Mapping corporate disclosures of climate risk

All companies listed on U.S. stock exchanges are required by the Securities and Exchange Commission (SEC) to disclose information that might help investors evaluate the value of a company’s stock. These annual disclosures—called 10-K filings—include information about how much debt and assets a company holds, whether they have any pending lawsuits, and what kind of other business risks (competitive, physical, reputational or regulatory) they may face. All this information is considered “material” in helping a shareholder make investment decisions. Sharing it publicly, so that all actors have access to the same information, is considered one of the cornerstones of a well-functioning market.

In 2010, the SEC issued interpretive guidance asking U.S. companies to disclose the risks and opportunities they face from climate change. The guidance is a first step in generating metrics to help investors evaluate the viability of existing business practices given the reality of a changing climate. However, since companies exercise their own judgment in how best to respond to “interpretive guidance”, current climate disclosures vary highly company to company. As a result, while a few groups have begun benchmarking different industry sectors around the quality of their disclosures, it is very difficult to get a holistic picture of the state of disclosures, or effectively compare disclosures within and across industry sectors—something that might allow investors to begin incorporating climate risk more robustly into their valuation models.

After tepid enforcement by the SEC of their own guidance rules, mainstream economic actors are increasingly calling for better disclosure. In a recent op-ed in the Washington Post, for instance, former Clinton Administration Treasury Secretary Robert Rubin writes, “I believe that such disclosures should be considered material and mandated by the SEC, not just requested by investors. If companies were required to highlight their exposure to climate-related risks, it would change investor behavior, which in turn would prod those companies to change their behavior.”

This website is a proto-type of data visualizations intended to provide investors, analysts, and financial regulators with a clearer picture of the state of climate related disclosures. Using natural language processing tools and network analysis software, we analyzed climate-relevant excerpts from publicly available 10-K filings of companies in the Russell 3000 index from 2009-2014. These excerpts (many constituting no more than a paragraph per year per company) were prepared by Jackie Cook of CookESG Research, with whom we collaborated and who has built a series of algorithms that scrape the relevant climate-risk sections out of 10-k filings across the entire Russell 3000 companies and makes these findings available and searchable by the public via the Climate Risk Disclosure website.

Our team convened over 2 days in Paris in conjunction with a “barcamp” organized under the auspices of a digital methods research seminar hosted by the Medialab of Sciences Po and CoreText of IFRIS. We focused on mapping the language of corporate climate disclosures between 5 major industry groups, including Oil and Gas, Electric Utilities, Food & Agriculture, Insurance and Garment & Apparel sectors – a total of 605 companies. The visualizations on this website are a first pass at finding strategies to render more visible the relationship between language and behavior to climate related risks across these 5 industry sectors.

Ian Gray, Medialab