The climate solution actually adding millions of tons of CO2 into the atmosphere

This story was co-published with MIT Technology Review.

Along the coast of Northern California near the Oregon border, the cool, moist air off the Pacific sustains a strip of temperate rainforests. Soaring redwoods and Douglas firs dominate these thick, wet woodlands, creating a canopy hundreds of feet high.

But if you travel inland the mix of trees gradually shifts.

Beyond the crest of the Klamath Mountains, you descend into an evergreen medley of sugar pines, incense cedars and still more Douglas firs. As you continue into the Cascade Range, you pass through sparser forests dominated by Ponderosa pines. These tall, slender trees with prickly cones thrive in the hotter, drier conditions on the eastern side of the state.

All trees consume carbon dioxide, releasing the oxygen and storing the carbon in their trunks, branches and roots. Every ton of carbon sequestered in a living tree is a ton that isn't contributing to climate change. And that thick coastal forest can easily store twice as much carbon per acre as the trees deeper inland.

This math is crucial to determining the success of California's forest offset program, which seeks to reduce carbon emissions by preserving trees. The state established the program a decade ago as part of its efforts to combat climate change.

But ecology is messy. The boundaries between forest types are nebulous, and the actual amount of carbon on any given acre depends on local climate conditions, conservation efforts, logging history and more.

California's top climate regulator, the Air Resources Board, glossed over much of this complexity in implementing the state's program. The agency established fixed boundaries around giant regions, boiling down the carbon stored in a wide mix of tree species into simplified, regional averages.

That decision has generated tens of millions of carbon credits with dubious climate value, according to a new analysis by CarbonPlan, a San Francisco nonprofit that analyzes the scientific integrity of carbon removal efforts.

The offset program allows forest owners across the country to earn credits for taking care of their land in ways that store or absorb more carbon, such as reducing logging or thinning out smaller trees and brush to allow for increased overall growth. Each credit represents one metric ton of CO2. Landowners can sell the credits to major polluters in California, typically oil companies and other businesses that want to emit more carbon than otherwise allowed under state law. Each extra ton of carbon emitted by industry is balanced out by an extra ton stored in the forest, allowing net emissions to stay within a cap set by the state.

As of last fall, the program had produced some six dozen projects that had generated more than 130 million credits, worth $1.8 billion at recent prices.

While calculating the exact amount of carbon saved by preserving forests is complicated, California's logic for awarding credits is relatively straightforward.

The Air Resources Board establishes the average amount of carbon per acre stored in a few forest types spanning large regions of the United States. If you own land that contains more carbon than the regional average, based on a survey of trees on your site, you can get credits for the difference. For example, if your land holds the equivalent of 100 tons of CO2 per acre, and the regional average is 40 tons, you can earn credits for saving 60 tons per acre. (This story will refer to each ton of CO2-equivalent as a ton of "carbon.") You must also commit to maintaining your forest's high carbon storage for the next 100 years.

These regional averages are meant to represent carbon levels in typical private forests. But the averages are determined from such large areas and such diverse forest types that they can differ dramatically from the carbon stored on lands selected for projects.

Project forests that significantly exceed these averages are frequently earning far more credits than the actual carbon benefits they deliver, CarbonPlan found.

This design also incentivizes the developers who initiate and lead these projects to specifically look for forest tracts where carbon levels stand out above these averages — either due to the site's location within a region, its combination of tree species, or both.

CarbonPlan estimates the state's program has generated between 20 million and 39 million credits that don't achieve real climate benefits. They are, in effect, ghost credits that didn't preserve additional carbon in forests but did allow polluters to emit far more CO2, equal to the annual emissions of 8.5 million cars at the high end.

Those ghost credits represent nearly one in three credits issued through California's primary forest offset program, highlighting systemic flaws in the rules and suggesting widespread gaming of the market.

"Our work shows that California's forest offsets program increases greenhouse gas emissions, despite being a large part of the state's strategy for reducing climate pollution," said Danny Cullenward, the policy director at CarbonPlan. "The program creates the false appearance of progress when in fact it makes the climate problem worse."

The Air Resources Board defended the program and disputed the central thesis of the study.

"We disagree with your statement that landowners or project developers are gaming the system or that there are inflated estimates" of greenhouse gas reductions, Dave Clegern, a spokesperson for the Air Resources Board, said in an email. Each version of the offset rules "went through our robust public regulatory review process," with input from the forestry industry, academia, government agencies and nonprofits, he added.

California's forest offset program is the largest in the country that is government-regulated. Other forest offset programs are voluntary, allowing businesses or individuals to purchase credits to shrink their environmental footprint.

