The Latest Trends in ESG as Deciphered by AI: An Introduction to the Latest Research
Report
The Latest Trends in ESG as Deciphered by AI: An Introduction to the Latest Research
Other
Paper Description
SHARE :.
INDEX
Introduction.
December 8, 2024,aiESGis the Chief Data Scientist atLi Chao(first author), and Chief ResearcherKeeley Alexander Ryotaand Chief Scientific Advisor.Hidetaro TakedaCEO.Sekigeki Oyoshi (god of wealth, music, eloquence and water)and the president of the board of directors.Shunsuke ManagiThe paper, co-authored by
The results of this research have been published in Business Strategy and the Environment (2024 Impact Factor: 12.5), a prominent academic journal in the field of ESG (Environmental, Social and Governance).
Paper Title: "ESG Tendencies From News Investigated by AI Trained by Human Intelligence" DOI (article link):.https://doi.org/10.1002/bse.4089
Research Points
In this study,Objective and quantitative analysis of how the tone of ESG-related news is changingThe text match pre-trained transformer (TMPT), a proprietary AI model, will be used for approximately two years from September 2021 to September 2023, and will be published by a U.S. news organization.Approximately 2.16 million articles in Englishanalysis.
The results revealed the following trends
Number of reports on environmental issues (E) is declining
Increasing interest in topics related to social issue(s)
Interest in ESG as a whole is on the rise
This change has important implications for companies. It will be increasingly necessary to accurately identify changes in social concerns and reflect them in business strategies. This study consists of the following steps
Figure 1: Flow of this study (prepared by aiESG)
This article describes the researchEasy-to-understand explanationsThen both,Services provided by aiESGWe will also provide information on If you are interested in ESG data analysis and strategy development, please read to the end.
Background and Objectives - Gap between ESG practices and public interest
In recent years, climate change and economic instability have led to air pollution and otherenvironmental problemand poverty and corruption.social problemare becoming increasingly apparent. In response to this situation, companies and governments haveEffective ESG (Environmental, Social, and Governance) Responseis required.
However, because ESG is such a broad and vague topic, it is not easy for companies and governments to accurately identify specific topics of interest to the public (e.g., air pollution or labor rights). The result,Difficult to adequately reflect the real voice of society when developing ESG strategiesThis is the challenge.
Therefore, in recent years, as a clue to the social evaluation of ESG,news reportThere is a growing focus on news, because news not only tells of market trends surrounding companies and governments,An important source of information that reflects the interests and attitudes of people in society as a wholeThis is because the
However, there has been no established method for quantitative and systematic analysis of ESG social evaluation using news.
In order to address this issue, this study aims to examine the past two years ofUnique AI analysis of vast amounts of news dataSh,Real-time visualization of social assessment on ESGThe goal was to make it possible for the company to Thereby,Changing tone of media coverage of environmental and social issuesand,Shifts in people's interestsIt will be possible to clarify the
In the next section, aiESG developedUnique AI Technologyand the analytical methods that utilize them will be explained in detail.
How to use AI to analyze news in 3 steps
In this study,Proprietary AI model "TMPT and utilizing theQuantitative analysis of ESG-related news trendsI did. The analysis followed the sequence shown in Figure 2. First,Analyzing the ESG relevance of newsUnique AI model that serves as the basis forDeveloped "TMPT."(1), and a large number of major media outlets, especially in the U.S.Gathering News Data(2). Then, utilizing TMPT,Analyze changes in news coverage and news toneand the evolution of society's interest in ESG (3).
Development of proprietary AI model "TMPT In this study, in order to accurately analyze the relationship between news and ESG-related topics, a unique AI model, "TMPT(The text match pre-trained transformer)TMPT has the following features.
Training data: 200,000 academic papersand constructed as training data for TMPT.Zero-shot learning*.Enables flexible labeling by
Differences from existing methods: Highly accurate identification of paraphrased expressions (e.g., "greenhouse gases" ⇔ "CO2", "GHG") that are difficult to identify with the conventional NER method*.
Percent correct: 85.73% (high accuracy for zero-shot learning)
Previous studies have focused on NER methods that have focused on specific industries or companies. However,Need for a model that can accommodate a more diverse industryDue to the growing number ofA wide range of topics can be identifiedThe TMPT has been developed to be the first of its kind in the world. TMPT has trained over 200,000 paper data and approximately 130 million coefficients,Accuracy significantly higher than previous modelsYou can determine the relevance of a news item by
*Analysis of ESG topics using AI technology:Eigenexpression Extraction (NER)andZero-shot learningWhat is the role of
With the growing importance of ESG (Environmental, Social, and Governance) data in recent years, methods that utilize AI technology to analyze large amounts of text data and gain useful insights are attracting attention. Among them,Named Entity Recognition (NER)andZero-shot learningplays an important role in helping companies organize and utilize ESG-related information.
NER is a technology that identifies proper nouns such as company and person names in text data. For example, while it can identify the word "greenhouse gas," it is difficult to comprehensively recognize synonyms such as "CO₂" and "GHG. The following is a brief overview of the ESG field.
