Brié, BjarneStouthuysen, KristofVerdonck, Tim2023-09-202023-09-202023http://hdl.handle.net/20.500.12127/7261We investigate the usefulness of emission information and climate-related annual report disclosures to financial analysts. We extract climate disclosures from annual reports using a state-of-the-art machine learning model, ClimateBert, and automatically identify and quantify the underlying disclosure topics using a structural topic model. For a large international sample of firms subject to a European nonfinancial disclosure mandate, we find that analysts face greater uncertainty while forecasting future earnings for firms with greater emission intensities and for firms that show weaker commitment to reducing emissions. We also show that the aggregate level of climate disclosure does not significantly reduce this uncertainty. However, we do find that disclosure on a specific set of topics improves analysts’ ability to forecast future earnings. Nevertheless, for firms with greater emission intensities, this relationship reverses. These findings suggest that, overall, current climate reporting practices are not informative to analysts since they do not mitigate the uncertainty inherent to the potential impacts of climate change.enClimate DisclosureSell-Side AnalystsClimate ChangeMachine LearningTextual AnalysisAnnual ReportsCarbon Emissions286361119751