CarbonPlan's study comes days after the Washington state legislature moved a cap-and-trade bill with an offset program to the governor's desk for approval. Oregon has also debated in recent months establishing a carbon market program that would emulate California's policy. In Washington, D.C., the Biden administration has signaled growing interest in harnessing forests and soil to draw down CO2. Businesses, too, increasingly plan to rely heavily on trees to offset their emissions in lieu of the harder task of cutting corporate pollution.

Forest offsets have been criticized for a variety of problems, including the risks that the carbon reductions will be short-lived, that carbon savings will be wiped out by increased logging elsewhere, and that the projects are preserving forests never in jeopardy of being chopped down, producing credits that don't reflect real-world changes in carbon levels.

But CarbonPlan's analysis highlights a different issue, one interlinked with these other problems. Even if everything else about a project were perfect, developers would still be able to undermine the program by exploiting regional averages.

Every time a polluter uses a credit that didn't actually save a ton of carbon, the total amount of emissions goes up.

Far from addressing climate change, California's forest offsets appear to be adding tens of millions of tons of CO2 into the atmosphere on balance, undermining progress on the state's long-term emissions goals.

"When you strip away all the jargon, you're left with a faulty set of assumptions that leave the door wide open to issuing meaningless offset credits," said Grayson Badgley, a postdoctoral fellow at Black Rock Forest and Columbia University, and the lead researcher on the study.

Cherry-Picking

CarbonPlan provided ProPublica and MIT Technology Review full and exclusive access to their analysis as it was being finalized. As part of that process, the news organizations sent the report to independent experts for review. The organizations also interviewed landowners, industry players and scientists and reviewed hundreds of pages of documents, including the project plans submitted by developers. CarbonPlan collaborated on the study with academic experts from the University of California, Berkeley, Columbia University and other institutions.

The study itself wasn't designed to assess whether developers or landowners are intentionally cherry-picking sites that stand out from regional averages, stating only that the system "allows for" developers to select such land. But the researchers themselves say that the level of excess crediting and the clustering of projects in certain areas suggest that industry players have gamed the system.

One form of cherry-picking identified by the researchers involves geographic boundaries. In the case of Northern California, the state's offset program established a dividing line that separates that coastal strip of redwoods and Douglas firs from an inland region that spans more than 28,000 square miles.

The board's rules state that tall mixed-conifer forests in the coastal region store an average of 205 tons of carbon per acre. For the neighboring inland region, the agency set the corresponding regional average at 122 tons per acre. The figure is lower because it includes more trees with less carbon, such as Ponderosa pines, which dominate the eastern end of the inland region and are all but absent on the coast.

But where the two regions meet, the forest on either side is virtually identical in many places, storing similar amounts of carbon. That means a project developer can earn far more money by choosing a site just east of the border, simply because they can compare the carbon in their forest against a lower regional average. For instance, maintaining a 10,000-acre forest of coastal redwoods and Douglas firs with carbon levels of 200 tons per acre could earn zero credits west of the line, or 624,000 credits east of it. The choice is between no money and more than $8 million.

To claim the most credits possible, for the full difference between the carbon on their land and the regional averages, developers or landowners must show that it's legally and financially feasible to log down to those regional averages. The averages are effectively a stand-in for the way that similar forests are typically managed in an area.

A dozen projects are located in Northern California, almost entirely lined up along the western edge of the inland zone where the carbon-rich trees are juxtaposed against the lower regional average.

"What we're seeing is developers are taking advantage of the fact that the big stuff and the scrubby stuff have been averaged together," Badgley said.

Once an offset project developer and landowner decide to work together, the developer will generally shepherd them through the process in exchange for a fee or share of the sales of the credits generated — an arrangement that can be worth millions of dollars.

One of the most prolific project developers in the California system is an Australia-based timberlands investment company called New Forests. The company and its affiliates have worked on eight projects located almost entirely along the advantageous side of the border, as well as six elsewhere. CarbonPlan, in a separate analysis done for the news organizations that wasn't included in the study, found that nearly all earned dubious credits, adding up to as much as $176 million worth.

A large share of those credits came from a single project outside California that profited from a glaring mistake in the rules. New Forests' affiliate, Forest Carbon Partners, helped the Mescalero Apache Tribe develop a forest offset project in New Mexico. The project earned 3.7 million credits worth more than $50 million, largely because it was located in an area where the Air Resources Board had set an erroneously low regional average.

Another form of cherry-picking involves tree species: Developers can seek out tracts with particular trees that store far more carbon than the surrounding region.

According to the study, one project in Alaska consists almost entirely of giant Sitka spruces, yet the local regional average was calculated from a wide mix of trees, including species like cottonwoods that store far less carbon. The project earned significantly more credits than it should have due to the flaws in the system, the study said. The project owner didn't return requests for comment.

