NLP and sentiment: A new tool for ESG investors
Natural Language Processing (NLP) can enhance ESG Investing in a variety of ways, from building alpha-generating strategies to monitoring controversy risk across portfolios.
Investors can make the most of the global trend in ESG investing using NLP technology, such as that developed by RavenPack, the company I work for.
RavenPack generates analytics by scanning online news, identifying entities, tying them to events, and then calculating sentiment. It has a rich ESG event taxonomy covering a wide range of controversies, from news about companies being fined for pollution, to labor disputes, to lawsuits.
This enables the large-scale tracking of controversies, covering millions of listed and non-listed companies, in multiple languages, from both local and international sources.
The power of nowcasting
One advantage of our data is that ESG sentiment can provide a complement to ‘traditional’ ESG ratings.
Many ratings are updated infrequently and a key advantage of using news sentiment is that a company’s ESG score can be nowcasted, providing the basis for real-time monitoring.
ESG investing has increased in popularity not only because a new generation of investors are more sustainability-conscious, but also because of growing evidence that highly-rated ESG companies tend to outperform their lesser-rated counterparts.
Our own research found that adding a news sentiment overlay can further enhance results. For stocks with a high ESG rating, those with positive sentiment tended to outperform those with negative sentiment.
Similarly for stocks with a low rating, those exhibiting negative sentiment tended to do comparatively worse than those with positive sentiment.
Investors’ growing focus on ESG factors has been responsible for the evolution of peculiar market phenomena.
After studying market reactions following negative ESG news events researchers from Monash University, isolated a recurring market pattern that could provide the basis for a profitable trading strategy.
Initial findings covered US stocks only, however, after broadening their study they found similar evidence globally, though the tendency was strongest in more efficient markets.
The ‘tick-shaped’ pattern started with the company’s share price falling steeply after the release of the negative ESG news. It then drifted higher over the subsequent medium-term. They described this as a ‘market overreaction’ due to investors increasing aversion to stocks tainted by negative ESG headlines. It was caused by the widespread offloading of such stocks due to managers trying to meet their clients’ ESG investment mandates.
The pattern could be used as part of an “Engagement Strategy” whereby the inciting incident would trigger activist investors to have a conversation with the management team of the company undergoing the controversy, to prompt them to take action to resolve the issue. This could potentially lead to the investor buying the drift higher, or simply keeping the company in their portfolio.
ESG and risk mitigation
In the realm of risk, research shows that a company’s ESG rating can materially impact its credit rating.
In one study RavenPack’s ESG event taxonomy was used to screen for company material credit events which were then cross-referenced with the same companies’ ESG scores. The study concluded that the lower the ESG rating, the higher the incidence of credit events.
One explanation was that poor ESG risk management tended to be indicative of poor risk management in general, leading to a lower credit-rating.
NLP-based analytics can be used in the early identification of emerging market themes that might materially impact a portfolio’s risk profile.
New market narratives often show up as ‘abnormalities’ in news data. Real-world examples include the radical policy shift on climate-change after the election of the Biden administration, the emergence of the “Black Lives Matter” movement, or the “MeToo campaign”.
By using keywords identifying these themes and topics, analysts can track their evolution and quickly evaluate how to protect investments.
Events relating to the topics can be codified into the AI’s NLP for future identification. This can lead to the development of highly actionable alerts and automated decision-rules. Novel events can easily be integrated for future tracking.
Big data use cases
There are many creative get-arounds analysts can use when conducting ESG analyses using Big Data and NLP-driven analytics.
One example might be how to measure employee satisfaction, a contributing factor to a company’s ‘Social’ performance. In most businesses employee satisfaction is measured via internal surveys or by using HR applications inaccessible to outsiders. An alternative might be to use Big Data and NLP to scan satisfaction reviews and ratings by employees on the jobsite Glassdoor.com.
The fallout from ESG controversies often leads to a ripple effect impacting many related assets. Sometimes this takes investors by surprise as hidden connections surface. To better prepare, investors can map company interrelationships using what are known as Network Graphs built on comentions in the news.
Network graphs highlight relationships with suppliers, customers, competitors, or other entities. They provide an idea of the strength, breadth and depth of ties. Journalistic digging can uncover hidden connections that it would be beyond the scope of a portfolio manager to find.
The tech giant Alphabet’s links with carmakers provides an example. In 2007 its network was composed of only one relationship – with GM, for financial reasons – fast-forward 10 years and it was connected to over 20 carmakers. This was because the company had become a technology provider to the autonomous car market, highlighting a shift in product strategy, with implications for its ESG profile.
The ESG investment revolution marks a historic turning point in global finance, and news analytics data gives investors a rich and comprehensive dataset to make the most of the new trend.
From providing strategies to generate alpha, to monitoring individual company ESG profiles, mapping interrelationships, to identifying emerging themes; news analytics data provides a wide coverage of the ESG space, making it an indispensable addition to any serious investor’s toolkit.
Peter Hafez (pictured), Chief Data Scientist, RavenPack
Peter Hafez is the head of data science at RavenPack. Since joining RavenPack in 2008, he’s been a pioneer in the field of applied news analytics bringing alternative data insights to the world’s top banks and hedge funds. Peter has more than 15 years of experience in quantitative finance with companies such as Standard & Poor’s, Credit Suisse First Boston, and Saxo Bank. He holds a Master’s degree in Quantitative Finance from Sir John Cass Business School along with an undergraduate degree in Economics from Copenhagen University. Peter is a recognised speaker at quant finance conferences on alternative data and AI, and has given lectures at some of the world’s top academic institutions including London Business School, Courant Institute of Mathematics at NYU, and Imperial College London.