The average correlation between the ESG scores assigned to the same company by different agencies is 0.54, ranging from 0.38 to 0.71 depending on the pairs of agencies considered. The correlation between the credit ratings assigned to the same bond by Moody’s and Standard & Poor’s is 99%. These two figures come from the same study: “Aggregate Confusion: The Divergence of ESG Ratings, ” published in the Review of Finance in 2022 by Florian Berg, Julian Kölbel, and Roberto Rigobon of MIT, who analyzed the methodologies of six major agencies. The authors explicitly cite the correlation in credit ratings to illustrate the extent of the divergence in ESG ratings. An investor comparing two credit ratings for the same bond obtains essentially identical information. An investor comparing two ESG scores for the same company obtains information that can diverge significantly.
“ESG” stands for Environmental, Social, and Governance. Three dimensions of corporate performance: environmental impact (emissions, resource consumption, waste management), social practices (working conditions, rights in the supply chain, diversity), and corporate governance structure (board independence, transparency, risk management). These three dimensions are aggregated into a single score. A company with excellent governance, good working conditions, and mediocre environmental performance may receive a higher overall score than a company with excellent environmental performance and weak governance. The three letters do not carry the same weight, and the weighting varies from agency to agency.
The aggregation mechanism is the crux of the matter. A high ESG score does not indicate which of the three dimensions is high. The analysis by Berg and colleagues identified five sources of divergence: the scope of measurement, the indicator used for each factor, the weights assigned to each dimension, company coverage, and analysts’ subjective assessments. The metric is the most significant source, accounting for 56% of the total divergence; the scope accounts for 38%; and the weights account for only 6%. Two agencies often measure different things even when they refer to the same factor.
The analogy is to a medical report that summarizes the results of a complete blood count, blood glucose, cholesterol, and transaminases in a single number. That number could be 72 for a patient with an excellent complete blood count and high transaminases, and 72 for a patient with the opposite profile. A doctor cannot use this number to decide on a treatment plan without knowing how it is calculated. An ESG fund with a medium-to-high score does not indicate whether the companies in the portfolio have reduced emissions, improved working conditions, or simply elected a more independent board of directors.
The European regulatory framework has begun to address this issue. The Corporate Sustainability Reporting Directive (CSRD), which will take effect for large European companies starting in fiscal year 2024, does not produce an aggregate score: it requires companies to report in accordance with the European Sustainability Reporting Standards (ESRS), which break down environmental, social, and governance aspects into specific indicators, with disclosure requirements regarding “double materiality.” In Italy, large publicly traded companies and those with more than 500 employees fall within the CSRD’s scope starting with the 2024 fiscal year. In 2025,ESMA began supervising ESG rating agencies, requiring transparency regarding their methodologies. This marks the first European attempt to regulate a market that had operated without public oversight since its inception.
The authors of the study themselves caution that this divergence does not render ESG measurement meaningless. The problem is not the concept itself, but the transition from raw data to an aggregate score, which introduces non-transparent methodological choices and makes the number difficult to interpret without knowing how it is calculated. The CSRD and ESMA’s oversight are moving in this direction: making the chain from data to score transparent, not abandoning the effort to measure sustainability.
This column explores climate-related terms whose technical meaning differs from common perception. “ESG” introduces a new mechanism: not a single ambiguous term, but a measurement system that aggregates incomparable dimensions into a single number, creating the appearance of comparability where structural divergence actually exists. The article on “science-based” had explored the difference between verified commitments and self-assigned labels. “ESG” adds another layer: even when the score is produced by an external body, the number may hide more than it reveals.
The next time you read “ESG fund” or “responsible investment,” there’s only one question to ask: Which of the three letters is driving the score? If the answer isn’t available, the number is less transparent than it seems.