Preserving especially carbon-rich forests is good for the climate, in and of itself. But when the trees in the project area bear little resemblance to the types of trees that went into calculating the regional average, it exaggerates the number of credits at stake, CarbonPlan's study found.

Mark Trexler, a former offsets developer who worked in earlier U.S. and European carbon markets, said the board should have anticipated the perverse incentives created by its program.

"When people write offset rules, they always ignore the fact that there are 1,000 smart people next door that will try to game them," he said. Since the board set up a system that "incentivizes people to find the areas that are high-density, or high-carbon, that's what they're going to do."

To estimate the extent of overcrediting in California's program, CarbonPlan calculated its own version of regional averages for each project. The researchers drew on the same raw data used by the Air Resources Board, but only used data from tree species that more closely resemble the particular mix of trees in each project area.

In total, 74 such projects had been established as of September 2020, when CarbonPlan began its research. CarbonPlan was able to study 65 projects that had enough documentation to make analysis possible. All received credits for holding more carbon than the regional average.

The researchers found that the vast majority of projects were over-credited, but about a dozen would have received more credits under CarbonPlan's formula. Those included two New Forests projects, which would have earned as much as an additional 165,000 credits.

The news organizations sent officials at the Air Resources Board a copy of the study and its detailed methodology weeks before publication. Clegern declined multiple requests to interview board staff and responded only in writing.

He did not address CarbonPlan's calculations. "We were not given sufficient time to fully analyze an unpublished study and are not commenting further on the authors' alternative methodology," he wrote.

The outside scientists who reviewed the research on behalf of ProPublica and MIT Technology Review praised the study.

"It's a really analytically robust paper and it answers a really important policy question," said Daniel Sanchez, who runs the Carbon Removal Laboratory at UC-Berkeley. While close observers are well aware of numerous problems with California's forest offset rules, "they're revealing a deeper set of serious methodological flaws," he said.

None of the reviewers pointed out any major technical or conceptual flaws with the paper, which has been submitted to a journal for peer review.

'A Significant New Commodity Market'

In early 2015, an offsets nonprofit hosted a webinar highlighting how Native American tribes could participate in California's program.

One speaker was Brian Shillinglaw, a Stanford-trained lawyer and managing director at New Forests who oversees the company's U.S. forestry programs. The company manages the sale of carbon credits, sells timber and on behalf of investors manages more than 2 million acres of forests globally, a portfolio it values at more than $4 billion.

New Forests also manages its affiliate, Forest Carbon Partners, on behalf of an institutional investment client it declined to name. Forest Carbon Partners finances offset projects and shepherds landowners through the process of applying for California's offset program.

"The bottom line is the California carbon market has really created a significant new commodity market," Shillinglaw said during his presentation. He said the program is something "many Native American tribes are very well situated to benefit from, in part due to past conservative stewardship of their forests, which can lead to significant credit yield in the near term."

Translation: Because many tribes have logged less aggressively than their neighbors, their carbon-rich forests were primed for big payouts of credits. Under Shillinglaw, New Forests or Forest Carbon Partners have helped to secure tens of millions of dollars' worth of credits for native tribes.

Among the 13 New Forests projects that CarbonPlan researchers were able to analyze, between 33% and 71% of the credits don't represent real carbon reductions. That's nearly 13 million credits at the high end.

"Although we cannot prove that New Forests acted deliberately on the basis of our statistical analysis, in our judgment there is no reasonable explanation for these outcomes other than that New Forests knowingly engaged in cherry-picking behavior to take advantage of ecological shortcomings in the forest offset protocol," said Badgley, the lead researcher.

New Forests managed the first official project in California's program, registering 7,660 acres of forest land on or near the Yurok Reservation, which runs more than 40 miles along the Klamath River near the top of that West Coast cluster of projects. The state issued more than 700,000 credits to the project for its first year, worth $9.6 million at recent rates.

State officials have pointed to the tribe's participation as a triumph of the program. In 2014, the board released a promotional video that showed the meticulous work of measuring trees in the Yurok project. James Erler, the tribe's then-forestry director, explained how offsets enabled the tribe to reduce logging. Near the end of the video, Shillinglaw appears in a sunlit forest, wearing a collared shirt and a New Forests-branded jacket.

"It's a beautiful watershed," Shillinglaw said over footage of a running stream and an elk standing before a thicket of trees. "This is the Yurok Tribe's ancestral homeland, and in part due to the carbon market will be managed through a conservation approach."

CarbonPlan estimates the project earned more than half a million ghost credits worth nearly $6.5 million.

Here's why the researchers say it was over credited:

The boundary dividing California's coastal and inland regions runs through the middle of the reservation. The carbon-rich forests on either side of that line are similar, filled with large Douglas firs like most of the coastal region. But more than 99% of the forest designated for preservation falls within the inland zone, where average carbon levels are much lower. The fact that the project was located in the most carbon-rich area of that zone enabled the landowners to earn an exaggerated number of credits.

