Early Alert System: One Way AI May Help Improve Investment Outcomes

Natural language processing can give investment managers an early warning about negative ESG news.

Artificial intelligence (AI) is taking the world by storm, and for good reason. AI can help automate mundane tasks, improve operational efficiency and enhance human decision-making. 

Some of its benefits extend to investing as well. We believe that powerful AI tools such as natural language processing (NLP) may help active managers deliver better investment outcomes by alerting them to negative environmental, social and governance (ESG) news before it becomes widely disseminated.

The Fallout When Bad News Breaks 

From environmental disasters like oil spills and chemical leaks to social issues such as child labor to governance failures like executive misconduct, negative ESG news can severely damage a company’s reputation, erode investor confidence and undermine financial performance.

Our research indicates that, while positive ESG news has quickly boosted company returns, negative ESG has done just the opposite. Companies with negative ESG events have underperformed their peers—not just on the first day the news broke, but for weeks after (Display). For stocks associated with negative ESG news, returns over the 10-day holding period shown annualize to –7.3% in excess return (–1.2% for fixed income); for stocks associated with positive ESG news, returns over the 10-day holding period shown annualize to 6.2% in excess return (0.2% for fixed income).

Natural Language Processing Changes the Game

Clearly, if asset managers could uncover ESG-related controversies faster than their peers, they’d have an investment edge. But that’s easier said than done. Manually sifting through mountains of corporate filings and news articles is not just cumbersome and time-consuming; it’s also a Sisyphean task that must be repeated daily.

That’s where NLP technology comes in.

NLP is a field of AI that enables algorithms to understand, interpret and generate human communication. NLP applications have evolved dramatically since their advent in the 1950s, progressing from simple pattern matching—such as recognizing the phrase “thank you” and responding with “you’re welcome,” or counting the number of occurrences of a given word in a text—to more sophisticated techniques that involve deep learning and contextual understanding.

The result is a technology that is highly accurate and nuanced in its understanding and generation of human language, with the power to process and analyze vast amounts of raw data quickly and efficiently.

Putting NLP to Work: Controversy Alerts

We’ve built a powerful NLP tool that serves as a complement to our fundamental analysis. Incorporating the latest large-language models, this tool transforms the daunting task of sifting through thousands of company and news reports into an efficient and insightful process that flags potentially material ESG issues in real time.

Our tool monitors companies around the world, screening global news in multiple languages for controversies across a wide range of topics, including modern slavery, child labor, discrimination, tax scandals, executive pay and corruption. Furthermore, the tool is scalable, screening for developing ESG controversies across hundreds of companies simultaneously and providing timely alerts before negative news impacts the market.

It understands financial jargon, is trained in ESG-specific knowledge, and can assess tone and context, with the goal of evaluating the sentiment around a particular ESG news item with respect to a specific company. The tool extracts relevant information from the news sources, creates a summary and highlights what it deems most important for our analysts to evaluate. And it includes links to the original articles, allowing analysts to verify the information quickly.

For example, our NLP tool flagged a large mining concern with operations in Central America, alerting us to alleged poor treatment of employees and community members that compelled local governments to file lawsuits. Eventually, the company’s local mining permits were pulled, and it had to abandon operations in the area after investing nearly US$10 billion in the project. Fortunately, our tool had alerted us to these risks ahead of the government’s punitive actions.

The Bigger Picture: Integrating Our NLP Tool

Of course, AI and other tools are no substitute for sound fundamental analysis, which is why negative ESG news alerts don’t necessarily keep us from taking a position. The news may represent headline risk with which we’re comfortable or that we find is already priced into valuations.

As always, it’s up to our investment teams to assemble relevant data, including NLP alerts, into a bigger picture. Ultimately, our NLP tool is just one resource—albeit a valuable one—within a broader fundamental analysis that helps us make an informed decision about a company.

What’s more, NLP allows our analysts and portfolio managers to devote less time to tracking down data and more time to thinking critically about it—and to putting it to use on behalf of our clients.