Offsets discredited

“There are three kinds of lies: lies, damned lies, and then, there is statistics” – popularised by Mark Twain.

After a slew of damning investigations about the failures and abuses of the offset industry, and a growing number of legal cases against companies reportedly misleading consumers with climate campaigns based on offsets, reputational damage(opens in new window) has become a serious risk for corporate carbon credit buyers. In an attempt to restore the credibility of the offset industry, two organisations with stakes in the long-term success of the carbon market released reports based on misleading data analysis. Their argument is that companies that use carbon credits are likely to decarbonise their operations further and faster than those that do not. Both reports received widespread publicity and have been used by academics(opens in new window) and industry supporters(opens in new window) to promote narratives that endorse the offset industry. However, a briefing(opens in new window) by SOMO and Data Desk, which analyses the methodologies and data sets used in these reports, reveals multiple reasons why the causal link between carbon credit buyers and emission reductions is statistically flawed. 

Companies purchase carbon credits to compensate for their emissions and to reach voluntary climate targets, such as achieving “net zero” or “climate neutrality”. Each carbon credit is generated by offset projects and is supposed to represent the equivalent of one tonne of carbon dioxide emissions (1tCO₂e). The carbon offset industry relies on carbon credits representing a “real” and “permanent” reduction of emissions. However, research has repeatedly shown that offsetting is a system with serious structural flaws and one that is easy to manipulate. The result is a market with many “phantom carbon credits(opens in new window) ” or “junk carbon credits(opens in new window) ”, which do not “cancel out” any emissions except on paper. In addition, land-based carbon offset projects in particular have been linked to conflicts over land, loss of access to livelihoods, forced evictions, and other forms of human rights violations.

Fundamentally flawed reports

The first of two misleading reports is All in on Climate: The Role of Carbon Credits in Corporate Climate Strategies(opens in new window) , and was released in October 2023 by Ecosystem Marketplace (EM). EM is an initiative of US-based Forest Trends, a non-profit organisation which advocates for market-based tools to address deforestation. EM claims(opens in new window) to provide “accessible and trustworthy information”.

The second report, Corporate Emissions/Performance and the Use of Carbon Credits(opens in new window) , was released in October 2024 by MSCI Carbon Markets, a US financial data company. This report updates a publication by Trove Research, an independent UK carbon market analytics firm that was acquired by MSCI in November 2023. MSCI Carbon Markets claims(opens in new window) to “bring clarity” to the carbon market.

Both reports attempt to prove a causal link between carbon credit use and corporate decarbonisation. EM bases its conclusion on just two years of emissions data (2020-2021) and reports that 59 per cent of carbon credit buyers reduce their emissions versus 33 per cent of non-buyers. EM also concludes that carbon credit buyers perform better on other metrics, such as disclosure of emissions and setting transparent long-term targets for decarbonisation. 

MSCI uses emissions data for six years (2017-2022) and calculates the median year-on-year reduction in percentages for each company across the period. It concludes an average 3.4 per cent reduction for users of carbon credits versus 1.5 per cent for non-users. 

Both reports use nearly identical datasets but make very different decisions about how to deal with the limitations. Their key data source is the Carbon Disclosure Project (CDP), a non-profit organisation that is the largest existing source of self-reported emissions data. Despite the importance of CDP’s dataset, it has several limitations, particularly when it is used as a representative sample of companies. The inappropriate ways in which EM and MSCI seek to mitigate or obscure these data limitations seriously undermine the conclusions of both analyses. 

There are at least eight reasons why the conclusions of the analyses that carbon credit buyers decarbonise further and faster than other companies are statistically flawed. See the full briefing(opens in new window) for details.

1. Selection bias

Disclosure via CDP is voluntary. Currently, almost 2,000 companies that have been identified(opens in new window) by CDP as major polluters do not report on their emissions and, as such, cannot be included in any analysis. Any conclusions drawn from CDP data must therefore be considered as biased towards companies that are incentivised and have the financial means to voluntarily report on their emissions and rely on their credibility.

