Ebook: Navigating AI in ESG & Risk Management: Unveiling the Mechanics
September 18, 2024
•
5 mins read
With all the buzz around Generative AI, it’s easy to forget that artificial intelligence (AI) has been driving innovation across industries for years. Environmental, Social, and Governance (ESG) and risk management are no different. However, at the rate AI is advancing and as the amount of raw data available for analysis continues to expand, the need to understand AI is more pressing than ever.
AI is transforming ESG, turning complex data into predictive insights and reshaping our approach to risk. But what does this mean for the industry, and how can professionals leverage this technology to maintain a competitive edge? The future of AI in ESG and risk management is not just a matter of technological advancement but a narrative of how we evolve with it.
This ebook dives into how AI works when applied to ESG, shares a few practical examples of what it looks like in real life, and offers a few predictions for what comes next.
Dive deeper into the mechanics of how AI works in ESG and equip your organization with the insights needed to enhance your ESG practices. Fill out the form below to access your copy.
SESAMm, an alternative data specialist at the forefront of AI-based text analytics, and EthiFinance, a non-financial analysis and consulting agency with expertise in sustainable development, are now working together to launch a new unique service: Identifying and scoring ESG controversies on European Small & Midsize Enterprises (SMEs).
When dealing with SMEs, a common issue with traditional data providers and rating agencies is the lack of coverage. Information gaps and opacity about a company can negatively affect asset managers in their investment decisions.
EthiFinance possesses strong expertise in rating SMEs – both listed and private – for the purpose of portfolio due diligence and monitoring. SESAMm’s TextReveal® tool leverages Natural Language Processing (NLP) to analyze more than 16 billion online articles and messages to track ESG topics.
“We have been working with SESAMm for months and have now reached a stage where we can deliver quality and depth of results needed to create reliable and relevant ESG alerts.”, shares Xavier Leroy, Head of Investor Services from EthiFinance “Our teams of ESG analysts have enhanced and refined these alerts for controversies into our methodology, thus ensuring value to our clients”.
Controversy detections are notoriously difficult to assess and evaluate properly, be it the language barrier, lack of data, or difficulty for machines to work with ambiguity. The democratization of artificial intelligence and data science, the exponential growth of data, and wider access to more computing power have provided a fertile ground for new solutions to address this issue.
“Natural Language Processing is a game-changer, especially when working on ESG. Combined with alternative data, it helps us crunch billions of articles in dozens of languages and detect controversies very quickly” explains Sylvain Forté, CEO of SESAMm. “By fine-tuning TextReveal® with EthiFinance’s expertise in ESG, we are able to reach excellent results with 100% accuracy on the detection of critical controversies.”
SESAMm’s data and tools are leveraged globally by asset managers like The Carlyle Group, to conduct ESG studies for the purpose of due diligence, company analysis, and risk management. The partnership with EthiFinance breaks new ground thanks to its unique expertise in sustainability, ESG know-how, and very specific focus on European midsized companies.
SESAMm and EthiFinance have also developed a specific virality indicator to help assess how ESG controversies are spreading on the web. “This technological edge relies on an enormous amount of data which we can access through SESAMm’s platform,” shares Leroy.
A worldwide scandal, significantly impacting a company’s market price and its ESG scoring, can start from just one article or post, sometimes in a very niche community. SESAMm’s data lake, with its billions of articles and documents, serves as a real-time curated proxy of the internet enabling the timely detection of such scandals and the subsequent insights.
“This level of speed and depth of information not only represents a headstart when it comes to the competition but also a safety-net for asset managers, looking to monitor and manage their portfolio” confides Forté. “This innovative approach can be applied to other regions, Europe is only a starting point.”
The two companies are now working together to expand this innovative approach to listed companies across Europe.
About EthiFinance
Founded in 2004, EthiFinance is a non-financial analysis and consulting agency that supports its clients in managing risks and opportunities related to sustainable development.
EthiFinance develops services tailor-made to meet the specific needs of its clients, including investors, banks, companies and not-for-profit organizations.
Since 2017, EthiFinance has been working closely with Spread Research, a French financial rating agency (registered with ESMA) and independent credit research provider. The two companies are now part of the Qivalio Group.
Qivalio is an innovative European group providing rating, research, and consulting services in the field of sustainable finance. The Group provides solutions for investors, companies and organizations to address the challenges raised by financing as well as social and environmental transformations.
SESAMm (www.sesamm.com) is an innovative company specializing in alternative data and artificial intelligence for investment. Its team builds analytics and indicators, such as Sentiment Analytics, ESG Indicators, and investment signals by analyzing billions of web articles and messages using natural language processing and machine learning. With its NLP platform TextReveal®, SESAMm addresses the entire value chain of alpha research. With 6 offices including Paris, New York, London, and Tokyo, SESAMm works with major hedge funds, banks, private equity firms, corporate and asset management clients around the world for fundamental or quantitative investment use cases, market analysis and competitive insights.