Figure 3: Explanation of NER and issues (created by aiESG, icons used: Freepik)
On the other hand,Zero-shot learningcan be used to properly classify words based on their characteristics, even if they have not been learned beforehand. For example, just as an AI can recognize an elephant based on features such as "long nose" and "big ears" even if it does not know the word "elephant," it can understand the relationship between "greenhouse gas" and "CO₂" or "GHG" in the context of ESG and appropriately determine relevant information.
Figure 4: Overview diagram for zero shot learning (created by aiESG, icon used: Freepik)
Thus, for companies to accurately analyze ESG data, they need techniques that extract clear proper nouns, such as NER, and techniques that capture context and features, such as zero-shot learningCombination of technologiesis effective, and advances in AI-based data analysis are expected to advance ESG strategy decision-making to a higher level of precision.
2.News Gathering Next,Two years from September 2021 to September 2023over a period of ..,Major U.S. MediaFrom, Collected ESG-related news to be used in the analysis of relevance to ESG topics.
Media covered: CNN, The New York Times, Forbes, etc. 354 U.S. news organizations
Target article:. Approximately 2.16 million(at sentence-end, falling tone) indicates a confident conclusionEnglish News
3. Trend analysis by TMPT Then, using the TMPTQuantitative analysis of trends in ESG-related newsI did. The analysis centered on,Relevance Analysisandsentiment analysisThe two methods are. A total of 12 indicators (topics) were used in the analysis, including environmental (3) and social (9), These topics have been the subject of analysis in many studies in the ESG field.
Figure 5: List of environmental and social topics used in this study
Relevance analysis (change in news coverage) First, news related to ESG isHow well reported is it?indicates object of desire, like, hate, etc.Time series evaluationI did.
TMPTto score how relevant the collected news data is to ESG-related topics,Variation in the amount of news coverage per topictracked.
For example, we analyzed in detail when themes such as "greenhouse gases" and "unemployment" increased or decreased.
Increase or decrease in the amount of media coverage indicates a change in the level of public interestindicators and provide clues to which areas firms and policy makers should focus their attention.
2.Sentiment analysis (change in news tone) Next, the news tells ustenor or drift of an argumentwere analyzed. The indicators used were, Positive (optimistic), negative (pessimistic), neutral (neutral)will be. These are by using a highly accurate AI (97.791 TP3T accuracy) optimized for financial news, News sentiments and opinions were analyzed chronologically.
TMPTand the emotions and tone contained in each news article,Positive, Negative, NeutralClassification into.
For example, we assessed whether the tone of news stories about "unemployment" and "corruption" has changed positively or remained negative.
This allows specific ESG themes to beUnderstand how you are perceived by societyCan be done.
Results - ESG News Trends and Changing Public Interest
As described above, we used TMPT to conduct a news relevance and sentiment analysis of news coverage on ESG.change in toneAnd,Shifts in people's interestswas revealed. In particular,Increased interest in social aspectsandDecreased interest in environmental aspectswere noticeable.
Increased interest in social aspects
The amount of coverage on the social front isOverall increaseand in particularCorruption" and "Poverty and Inequality."News about thesurgeI did.
Social issues such as "forced labor" and "unemploymentThe argument about thePositive trendsand shows that theseImproved social problemsThe results of this study suggest that the "new" market may be in the process of being established.
However,Access to improved drinking water sources."The tone of the news about the news continues to benegativeand that challenges remain.
2.Decreased interest in environmental aspects
Energy Use, Air Pollution, Greenhouse Gases.The amount of news coverage aboutdecreasing trendin particular.The topic of "air pollution" saw a significant drop in coverage.I did.
In addition, reports of "air pollution" areThere is a lot of negative tone.From,Interest in the topic is declining as pessimism spreads among peopleIt can be seen that
These results are,Society's overall interest is shifting from environmental (E) to social (S) topicsThis suggests that
3.Overall ESG tone shifted to positive
Relevance analysis and sentiment analysis will help to determine not only the environmental/social aspects of the project,Changing interest in ESG as a wholeThe report also clarified about the
The amount of ESG-related news coverage is determined by relevance analysis.Overall increaseThe results of the survey confirmed that
Also in the sentiment analysis,Overall ESG tone shifts toward optimismThe results of the survey revealed that the company is doing so.
in this wayWhile interest in environmental aspects is declining, interest in social aspects is increasing and the tone of the news is changing positively.We found out that In addition, the amount of coverage of ESG-related news overall has increased and the tone of the debate has shifted to the positive,Society's interest in ESG as a whole is growing even moreThis is considered to be the case.
Consideration - New Possibilities for ESG Valuation Using AI
This study is based onFirst quantitative analysis of ESG-related news coverageThe results of the study are important and provide many insights into corporate decision-making. Until now, ESG assessments have been based on corporate disclosures and annual reports, but this studyA new approach that leverages media coverage to capture public interest and evaluation in real timeproposed. In the following, we discuss the key points of this research and potential future applications.
