Furthermore, large corporations that operate within the European Union are legally required to disclose their greenhouse gas (GHG) emissions. More troubling, a recent analysis by the nonprofit CARB shows that nearly 50% of these detailed reports contain no usable GHG data at all. This finding raises questions about the credibility and broader impacts of sustainability reporting, as well as the adequacy and quality of currently available reporting frameworks.
This is the first analysis to use a dataset of these sustainability reports. These disclosures were randomly selected from companies included in the MSCI World Small Cap index and the German DAX. Jacob Beck from the PHL led the annotation effort specifically with greenhouse gas emissions, working to fill these major data gaps.
Insights from the Research
Dr. Andreas Dimmelmeier, a sustainable finance researcher with the GreenDIA consortium, expressed that the complexity of reporting greenhouse gas emissions is not unexpected. His thoughts help to illuminate the hurdles companies are up against in order to be able to provide meaningful data.
“The hard-to-resolve cases stem not only from complex and partly inconsistent reporting protocols, but also from missing context and incomplete disclosures in company reports. Many companies in our sample did not disclose emissions according to established reporting and calculation frameworks.” – Dr. Andreas Dimmelmeier
To facilitate greenhouse gas emission extraction, the research team, led by Beck, has created a benchmark dataset. This resource aims to assist researchers and practitioners in measuring progress more accurately and closing critical data gaps on the path to achieving net zero emissions.
The Role of Technology in Data Extraction
The process of pulling GHG emissions data is no simple task, as many barriers must be navigated. Dr. Malte Schierholz, one of the researchers who’s been part of this effort, described the first challenge. The team had to take GHG emissions value outputs saved as PDFs and convert them into a clean, usable table format.
“The basic task was to extract GHG emissions values from PDF files into a table,” – Dr. Malte Schierholz
Though technology in automatic extraction methods has made great strides, Schierholz warns not to place undue dependence on technology. He observed that measurement errors are all too possible and underlined the value of careful peer-review practices.
“With automatic extraction methods, it’s easy to fully trust the LLM’s output and overlook measurement errors that occur frequently.” – Dr. Malte Schierholz
Throughout the annotation process, the research team became involved in collaborative and creative discussions to address discretionary or ambiguous cases that emerged in annotation.
“In the end, some ambiguous cases still required expert group discussion.” – Jacob Beck
The Path Forward
The publication resulting from this research, titled “Addressing data gaps in sustainability reporting: A benchmark dataset for greenhouse gas emission extraction,” appears in the journal Scientific Data. Written by Jacob Beck and others, it’s filled with important recommendations for improving both the comparability and quality of sustainability disclosures.
We hope this resource provides inspiration and transparency for how to measure progress. Its goal is to reduce greenhouse gas emissions in the most efficient way possible. By improving data quality and consistency, it intends to enable companies to make better informed decisions while providing stakeholders with better information.
The publication’s DOI is 10.1038/s41597-025-05664-8. It’s a welcome development, both for the progress it represents in addressing the challenges of sustainability reporting and the political will that led to it.