Years ago, it was generally thought that no one really read corporate sustainability/CSR reports. About four years ago, we realized that was no longer true and the question became “who is reading our ESG/sustainability reports, and what are they looking for?” Today – and in the future – we have to add “What is reading our ESG/sustainability reports and how are the reports being interpreted for humans?” Of course, I am referring to AI. This paper in the Journal of Banking & Finance discusses
“ClimateBert CTI, a deep learning algorithm, to identify climate-related cheap talk in MSCI World index firms’ annual reports… ClimateBert CTI enables researchers and stakeholders to discern between genuine and unspecific climate commitments and to shed light on the quality of firms’ disclosed climate-related pledges. This allows for subsequent large-scale analyses between firms’ climate commitments and their alignment with actual actions and outcomes. Specifically, we utilize ClimateBert CTI to develop a ‘cheap talk index’ that quantifies the share of superficiality in companies’ climate commitments.”
The authors applied the algorithm to climate commitments of 14,618 annual reports from MSCI World index firms spanning 2010 to 2020, analyzing how the companies discussed TCFD, the SBTi, and the CA100+. Results from ClimateBert indicate that:
“… companies with higher levels of cheap talk are associated with higher negative environmental news coverage and with higher emissions growth. These relationships suggest that companies with a high level of cheap talk prioritize maintaining a positive public perception over making meaningful changes to their business practices. Consequently, this can impede progress toward global emissions reduction targets, as companies that engage in cheap talk fail to take the necessary steps to mitigate their climate impacts and manage their climate risks.”
One should question the data on which ClimateBert was trained, and the learning instructions it was given to interpret the corporate reports – especially if the algorithm catches on. The other thing I know you are asking – what about ClimateErnie? I’m more interested in ClimateCookieMonster or ClimateAnimal myself.
If you aren’t already subscribed to our complimentary ESG blog, sign up here: https://practicalesg.com/subscribe/ for daily updates delivered right to you.
