UCSD: Identify “Greenwashing” Funds using NLP

The purpose of this research is to determine how funds with “ESG” in their name define ESG in the prospectus. We will use natural language processing (NLP) to extract common definitions in order to create a list of terms or phrases that accurately define ESG funds. If enforced by the U.S. Securities and Exchange Commission (SEC), this will contribute to “truth in labeling” of mutual funds and ETFs and help eliminate misleading marketing, fund naming, and prospectus language. We intend to apply the DistillBERT model using NLP methods that are able to identify whether funds are “real” ESG funds or just greenwashing funds based on the prospectus and other grading information provided by firms. Finally, as prospectuses ultimately proved ineffective as a tool to distinguish good from bad ESG funds, we will provide As You Sow with recommendations and suggested improvements on appropriate phrases that likely should appear on ESG named fund prospectuses.