Sign up for our newsletter - Navigating the horizon of business technology​
AI / Deep learning

AI is tackling the lack of transparency in ESG investing

Applying AI to the wealth of unstructured data on the internet is allowing analysts to counteract corporate 'greenwashing' and help ESG investors.

Ethical investing took off in 2020, with a record $1.7trn of assets held in sustainable funds at the end of the year, according to data provider Morningstar. But the rush to pump money into companies with robust environmental, social and governance (ESG) profiles has also raised questions about what falls under the umbrella of ‘sustainable’ investing. A host of AI-driven technology is coming to the rescue, promising to increase transparency in ESG investing and counteract the reliance on self-reporting, which has facilitated corporate greenwashing.

Alternative data sources used in ESG investment analysis include geospatial imagery. (Photo by Blue Swallow Imagery/Shutterstock)

The number of sustainable investors has ballooned in recent years – the proportion of the population stating that they are “very interested” in sustainable investing more than doubled between 2017 and 2019, according to Morgan Stanley. The pandemic has exposed enduring social inequalities and brought a renewed focus on the environment; this, alongside booming ESG funds, has shifted more attention to the subject.

ESG reporting in the internet age

The growing interest in ESG investing has prompted many listed companies to publish sustainability reports alongside their regular financial updates. But amid suspicions of greenwashing, a term coined in the 1980s to describe dubious environmental claims by corporations, a new class of ESG data provider has sprung up.

White papers from our partners

These providers use AI to collect and analyse the growing wealth of non-financial data available on the internet and beyond – from weather and satellite imagery to social media posts – to help investors determine the real social and environmental impact of investible companies. It is a booming market: the ESG data industry is projected to be worth $1bn this year.

These alternative data sources are introducing transparency into what has historically been an opaque market, says Bhavna Rawlley, vice-president of capital markets at Accenture Singapore. “What AI and the whole process of gathering data is trying to do is to bring in that standardisation,” she says. “You can actually validate [ESG reporting]… if companies are reporting fluff instead of outcomes based on hard parameters.”

This transparency has been made possible by the proliferation of unstructured data on the internet and the accompanying advances in AI, says Thomas Kuh, head of index at ESG data provider TruValue Labs. “[AI] technology is fundamentally transforming the whole chain of value in the information flow for ESG investing,” he says. “The amount of potentially relevant information that’s out there is so vast, it simply cannot be appropriated by human beings without machines.”

The result has been a shift in power towards stakeholders, who now wield power over what is material to company value and are increasingly vocal about the need for a robust ESG strategy. “Corporations really controlled their own narrative for a long time, but they don’t any more,” says Kuh. “We’ve got advancements in computer technology and growth in AI. This combined with the internet is providing a voice and agency for stakeholders… [and has] led to a democratisation of how companies are perceived.”

Diversity in ESG data and analysis

The ESG data and analytics market is nascent: a wide range of traditional ratings providers and innovative new start-ups are fighting to establish their respective approaches. This has led to a proliferation of – often conflicting – indices that are confusing investors.

A study by researchers at MIT found the correlation between five prominent ESG ratings agencies was 0.61, meaning there is little consensus on which companies perform well. For comparison, credit ratings from Moody’s and Standard & Poor’s have a correlation of 0.99.

In January, the European Securities and Markets Authority called on the European Commission to introduce “appropriate regulatory requirements to ensure [the] quality and reliability” of ESG ratings and assessment tools.

Kuh argues that regulators should not rush to impose standards on the industry, however. “You don’t want a regulatory process that, in essence, forces the ratings to be highly correlated the way credit ratings are,” he says. “It would be comfortable for many if ESG ratings correlated the way credit ratings do. On the other hand, highly correlated credit ratings did not help us out in 2008.”

[Keep up with Tech Monitor: Subscribe to our weekly newsletter]

In fact, the diversity of data and AI models is an advantage for asset managers, Kuh argues, as it allows them to develop their own proprietary approach based on their firm’s ESG investment strategies.

Where regulation may be needed, many believe, is in eliciting greater transparency in how these AI-driven models and products are constructed. Research from the Centre for Innovation Management found that when discrepancies between sustainability reports and real ESG performance came under scrutiny, greenwashing declined.

Governments around the world are beginning to take action. Europe is leading the way, with its impending Sustainable Finance Disclosure Regulation to solidify reporting requirements for the financial market. Meanwhile, the UK’s Financial Stability Board has established a Task Force on Climate-related Financial Disclosures (TCFD) to develop recommendations for more effective climate-related disclosures by companies. Its standards are already inciting change around the world, with the proportion of disclosures adhering to its recommendations increasing by an average of six percentage points between 2017 and 2019.

AI has considerable potential to make ESG investment more effective, says Charles Radclyffe, CEO of EthicsGrade, a start-up using natural language processing (NLP) to build an ESG rating model that scrutinises companies’ ethical use of AI. But it will only help investors match their money with their values if the analysis itself is transparent too.

“Proprietary indexes and ratings are very attractive, but only if we are very clear about the transparency of what goes into that and the fact that we are reflecting values from an investor or from a brand,” he says. “We absolutely should agree standards and specifications, so we can compare apples with apples.”

The issue at the moment is that ESG is treated as a check-box exercise, he adds. “When it comes to ethical investing, people are simply asked do you want sheer performance or are you an ethical investor, as if those two statements are somehow mutually exclusive,” he says. “We’ve got to personalise ESG ratings… like Tinder, whereby the match that you get to see is the one that’s the closest match to what you’re looking for.”

Home page photo Song About Summer/Shutterstock.

Amy Borrett

Amy Borrett is the resident data journalist at Tech Monitor.