At least one person involved in the Yurok Tribe's forest offset efforts was aware of how geographical choices swing the credits that can be earned.

Erler said during a 2015 presentation at a National Indian Timber Symposium that the tribe had the "distinct pleasure" of having the boundary run through its territory.

"You can take the same inventory data and apply it to the California Coast" — the region to the west — "and it doesn't come out with the same numbers as you do if you cross the street," Erler said at the conference, captured in a YouTube video posted to the Intertribal Timber Council's channel. "Vegetation may be the same, but it changes."

Badgley said that while the researchers can't speak to the intentions of any actors involved, it's clear that this project "benefited from over-crediting and that the Yurok Tribe's forester was aware how the specific aspects of the protocol rules our study criticizes led to beneficial outcomes."

Erler didn't respond to a list of emailed questions.

In an emailed statement, Yurok spokesperson Matt Mais said that the property was the only land the tribe had available to enroll at the time and strongly denied the tribe engaged in any sort of gaming of the system. He didn't respond before press time to a subsequent inquiry asking why the rest of the tribe's land wasn't available for the offset program.

Over the last decade or so, the tribe has slowly reacquired tens of thousands of acres of its ancestral territory, in and around the watershed of Blue Creek and other streams that sustain migrating salmon, from the Green Diamond Resource Company, a major Seattle-based timber business. The complex multistep land deals were done in partnership with the nonprofit Western Rivers Conservancy and financed through government grants, philanthropic donations and the sale of the tribe's offset credits.

"As we have recovered additional forestlands, we have enrolled additional acreage in California's climate programs in support of our Tribe's strategic goals including protecting salmon habitat, sustaining the revitalization of our cultural lifeways, and facilitating economic self-sufficiency," Mais wrote.

"It's insulting to claim that the Yurok Tribe has 'gamed' or 'exploited' California's climate regulations," he added. "Equally important, it's concerning that elite institutions now criticize us for legally and ethically using a program that was created to protect mature forests and then using those funds to purchase and restore more forest land that was, at one point, ours."

New Forests defended its practices in emailed responses to questions, arguing its projects have preserved existing carbon stocks and removed CO2 from the atmosphere through subsequent tree growth "as confirmed via third-party verification."

In a statement, the company said it has worked on projects in numerous areas, not just along the program's regional boundaries. The company said its projects "have protected and will enhance carbon storage on hundreds of thousands of acres of forests," adding that one project with the Chugach Alaska Corporation enabled the permanent retirement of a significant portion of the coal reserves in the Bering River Coal Field in southeastern Alaska.

New Forests follows the board's "scientifically-accepted regulations to both the spirit and letter of the program," the company said in a subsequent statement. "New Forests is proud of the forest carbon projects we have developed under California's climate programs — they have generated positive environmental impact and furthered the economic and cultural objectives of the family forest landowners and Native American tribes with whom we have worked."

New Forests didn't respond to numerous additional inquiries, including direct questions about whether it was gaming the rules of the program.

In an emailed response, CarbonPlan stressed that its paper criticizes the design of the program — not the Yurok Tribe or other landowners. Nor does it allege anyone has broken the rules. Its analysis doesn't consider or depend on the intent of any forest owners, who can benefit from flaws in the rules whether they intended to or even know about them.

"We recognize the injustices experienced by the Yurok Tribe, including the seizure of their historical lands by the United States government and its citizens," the nonprofit stated. "We also recognize the Yurok Tribe's legitimate interest in securing resources to repurchase lands that previously belonged to the Tribe and its people."

An Open Secret

Chris Field, an environmental studies professor at Stanford University, was co-author of a 2017 study that found California's program was helping to prevent emissions on balance by reducing logging. About 64% of the 39 projects studied were "being actively logged at or prior to project inception."

Field said the state program is "relatively well-designed to address key issues," but said it can and should be improved.

He added that there are firm limits on the role that offsets can play in California. From now through 2025, state polluters can only buy offsets to cover as much as 4% of their carbon emissions; from 2026 to 2030, that ceiling rises to 6%.

But those numbers understate the critical role of offsets in California's cap-and-trade program, viewed by some as a model for market-based climate policy.

Under that program, California sells permits that allow certain industries to emit greenhouse gases, with each permit worth one metric ton of CO2. The state also regularly gives away a certain number of permits to various regulated companies. The total number of permits, called a "cap," declines over time.

Polluters can also purchase permits from other companies with extras to spare, which constitutes the "trade." Or they can buy carbon offset credits, which cost slightly less than permits.

To participate in the offset program, landowners must hire technicians to survey the trees on their land, then take data such as tree type, height and diameter and plug it into equations to estimate the carbon stored per acre.

