The pitfalls of (non-definitive) Environmental, Social, and Governance scoring methodology

Özge Sahin, Karoline Bax, Sandra Paterlini, Claudia Czado

Research output: Contribution to journalArticlepeer-review

10 Scopus citations

Abstract

Evaluating companies' sustainability performance embraces environmental, social, and governance (ESG) activities. Data providers assign companies ESG scores as a quantitative measure based on available information. Refinitiv (previously ASSET4) is a key data provider whose scores are used extensively by researchers and companies; however, their ESG scoring methodology allows the scores from the five most recent years to change post-publication without any announcements. Such ESG scores are called non-definitive. Then, ESG research findings and companies' sustainability performance using the ESG data from the same data provider might be inconsistent. Optimization and exploratory data mining approaches show that it is possible to change ESG scores to exhibit stronger risk dependence. We discuss how the initial disclosure of ESG information and updating the published ESG information alter the way ESG scores are computed in a given industry group, impacting ESG research findings significantly. Moreover, the initial disclosure of ESG information and an update in the published ESG information might allow some companies to appear more sustainable, even though nothing has changed. Finally, our work indicates the criticality that should be addressed to improve comparability within research studies and companies' sustainability performance relying on data from the same ESG providers.

Original languageEnglish
Article number100780
JournalGlobal Finance Journal
Volume56
DOIs
StatePublished - May 2023

Keywords

  • Corporate social responsibility
  • Data mining
  • Data quality
  • ESG scores
  • Risk
  • Sustainability performance

Fingerprint

Dive into the research topics of 'The pitfalls of (non-definitive) Environmental, Social, and Governance scoring methodology'. Together they form a unique fingerprint.

Cite this