This limitation affects both reports in different ways. For EM, the sample of 7,352 companies analysed consists only of those that have reported emissions data via CDP in both 2020 and 2021. 

MSCI, in an apparent attempt to mitigate this issue, starts with its own ACWI Investable Market Index(opens in new window) that covers 8,844 companies. However, the report’s use of six years of emissions data for each company quickly introduces an even more severe restriction, leaving just 2,936 companies for which CDP data is available across the entire period. Consequently, this group’s distribution is, again, likely to be biased toward companies that are already incentivised to report voluntarily.

2. Irregularities with carbon credit data

The self-reported nature of the CDP data especially affects the information on carbon credit use, which EM itself notes contains “many irregularities […] likely due to confusion around types of credit schemes, project methodologies, and distinctions between credit purchases and origination.” The EM report notes that at least 21 companies misclassified the buying of certifications under obligatory compliance mechanisms as voluntary action. 

While EM states that they amended this aspect of the CDP data, no such process is discussed in the MSCI report, despite considerable detail being provided on the methodology. This, in turn, raises questions as to whether MSCI’s analysis properly distinguishes between voluntary and obligatory action.

3. Categories tainted by insufficient information on companies

Many companies that submit data to CDP fail to provide answers to the questions on carbon credits. When EM compares its own data on carbon credit buyers with that of CDP, it finds that “36 percent of CDP respondents that are using carbon credits do not disclose their purchase or origination of credits”. 

The implications of this finding for both reports are significant. If more than one-third of carbon credit buyers do not report this to CDP, then the “non-buyers” category in both reports is likely to have included a substantial number of companies that have, in fact, bought credits, calling the validity of the metrics generated using the two categories into question.

4. You can’t reduce what you don’t report

Rates of disclosure in the CDP data under the different Scopes ( 1, 2, and 3 ) of emissions differ significantly between carbon credit buyers and non-buyers, with many companies not submitting any Scope 3 data. This represents the greatest uncertainty in the findings of both reports. 

According to MSCI, 80 per cent of companies in their sample did not disclose any Scope 3 emissions in 2017, falling to 59 per cent in 2022. To address this, MSCI analyses Scopes 1 and 2 separately from Scope 3 emissions. Yet, more issues appear when looking at the categories that make up Scope 3. In 2022, 35 per cent of companies in the MSCI sample had disclosed emissions related to business travel, but just 17 per cent had disclosed emissions related to the transportation and distribution of their products or to their ultimate use and disposal. This latter category is extremely important for some companies, notably in the oil and gas industry, where the use of their products represents the vast majority of their total emissions. 

EM’s conclusion about the proportion of non-users reducing their Scope 3 emissions could differ substantially as a result of the missing data. For example, a financial institution that reported significant Scope 3 emissions from investments in 2020 but subsequently divested these assets would be counted by EM as decarbonising, whereas a company performing exactly the same actions without ever reporting any Scope 3 emissions would not. You can’t reduce what you don’t report.

5. Data distorted by mergers and acquisitions

Any analysis of corporate emissions data over time is directly affected by changes to the company structure as a result of the purchase or sale of assets and through mergers and acquisitions. Afterall, the emissions associated with the assets being sold become the responsibility of the buyer. This means that a company’s reported emissions in one year may be significantly higher or lower than in the previous year, even if its underlying business model and production processes remain unchanged. These distortions are likely to be relevant for the oil and gas, mining, and heavy industry sectors, all of which are well-represented in the CDP dataset. 

The EM report takes no steps to address this issue. MSCI applies the blunt instrument of excluding companies whose median year-on-year emissions change over the period is greater than ±30 per cent. This, in conjunction with the requirement for six years of emissions data, reduces the sample size for MSCI’s main analysis to just 2,665 companies. This makes the overall picture incomplete and excludes many highly polluting companies.

6. Discrepancies in emissions reported

The EM report lists the 50 most polluting companies that do not use carbon credits, but the figures provided by EM appear not to match what those companies themselves report. 

