Evaluating a company’s exposure to environmental, social and governance (ESG) risk is an investment essential. But accessing reliable and timely information can be a challenge, given corporate transparency is uneven, and consistent global regulatory standards are still a work in progress. Now, data science is helping to provide the information investors need.
A skilled artificial intelligence (AI) and data science team can provide new information and ways of flagging, early on, whether a company may have a brewing ESG controversy that could send its stock price plunging, or whether a company may, in contrast, be developing new technologies that could help solve environmental issues.
AI can dig beneath the surface to identify with greater accuracy fresh, high quality, unique and complex data signals – creating new metrics, and drawing on alternative information sources to allow customized modelling. As such, AI is adding a useful new layer of information that can be utilized by a variety of funds, not just those labelled as “sustainable.”
And while these data science signals are just one small addition to the larger investment decision-making process, they represent a big step forward, offering investors the enticing possibility of an early lead in bringing important corporate issues and opportunities to the surface before they have a large material impact. They can also be customized, offering a different approach to ESG rating providers (so called “third party solutions”) that tend to rely on imprecise industry averages.
Shortfalls of existing data: Filling the gaps
AI can assess ESG risk, identify leaders and help fill the gaps in ESG data self-reported by companies themselves. Gap-filling matters because disclosure varies widely. Emerging market companies, for example, tend to disclose less than those in developed markets, small caps disclose less than large caps (it can be expensive), while companies in Europe disclose more than in other regions.
Disclosure also varies by sector and within sectors, as some companies may withhold information intentionally. Even companies releasing social responsibility reports do so infrequently and may not include early-stage problems. The gaps extend to third-party data providers, which don’t comprehensively cover all that investors want to know.
AI can deduce the missing pieces by, for example, using available data points that are correlated with an unreported ESG factor, producing outputs that allow investors to make systematic comparisons.
Here are a few real-life success stories:
Issue 1: Uncovering ESG risks before they become news
The data science solution: AI can be used to analyze complaints lodged by watchdog non-governmental organizations (NGOs), whose expertise puts them at the forefront of emerging concerns in their area, such as climate change, child labor and privacy rights. Using information gathered from the web, data scientists sift through NGO complaints to systematically identify, and overweight, material ESG risks most likely to subsequently hurt stock prices.
One case regarding very negative NGO complaints about the dangers of a chemical product appeared eight months before a court ruling and international news headlines on the problem. The company’s stock market value dropped more than 44% over the 14 months following the initial NGO signals. Complaints to NGOs have also helped identify the ESG issues that are becoming more important. For example, from 2018 to 2021, the financial sector stood out for the number of complaints lodged by NGOs around financing related to climate change (Exhibit 1).
NGOs frequently flagged the financial sector on climate change concerns
Exhibit 1: Climate change complaints by sector (%)
Source: SIGWATCH, 2018-2021.
AI analysis has also helped determine the topics within the issue of climate change that have attracted the most NGO complaints globally in recent years. Water is the top issue that has risen to the fore (Exhibit 2).
Shifts in climate-related NGO complaints suggest the types of negative news that may hit stock prices of poorly performing companies
Exhibit 2: Change in topics discussed in NGO complaints tagged “climate change” (2018–2021)
Source: SIGWATCH, 2018-2021.
Issue 2: Identifying tomorrow’s leaders in ESG innovation by patent filings and R&D
Green technologies have become a profitable business, with many companies now claiming to be sustainability innovators. But are they? Capital expenditure spending isn’t likely to capture the quality, or size (if it’s a percentage of a small total) of green tech activity.
The data science solution: AI allows data scientists to track and assess green tech patent filings and corporate research and development (R&D) investments, helping determine the companies with the most promising green tech ideas. The quality, rather than quantity, of patents filed matters most. Data scientists consider a number of quality measures, aggregating them into a signal that identifies sustainable leadership.
One leader, identified in the packaging industry, filed patents to reduce the amount of material used and to incorporate sustainable materials. Investigative AI processes also spotted an oil and gas company filing high-quality green tech patents, presenting an investment opportunity before the market had priced in the growth potential.
Issue 3: Using AI to assess workforce diversity below the surface
The commonplace “diversity signal” – a metric that shows, for example, whether a company has a diversity program in place or not, or the percentage of female employees – can contribute to greenwashing/social-washing and hide the true picture of operational sustainability.
The data science solution: Data science is helping to clarify how diverse a company really is, in an ESG area where self-reported metrics are often incomplete. To deduce diverse hires, AI can analyse public employment profiles and other alternative data sources. Staff turnover is another salient issue companies may not disclose in a timely manner. AI can use training expenditure and reliance on temporary employees, among other data, to predict this financially material issue. Data analysis can also uncover whether employees of a particular gender, race or ethnicity are concentrated in certain areas or in low-ranking job functions rather than reaching senior levels (undermining claims to true diversity).
Local demographics can be factored in, since diversity looks different in different places. The analysis can be done for an enterprise or a single line of business.
In the tech industry, for example, data science helped uncover that a particular company had a decidedly mixed track record on gender diversity. Even as the percentage of women employees rose over the last decade, women’s average seniority declined.
The data signals may also trigger an asset manager to engage a company in conversation on social matters, if the company is claiming something that is not really the case.
Conclusion: A deeper view of company fundamentals
Assessing financially material ESG risks and opportunities requires going beyond commercially available ESG ratings. Machine learning and AI give active asset managers the insights needed to identify likely sustainability leaders and ESG laggards before the market has priced in these insights.
The process enriches ESG understanding in a thoughtful, deliberate and differentiating way, allowing for more granular investment decisions that can bolster a wide range of ESG-integrated investment strategies.
09wy220111132431