SRN | Assessing Corporate Sustainability with Large Language Models: Evidence from Europe accepted at Nature Communications

We are happy to share that our paper, Assessing Corporate Sustainability with Large Language Models: Evidence from Europe, has been accepted for publication in Nature Communications.

For the Sustainability Reporting Navigator (SRN), this publication represents more than a successful research project. It is an important step toward a central goal that has motivated the SRN and its team from the beginning: making sustainability information accessible, comparable, and useful to a broad audience.

As we also see in the study, sustainability reporting has increased over the last decade. Companies now publish extensive disclosures on environmental, social, and governance (ESG) issues. Yet much of this information remains locked away in lengthy pdf reports, which makes sustainability data often difficult to compare and analyze.

Our research explores how recent advances in artificial intelligence can help address this challenge. Using large language models, we developed a framework that automatically extracts quantitative ESG indicators from corporate reports. Applying this framework to reports from the 600 largest listed companies in Europe over the period 2014–2023 allowed us to systematically track sustainability disclosures and performance across 501 ESG indicators.

Why is this exciting?

First, it demonstrates that large language models can be used not only to generate text, but also to extract structured sustainability information from complex corporate disclosures with a high degree of accuracy.

Second, it opens the possibility of creating sustainability datasets at a scale that would be virtually impossible through manual collection alone. Researchers, investors, policymakers, and civil society increasingly require timely and comprehensive information on corporate sustainability. AI-based methods can help meet this demand.

Third, the findings illustrate the importance of transparency itself. One of the study's key insights is that changes in reported sustainability performance are often intertwined with changes in disclosure practices. Understanding what companies report, and what they do not report, is therefore essential for interpreting sustainability outcomes.

Click here to read the working paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5361703

You can also explore and download the data, the code and framework here: https://www.srnav.com/insights/assessing-corporate-sustainability-with-large-language-models%3A-evidence-in-europe