Most of the credits are distributed during the initial stages of a project, which can help to repay setup costs. Projects can also earn additional credits over time as the trees grow and absorb CO2, but those credits accrue slowly, and are dwarfed by the initial credits given to forests with more carbon than the regional average.

The type of forest projects that CarbonPlan analyzed account for 68% of all credits issued by the Air Resources Board since the program's launch, far eclipsing other types of offsets like capturing methane from dairy farms or coal mines, CarbonPlan found.

Cap and trade is designed to slash the state's carbon footprint by 236 million tons of CO2 over the next decade, about a third of the cumulative reductions needed to meet the state's emissions targets over that time.

Barbara Haya, who leads the Berkeley Carbon Trading Project at UC-Berkeley and is a co-author of the CarbonPlan study, calculated that up to half of those cap-and-trade emissions cuts could come via offsets.

Haya said these cherry-picking practices have been an open secret. The study is "revealing to everyone what a lot of people in the industry understand," she said.

Conservation vs. the Climate

Supporters of forest offsets say no system is perfect, and that focusing solely on the carbon math overlooks the incentives offsets create for protecting forests.

Field said offset systems should balance two goals: ensuring real emissions cuts, and creating ways to fund forest conservation. If CarbonPlan's study shows projects are gravitating toward high-carbon forests, then those are exactly the types of trees you'd want to save "if you have a conservation agenda," he said.

Cody Desautel, president of the Intertribal Timber Council, a Portland-based nonprofit consortium of native tribes, said that offset programs have provided critical financial flexibility for tribes. They've allowed them to buy back historic land, build needed infrastructure, create jobs for members or simply save up money for financial security. But above all, they've created incentives to manage forests in sustainable ways, he said.

"Tribes are very conservation-minded," said Desautel, who is also the natural resources director for Washington's Confederated Tribes of the Colville Reservation, which operate an offset project under California's system. "Their practices are largely based on what's best for the ecosystem, not what makes the most sense economically. And there's never been any value to that management approach in the past. These carbon projects provide an opportunity to value that."

He added, "If there's no value to owning forest land, it probably won't be forest land long into the future."

The Yurok Tribe's offset projects have clearly helped in these sorts of ways, even if they didn't provide the full promised carbon benefit.

The tribe has said it is using the acquired land and funds to restore its old-growth forests, produce traditional foods and basket-weaving materials, create a salmon sanctuary and improve habitat for endangered or culturally important species like the coho salmon, northern spotted owl, blacktailed deer and Roosevelt elk.

"Our partnership with New Forests will provide the Tribe with the means to boost biodiversity, accelerate watershed restoration, and increase the abundance of important cultural resources like acorns, huckleberry and hundreds of medicinal plants that thrive in a fully functioning forest ecosystem," Thomas P. O'Rourke Sr., then-chairman of the Yurok Tribal Council, said in a statement at the time.

But if the societal goal is preserving forests, it would be simpler and more effective to describe it accurately and fund it directly, said Haya, the UC-Berkeley expert. As soon as these forests get tied up in an offset program, the carbon math does matter, because every additional ton purportedly preserved in trees enables polluters to purchase the right to generate an additional ton of CO2.

Forest offsets appeal to the public partly because of what academics call "charismatic carbon" — they offer a feel-good story of environmental and social good.

"Any good conservation advocate would tell you there's a desperate need for more funding, and we agree entirely," CarbonPlan's Cullenward said in an email. The "problem isn't that conservation is bad, it's that the system of carbon offsets channels these real needs and sincere hopes into a system that grinds it all up and spits out garbage on the other side."

"The Best Bang for the Buck"

California's Air Resources Board approved the forest offset program's official rules in 2011, after years of discussions with dozens of experts, including government scientists and staff from conservation groups.

In adopting it, the agency relied heavily on Climate Action Reserve, a nonprofit that created programs with voluntary offset credits. The nonprofit, which continues to advise the agency, led an effort to calculate regional carbon averages as part of an initiative to update its voluntary offset rules.

To do so, the nonprofit used data from the U.S. Forest Service, which surveys tens of thousands of forest plots nationwide. The nonprofit grouped data from different tree species and combined data from various geographic zones into larger regional areas called supersections. This simplification allowed the Climate Action Reserve to create a set of common baselines that estimated the amount of carbon stored in typical privately owned forests. The baselines take into account such forest uses as logging.

But the use of these broad averages obscured real differences on the ground. Some industry insiders and researchers began to notice that landowners and developers routinely located their projects in areas where the specific tract of forest differed greatly from the regional averages.

Zack Parisa, chief executive of the carbon offsets company SilviaTerra, previously consulted for project developers and landowners enrolling forests in California's system. But he said he stopped out of frustration, after seeing the ways it was regularly being gamed, including the cherry-picking techniques CarbonPlan highlighted.

