ESG Ratings Agencies (MSCI, Sustainalytics, CDP) vs ESG Reports: Why a Company Can Score High and Report Badly (or Vice Versa)

Aug 29

8 min read

Tesla is, depending on who you ask, an ESG leader or an ESG risk. MSCI has rated it favorably, largely on the strength of its vehicles' contribution to cutting transport emissions. Sustainalytics and S&P Global have taken a noticeably harder line, flagging workplace discrimination claims, governance concerns, and public disclosure practices. Same company, same year, same publicly available information, two structurally opposite conclusions. Elon Musk has publicly called ESG ratings "a scam." He's wrong that they're meaningless, but he's pointing at something real: the scores genuinely do disagree with each other, constantly, and understanding why is more useful than picking a side.

This isn't a data quality problem that will resolve itself. A landmark 2022 study by MIT Sloan researchers Florian Berg, Julian Kölbel, and Roberto Rigobon, published in the Review of Finance, found that the average correlation between six major ESG rating providers sits at roughly 0.54. Compare that to credit ratings, where Moody's and S&P agree with a correlation above 0.99, and the scale of the disagreement becomes obvious. This article explains why that gap exists, using three specific providers, MSCI, Sustainalytics, and CDP, that show up on almost every sustainability report's back page, and three real, current company examples that show the pattern in practice.

Why the Same Company Gets Different Grades: Three Different Questions

The simplest way to understand ESG rating disagreement is to notice that the major providers aren't actually answering the same question. MSCI asks whether a company is managing its financially material ESG risks better than its industry peers, an industry-relative question. Sustainalytics asks how much of a company's economic value is at risk from ESG factors it hasn't adequately managed, an absolute question, not measured against peers at all. CDP asks something narrower still: how complete and credible is a company's self-reported environmental disclosure. A Sustainalytics "Low Risk" rating and an MSCI "AA" rating aren't two answers to the same question. They're answers to two different questions that happen to get compressed into a single number and placed side by side on an investment platform.

Key takeaway: before comparing any two ESG scores, check what each one is actually designed to measure. A company being "better than its industry peers" and a company having "low absolute unmanaged risk" are genuinely different claims, and a company can be true on one and false on the other at the same time.

MSCI: Relative Performance, Not Absolute Performance

MSCI ESG Ratings score companies on a seven-tier letter scale from AAA down to CCC, and the methodology is explicitly industry-relative: MSCI assesses how well a company manages the specific ESG risks and opportunities its own industry has been determined to face financially, then ranks it against other companies in that same industry, not against companies in general. That single design choice explains a lot of counterintuitive results. An oil and gas company can receive a respectable MSCI rating not because its absolute environmental footprint is small, it usually isn't, but because MSCI is asking whether it manages the specific, financially material risks of being an oil and gas company better than other oil and gas companies do.

Key takeaway: an MSCI rating tells you how a company stacks up inside its own industry, not how sustainable the company is in any absolute sense. A high MSCI rating in a carbon-intensive industry is not the same claim as a high MSCI rating in a low-impact industry, even though both might show up as the same letter grade.

Sustainalytics: Absolute Risk, Not Peer Comparison, and Not Just What Companies Disclose

Sustainalytics, now operating as Morningstar Sustainalytics, deliberately does not attempt to score a company's overall sustainability performance or ethical standing. Its ESG Risk Rating measures one specific thing: the amount of a company's enterprise value that's at risk from unmanaged environmental, social, and governance factors, scored on a numeric scale where, notably, a lower number is better, the reverse of what most people intuitively expect from a "rating." Because it's a risk measure rather than a peer-relative score, a Sustainalytics rating doesn't rank a company against its industry the way MSCI does, and critically, Sustainalytics weighs analyst research and controversy monitoring heavily enough that its score doesn't simply reproduce whatever a company chooses to publish. A material issue a company stays silent on isn't treated as neutral; Sustainalytics can treat that silence itself as a signal of weak management.