Shift in public interest in ESG
The study analyzed more than 2.16 million English-language news stories from 354 major U.S. media outlets over a two-year period from September 2021 to September 2023. The results revealed the following trends
Increased news coverage of social issuesand increased interest in topics such as "poverty and inequality" and "corruption.
News coverage of environmental issues declinesThe amount of media coverage on "energy use," "greenhouse gases," and "air pollution" declined.
The overall tone of ESG-related news isShift in a positive directionThe tone of the news, especially regarding social aspects, changed positively.
These results are,ESG evaluation is shifting from "Environmental (E)" to "Social (S)This suggests that the especially in the wake of the new coronavirus pandemic and economic instability,Increased interest in social issues and corresponding changes in media coverageThis is considered to be the case.
2. New approaches through the use of AI
In this study,Proprietary AI "TMPTto automate the analysis of ESG-related news. The application of this technology will enable companies to develop more sophisticated ESG strategies and risk assessments.
Real-time understanding of media trends
Traditional ESG assessments have relied on company reports and disclosure information,Utilizing news enables strategy formulation based on the "true feelings of societywill be.
This allows companies to accurately understand consumer and investor expectations and to be more flexible in their ESG responses.
Utilization in Integrated Reporting
TMPT is an excellent AI model for text analysis and can be used to analyze texts other than news.
Take TMPT, for example.Analyze a company's Integrated Report (IR)It is possible to As with the news data in this study, relevance analysis can be performed on the IR text data to quantify how well the IR statements relate to ESG topics. Through this, it is possible to determine which ESG topics are prioritized and focused on by the company.Quantitative scoring of trends and policies (+ ability to perform)It is possible to do so.
This will enable stakeholders, including investors, to visually judge a company's ESG strategy and make appropriate investment decisions. Furthermore, companies that score highly on ESG strategies invite further investment from investors, enabling them to build strong relationships with investors and a solid financial foundation.
In other words, IR analysis based on the TMPT is a useful strategy not only for stakeholders, including investors, but also for companies.
Quantitative Assessment of ESG Risks
If the methods of this study are applied,ESG risks can be scored not only on a company-by-company basis, but also by region and productIt is. It is expected to provide more detailed ESG risk information to companies and consumers.
Example: Analysis of news data on the amount of resources used in a product and the likelihood that the product poses a child labor risk.
3. why is this research important?
The significance of using AI as an ESG assessment method goes beyond mere technological innovation. The results of this research are,It provides a new perspective to help companies and investors make decisions that are more in line with actual conditions.
Real-time visualization of ESG "social evaluation
Traditionally, companies have not had the means to understand ESG "social evaluation" in real time. However, by utilizing the methodology of this study,Immediately catch changes in public interest and evaluation through media reports and reflect them in strategieswill be.
2.Demonstration of performance of high-precision AI models
In this study, we demonstrated the effectiveness of ESG trend analysis using AI. In the future, this methodology will be utilized,
Analysis of the company's integrated report
ESG Risk Assessment by Region and Product Such as,It is expected to be applied to more advanced ESG analysis.
3.Further development of academic research and corporate strategy on ESG
This study is,Media coverage is important for understanding public interest in ESG and market trendsWe have shown that With the dissemination of this knowledge,
Promote academic research on ESG
Acceleration of the trend for companies to develop flexible ESG strategies based on the "voice of society
is expected to be
4. future prospects
The analysis in this study was conducted on English-language news. However, in the future, similar analysis can be conducted in non-English speaking regions to measure interest in ESG, and in the futureIt would also be applicable in the Japanese market.
AI analysis of ESG reporting in Japan could help domestic companies optimize their strategies.
It is imperative to understand the changes in ESG concerns of Japanese consumers and investors in real time, and to utilize these changes in corporate sustainable growth strategies.
In the future, by applying the methodology of this study to the Japanese market,It will be possible for Japanese companies to enhance the sophistication of their ESG assessments and establish sustainable management strategies that are more in line with actual conditions.
Conclusion
In this study, we objectively and quantitatively demonstrated temporal trends in the relevance of ESG topics and changes in people's perceptions based on news reports, using our proprietary high-precision AI.
This study,Rising ESG trends do not necessarily coincide with attention to individual itemsindicates that the interest level of each item is changing in a way that is influenced by social conditions. It should be noted that each of these items has its own level of interest in a way that is influenced by social conditions.
Based on the results of this study, aiESG willFurther sophistication of quantitative ESG assessmentWe will continue to work toward this goal. In the future, we will work to develop a corporate ESG strategy thatRisk Aversion, ,Supply Chain OptimizationIn addition, the disclosure of information about your company will causePredicted impact on stock priceIt will be possible to develop services that support the
Based on this research, aiESG provides ESG analysis services related to the supply chain of manufactured goods. If you have any questions or concerns about corporate disclosure and sustainability information disclosure, please feel free to contact us.