Greenwashing of the ESG Industry: Using ML to Demystify ESG Fund Ratings
Environmental, Social, and Governance (“ESG”) investing has gained interest from all investor types, perhaps led by the millennial client, over the last several years. However, the sector has taken center stage in light of the COVID-19 pandemic as firms and investors embrace a new normal. Investors are holding companies responsible, more so than ever, to sustainable practices around supply chain management, biodiversity impact, and treatment of employees; they expect transparency around reporting.
The Celent Wealth Management team plans to explore the broader topic of ESG in greater depth through its upcoming reports available to Celent clients. In the meantime, a sub-topic of ESG that caught my interest is the challenges around ESG integration and ESG ratings.
ESG integration is one component of assessing the economic value associated with ESG factors. It is the incorporation of ESG into the investment decision and analysis process to understand the value of that fund on a portfolio. Simple enough, right? Sure, but where the waters get murky is deciphering the data of these funds.
Since the financial crisis, the investment industry has faced unparalleled scrutiny over nearly every operational aspect with the overarching theme of transparency, investor protection and accountability. ESG funds or ESG-centric funds, whose proposition is responsible and sustainable investing, are undoubtedly no exception.
Most wealth and asset managers regularly cite the inclusion of ESG analysis in their investment selection process through one or several of these five approaches: exclusionary screening, positive screening, ESG integration, impact investing and active stewardship.
There are a host of challenges that come with assessing the value of an ESG fund. The lack of standardization (apples-to-apples comparison of ESG definitions, for example) and a lack of quality data (i.e. data input inconsistencies, untimely data, unaudited data, reliance on self-reported data) are the primary issues.One rating agency can value a firm, such as Tesla, with a heavily weighted score towards its environmental impact, while another agency may find Tesla’s treatment of employees to be more important thereby resulting in contradictory ratings. Investors are confused and rightfully so – the mismatched scoring data across the ESG industry is dramatic. “Greenwashing” is a problem in the industry where some funds may be aligned with ESG principles while others are not. How do you quantify a subjective factor – what is more important to one set of clients may be less important to another.
Welcome: Big Data and Machine Learning (“ML”). Though still at an early stage, in terms of its use in the ESG industry, ML can distill millions of datapoints. While not an end-all-be-all solution, the break-down of bespoke client preferences and rapid machine-led analysis of minute unstructured or alternative data, such as social media sentiment or satellite data, can lead to salient ESG signals, understanding client sentiment, and potentially innovative benchmarking. ML can move the industry towards a transparent ratings framework based on SASB (Sustainability Accounting Standards Board), provide real-time data signals, and consistent, objective reporting.