Key takeaway: a Sustainalytics score is checking something a company's own report often can't show you: what independent analysts and news-driven controversy tracking reveal that the company's own disclosure doesn't mention. This is a large part of why a beautifully produced sustainability report and a poor Sustainalytics score can coexist without contradiction.

CDP: Grading the Disclosure Itself, Not the Company's Overall Sustainability

CDP is frequently mentioned in the same breath as MSCI and Sustainalytics, but it's doing something narrower and structurally different from both. CDP scores companies A through D-, based entirely on their responses to CDP's own questionnaires covering climate change, water security, and forests, and nothing else, meaning CDP has no view at all on a company's labor practices, board diversity, or human rights record. It's also, importantly, a self-reported system: the underlying data comes from what the company itself submits in response to CDP's questions, not from independent monitoring, though CDP does apply strict scoring rules, including a set of Essential Criteria a company must clear regardless of its raw point total to reach the top grades, and 2025 saw the CDP A List grow past 800 companies, more than double its size just a few years earlier.

Key takeaway: a strong CDP score is a real, useful signal about the quality and completeness of a company's climate disclosure specifically. It is not a general ESG endorsement, and it is only as reliable as what the company chose to submit, since CDP's core methodology is built on self-reported answers rather than independent verification.

The Academic Case: Why the Disagreement Isn't Going Away

Berg, Kölbel, and Rigobon didn't just measure the disagreement between rating agencies, they broke down what was causing it. Their analysis attributed roughly 56 percent of the divergence to measurement differences, meaning agencies use genuinely different indicators to assess the same underlying issue, for example one agency scoring "labor practices" using reported turnover data while another uses legal case history. Scope differences, meaning agencies simply choosing to measure different sets of issues in the first place, accounted for roughly 38 percent. Weighting differences, how much importance an agency assigns to a given issue once measured, made up only around 6 percent, the smallest factor, which is itself a useful finding: most of the disagreement isn't about priorities, it's about which yardstick gets used and which topics are even on the yardstick at all. A 2025 OECD review found this hasn't meaningfully improved since the original 2022 study, with pairwise correlations between major raters still ranging from roughly 0.38 to 0.71 depending on which ESG pillar is being compared.

Key takeaway: ESG rating disagreement isn't a temporary data problem waiting to be fixed. It's a structural consequence of different organizations measuring different things in different ways, and the gap has persisted for years without narrowing.

Three Real Companies, Three Real Divergences

Tesla, in the automotive and clean energy space, is the clearest illustration of MSCI's relative, product-focused lens colliding with Sustainalytics' absolute, controversy-weighted one. MSCI's rating leans heavily on Tesla's contribution to reducing transportation emissions through its vehicles and energy products. Sustainalytics and S&P Global have separately flagged reported workplace discrimination issues and governance and disclosure concerns, pulling their assessments in the opposite direction. Tesla's own sustainability reporting emphasizes the emissions-avoidance side of its story, which is real, but it's only one half of what the two rating philosophies are each independently measuring.

Amazon, in e-commerce and logistics, shows the same pattern between a different pair of agencies. In 2025, S&P Global placed Amazon in the top quartile of its industry for environmental management, while ISS ESG ranked it in the bottom third of the same industry, largely over differing assessments of packaging waste, warehouse worker conditions, and Scope 3 logistics emissions, the value chain emissions category covered in depth elsewhere in this series. Both agencies had access to largely the same underlying disclosures. They weighted the components of "environmental management" differently enough to land in opposite quartiles.

TotalEnergies, in oil and gas, shows how a company's own narrative and an external rating can pull apart around a single strategic question: how much credit should a fossil fuel company get for its renewable energy transition, relative to how much penalty it should carry for continued fossil fuel production. Refinitiv scored the company in the 70th percentile of its industry, weighted more toward its renewable transition strategy. MSCI scored the same company in the 25th percentile, weighting the continued scale of its fossil fuel production more heavily than its transition commitments. The company's own sustainability reporting foregrounds the transition story; the rating gap shows how differently that story can be priced by two different methodologies looking at the same facts.

