This report is the product of collaboration between Amundi, a global asset manager, and the International Finance Corporation (“the IFC”) on the analysis and availability of environmental, social and governance (“ESG”) data for investments in emerging markets. It focuses on the use of artificial intelligence (“AI”) for that purpose.

As background, the report identifies the growth in ESG investing, the positive correlation between ESG integration and financial performance and, in emerging markets, ESG integration as an investment strategy with the potential to achieve development outcomes that align with the United Nations Sustainable Development Goals. However, the availability of reliable ESG data is a challenge for investors who want to integrate ESG considerations when assessing emerging market investment opportunities.

To address this challenge, the report discusses how AI could help investors analyse “unstructured data”, to enhance investing in emerging markets, examples of such data being news articles, civil society reports, ESG disclosures, annual, integrated and sustainability reports and bonds prospectuses. This data is currently underused in analysis of corporate ESG performance. AI applications, enhanced by machine-learning algorithms and cloud computing, have led to innovations in the analysis of unstructured data on a large scale through the use of natural language processing (“NLP”) techniques.

The report contains a detailed description of an experiment conducted by Amundi and the IFC to test an ESG-domain-specific NLP application (“esgNLP”) to support the analysis of the ESG performance of emerging market financial institution (“FI”) issuers of fixed income bonds. For this, Amundi’s controversy and ESG scores for a group of FI issuers of hard currency debt are compared with an NLP analysis of short-term ESG controversy and long-term ESG performance signals that have been extracted from unstructured data.

The report records a number of findings from the experiment. It establishes esgNLP’s potential as an additional ESG data analysis and scoring tool. It demonstrates the value of unstructured data as a source of insight into ESG performance for emerging market issuers. It demonstrates the potential for AI solutions such as NLP to unlock value from such unstructured data and for algorithms, such as esgNLP, to analyse large amounts of unstructured text from public sources, thereby allowing for a rapid, comprehensive and at-scale analysis.

Conducting an analysis by document type also allows for the comparison of “sentiment profiles” (positive, negative or neutral) of different sources of information and types of documents. The difference between sentiment profiles of issuer- disclosed information and of analyses of text from media and other sources supports the case for greater transparency in material non-financial disclosures as well as independent verification and assurance of that information.

Issues to be addressed with the esgNLP model include complexity posed by non-English text in emerging market disclosures, transparency around how the model works, refinements to manage data bias and managing model drift, being the deterioration of model performance due to changes in data.

The report recommends cross-industry collaborations between “big finance” and “big tech” to address ESG integration and support the development of sustainability solutions like esgNLP.

1. Integrated environment, social and governance (“ESG”) investment strategies can support efficient resource allocation to economic sectors. This has the potential to contribute to development outcomes that are aligned with the Sustainable Development Goals (“SDGs”) which, themselves, have spurred the importance of ESG issues.
2. Emerging markets present the greatest opportunities for investors to achieve impacts through the SDGs because their development needs are the most significant. Institutional investors can play an important role in achieving the SDGs by aligning their investment goals with ESG and impact outcomes that are connected to the SDGs in emerging markets.
3. The funding requirements to meet the SDGs are much larger than the resources allocated to emerging markets. Although the global ESG fund universe has tripled since 2015, most of the growth has been in the developed world.
4. Emerging markets have a potential for higher returns, but investors have historically regarded those markets as riskier than the markets in developed economies. Reasons given for this include political instability, poor transparency, weak legal systems and creditor rights frameworks and data gaps.
5. Access to data is the primary challenge faced by investors who want to integrate ESG considerations when assessing emerging market investment opportunities. This is due to a lack of harmonisation of reporting standards, taxonomies and extra financial indicators, gaps in publicly available ESG data and reliance on ESG data providers whose ratings and methods have been criticised for being incomplete and inconsistent.
6. Certain data sources, namely “unstructured data”, being data sourced from various entities beyond just the issuer, are underused for ESG assessments of issuers in emerging markets. Such data, if correctly analysed, could strengthen an ESG assessment as it gives investors access to new types of information, providing perspective to data already available.
7. Artificial intelligence (“AI”) and, specifically, natural language processing (“NLP”), can enable investors to access unstructured data sources in an efficient way. The rapid and comprehensive analysis of data on a large scale that these methods provide, enables investors to direct resources and technical capacity, set priorities, develop mitigation measures, meet stewardship goals and define the scope of an engagement program.
8. NLP techniques such as sentiment analysis, name entity recognition and text summarisation from unstructured data enables more consistent due diligence through the comprehensive identification of ESG risks that it provides. AI and NLP unlock the value of unstructured data as a source of insight into ESG performance for emerging market issuers.
9. Investors are making use of AI and emerging technologies to support ESG data collection and analysis for both developed and emerging markets. Advances in AI technologies can help investors overcome ESG data challenges and thereby play a transformative role in unlocking emerging markets for greater investments in the short and medium term.
10. This report contains the following resources:
    • An appendix, tables and charts relating to the AI NLP model and the outcomes of its use in the experiment the subject of this report.
    • A list of sources used in this report.
    • A table containing a discussion of the Sustainable Finance Disclosure Regulation introduced by the European Commission.