomly from a pool, paid a set fee, and rewarded for accurate auditing. Independent testing showed that this system resulted in significantly more truthful audits. So, there is a strong argument for removing both direct payments and the ability to select firms both in the relationship between developers and certifiers and between certifiers and auditors. Spot checks of auditor activity, along with harsh penalties for those discovered engaging in fraudulent accounting, could keep the market honest. Furthermore, as the market expands, so will the benefits of an honest reputation for auditors.

Second, forest carbon can be difficult, if not impossible, to measure accurately, and some companies are now in the process of creating proprietary datasets on the potential carbon content of forests, which they propose to charge customers to access – essentially, aiming to compete in a market for information that will ostensibly provide customers with advantages in the offset market. This approach cannot work. On the contrary, it’s critical to have one universal standard for methodologies and measurements to ensure that these measurements and methodologies are conservative and to ensure that there is a clear institutional process for validating and standardizing new measurement methods as they are developed. This is because to trade a commodity, we need a standardized means to measure that commodity across the market. Think of the market in, for example, gold. There are standard ways to measure the quality (in karats) and weight of any item of gold. Then, there is the actual price of gold, at any specific quality, which is set by the market. So, there's value to proprietary information about the state of the market and the way prices might evolve, but if we make the ways of measuring quality or weight proprietary, too, then there is never any way to determine for certain the amount of gold in any item and no way to resolve for certain the price of that item at any moment. Additionally, the primary goal of generating carbon credits is to avert climate breakdown, so it is vital that every carbon credit traded genuinely represents a reduction of at least 1 tonne of carbon emissions and that the methodology used to calculate this is transparent. Although the measurement of the carbon content of any forest will never be perfect, a conservative methodology can create a "floor" that creates a statistical near certainty of "at least" a specific level of content. Together, these arguments create a strong case that all the data and supporting methodology that support the credibility of offsets should be made public and free to access. The GEOTREES project is attempting to do exactly this.

Third, we can take advantage of recent innovations in social science research techniques to prove additionality. Essentially, proving additionality requires credibly proving a counterfactual: that emissions would have been higher without the forest carbon project under consideration. Existing projects typically attempt to do this by using a single "control" area, which they contrast with the area of the intervention. The problem with this approach is that the control area often differs substantially from the intervention area, and that can mean that it is exposed to a different rate of deforestation and doesn’t provide a valid counterfactual. This is a problem that can be solved; academic economists have recently become very adept at creating plausible counterfactuals. The answer here might be to use the same technique as Jayachandran et al. have in their paper: a randomized, controlled implementation of the intervention in which an area was divided up into a large number of smaller parcels, and control and treatment were randomly assigned to these parcels. In the paper, the randomized implementation was intended to prove that payment for ecosystems can work in a forest context, but the same methodology can be used to prove the additionality of an emissions reduction project and the validity of credits generated by that project. Given that the intention of the intervention is to maximize carbon sequestration, parcels of land would not need to be divided 50/50 between treatment and control, as they were in the experiment; instead, careful design of randomization and shape of parcels should maximize treatment area while still providing sufficient statistical power to prove additionality. Jayachandran et al. also used the evaluation structure to test for leakage (a phenomenon in which when deforestation is prevented in one area, it moves to adjacent land) and, at least in their intervention, found that there was very little.

Fourth, we should be less concerned about another regulator priority: permanence. Forests are by nature vulnerable, and it is likely that when payments stop, as they almost certainly will at some point, deforestation will continue at similar rates – and indeed, this is what Jayachandran et al. found in their experiment. However, given that carbon pollution is cumulative, that the damage caused by climate breakdown is increased with higher atmospheric concentrations, and that we may be soon reaching concentrations that will lead to dangerous tipping points, there is substantial value in delaying emissions, even where they are not eventually prevented. Here again, Jayachandran et al. led the way, having made calculations of the benefits of the emissions that were delayed by their experiment. There’s no reason not to construct and trade credits on the basis of these delays, which will have some fraction of the value of permanent prevention of emissions.

Last, where payments are promised to low-income communities to compensate for the economic activities they renounce to preserve the forest, and where buyers of credits are often buying because of ethical concerns, it is vital that these communities actually receive those payments. In states with very low-capacity, dysfunctional, or absent government, it is not certain that this will happen; for example, a badly regulated, remote, and unaccountable developer might decide to keep an excessive proportion of the payments, as appears to have happened in the Kariba project, or a dysfunctional or hostile authority, which is not uncommon in the fragile states that are home to many of the world’s most valuable forest carbon sinks, might keep payments for itself and forcibly remove residents from the forest area. Recent work by Callen et al. provides one solution; they conducted a randomized evaluation of an intervention that used digital payments and blockchain verification to send funds to women in Afghanistan who were subject to hostile Taliban rule. They found that even previously tech-illiterate, low-income women swiftly learned to use the payment system, and no evidence was found of diversion by the authorities. Again, some of the techniques used to test this intervention as a proof of concept can also be used to prove that implementation is actually working as planned.

Implementing better-designed forest preservation projects that can provide credible VCCs does not, of course, prevent the fundamental paradox of voluntary offsets: they can be, and are, used to enable citizens in rich countries to justify continued high-carbon-emitting activities and to enable companies in these countries – particularly fossil-fuel companies – to justify continuing to sell high-carbon-emitting products and services. At this stage of the climate crisis, we need both to preserve and restore forests and to put a hard stop to unnecessary high-carbon-emitting activity, using offsets only to compensate for unavoidable carbon emissions. So economists also need to be thinking about how to implement effective regulation to prevent these wasted emissions.