How reliable are corporate climate disclosures? Our analysis of a decade of sustainability reports reveals a troubling pattern.
Corporations are among the largest contributors to global greenhouse gas emissions. In the United States alone, the industrial sector was responsible for 30% of the country's 2022 total emissions—approximately 2 billion tonnes of CO₂ equivalent.
Stakeholders have called for greater transparency, and large public firms have responded by increasing voluntary disclosures through corporate social responsibility (CSR) reports. But these reports remain largely unregulated and unaudited.
We examined every available CSR report from 724 S&P 500 companies over more than a decade. What we found raises serious questions about the reliability of self-reported emissions data.
"58% of all emissions with available forward-looking reporting were revised."
From 2010 to 2020, the percentage of firms revising their reported emissions has remained stubbornly high—ranging from 50% to 70% each year.
Despite growing attention to climate disclosure, the revision rate hasn't decreased. The average across all years: 58%.
During the same period, financial metrics like revenue were restated in less than 2% of reports to the SEC.
Total emissions that went under-reported from 2010 to 2020
More than the total 2020 emissions of Venezuela, Nigeria, Qatar, and Kuwait combined.
Over the decade, the cumulative amount of emissions that went unreported has grown substantially.
By 2020, firms had understated 135 million tonnes of emissions—more than twice the 57 million tonnes that were overstated.
This likely represents a lower bound. There's no requirement for firms to revise, meaning many inaccuracies may never be corrected.
We investigated multiple potential explanations for these revisions. The results were striking—not for what they revealed, but for what they ruled out.
Observable firm characteristics like size, industry, and institutional ownership? Not predictive of which firms would revise.
Third-party assurance of CSR reports? No significant relationship with revision likelihood.
Changes to measurement methodology? Only 5.4% of revisions cited this as a reason.
We hand-collected footnotes from CSR reports to identify how firms explain their revisions.
83.7% of revisions came with no acknowledgment or explanation whatsoever.
Major data providers like Trucost, CDP, and Refinitiv don't systematically correct for these revisions either.
Search and filter the complete dataset of emissions revisions
Loading data...
Our sample consists of all 724 companies included in the S&P 500 Index at least once between 2010 and 2020. We collected 3,657 CSR reports and identified 266 firms that reported historical emissions data at least twice, enabling comparison across years.
For each firm-year observation, we tracked whether the originally reported emissions value was later revised in subsequent reports. We defined a revision as any change between the initially reported value and the final reported value for that year.
To ensure accuracy, we conducted multiple verification passes and excluded obvious typographical or scale errors. The methodology represents a conservative approach—actual revision rates may be higher.
Access the full dataset and published paper
Cohen, L., Rouen, E., & Sachdeva, K. (2025). Widespread revisions of self-reported emissions by major US corporations. Nature Climate Change. https://doi.org/10.1038/s41558-025-02494-9