Deutsche Bank’s AI Digs Through ESG Disclosures
Fast growing interest in the environmental, social and governance (ESG) aspects of investing has led to a data explosion over the past decade. In the US alone, the growth has been spectacular: four-fifths of American companies now publish reports on corporate social responsibility. That’s quadrupled in seven years. In general, the amount of data reported to the SEC has increased five-fold since the financial crisis. While we enthusiastically welcome the increased transparency and trust more data can bring, sifting through all that information is difficult, and investors seem to have particular trouble translating ESG disclosures into decision useful information, especially as so much of it is unstructured textual, qualitative disclosure, rather than comparable quantitative data.
The latest developments in AI and machine learning can help dig through this data. Data scientists at Deutsche Bank’s Data innovation Group (dbDIG) have developed an artificial intelligence tool, a-Dig, which uses Natural Language Processing to learn how to infer context as it sifts through non-financial information. This is an example of next-generation AI, moving beyond simple sentiment analysis which can be subject to significant levels of “false positives” and “false negatives”.
The results of a-Dig show that ignoring qualitative information is a big mistake, but that to take advantage of it at scale today, machine learning and artificial intelligence is required. With vast amounts of data increasingly available, this type of big data analysis is quickly becoming an important part of investing.
Our take? Making data more readily consumable, comparable and discoverable is, from the perspective of companies, the key to standing out from the crowd. In relation to ESG disclosures in particular, this means that there is a need to work to make definitions more consistent and more comparable and to ensure that those disclosures are digital in the first place.
More details here.
