ESG ratings disagree: why, and how to read them

Published 13 November 2026 · Responsible finance

A fund factsheet says the portfolio has an "AA" ESG rating. A company's annual report quotes a "low risk" score from another provider. A news story says the same company is a sustainability laggard according to a third. All three can be accurate. ESG ratings are not measurements of a single, agreed quantity, the way a credit rating estimates the probability of default. They are opinions built from different questions, different data and different weights, and they often disagree. Understanding why is the first step to using them sensibly.

How large is the disagreement?

The most cited study on the subject is "Aggregate Confusion: The Divergence of ESG Ratings" by Florian Berg, Julian Kölbel and Roberto Rigobon, published in the Review of Finance in 2022. The authors compared the ratings that six well-known providers gave to the same companies: KLD, Sustainalytics, Moody's ESG (formerly Vigeo-Eiris), S&P Global (RobecoSAM), Refinitiv (Asset4) and MSCI.

The ratings were only loosely related. In a summary of the research published by MIT Sloan, the average correlation between the agencies' ratings is given as 0.61, against 0.99 for the credit ratings of Moody's and Standard & Poor's. Put plainly: if two credit agencies look at a company, they almost always reach the same view of its creditworthiness. If two ESG agencies look at it, a company in the top group for one can sit in the middle, or lower, for the other.

Three sources of divergence

The study's main contribution is to explain where the disagreement comes from. The authors mapped each provider's hundreds of indicators onto a common list of categories, then split the divergence into three parts.

• Scope: the agencies do not assess the same things. One counts lobbying or animal welfare, another does not. Scope accounts for 38% of the divergence in the published paper.
• Measurement: when they do assess the same thing, they measure it differently. For "labour practices", one provider may look at staff turnover, another at the number of labour-related lawsuits, a third at whether a policy exists. Measurement is the biggest factor, at 56%.
• Weight: the agencies give different importance to the same categories when they add them up. Surprisingly, this matters least: 6%.

The authors also found what they call a "rater effect": a provider's overall view of a company tends to colour the way it scores that company in individual categories. A firm seen as good overall tends to be scored well across the board by that provider, which suggests the categories are not assessed in complete isolation.

Risk to the company, or impact on the world?

A further source of confusion lies upstream of the data: the question the rating is trying to answer. Many widely used ratings measure how exposed a company is to environmental, social and governance risks, and how well it manages them, from the point of view of the investor's money. Others try to measure the company's effect on the environment and society. The two can point in opposite directions. A company with large emissions in a sector where regulation is unlikely to bite soon may score as a low financial risk; seen from the climate's side, it is the opposite.

European rules call these two dimensions "double materiality": financial materiality (the effect of sustainability issues on the company) and impact materiality (the effect of the company on people and the planet). A rating that addresses only one of them is not wrong, but it answers a narrower question than many users assume. This is the same distinction we drew between company impact and investor contribution in what impact investing measures.

What the EU regulation changes from July 2026

The European Union has decided to supervise the providers. Regulation (EU) 2024/3005 on the transparency and integrity of ESG rating activities was adopted on 27 November 2024 and has applied since 2 July 2026. Its main features:

• Authorisation by ESMA. Providers that issue ESG ratings in the EU, whether published or sold to regulated financial firms, companies subject to EU reporting rules or public bodies, need to be authorised and are supervised by the European Securities and Markets Authority. Providers established outside the EU need an equivalence decision, endorsement or recognition. Providers already operating had to notify ESMA by 2 August 2026 and apply within four months of 2 July 2026. Small providers can benefit from a lighter registration regime for a period.
• Methodologies on the website. Providers must publish, in a dedicated section of their website, the methodologies, models and key assumptions they use, including whether a rating addresses financial risk, impact, or both.
• Separate E, S and G ratings, rather than a single aggregated score. A provider may still publish a single ESG rating, but it must then disclose the weighting of each of the three pillars and explain the weighting method.
• Separation of activities. Providers may not also offer consulting to investors or companies, credit ratings, audit or sustainability assurance, among other activities. Benchmarks, investment services, banking and insurance are possible only under specific conditions, which address the conflict of a provider grading the companies it also advises.

The regulation deals with transparency and integrity. It does not impose a common methodology, and it is not designed to make the ratings agree. Divergence will remain, but users should be able to see where it comes from.

Five questions to ask about any ESG score

1. Which question does it answer? Financial risk to the company, impact on society and the environment, or both? The provider's methodology must now say.
2. What is in scope? Look at the list of issues covered. A score that ignores the issue you care about, such as deforestation or tax practices, cannot tell you anything about it.
3. Is it relative to the sector? Many ratings rank companies against their industry peers. The "best" oil company may have a top rating without being low-carbon in absolute terms.
4. Where does the data come from? Company reports, questionnaires filled in by the company, news and controversies, or estimates? Gaps are often filled with estimates, which can favour the largest companies that publish the most.
5. What do other providers say? If two or three ratings point the same way, the signal is stronger. If they diverge, the reasons are usually in the methodologies.

Using ratings without being used by them

ESG ratings are useful as a starting point: they gather data no individual investor could collect, and they flag controversies. They are a poor substitute for reading what a fund actually does. A sustainability label on a fund relies partly on these scores, which is why the fund-name rules and a manager's own policy, examined with our 10-point checklist, matter more than any single grade. And when a fund claims to push companies to improve, its proxy-voting record is a better test than the score of its portfolio.

This article is general information, not investment advice. Rules are described as published at the date above; check the current texts before relying on them.

Sources

• Florian Berg, Julian F. Kölbel and Roberto Rigobon, "Aggregate Confusion: The Divergence of ESG Ratings", Review of Finance, vol. 26, no. 6, 2022, pp. 1315-1344, doi:10.1093/rof/rfac033.
• MIT Sloan School of Management, "Why ESG ratings vary so widely (and what you can do about it)", Ideas Made to Matter.
• Regulation (EU) 2024/3005 of the European Parliament and of the Council of 27 November 2024 on the transparency and integrity of Environmental, Social and Governance (ESG) rating activities, EUR-Lex.

Related: what impact investing measures · fund-name rules · proxy-voting records.