A single company can look like a sustainability leader to one rating agency and a laggard to another, in the same month, using the same public disclosures. That is not a data error; it is the predictable result of rating agencies asking different questions, weighting the answers differently, and defining ESG itself in ways that only partly overlap. As regulators, investors, and boards lean harder on these scores to guide capital allocation, understanding exactly where the methodologies split apart has become essential rather than academic.
The scale of the disagreement is well documented in academic research rather than anecdote. A widely cited study published in the Review of Finance examined six major ESG rating providers, including MSCI, Sustainalytics, Moody's ESG, S&P Global, Refinitiv, and KLD, and found overall correlations between their ratings ranging from just 0.38 to 0.71, with governance scores showing the weakest agreement of the three pillars, averaging around 0.30. A related decomposition of that same divergence found that differences in how agencies measure a given issue account for 56% of the disagreement, differences in what they choose to include in scope account for 38%, and differing pillar weightings explain the remaining 6%, meaning most of the gap comes from judgment calls embedded deep inside each methodology rather than from which topics agencies decide to cover at all.
Agencies also disagree on how much raw data to pull in before scoring even begins. S&P Global's own published Corporate Sustainability Assessment methodology draws on roughly 1,000 underlying data points per company across a double-materiality framework, weighing both a company's impact on the world and ESG factors' impact on its financial performance, an approach that produces a fundamentally different score than a provider working from a narrower, risk-only data set.
Weighting compounds the problem. Independent analysis of provider methodologies shows that some raters apply roughly equal weight across environmental, social, and governance pillars regardless of sector, while others dynamically reweight the three pillars by industry materiality, so an oil and gas company's governance failings might barely move one agency's score while dominating another's. A company can therefore improve its environmental disclosures significantly and still watch its rating fall at a provider that puts more weight on governance or controversies during the same period.
Governments are no longer treating this divergence as a private-sector quirk. Under EU Regulation 2024/3005, which enters into force on 2 July 2026, the European Securities and Markets Authority becomes the sole EU supervisor of ESG rating providers, requiring them to be authorized, to disclose their methodologies, models, and key assumptions publicly, and to submit to ESMA's ongoing powers of investigation and sanction. The regulation notably does not force agencies onto a single scoring methodology; it targets transparency and governance instead, which means divergence itself is expected to persist even after the rule takes full effect, just with clearer documentation behind it.
Capital allocated under some sustainable investing label continues to grow despite the confusion. The Global Sustainable Investment Alliance's most recent biennial review put fund assets using responsible or sustainable investment approaches at USD 16.7 trillion, an increase of nearly USD 5.5 trillion, or 49%, over the prior two years using comparable data. Yet trust in the underlying scores has not kept pace with that growth: PwC's Global Investor Survey found that only 33% of institutional investors rated the quality of ESG reporting they see as good, with under half saying they trust the information behind ESG ratings and scores, and 74% saying a single globally consistent set of ESG metrics would help them make better-informed decisions. The consequences of that gap have played out publicly. When Tesla was dropped from a major ESG index over governance and labor concerns, Reuters reported that MSCI simultaneously rated the company as an average performer while Sustainalytics classified it as medium risk, the same company earning three different verdicts from three respected raters at once. Until scope, measurement, and weighting converge, or at minimum become fully transparent under rules like the EU's new regime, companies and investors should expect ESG scores to keep telling different stories about the exact same set of facts.