How can artificial intelligence, AI, contribute to the work of responsible investment? That was the main question when ICMIF member Länsförsäkringar’s Research Fund (Länsförsäkringar Forskningsfond) started its research project “ESG prediction”.
The research examines how AI could be used as a tool to handle unstructured sustainability information and contribute to the investment process within Länsförsäkringar Liv’s asset management, in particular by predicting companies’ sustainability risks.
“Previous research shows that there are many methods for measuring companies’ sustainability work, but the research differs and there is no major consensus between the methods. In addition, the values are often lumped together into an overall assessment of companies which have been analysed, which complicates the interpretation between acting and individual evaluation criteria,” says Kristofer Dreiman, Head of Responsible Investments at Länsförsäkringar Liv.
The research on AI and responsible investments is funded by the Länsförsäkringar Research Fund and conducted in collaboration between employees from Länsförsäkringar Liv, Stockholm University, Gävle University and via the Centre for Research on Economic Relations (CER) at Mid Sweden University in the field of economics and AI. In the initial phase of the research project, the relationship between companies’ sustainability ratings (also known as “ESG ratings”) and the presence of involvement in controversies was examined. For example, the researchers found a low correlation between the E-ratings (the E-value in ESG*) and the risk of an environmental controversy, which means that companies with relatively high E-ratings still have a high risk of involvement in environmental controversies.
Länsförsäkringar believes that previous research and the new results confirm the need for new, alternative methods that can differentiate companies based on maturity in sustainability work and sustainability risks. In addition, there is now much more actual sustainability information about companies available, albeit of varied quality and coverage. More sustainability information will also be available in the future as a result of new EU regulations for a sustainable financial market.
Kristofer Dreiman says: “This is where the idea of AI and machine learning as a tool in the work with responsible investments came from. After the initial phase, the project is now investigating how it is possible to use AI and machine learning to find patterns and connections between, for example, environmental indicators and environmental controversies. With the help of historical data and patterns, these values could predict companies’ sustainability risks. These risk levels can be analysed against which preventive measures companies take. The goal of the ESG prediction project is to be able to identify companies with high sustainability risks, explain why and then use the results as a basis for further analysis or advocacy work.”
* ESG values are provided by various methods for evaluating and measuring companies’ commitments and actions linked to environmental, social and (corporate) governance-related factors.