Parisa said he doesn't blame landowners or project developers, who are acting out of rational self-interest.

"If someone shows up and is offering a contract to buy carbon and it doesn't require them to change anything about how they manage the forests, that's free money and they'd be stupid not to take it," he said.

"I'm not hunting for a villain here," Parisa added. "Of course they look for the best bang for the buck."

In addition to New Forests, other developers also worked on projects where favorable boundaries and forest types boosted the credits that could be earned, according to CarbonPlan. Those include Bluesource and Finite Carbon, which BP purchased a majority stake in late last year. The researchers found that those two developers' projects, taken together, generated up to 24 million credits that don't represent actual carbon reductions.

New Forests, Finite Carbon, Bluesource and other subjects of this article were provided the full study and an accompanying paper describing its methods.

Finite Carbon declined to address detailed questions, but stressed that the Air Resources Board and an independent auditor found that their projects were in compliance with the rules.

In a statement, the company said there were "unanswered questions" about the CarbonPlan study's methodology, adding, "however we cannot comment further on it as the underlying raw data is not currently available for public review."

Emily Six, the marketing and communications manager for Bluesource, denied the company had gamed the rules in any way.

In an email, Six said California's program actually undercounts the carbon preserved through projects by not crediting the amount stored in other parts of the forest like soil, shrubs and foliage. She also stressed that without offsets, some landowners could have chopped down their forests to carbon levels well below the regional average.

"Deliberately overstating climate benefits would run counter to our very purpose for existence," she wrote. "Bluesource exists to improve the world by improving the environment."

The experts who wrote the original offset rules relied on the only national forest dataset available, from the U.S. Forest Service's Forest Inventory and Analysis Program, said Constance Best, co-founder of the Pacific Forest Trust. The conservation nonprofit was closely involved in the creation of the early program and participated in it.

Best said it was necessary to create carbon averages for larger regions and forest types because there wasn't enough fine-grained data to ensure accuracy at highly local levels. She disputed CarbonPlan's claim that its researchers had created a better way of calculating regional averages, since their method required relying on a smaller number of forest plots.

"The reason some super sections are large is to assure the data is more accurate," Best said in an email. "So their solution creates more problems."

In a separate note, she said: "The paper you shared has a strong editorial bias that undermines its findings and makes me question their data and analysis. It deliberately exaggerates what they present as smoking gun over-credited projects."

In an emailed statement, CarbonPlan acknowledges that using fewer forest plots entails some uncertainty. But the researchers stressed they clearly accounted for it by providing a range of results, and maintained their findings are more accurate because they considered the specific mix of tree species in each project. CarbonPlan also shot back at the allegation of bias: "Having done our work on the basis of extensive public program records, and with fully reproducible methods, data, and code, we are confident that other researchers are capable of judging our paper on its merits."

While the board has updated regional averages based on more recent forest data, critics say efforts to address more fundamental problems have been thwarted.

Researchers and activists also worry about the close ties between the Air Resources Board and the groups that now profit from the program.

For example, whenever a landowner wants to enroll a forest tract in California's program, they open an account at Climate Action Reserve or two other nonprofits that have received the board's blessing to review the documents.

If the project clears the Climate Action Reserve's review and a subsequent audit by the state board, the nonprofit charges 19 cents for every credit issued. For one of the largest projects in the program, for instance, that would have added up to more than $1 million.

It "strikes me as a massive conflict of interest for an organization — whether nonprofit or not — that designed the system to have a financial stake in its operation," David Victor, a professor at the University of California, San Diego, who has closely studied international offset systems, said in an email. (Victor recently co-authored the book "Making Climate Policy Work" with Cullenward.)

"In any other market, putting the market players in charge of key elements of its design would lead to 'hollers'" over the conflicts of interest, Victor said. With the forest offset program, "everyone seems fine or even happy about the arrangement."

Climate Action Reserve didn't respond to multiple requests for comment.

'Too Good to Be True'

Hardy, drought-tolerant softwoods like junipers and pinyon pines dominate in the hot, dry landscape of central New Mexico, with smatterings of taller Douglas firs and spruces in the cooler, higher reaches of the mountains.

But under the initial rules of California's program, those forests were considered to contain no carbon whatsoever.

The error stemmed from the fact that there was no available Forest Service data in that part of New Mexico when the Climate Action Reserve calculated regional averages, said Olaf Kuegler, a Forest Service statistician who provided technical assistance to the nonprofit on the federal database.

Consequently, the Climate Action Reserve set the regional average for an area stretching nearly 34,000 square miles at zero, which meant anyone who owned a few dozen trees could earn carbon credits.

