The sheer volume of data available tracking companies’ environmental, social and governance standings has exploded in recent years. But the ESG conclusions that can be drawn from the information are still limited, according to a white paper published Thursday by Generation Investment Management.
“Today’s ESG data has real limitations,” wrote the paper’s authors. “The risk is that it puts the spotlight on what is available, rather than what is most important.”
Generation, the sustainable manager founded by Al Gore and David Blood, argues in the paper that the data that’s now available can be vastly improved using some readily available tools such as machine learning.
“We need better data that connects with the scale and urgency of the problems we face,” said Felix Preston, the firm’s director of research, in an interview.
Generation, which uses ESG data sources as one part of its fundamental investment process, acknowledges that organizations want data that can be easily used to screen companies.
“We can see the growing demand for simple ESG data that can be easily plugged into financial models at scale,” the authors wrote. “We believe that bridging the gap between the value of rich contextual information and the need to ‘plug values into the spreadsheet’ is where ESG data needs to go.”
The volume of ESG information is exploding, as data in general increases exponentially. For example, in 2004, about 300 companies — mostly European — publicly reported annual greenhouse gas data. Now about 7,000 global companies report the data, according to Generation. Vendors are collecting data on environmental and social risks in supply chains and breaches of environmental regulations and allegations of corruption.
Even so, says Generation, the information is spotty and doesn’t indicate whether a company is changing or what’s on management’s mind.
[II Deep Dive: Can a Hedge Fund Make the World Better?]
“ESG metrics are imperfect proxies for the environmental and social impact of companies and how they are governed,” the authors wrote. “Take the climate crisis. Last year’s greenhouse gas emissions data for a company’s operations is a commonly used metric. A more complete picture of how the company is doing would depend on understanding its supply chains, interactions with customers, opportunities to innovate and more.”
Preston said the kind of contextual information that Generation’s investment teams discuss and implement in their processes is broadly available. Machine learning could help incorporate it.
“Start weaving [that] into mainstream data,” he said. “At the moment in the ESG world, it seems like [machine learning] is focused on filling gaps and standardizing. But it could be used to bring in all this rich context that would be beneficial and better reflect sustainability risks and opportunities that companies face.”
Generation said that in addition to machine learning, technologies like satellite imaging can provide information on companies’ progress on certain initiatives. The asset manager also called for third parties to verify ESG information on areas such as supply chains and the use of raw materials. It also recommended that single data points on companies be broken down into their component parts.
“For firms collectively allocating trillions of dollars across every sector in the economy, the desire for a single-point-of-truth platform for ESG is irresistible,” according to the paper. “Ensuring the underlying data provide better proxies for the impact of both what companies do and how they operate is therefore an urgent task.”
