As part of an ongoing effort to systematically and transparently incorporate ESG in its credit ratings review, a global financial services company turned to Amenity to identify where and when ESG factored as a material credit consideration in its published research documents.
The client’s first challenge in addressing ESG was the volume of documents that they would need to review—over 19,000 reports on private and public sector entities from the previous year. Each document would have to be tagged based on a complex system that involved a multi-tiered list of ESG considerations and other rules.
The second challenge was the extensive QA process that would need to be completed on each document. If done manually, this endeavor would require a considerable number of analyst hours to perform. Analysts well-versed in ESG would have to pore over each document to identify where ESG factors were mentioned and in what context—to what degree these factors affected an organization’s credit rating, whether the impact represented a risk or opportunity, and so on.
This particular client required a more complex classification system to meet their specific needs. Amenity first fine-tuned the client’s original classification framework. With the new framework in place, the client was able to run historical data easily and in a scalable way, allowing for quick, iterative adjustments. Amenity then configured and ran one of two contextual NLP models based on the dataset type, public or private.
To prevent the model from degrading, Amenity implemented a dynamic model development and validation process—tests to ensure model accuracy and that the model improves over time—which significantly reduced the QA process. The resulting solution has been in use for four consecutive years, and it currently delivers an unprecedented accuracy rate of over 96%, as measured against a control sample of data annotated by the client.
See how Amenity’s text analytics can be applied to your unique workflow.
See for yourself. Try it free with no strings attached to start uncovering insights.