It's a word that most of us in the U.S. despise, almost as much as the word taxes. It's probably because, like taxes, we can't escape its wallet-draining effect when it increases. Maybe the way we feel about it is because the last time the U.S. economy deflated—giving us relief from it—was in the 1930s, when "Prices dropped an average of nearly 7% every year between the years of 1930 and 1933," according to Investopedia. But I digress.
We won’t go into how inflation works, but how the government calculates it—and how its categories affect it—has always been consistent. At least it was until the COVID-19 pandemic hit, that is.
What NLP text mining reveals about the U.S. economy inflation-rate factors and the online conversations about them
To ensure we're on the same page about how we came to the forthcoming information in this use case, let's cover a couple of basics on NLP text mining and inflation rate indexes.
What are NLP and text mining?
Natural language processing (NLP), an A.I. technology, automates the data analysis of mined textual, unstructured data. It includes natural language understanding and natural language generation to simulate a human’s ability to create language, and it’s a component of text mining that performs a special kind of linguistic analysis by deep learning algorithms so a machine can “read” text. Apps like Grammarly or Wordtune analyze text to improve a written text, for example, and chatbots use this technology to interact with customers. Text mining, or text analytics, is the process of examining big data document collections. It’s a computer science discipline that converts unstructured text data in documents and databases into normalized, structured data and datasets for analysis by machine learning models. Deep learning machine-learning algorithms then analyze this data, analyzing semantics and grammatical structures, to gain new insight or aid research from human language. Together, NLP and text mining are like a search engine on steroids.
The Consumer Price Index (CPI)
According to this Forbes Advisor article, "The two most frequently cited indexes that calculate the inflation rate in the U.S. are the Consumer Price Index (CPI) and the Personal Consumption Expenditures Price Index (PCE)." For this article, however, we'll only use the Bureau of Labor Statistics (BLS) method of CPI inflation calculation as a reference. CPI observes a specific group of commonly-purchased goods and services to gauge how prices fluctuate. These foods and services include:
Apparel: Women's and men's clothes, jewelry, etc.
Alcoholic beverages: Beers, wine, liquor, etc.
Energy and commodities: Gasoline, natural gas, electricity, etc.
Food: Items bought by the average consumer, such as breakfast cereal, milk, meat, fruits, vegetables, etc.
Housing and shelter: Rent, housing insurance, bedroom furniture, hotel or motel accommodation costs, etc.
Medical care services: Physicians' services, prescription drugs, medical supplies, etc.
New and used vehicles: Trucks, vans, sedans, SUVs, etc.
Tobacco and smoking products: Tobacco-related items, such as cigarettes, cigars, bidis, kreteks, loose tobacco, etc.
Transportation services: Airline fares, vehicle insurance, etc.
NLP text-mining process: web mentions matched to CPI categories
Using SESAMm's web text analysis engine TextReveal®, we analyzed textual data relating to the inflation topic within the U.S. from 2017 until now. For this analysis, we defined co-mentions as the articles and social media posts that mention "inflation" and at least one of the CPI categories. Note: Although we can analyze more than 100 languages, we focused on English in this case. Also, we didn’t conduct a sentiment analysis from the information extraction.
Figure 1: Inflation co-mentions by category and percentage.
From 2017 to 2019, inflation co-mentions within the U.S. are relatively stable (see Figure 1). But this trend changes with the first shift in 2020, continuing its rapid growth and peak by the end of 2021 due to this surge of inflation reaching record levels.
What was one of the main drivers of the inflation surge? Used cars.
3 used-car and inflation trends uncovered through NLP Text Mining
According to the U.S. Bureau of Labor Statistics, the cost of used vehicles was one of the main drivers of the inflation spike. How did used cars contribute to inflation? The chain of events occurred like so: The increased used-car demand was fueled by a new-vehicle supply shortage caused by a chip shortage generated by supply-chain interruptions due to the COVID-19 pandemic.
As the pandemic-induced supply-chain interruption unfolded, used-car trends developed. Here are three we found in our data mining research:
Trend 1: Co-mentions percentage for used vehicles more than doubled
Figure 2: Used vehicles co-mentions increase percentage-wise.
Based on the percentage of co-mentions compared to other topics, the used-car topic moves from the number eight spot to the number four spot in 2021 (see Figure 2).
Figure 3: Used-car co-mentions begin in early 2021 and exceed those for new cars.
Before 2020, mentions were relatively steady. However, we observe an increase in used-vehicles mentions caused by disruptions in supply chains leading to chip shortages (see Figure 3) as early as January 2020. These shortages led to a decrease in new vehicle inventory. The Statista report, indicating an increase of the used vehicle value index by 49 points compared to the price index recorded in 2020, supports our findings.
Trend 2: Used vehicle prices rose with used-car co-mentions
Figure 4: In 2020, inventory spikes as production and sales plummet, affecting inflation.
