Mapping the Issue-Space of Climate-Related Risks in Corporate 10-K Reports

An (initial) analysis of semantic affiliation between 600 companies in the Russell 3000 Index and their risk language

This website is part of an emerging study looking at how “climate-risk” is described, measured and turned into an issue for political and economic concern. In this first branch of the project, we look at narrative data from over 600 corporate disclosure fillings of companies traded on the U.S. stock exchange. These filings (called 10-k forms) are federally mandated reports where companies are required to share with the public information that might help investors decide whether or not to buy stocks in a given company.

We took climate-related excerpts from the 10-k filings (available at climateriskdisclosure.org) and ran this narrative data through some natural language processing and network spatialization software. We then turned the results into a series of visualizations that reveal different trends in how companies are talking (or not talking) about climate-risk in their 10-k forms. These visualizations include:

  1. Basemap: a network visualization of the top 250 climate-issue terms used by companies in 5 major industry sectors of the Russell 3000.
  2. Terms Table: A break down of each of these terms, year by year, and their usage by industry sector.
  3. Companies: radial diagrams of each company in our dataset that provide quick profiles of what kind of issues related to climate change are of concern for each company, and how much they are writing about their approach to these concerns in their 10-k filings.

Visit the About section to learn more about 10-k filings and rules of disclosure.
Visit the Methods section to understand how we isolated the different excerpts from the disclosure reports.


This is an early-phase project of the Medialab, Sciences Po and CookESG Research.
With special thanks to participants of the seminar Data, Digital Methods and Mapping Social Complexity.