The puzzle
Three rating agencies looked at the same company in March 2025. MSCI downgraded it. Sustainalytics raised its risk classification. ISS ESG withdrew its climate-leadership designation. Each move was directionally negative but methodologically distinct, and none of the three agreed on which dimension of the company's strategy had deteriorated. The company was BP. The same pattern, though less starkly, has appeared at Shell, TotalEnergies, Glencore, and — in the financial sector — at HSBC and Standard Chartered.
The standard reading among boards has been that ratings divergence is noise. The aggregators disagree because their methodologies are immature; in time the methodologies will converge; meanwhile the noise can be reported around. That reading is wrong. The divergence is not noise. It is signal — and the agencies are picking up different facets of the same decision precisely because their methodologies are different. A chair who reads only the consensus across raters loses the most analytically valuable information the ratings produce.
This briefing sets out what the three major raters are actually measuring, why their answers diverge, and how a board should read the divergence as part of its own diagnostic.
The methodology gap, plainly stated
MSCI ESG Ratings. MSCI's methodology is industry-relative and financial-materiality-led. The rater begins by identifying, for each industry, the ESG issues most likely to affect medium-term financial performance — a process MSCI calls Key Issues identification. For an integrated energy company, the Key Issues weighting will tilt heavily towards carbon emissions, transition risk, and biodiversity exposure, with social and governance factors filling out the remainder. The company is then scored on each Key Issue against industry peers, weighted, and assigned a letter rating from CCC to AAA. The rating expresses the company's exposure-and-management profile relative to its industry peers. A downgrade from AA to A on the BP February 2025 reset reflects MSCI's reading that BP has moved closer to the industry median on transition opportunity — not that BP has become absolutely worse, but that BP's strategic distinctiveness on transition has narrowed.
Sustainalytics ESG Risk Ratings. Sustainalytics' methodology is an absolute-scale unmanaged-risk measurement. The rater begins by identifying, for each company, the material ESG risks the company is exposed to (Exposure) and then assesses how well the company manages those risks (Management). Unmanaged Risk is the residual — the portion of the exposure the company has not addressed through its policies, programmes, and disclosed performance. The company is then placed into a Risk Category from Negligible to Severe. The Sustainalytics methodology is not industry-relative in the same sense as MSCI's; a steel company with a very high absolute carbon footprint can still receive a Low Risk rating if its management practices fully address the exposure, and a company with a comparatively small footprint can receive a High Risk rating if its management is judged to be lagging. A move from Medium to High Risk on the BP February 2025 reset reflects Sustainalytics' reading that BP has expanded its present-day exposure (more upstream production) without expanding its management response — increasing the unmanaged residual.
ISS ESG. ISS ESG operates a Corporate Rating that is closer in form to MSCI's but with a more policy-aligned weighting — the rater is owned by Deutsche Börse, and its methodology incorporates explicit reference to international policy frameworks including the UN Sustainable Development Goals, the Paris Agreement, and the OECD Guidelines for Multinational Enterprises. ISS ESG also operates a Climate Awareness Scorecard and a separate Climate Leadership classification used by certain mandated funds. The withdrawal of BP's climate-leadership classification is procedurally distinct from a rating downgrade — it is a categorical exclusion from a defined "leadership" cohort, applied when a company's commitments fall below a stated threshold. The withdrawal does not, on its own, change the underlying Corporate Rating.
The three methodologies, in summary:
- MSCI — industry-relative, financial-materiality-led — answers "how does this company compare to its peers on the issues most likely to affect medium-term financial performance?"
- Sustainalytics — absolute-scale, unmanaged-risk-led — answers "how much ESG risk does this company carry that it has not addressed?"
- ISS ESG — policy-framework-aligned — answers "how does this company score against international policy frameworks, and does it qualify for defined leadership cohorts?"
These are not three answers to the same question. They are three answers to three different questions. The divergence is what the methodologies are designed to produce.
Why divergence persists
A reasonable chair, reading the above, will ask why the rating agencies have not converged on a single methodology. Three reasons.
One — the questions are genuinely different. A pension fund using ESG ratings to construct a transition-aligned index portfolio needs the MSCI question — relative performance among the available investible universe. A reinsurance underwriter using ESG ratings to assess catastrophe exposure needs the Sustainalytics question — absolute risk that has not been managed. A sovereign wealth fund constrained by an exclusion list tied to the Paris Agreement needs the ISS ESG question — policy-framework alignment. Methodological convergence would force one of these users to use a rating designed to answer a question they are not asking.
Two — the regulatory framework has not yet forced disclosure of methodology. Credit-rating agencies are subject to detailed methodology-disclosure requirements under the EU's CRA Regulation and the SEC's Regulation NRSRO. ESG-rating agencies are not yet subject to comparable requirements. The FCA has been consulting on ESG-rating-provider oversight since 2023; the EU adopted a Regulation on ESG-rating activities in 2024 with transition provisions running into 2026; the SEC has stayed its proposed climate-disclosure rule pending litigation. The regulatory framework that would force methodological transparency, and might thereby force convergence, is not yet in place.