Because of the pandemic, car production nearly stopped along with the sale of cars, which created two situations: 1. high inventory to sales ratio and 2. historically low car production (see Figure 4). Vehicles sales picked up later, but car production was still suffering because of supply-chain disruption. That meant the inventory to sales ratio dropped to virtually zero.
So consumers with little-to-no options for new vehicles turned to used cars, increasing their demand and therefore increasing their prices. We confirm this hypothesis with increasing mentions within the used-vehicles topic, coinciding with an inventory volume decrease. All in all, used-vehicle prices rose 40.5%.
Trend 3: The COVID-19 pandemic and new vehicle inventory shortage increased demand
A smaller new-vehicle inventory wasn't the only reason consumers sought out used vehicles. They also wanted used cars because of the pandemic.
Figure 5: The pandemic and new-vehicle supply shortage became bigger reasons for consumers to seek out used cars over cost.
For 2020, we observe that consumers avoided public transportation by rising co-mentions between pandemic-related mentions and the demand for secondhand vehicles (see Figure 5).
Used-car and inflation trends summary
We can summarize the used-car and inflation trends with one phrase: It's a used-car seller's market. For example, online retailers like Carvana have leveraged these factors to grow significantly. In contrast, due mainly to significant supply chain disruptions, motor companies have had the opposite effect, with the Automotive industry projected to lose $210 Billion. Judging by the number of mentions in public web forums and social media, the chip shortage and used-car boom affected General Motors, Ford, and Toyota the most (see Figure 6).
Figure 6: General Motors, Ford, and Toyota suffered pandemic-related shortages the most based on co-mentions.
About SESAMm and TextReveal’s® NLP Text-mining Capabilities
SESAMm is a leading company in alternative data and artificial intelligence, delivering global investment firms and corporations descriptive, prescriptive, or predictive investment analytics worldwide. TextReveal is SESAMm's premiere NLP text-mining product, a solution that allows you to fully leverage NLP-driven insights and receive high-quality results through data streams, modular API and dashboard visualization, and signals and alerts. In other words, we organize, categorize, and capture relevant information from raw data for you.
In October 2022, this multinational integrated energy and petroleum company found itself in the crosshairs of two non-government organizations (NGOs), accusing it of "exploiting a gas field used to manufacture kerosene used by Russian planes in their bombings in Ukraine," according to the French daily Le Monde. The accusations expressly point out the 16 March 2022 strike, which killed around 600 civilians taking shelter at a Mariupol theatre.
The company? TotalEnergies.
The NGOs? Razom We Stand (Ukraine) and Darwin Climax Coalitions (France).
Also, in October, TotalEnergies posted a third-quarter net profit amid these allegations. "The French group reported an adjusted net income of $9.86 billion, compared with $4.77 billion for the same period in 2021 and $9.8 billion in the second quarter of this year," per Reuters.
Should investors and asset and portfolio managers be concerned? Is this one instance of allegations a nothing burger, or is it a sign—one of many red flags—to evaluate? Let's find out in this edition of Alternative Data Trends: TotalEnergies.
After the spike in controversies in March 2022, TotalEnergies's web mentions continued to increase. And as a reflection of the mentions volume increases, the company's polarity decreases, moving in the direction of negative sentiment. News events triggering these movements include the legal case for allegedly fuelling Russian bombers and refinery strikes over wages.
Let's take a closer look at TotalEnergies's environmental risks, specifically for general environmental strategy, climate change and atmospheric pollution, and biodiversity. The topic of climate change and atmospheric pollution takes the largest share of mentions. Web data driving this volume include:
TotalEnergies's social risks reveal even more controversies, looking at human rights, human capital, and customer relations topics. Human rights and human capital topics almost equally dominate topic mentions. Controversies highlighting the human rights topic include:
Finally, let's look at TotalEnergies's governance risk, specifically within topics for influence strategy and communication, corruption, and board of directors (executive and company management). The influence strategy and communication topic dominates the mentions volume. These controversies relate to greenwashing, such as:
Recapping TotalEnergies's ESG risks and public perception
We only scratched the web data surface for information on TotalEnergies, but even with this overview, you can probably see a pattern. Should investors, asset managers, and portfolio managers be concerned about the latest allegations against TotalEnergies? Maybe. At the very least, the data shows you should dig into the research and ask more questions to mitigate any risks in your portfolio companies.
Reach out to SESAMm
SESAMm is a leading NLP technology company serving global investment firms, corporations, and investors, such as private equity firms, hedge funds, and other asset management firms, by providing datasets or NLP capabilities to generate their own alternative data for use cases, such as ESG and SDG, sentiment, private equity due diligence, corporation studies, and more.
For questions, access to the full report, or to request a TextReveal® demo, contact us here:
Stay ahead with the latest in ESG and AI intelligence
Join our mailing list to receive new reports, event invites, and updates from SESAMm directly to your inbox.
.png)