Kuegler said he wasn't aware of the mistake until early or mid-2014, when Air Resources Board employee Barbara Bamberger asked him about it. Bamberger, who leads the board's work on forest offsets, later highlighted the error during an October 2014 webinar on offsets.

During her presentation, Bamberger said the board was updating the regional averages in ways that could lead to major changes in certain areas.

"This may be due to the fact that no data existed for some years in the original span from years 2002 to 2006," she explained. "For example, in New Mexico data wasn't collected until the end of that period."

Almost exactly one year after Bamberger's presentation, New Forests' affiliate filed the paperwork for a nearly 222,000-acre project in New Mexico, stretching across the Mescalero Apache Tribe's nearly half-million-acre reservation about ninety minutes west of Roswell. More than a third of the project's trees were carbon-rich Douglas firs, according to the project's paperwork. Shillinglaw signed the forms.

The erroneously low carbon calculation allowed the developer to claim they could have heavily logged the forest, boosting the amount of credits they could earn.

The project earned 3.7 million credits for its first year, worth more than $50 million.

When the California board's updated rules went into effect two weeks later, it set a far higher regional average for most of the project area. If that standard had been in place earlier, it would have eliminated nearly every credit the project earned, CarbonPlan found. The project generated more ghost credits than any other in the nonprofit's study, based on its more conservative calculations of regional carbon averages.

The Mescalero Apache Tribe's president at the time, Danny Breuninger Sr., said the tribe welcomed the project.

"None of us had heard about the carbon credit program, and in a way it sounded too good to be true," he said. "But it was a great deal. It worked out great for us."

Breuninger referred further questions to the tribe's current president, Gabe Aguilar. Neither Aguilar nor the tribe's attorney, Nelva Cervantes, responded to repeated inquiries.

In a statement, the Air Resources Board said the project met all the requirements of the program at that time. The fact that the board was in the process of developing new regional averages using data that didn't previously exist didn't make the earlier figures "invalid or erroneous," it added.

'A Second Wave of Colonization'

Ghost credits matter because they allow other companies to purchase the right to continue emitting real greenhouse gases.

Credits from the Mescalero Apache Tribe's project were sold to PG&E, Chevron and a company that drills for oil in Kern County, California, according to the latest figures available.

The Yurok Tribe's 7,660-acre project generated credits that were obtained by a variety of energy companies like Calpine, PG&E and Shell.

Some tribal members are deeply uncomfortable with the idea of selling offsets to companies like this even if they are legitimate, fearing they're effectively profiting from pollution.

The offsets, by definition, allow California companies to continue producing more CO2 than otherwise allowed — as well as the toxic pollutants like soot and heavy metals that frequently accompany such emissions — often near poor neighborhoods. Communities near refineries, cement kilns and power plants have frequently opposed offset programs.

Thomas Joseph, an activist and a member of the Hoopa Valley Tribe in California, said offset developers target tribal projects because tribes are in "dire need of revenue" and own vast tracts of mostly intact forest. He said his tribe has resisted multiple pitches from developers. "For us to use this as a means to allow corporations to continue to pollute," he said, goes "against our cultural values." He added, "I see it as a second wave of colonization."

Desautel, the Intertribal Timber Council president, sees it differently. When the issue comes up among tribal members, he explains that polluters under cap and trade need to pay either the state for permission to pollute, or landowners through carbon offsets.

"The check is getting written one way or the other," he said. "It's just a question of where it goes and what's being accomplished with that funding."

SilviaTerra's Parisa said that landowners and project developers will continue to respond to the incentives created in the program, in ways that overstate climate progress, until the program itself changes.

"We need better rules," he said. "Let's make sure the dollars we spend actually change things.

"Forests really can be a part of the solution for the climate, but we haven't gotten it right yet."

How We Calculated Offset Credits and Their Monetary Value

The value of the credits throughout these stories was calculated using the fourth quarter 2020 average price for all offsets in California's system ($13.67). The actual amount of money earned by landowners, developers and other actors in the system will depend on when the credits were sold, how many were sold and how many they had to contribute to the program's insurance "buffer" pool. These are private transactions, and the specific terms aren't provided to outsiders.

When the Air Resources Board issues credits to a project, about 20% of those credits go into the pool. Credits in the pool can never be sold, but act as a kind of backstop in case of wildfires, drought and other events. For instance, if a fire burned up part of a forest located on a project site, resulting in one million tons of CO2 released into the air, then a million credits would be removed from the buffer pool to account for the loss.

Our story includes buffer pool credits when describing the credits issued to a group of projects, as all of those credits affect the integrity of California's offset program. We exclude buffer pool credits when calculating the credits earned by any single project and their monetary worth.

