Scope 3 Emissions: Data Quality and Machine Learning Prediction Accuracy

Value of the Information & Strength of the Argument

The webinar provides valuable insights into the quality and reliability of Scope 3 emissions data, a critical issue for climate finance. The argumentation is solid, based on a rigorous comparison of three major data providers and a comprehensive machine learning analysis. The speakers clearly explain their methodology and findings, making a strong case for the need for improved data quality and the potential of machine learning to fill gaps. The study’s novelty lies in its detailed examination of data divergence and the category-level prediction approach, which significantly improves accuracy.

Scientific Rigor, Source Quality, Title Accuracy

The research is scientifically rigorous, with a clear methodology and use of multiple datasets. The sources are primarily the data providers themselves, and the study builds on prior work published in Energy Economics. The title accurately reflects the content, and the presentation is well-structured. The webinar is an original study, not yet peer-reviewed, but the authors are reputable academics. The analysis of data quality and prediction accuracy is thorough, and the conclusions are well-supported by the evidence presented.

The title accurately reflects the content, focusing on data quality and machine learning prediction of Scope 3 emissions.

8/10

The webinar presents original research with a clear methodology, using multiple datasets and rigorous statistical analysis. The speakers are established academics in climate finance. The study is not peer-reviewed yet but is based on solid empirical work.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 02:00Introduction to the webinar and the research questions.
• 05:00Explanation of Scope 3 emissions and the importance of data quality.
• 10:00Overview of the data providers and the matching process.
• 15:00Discussion of divergence in raw emission values and rankings.
• 20:00Analysis of composition and completeness of Scope 3 categories.
• 25:00Introduction to machine learning models used for prediction.
• 30:00Presentation of prediction results, including improvements by category.
• 35:00Discussion of limitations and implications for investors.
• 40:00Q&A session with audience questions.
• 45:00Concluding remarks and future research directions.

• Energy Economics paper on machine learning for Scope 1 and 2 emissions — Prior work by the authors that inspired this study.
• Bloomberg, Refinitiv Eikon, ISS datasets — Data providers used for the analysis.

• CDP (Carbon Disclosure Project) reports — Firms report emissions to CDP, which is used by data providers.

• ISS data adjustments — ISS uses proprietary models to adjust reported values, leading to significant divergence from other providers.