Here's our findings:

1. Among the four most prominent international carbon credit marketplaces, the two largest and most prevalent, Verra and Gold Standard, have the lowest average project quality.

Figure 2. Project quality average ratings across each of the four major registries showing that CAR has the highest average quality by a marginal amount, but also has the lowest percentage of total issuances given a quality rating (plot data available here).

Here we note that the vast majority of projects are not rated (we used various imputation methods to obtain complete data across all project) and as such few conclusions can be drawn definitively, but this by itself highlights the need for ratings agencies to offer full market coverage. This may also suggest that the majority of projects on registries are not actively traded and as such, there is little value in assigning a rating to many projects.

Further, while the Gold Standard and Verra registries do tend to have the lowest project ratings on average (e.g. the lower likelihood that the project's stated carbon benefits are in fact real), the average quality gap is quite small (0.01 to 0.09, based on our project quality metric.)

2. Among the 10 best rated and 10 worst rated categories of projects, Wetland Restoration stands out as the highest average quality ratings, but among the lowest total issuance volume.

Figure 3: Wetland Restoration and Avoided Grassland Conversion stand out as the top two project categories with the highest average ratings of projects fitting those methodology types (plot data available here).

Many of the historic renewable energy-focused carbon projects stand out with lower average ratings likely driven by concerns about additionality, namely concerns that the carbon revenues weren't meaningfully determinative of whether the project was going to proceed or not. The number of projects of this type has also decreased since the 2020 decision of Gold Standard and Verra to exclude middle-income countries' grid-connected renewable energy projects from crediting.

Wetland Restoration stands out with potential. This category includes carbon projects related to mangroves, seagrass, and peatlands. It is clear to see why there is burgeoning interest in these types of projects (also called “blue carbon”) as ecosystems such as mangroves hold ~4x the carbon sequestration potential as any terrestrial ecosystem. In addition, mangroves and seagrass serve important ecosystem functions for marine animals and the livelihood of coastal communities.

Several notable and potential projects in this space include:

• The Blue Carbon Project Gulf of Morrosquillo in Cordoba, Colombia which was Verra's first blue carbon conservation project protecting and strengthening a 7,500 hectare site that Apple proudly sponsored
• The Indus Delta Blue Carbon project in partnership with the Government of Sindh to reforest 350,000 hectares of coastal lands in Pakistan that just recently started receiving credits
• The recently signed, non-binding MOU between Japanese firm Marubeni and the Philippines Department of Environment & Natural Resources to potentially reforest 40,000 hectares of Negros Occidental

3. Among the top 15 project developers by issued volume, only South Pole and Anew are developing in multiple project categories.

Figure 4. Thirteen of the top fifteen developers by total issued volume of carbon credits have either a sole focus in forestry or renewable energy. It is only the private-equity backed South Pole and Anew that are developing across a diverse range of project scopes (plot data available here).

The vast majority of project developers are focused on a single project scope. South Pole, with funding from Temasek, and Anew, with funding from TPG, are the only top developers with a diverse range of project scopes. Private equity is clearly moving into the carbon credit space with Rubicon Carbon seeking to raise $1B to build a platform and financial instruments to support buyers of carbon credits. Oak Hill Advisors and partners recently bought $1.8B of US forest land to use for carbon offsetting, following a $500M joint venture for the same effort only ~12 months before.

We believe there will be further consolidation of project developers and the creation of multi-scope project development entities that use similar back-office functions to produce project development documents and benefit from the diversification of project scopes among their portfolio of credit issuances.

4. There is no evidence that higher-quality projects are retired (e.g. claimed) at higher rates than lower-quality ones, suggesting a lack of transparency for buyers.

Figure 5. Plot of every retired project based on quality score and percentage of the project that was retired as of August 2022. No predictive power was found between quality score and the degree to which the project was retired (plot data available here).

While we had hoped to find a correlation here, there was none, even when looking just at the top 40 or top 100 projects by total issued volume. Given 3rd party rating agencies to support the offset buy-side (e.g. Sylvera) are only a few years old, it is perhaps to be expected that quality was difficult to discern on what project volumes to buy and retire throughout the decades-long history of the carbon offset market. We leave the exploration of this relationship across time as an area of investigation for future researchers.

It is also important to note that price is not recorded within the publicly available registry datasets, which may be causing unobserved variable bias in this relationship. Certainly, if registries were to require price disclosure at retirement, it would help level the playing field for project developers to discover transacted prices to benchmark for their pipeline and for corporate customers to know they are paying a fair price for a comparable offset. This would also reveal the incredibly high transaction fees (>15% in many cases) from offset brokers extracting value in trading and speculating on price changes in the period between issuance and retirement.

However, importantly, if price is the unobserved variable driving the non-relationship between project quality and project retirement rates, that is prima facie evidence for the lack of fungibility in the offset market. That is, it would be evidence that not all offset claims are equal and interchangeable, in the same way that if you found one person agreeing to sell a share of Berkshire Hathaway for $10,000 and another for $25,000, you’d reasonably conclude that not all shares are equal in some meaningful way.