How We Put Together Our Maps

The first map depicts the common practice (per-acre regional average) for the Northern California Coast Supersection, Redwood/Douglas-fir Mixed Conifer assessment area (high site class), and the Southern Cascades Supersection, Mixed Conifer assessment area (high site class), based on the Air Resources Board's 2015 Compliance Offset Protocol. The second and third maps show what is effectively the common practice for each ecosection within the Southern Cascades Supersection; CarbonPlan calculated the numbers using data from the Air Resources Board and the U.S. Forest Service Forest Inventory and Analysis Program. The third map omits one additional project on the California-Oregon border. That project had initial carbon levels below the regional average, and was not included in CarbonPlan's study.

How We Got the Story

ProPublica and MIT Technology Review decided to collaborate on this project because of our respective track records of reporting on carbon offsets. In 2019, ProPublica reporter Lisa Song wrote about problems with international forest offsets and California's cap-and-trade program. Separately, Technology Review editor James Temple spent much of 2019 and 2020 reporting on the promises and challenges of carbon removal efforts, including the Air Resources Board's compliance carbon offset program. Both Song and Temple had independently interviewed several co-authors of the CarbonPlan report for their respective stories.

In late 2020, when CarbonPlan was partway through its analysis, study co-author Danny Cullenward pitched the study as a story to Technology Review. Temple then contacted Song to discuss a reporting partnership. We decided that such a complex, technical story would benefit from a newsroom collaboration.

Cullenward, a lecturer at Stanford Law School and CarbonPlan's policy director, had studied California's climate policy system for years. In 2019, Cullenward and ecologist Grayson Badgley, his former colleague from the Carnegie Institution for Science, decided to analyze the state's offset program in a comprehensive way after attending a workshop where they learned more about how the program's rules were designed. (Cullenward is also vice-chair of the Independent Emissions Market Advisory Committee, a group of experts convened by the California Environmental Protection Agency to advise the Air Resources Board on cap and trade. Cullenward said his work at CarbonPlan doesn't speak for the committee.)

In early 2020, Cullenward joined the startup CarbonPlan. The nonprofit assesses the scientific integrity of carbon removal efforts. That includes various types of carbon offsets, as well as emerging technologies that remove CO2 from the air. CarbonPlan receives project-specific funding from companies and other organizations. For instance, Stripe paid CarbonPlan to evaluate different carbon removal options.

Microsoft also paid CarbonPlan to study how climate change would affect the ability of forests to mitigate global warming. CarbonPlan used part of that funding to digitize the forest carbon offset project documents in California's program. Badgley, a postdoctoral fellow at Black Rock Forest and Columbia University, digitized those records and was paid as a consultant by CarbonPlan.

CarbonPlan then used separate unrestricted funding (from various individuals and foundations) to study those projects, working with Badgley and other scientists including Barbara Haya, who leads the Berkeley Carbon Trading Project at UC-Berkeley.

Its study is focused on the primary form of forest offsets in California's program, called Improved Forest Management. These IFM projects reward landowners for managing their forests in ways that prevent further emissions or absorb more carbon over time.

In part because the study hadn't been submitted to a scientific journal, which would include a formal peer review process, we took added steps to check its quality. First, we did a gut check and interviewed several forest experts about the report's premise. CarbonPlan didn't yet have final numbers on the scope of the over crediting, but academics we talked to said that using regional carbon averages created the possibility of awarding excess credits and incentivizing cherry-picking.

Weeks later, when CarbonPlan completed a draft, we sent it to several outside scientists for a detailed review, including Heather Lynch, Professor of Ecology & Evolution at Stony Brook University, and a member of ProPublica's data advisory board; Dan Sanchez, who directs the Carbon Removal Laboratory at UC-Berkeley; and David Valentine, Chair of the Department of Natural Resources and Environment at the University of Alaska-Fairbanks.

These scientists are all experts on forests, climate change, the carbon cycle and/or carbon removal. They all have at least a general understanding of California's offsets, but do not work for offset developers.

We also sent the study to a fourth scientist, Hunter Stanke, a Ph.D. student in the School of Environmental and Forest Sciences at the University of Washington. Stanke developed the rFIA software that CarbonPlan used in its analysis. The software analyzes raw data from the Forest Service's Forest Inventory and Analysis Program, often used by academics, government agencies and timber companies for purposes unrelated to offsets. Before the newsrooms sent Stanke the study, he had provided technical assistance on rFIA to the lead author of the CarbonPlan study, but he wasn't aware CarbonPlan was using the software to study offsets.

All four scientists praised the study and its methodology. They asked for clarification on several technical details, which we sent to CarbonPlan. The nonprofit incorporated some minor suggestions into its final draft, but said the changes didn't alter the overall findings.

CarbonPlan also conducted several analyses of its raw data on behalf of the reporters, including calculations of the level of excess crediting in projects that specific developers worked on.