Machine Learning in ESG Risk Management: Machine Learning for Sustainability Disclosure in EU Pulp and Paper Industry under the CSRD : a literature review

Abstract

The Environmental, Social, and Governance (ESG) factors are central to the sustainability disclosure practices and risk management in the European pulp and paper industry (PPI). Especially, under the requirements of the Corporate Sustainability Reporting Directive (CSRD), this thesis examines how Machine Learning (ML) can strengthen ESG Risk Management in resource-intensive sectors. The thesis applies a literature review and qualitative documentary analysis to systematically explore the theoretical potential and applications of ML techniques for identifying and assessing ESG risks in the PPI. The thesis further explores the opportunities and challenges in integrating AI-driven approaches to ESG risk management. The analysis is guided by three central theories: Stakeholder Theory, Risk Management framework (Hopkins), and Socio-technical System theory (STS). The results demonstrate that ML can significantly improve the accuracy, timeliness, and financial relevance of ESG reporting while also enabling innovative risk identification for carbon emissions, water resources, supply chain dependencies, and economic distress indicators. Nonetheless, the pain points are insufficient data structure, misalignment between environmental and financial metrics, and ethical concerns, which are limiting the full implementation of AI/ML in the industry. The work is limited to a qualitative, conceptual exploration using thematic coding and secondary data, without technical modelling or primary interviews. As the thesis concluded, Machine Learning serves not as a tool for ESG frameworks but as a critical enabler in aiding more reliable, efficient, and forward-looking sustainability reporting in the PPI.