5. High-volume retirees, that is, the largest global corporations claiming the benefits of carbon credit projects, do seem to filter and select for higher-quality projects on the whole.

Figure 6. Plot and table of the top 40 offset purchasers by total volume showing some predictive power that the higher the purchaser volume of retirements, the higher quality their portfolio on average. (plot data available here).

These figures show the behavior of the largest end retirers in our dataset, namely those that voluntarily self-identify in the international registries when they claim and retire a credit in the registry. It must be noted that the data quality across all the retirement data, however, is extremely poor with numerous issues:

• There were several large retiring entities that were for specific projects/initiatives for which we couldn’t identify the ultimate sponsor company that we have excluded from the above chart.
• The 2nd largest retirer is “Third Party”. We believe this happens when a carbon brokerage firm (who often also does origination) retires large tonnage on behalf of a large, unidentified third-party enterprise client. We believe this practice should be changed by the registries to drive further transparency.
• There is also far too much “missing data”. A shocking 16% of the entire retirement dataset by row count & 50% by volume contains no information on who retired it to make a claim, simply that there was a retirement by an unidentified party on a (mostly) identified date.
• We had to spend several weeks cleaning & standardizing retirer names to single recognizable entities. This was a combination of manual work and automated parsing/regex. We wish the registries required disclosure of retirer identity in a modern web form with data entry checks or a drop-down, so work like this doesn’t have to be done at the data layer.

Ultimately, we see no reason that a company retiring an offset to make a claim should not be clearly and consistently documented by name in the registry.

Among the lowest-quality, high-volume retirers we see many market participants making “carbon neutral” claims based on extremely suspect offsets. For instance, Primax Colombia’s offset purchases have been widely reported as a tax avoidance manuever using low-quality offsets to avoid a $5/ton carbon tax levy. For the other low-rated retirers, it is hard to discern whether the choice of credits is out of lack of transparency and awareness, or simply driving to achieve a carbon neutrality/net zero claim statement with the lowest total cost.

On the positive side, Volkswagen, Disney, and Eni are all among the top 10 retirers by volume and retiring with portfolio quality above what would be expected from the overall trend in the data. Whether they invested in dedicated carbon offset purchasing teams or simply allocated higher budgets to these efforts than others, we would love to highlight & champion similar companies like this. Among the best rated of small-scale buyers in our dataset is Ducks Unlimited, a non-profit supporting wetland restoration efforts to protect North American waterfowl, which has an active carbon program to protect and restore wetlands and grasslands.

Overall, we find it extremely positive that the largest purchasers in this space on average distinguish quality between carbon assets and believe this means that high-quality carbon assets have long-term price appreciation upside. As the market continues to deepen in volume and the tooling ecosystem continues to grow, we expect to see more discernment and diligence in the carbon procurement process.

Conclusion and next steps

The voluntary carbon market has many flaws in its implementation today, but it is critical to create a market that allows for payments to projects that genuinely protect and restore critical ecosystem services that sequester carbon. We have so little time and carbon budget left. The IPCC report makes clear how large and critical the ecosystem and protection opportunity is for 2030. True nature-based removals are provably scalable, self-maintaining & environmentally beneficial. While technological solutions like direct air capture should be part of the overall planetary carbon removal regime, their current cost is still $500+/ton and at minuscule scale in comparison to nature-based estimates of ~$20-100/ton and at massive existing scale.

To have a chance to limit warming to 1.5C by 2030, nearly every viable ecosystem preservation and restoration project must be funded and green-lit while still holding out hope for the long-tail possibility that energy-intensive, industrial carbon removal technologies will catch up. Projects with local communities at the center, high carbon sequestration rates, 100+-year storage potential, and demonstrable co-benefits make the best candidates for projects, with blue carbon showing early promise as a potential game changer.

Blue carbon ecosystems (mangroves, seagrass, and peatlands) store some ~10-24B metric tons of carbon, while smaller in scope that the total carbon storage of the world’s terrestrial forests and vegetation (~450-650B tons), these coastal ecosystems sequester carbon at ~3x the rate of the most productive tropical forests and over ~7x the rate of temperate forests. They can also store ~3-5x more carbon per hectare than tropical forests, locking the carbon away in the soils as opposed to above ground biomass, which always carries forest fire release risk, which will be an increasing concern on an ever-warming planet.

They are under immense development pressure too. According to NOAA, Since the 1940s, 30-50% of mangrove ecosystems have been lost and since the 1990s, 50% of seagrass meadows have been destroyed. As a result, while these ecosystems cover just the thin coastal fringe of ~5% of the tropical forest area, they account for ~20% of emissions from deforestation. These losses are especially acute in Southeast Asia which has the highest deforestation rate of any region globally, yet holds ~15% of the world’s tropical forests and ~33% of the world’s remaining mangrove forests. As a result, Bain and others see an >$10B opportunity in Southeast Asia alone with some of the highest-quality, investable carbon offset project potentials.