Three — divergence is commercially functional for the raters. MSCI, Sustainalytics, and ISS ESG sell different products to different buyers. A regime in which all three converge on the same methodology would commodify the rating. The current divergence preserves product differentiation.
The BP case as canonical 2025 example
The clearest worked example of the divergence pattern is the BP February 2025 reset. Auchincloss withdrew the forty-percent oil-cut target, raised oil-and-gas capex, and reduced renewables spend. Within four weeks, the three major raters moved as follows.
MSCI — downgraded BP from AA to A. The rater's published rationale referenced reduced industry-relative leadership on transition, narrower distinctiveness from peers on Key Issues including Carbon Emissions and Opportunities in Clean Tech.
Sustainalytics — moved BP's ESG Risk Rating from Medium Risk to High Risk. The rater's published rationale referenced expanded present-day exposure (higher upstream production) and a management response judged not to have expanded proportionately.
ISS ESG — withdrew BP's Climate Leadership classification while making no immediate change to the underlying Corporate Rating. The rationale referenced fall below the threshold of stated emissions-reduction commitment associated with the leadership cohort.
A chair reading the three moves as a single negative signal would conclude that BP had become "worse" on ESG and respond accordingly. A chair reading the three moves as three separate diagnostic outputs would conclude something more useful: that BP had moved closer to its industry peers (MSCI), that BP had expanded its unmanaged-risk surface (Sustainalytics), and that BP had stepped out of a defined leadership cohort without immediately changing its underlying corporate score (ISS ESG). Each finding is true. Each finding is informative. None of the three is the same as either of the others.
The institutional consequence of the three-way pattern was material. LGIM's internal climate policy uses the MSCI rating as one input into its engagement and divestment decisions; the downgrade triggered a formal engagement process. Norges Bank's ethics council uses its own absolute-risk criteria, which align more closely with the Sustainalytics methodology; the council opened a preliminary review. Mandates linked to the ISS ESG climate-leadership classification — primarily certain European pension funds and a small number of Article 9 funds under the EU's Sustainable Finance Disclosure Regulation — were forced to rebalance.
The ratings did not make any decision. They made decisions legible to the actors whose decisions mattered. Each rater triggered a different set of consequences in a different set of mandates.
How a board should read the divergence
The instinct of most boards confronted with rating divergence is to pick the most favourable rater and treat the others as outliers. The instinct is wrong. Three principles.
Principle 1 — read all three. Do not pick. The three raters are answering three different questions. Picking the most favourable rater is equivalent to deciding, in advance, which question is the one your stakeholders will use. They will not all use the same one. The pension funds and asset managers who hold your stock will, in aggregate, consult all three.
Principle 2 — read divergence as a diagnostic. When MSCI downgrades and Sustainalytics raises risk, the methodology-driven distinction is doing analytical work. The MSCI move tells the board something about its industry-relative position; the Sustainalytics move tells it something about its absolute exposure. If the two moves agree directionally — both negative, both positive — the case is over-determined. If they disagree directionally, the case is split, and the split is the diagnostic. A chair should be able to articulate the disagreement.
Principle 3 — track the trajectory, not the level. A single rating snapshot is less informative than the trajectory across two or three reporting cycles. MSCI revises Key Issues weightings periodically. Sustainalytics revises its Exposure scoring as new sectoral data arrives. ISS ESG adjusts its climate-leadership thresholds in response to policy framework changes. A board reading the trajectory across cycles will see methodology shifts that a single-snapshot reading misses.
Practical step. Most large-cap companies have an investor-relations function that already tracks rating moves. Few have a board-level reading discipline that distinguishes the methodological reasons for each rater's reading. The shortest path from the present state to the diagnostic state is a one-page board memo, refreshed quarterly, that reports the level and the trajectory of each of the three ratings, with a one-line plain-English explanation of any change.
Three implications for chairs
Implication 1 — divergence is the signal. A chair who reads only the consensus across raters loses the most analytically valuable information the ratings produce. The diagnostic value sits in the disagreement, not in the average.
Implication 2 — the rating drives mandate flows the company cannot see directly. Engagement processes, divestment lists, leadership cohorts, exclusion criteria — each is triggered by a different rater, in a different mandate, in a different jurisdiction. The board that does not read all three is a board that cannot anticipate which mandate flows are about to move.
Implication 3 — methodology will be regulated, not standardised. The FCA, EU, and SEC are converging on disclosure-of-methodology requirements rather than standardisation. The regulatory direction of travel makes divergence more legible, not less divergent. A board's reading discipline will, by 2027, need to be more sophisticated than a board's reading discipline in 2024 — not less.