As environmental, social and governance (ESG) investing has evolved rapidly in recent years, so too has the way asset managers analyse and assign value to ESG factors.
Traditional ESG ratings are increasingly viewed as inadequate, since company-reported data is subject to greenwashing and annual ratings are not timely enough for investment decision-making.
In the Information Age, corporations are no longer the sole authors of their own narratives because the internet has given stakeholders in companies a vehicle for distributing information that impacts the market's valuation of companies. Today's superabundance of information calls for a new ESG analytical model built around the application of artificial intelligence. Technology is the only way to ascertain signals embedded in unstructured data in order to gain deeper insights into sustainable investment.
The internet and the resultant explosion of data has facilitated the conversation about ESG issues by mobilising corporate stakeholders and fostering conditions favourable to mainstream adoption of ESG investing. Technology enables us to analyse volumes of information far beyond the capacity of humans. By connecting the dots, we can begin to understand why ESG has flourished in the Information Age, where to look and how to deepen our understanding of the relevance of ESG factors to investment.
ESG research has gone through three stages characterised by the prevailing data and technology:
- Scarcity: Limited ESG information, primarily from print media clippings, was manually collected and compiled into reports for clients (1970s-1990s).
- Abundance: Data growth driven by the internet and company disclosure led to the development of ESG ratings designed to provide industry peer rankings (2000s-2010s).
- Superabundance: AI applied to proliferating unstructured data provides research firms the ability to generate dynamic ESG scores reflecting stakeholder sentiment (2010s-present).
In today's era of superabundance, analysts are no longer as dependent on company-disclosed data about ESG behaviour - nor can they afford to be. Instead, the challenge is to discern meaningful, material and potentially predictive information from the enormous volumes of unstructured data. Just as researchers embraced the internet in the era of information abundance, they must now embrace the defining technologies of the era of superabundance: artificial intelligence (AI), natural language processing and machine learning. They are the most efficient and accurate way to analyse the sheer quantity of data in the world today.
ESG research is in the early stages of a foundational transformation as it adapts to the superabundance of unstructured data. Most traditional ESG rating agencies are responding to this phenomenon by incorporating AI into their traditional analyst-based fundamental research model. Technology can make their legacy infrastructure more efficient at gathering information. That said, if it simply improves the efficiency of a model that has been criticised for introducing analyst subjectivity and is at risk of becoming outmoded, is this progress? Using AI for data mining alone misses the opportunity that the technology offers for uncovering intangible value and deepening our perspective on how to make ESG data most valuable for portfolio construction.