Skip to main content

Man vs. machine: A tale of two sustainability ratings systems

Ai vs. human analyst ratings for ESG
GreenBiz photocollage, via Shutterstock

On May 11, 1997, globally acclaimed chess master Garry Kasparov battled IBM’s Big Blue supercomputer in the most widely watched chess match. The BBC reported that the match between man and machine had "suspense beyond anything the world had ever known," and was a test of whether the IBM team could teach the machine to "surpass the brilliance of the human brain."

The match was also a test of whether artificial intelligence could outsmart humanity.

On one side, world champion Kasparov — who, at 34, was playing at his very best — had soundly beaten IBM’s Big Blue just a year earlier. On the other side, IBM’s Big Blue could process 200 million moves a second. The match was a nail-biter. In the deciding sixth match, Big Blue overtook Kasparov in fewer than 20 moves to win the match, shocking the world and ushering in a new wave of opportunities and applications for AI.

The next high-profile test between man and machine came 14 years later when, on Feb. 16, 2011, IBM’s supercomputer Watson outdueled champions Ken Jennings and Brian Rutter on the TV show "Jeopardy." The system that powered Watson "took about 20 researchers three years to develop," said David Ferrucci, who led the IBM Watson Research Center.

Ferrucci added, "Watson could process the equivalent of a million books a second." The match was close at times, but in the end the machine prevailed. Watson is deployed in healthcare to support cancer diagnosis, and Ferrucci seems even more excited. "Today it’s no longer purely a computing system designed from Q&A but can 'see', 'hear', 'read', 'talk', 'taste', 'interpret', 'learn' and 'recommend.’"

Two companies, Morningstar and Truvalue Labs, represent two distinctly different ways to assess companies on ESG data.

A similar battle is brewing in the $500 million global market for corporate sustainability research and ratings.

Arguably the two leaders from each side are Sustainalytics, now owned by Morningstar and representing the analyst-based sustainability research and ratings approach; and Truvalue Labs, which harnesses AI, machine learning and natural language processing to provide investors and companies with corporate sustainability performance insights and assessments.

The differences between Sustainalytics and Truvalue Labs’ methodology are depicted below.

Analyst vs. AI driven ESG

In 2018, sustainable investing worldwide eclipsed $30 trillion in assets under management, according to the Global Sustainable Investment Alliance, with $12 trillion originating in the United States.

But how did sustainable investing hold up during the recent market downturn?

According to Morningstar’s head of sustainability research, Jon Hale, mutual funds that integrate environmental, social and/or governance (ESG) factors registered record growth in the first quarter of 2020, beating the previous record in Q4 2019.

"Sustainable funds in the United States set a record for flows in the first quarter [of 2020]," wrote Hale. "ETFs, passive funds, and iShares dominate as U.S. ESG funds gathered $10.5 billion in the first quarter."

Clearly, sustainable investing is no longer on the margins — it’s gone mainstream in a big way.

Machine-based approach: Truvalue Labs

Truvalue Labs applies AI to uncover opportunities and risks hidden in massive volumes of unstructured data, including ESG behavior that can have a material impact on company value. It also integrates SASB’s materiality guidelines, which is the gold standard for investors seeking to evaluate a company’s performance from a financial materiality prism.

"The world is awash in unstructured data," wrote Thomas Kuh, head of index at Truvalue Labs. "Roughly 90 percent of the data in the world was generated over the last two years alone."

This superabundance of unstructured information, says Kuh, provides an important source for ESG signals and makes it imperative to "incorporate an ‘outside-in’ view of ESG data to complement companies’ ‘inside-out’ reporting."

As such, Truvalue Labs made the strategic decision to exclude from its analysis a company’s self-produced ESG information in order to, it believes, obtain an unvarnished impression of companies’ performance.

"Many sustainable investors are skeptical of AI and the idea that we can trust machines. We make a point of emphasizing that the role of AI is to do things humans can't do efficiently in order to make humans work smarter. And good AI depends on being deployed intelligently by humans. Not all AI is created equal," said Kuh.

Truvalue Labs’ sustainability performance assessments take into account the sentiments of a wide array of corporate stakeholders. Its coverage of ESG research and ratings spans more than 16,000 companies and it applies its proprietary, customizable AI algorithms to more than 100,000 sources.

Analyst-based approach: Sustainalytics

Sustainalytics has 25 years of experience covering more than 40,000 companies on hundreds of ESG performance indicators. Its 650 employees provide sustainability ratings for more than 20,000 companies that can be customized to clients’ specifications.

In April, Morningstar acquired Sustainalytics. "Modern investors are demanding ESG data, research, ratings and solutions in order to make informed, meaningful investment decisions," said Morningstar’s CEO Kunal Kapoor. "From climate change to supply-chain practices, the nature of the investment process is evolving and shining a spotlight on demand for stakeholder capitalism. Whether assessing the durability of a company’s economic moat or the stability of its credit rating, this is the future of long-term investing."

Morningstar and Sustainalytics have worked together to provide investors and companies with new ESG analytics, including: the industry’s first sustainability rating for funds, based on Sustainalytics’ ESG ratings; a global sustainability index family; and sustainable investing analytics that includes carbon metrics and controversial product involvement data

Implications for companies and investors

Warren Buffet said famously several years ago: "It takes 20 years to build a reputation and five minutes to ruin it. If you think about that, you’ll do things differently." Today, with the proliferation of social media, a company’s reputation can be bolstered or tarnished in a matter of seconds. Against this backdrop, what are savvy investors and corporations to do to prepare for the new normal?

Truvalue Labs’ AI approach provides several advantages:

  • AI reduces the potential for ESG analyst biases to tilt assessments and ratings.
  • AI can be applied to more companies more quickly, while analyst-based ESG information is generally updated quarterly or even annually, depending on the provider.
  • Because Truvalue Labs’ AI does not rely on self-reported data, it is not as constrained by uneven availability of company-provided data.

Sustainalytics’ analyst-based approach provides its own advantages:

  • It enhances a company’s disclosed ESG information with third-party insights.
  • Experienced analysts can provide invaluable judgment on a company’s culture and trends over time.
  • A company’s ESG information can be validated by an assurance provider to enhance accuracy and reliability.

So, which approach works better today? I believe the truth, as with most things, is in the middle. What works best in today’s market is a side-by-side comparison between analyst-based research and ratings such as from Sustainalytics alongside AI technology from Truvalue Labs.

My bet, however, is that within the next 24 months, machine again will prevail over human and AI will become the industry standard for assessing a corporation’s sustainability performance and impacts. This will, in turn, have broad implications for the future of corporate sustainability reporting and performance measurement.

It’s indeed a brave new world.

More on this topic