Key takeaway: in all three cases, the company's own report isn't lying, and neither rating agency is wrong. Each is applying a genuinely different, internally consistent methodology to the same underlying set of facts, and the resulting spread is the honest output of that difference, not evidence that one number is correct and the others are mistakes.

When Self-Reported Diverges From Independently Checked

CDP's self-reported model creates a specific kind of gap worth watching for directly. In June 2025, Carbon Market Watch and the NewClimate Institute published an independent review of major technology companies' climate strategies and rated Meta, Microsoft, and Amazon as "poor," with Alphabet separately scored "poor" specifically on its emissions-reduction targets, even as several of these same companies maintain strong standing on self-reported climate disclosure platforms and have publicized ambitious net-zero commitments. The independent reviewers pointed specifically to the surge in energy demand from AI infrastructure growth as a factor making existing pledges harder to credibly achieve, a piece of context a company's own forward-looking climate targets don't always foreground.

Key takeaway: the widest gaps tend to appear exactly where a scoring system relies most heavily on what a company chooses to submit itself. An independent assessment built from outside data, rather than a company's own questionnaire responses, is answering a meaningfully different, and sometimes less flattering, question.

Regulators Are Now Watching the Raters, Not Just the Companies

This disagreement problem has become significant enough that the EU has started regulating the rating agencies themselves, not just the companies they rate. Regulation (EU) 2024/3005, the EU's ESG Ratings Regulation, entered into force on 2 January 2025 and applies from 2 July 2026. It requires ESG rating providers operating in the EU to be authorized, to disclose their methodologies, and to manage conflicts of interest, including separating ESG rating activities from other business lines like credit ratings or consulting. It's a direct regulatory acknowledgment that the opacity behind how these scores are built has itself become a market risk, not just an academic curiosity.

Key takeaway: the fact that regulators are now writing rules for the rating agencies, rather than only for the companies being rated, is itself strong evidence that the disagreement problem discussed in this article is recognized as systemic, not anecdotal.

How to Actually Use These Scores

Never compare scores across providers directly. A Sustainalytics number and an MSCI letter grade cannot be meaningfully placed on the same scale, because they're not measuring the same construct.
Check whether the score is relative or absolute. MSCI tells you about industry-relative standing. Sustainalytics tells you about absolute unmanaged risk. Knowing which one you're looking at changes what a "good" score even means.
Remember CDP is climate, water, and forests only. A strong CDP score says nothing about labor practices, board composition, or human rights, and it's built on what the company itself chose to report.
Treat agreement as the strongest signal. When multiple independent providers land on the same conclusion about a company despite using different methodologies, that convergence is more informative than any single score on its own.
Read the company's own report for mechanism, and the rating for challenge. A company's report explains what it's doing and why. A divergent rating is often flagging exactly the part of that story an outside analyst isn't fully convinced by.

References

Berg, F., Kölbel, J. F., & Rigobon, R., "Aggregate Confusion: The Divergence of ESG Ratings," Review of Finance, 2022
MSCI, MSCI ESG Ratings Methodology, msci.com
Morningstar Sustainalytics, ESG Risk Ratings Methodology, sustainalytics.com
CDP, Climate Change Scoring Methodology and A List Criteria, cdp.net
OECD, ESG Ratings and Investment Strategies / 2025 update on rating divergence, oecd.org
Regulation (EU) 2024/3005 of the European Parliament and of the Council on the transparency and integrity of ESG rating activities, in force 2 January 2025, applying from 2 July 2026
Carbon Market Watch and NewClimate Institute, Corporate Climate Responsibility Monitor, June 2025
S&P Global, ESG Scores and Ratings, 2025 sector benchmarking, spglobal.com