For example, according to EM, Şişecam, a Turkish glass manufacturer, has the highest emissions in the group of companies that do not use carbon credits. EM claims Şişecam emitted more than 3 billion tCO2e in 2021 – more than the emissions reported in the same year to CDP by Delta Airlines, Total Energies, and Shell combined! While Şişecam did not report its Scope 3 emissions at the time, the company’s 2024 CDP submission estimates its total emissions for that year – including Scope 3 – at 11.7 million tCO2e, a figure more than 250 times smaller than what EM reported. A similar situation applies to several other companies in the list, including Kongsberg Automotive, which EM claims is emitting at levels far exceeding those reported by Kongsberg.(opens in new window)

EM did not respond to SOMO’s request for further clarification of these figures. EM’s reporting of emissions that differ so significantly from what the companies themselves report raises questions about both the data used in the analysis and the rigour with which the report as a whole has been put together.

7. Accounting tricks for lowering Scope 2 emissions

Both EM and MSCI use market-based emissions estimates for at least some of the analysed companies, despite this being widely criticised(opens in new window) for allowing companies to report lower Scope 2 emissions without necessarily decreasing their consumption of fossil fuel-generated electricity.

When calculated on a market basis, Scope 2 emissions reflect the contractual arrangements a company has made to purchase electricity, for example through renewable energy certificates or power purchase agreements. These are calculated using the emissions intensity of the company’s contracted electricity supply, which in cases of renewable energy will generally be zero, even where the electricity actually consumed by the company is from a mix of sources on a national or regional grid. Analysis(opens in new window) by the Financial Times has shown how major US tech companies use such certificates to substantially lower the emissions they report. 

MSCI clarified to SOMO that “the significant majority of scope 2 emissions [used in their analysis] should ignore the carbon reductions from renewable energy certificates purchases”. However, they gave no more explanation on how this was (or was not) verified.

8. Sectoral discrepancies

The MSCI report presents a number of its findings by industry. This makes it clear that the analysis of users and non-users of carbon credits only holds for certain sectors. Among financial firms and communications services companies that are significant users of carbon credits, for example, the analysis finds a statistically significant difference in emissions reductions. However, when we turn to sectors of the economy with a more substantial material footprint, these gaps appear to shrink. For companies in the energy, real estate, consumer staples, utilities, and information technology sectors, which make up one-third of MSCI’s sample of 2,665 firms, the difference between carbon credit users and non-users is statistically insignificant. 

Misleading narratives that exacerbate the climate crisis

These eight serious flaws in the analysis and the methodologies used by EM and MSCI mean that their headline conclusion is not supported by the data. 

When data analysis is compiled and presented by companies with vested interests in the conclusions, they can easily mislead and manipulate. The fact that the authors of both reports examined in this article are entities owning or representing commercial interests in the offset industry they promote raises serious concerns. 

By presenting only the data that aligns with the industry narrative and ignoring counter evidence and limitations of the datasets they use, organisations such as EM and MSCI mislead consumers, citizens, and policymakers. Reports such as the two analysed in this article create false perceptions of what carbon credit buyers and the offset industry actually do: divert attention from the continued extraction and consumption of fossil fuels. In this context, these reports are an attempt to push a narrative to rescue an industry that is failing on all accounts to deal with the climate crisis. 

It is clear that oil and gas companies buying carbon credits are not climate successes but, on the contrary, they are climate failures. The failure to phase out fossil fuels in the coming years will have a defining impact on our collective ability to handle the already impactful climate crisis. Since the beginning, the offset industry has sought to control the narrative by suggesting that offsets “are the best option” to address the climate crisis. Carbon offsets are an obstacle to structural change because they allow the fossil fuel-dependent economy to continue under an illusion of climate action. It is time to abolish this industry and focus on real and just climate action. Offsets are definitely not a solution. Numerous initiatives for achieving real emissions reductions already exist, and many climate justice actions and proposals are being developed by governments, groups, and movements worldwide. Addressing the climate crisis requires transforming the economic system. And for that, collective and just action in solidarity with those on the frontlines of protecting it needs to be placed at the centre.