ESG Ratings: How MSCI, Sustainalytics and ISS Build Scores and Why They Disagree
Follow a rating from raw data to a letter grade, see how industry-relative grading works, and understand why the big agencies rate the same company differently.
What you will learn
- The building blocks of an ESG rating: key issues, weights, exposure and management scores.
- How MSCI, Sustainalytics and ISS ESG differ in scale, logic and meaning.
- Why industry-relative grading lets an oil company be rated "AA".
- The evidence on rating disagreement and how to use ratings despite it.
Explanation
An ESG rating is an opinion produced by a data company, not a credit rating and not a measure of ethics. Three providers dominate European funds.
MSCI ESG Ratings
MSCI assigns each company to an industry and selects the key issues that are financially material there ("carbon emissions" for utilities, "privacy and data security" for software). Each key issue gets an exposure score (how much risk the company faces) and a management score (how well it handles it). Issues are weighted, combined into a 0–10 score and normalised against industry peers, giving a letter from AAA (leader) through AA, A, BBB, BB, B to CCC (laggard). The rating is industry-relative: an "AA" oil producer manages its risks better than most oil producers; it is not a low-carbon company.
Sustainalytics ESG Risk Rating
Sustainalytics (owned by Morningstar) measures unmanaged ESG risk on an absolute scale. It starts with the company's exposure to material issues, subtracts the part that is managed, and reports the remainder as a number. Lower is better. The number falls into five categories: negligible (below 10), low (10–20), medium (20–30), high (30–40) and severe (40 and above). Because the scale is absolute, a tobacco company can never reach "negligible", however well it is run.
ISS ESG Corporate Rating
ISS ESG grades companies from A+ to D- on industry-specific indicators and awards "Prime" status above an industry-specific threshold. It leans more towards double materiality and runs separate norms-based screens (lesson 3).
| Provider | Scale | Logic | Best possible for a coal miner? |
|---|---|---|---|
| MSCI | AAA–CCC | Industry-relative, financial materiality | AAA is possible in theory |
| Sustainalytics | 0–100+ risk points, lower is better | Absolute unmanaged risk | Stays "high" or "severe" |
| ISS ESG | A+ to D-, "Prime" threshold | Industry-specific, more double materiality | Prime unlikely |
Why the agencies disagree
A study published in the Review of Finance in 2022 (Berg, Kölbel and Rigobon, "Aggregate Confusion") compared six rating providers and found an average pairwise correlation of only about 0.54, ranging from 0.38 to 0.71. The authors found that the main source of divergence was measurement (different indicators for the same issue), followed closely by scope (different issues included), while weights played a minor role. Practical consequence: two funds with "top-rated" holdings can hold very different companies.
How to use ratings anyway
- Read a rating as "risk to the company, judged by one provider", not as a moral grade.
- Compare fund-level average ratings only when both funds use the same provider.
- Check sub-scores: a strong total can hide a weak environmental pillar.
Worked example
Hypothetical example: a fictional cement producer, Kalkstein AG, is rated by two invented agencies that copy the MSCI-style and Sustainalytics-style logic. All numbers are made up.
Agency 1 (industry-relative, 0–10 per issue). Four key issues with weights and scores:
| Key issue | Weight | Score (0–10) | Weight × score |
|---|---|---|---|
| Carbon emissions | 30% | 4.0 | 1.20 |
| Water stress | 20% | 7.0 | 1.40 |
| Labour management | 25% | 6.0 | 1.50 |
| Corporate governance | 25% | 5.0 | 1.25 |
Step 1 – weighted score: 1.20 + 1.40 + 1.50 + 1.25 = 5.35.
Step 2 – industry normalisation. If the worst cement company scores 2.5 and the best 8.5: position = (5.35 − 2.5) ÷ (8.5 − 2.5) = 2.85 ÷ 6.0 = 0.475, slightly below the industry middle, roughly "BBB" on a seven-letter scale: average among cement producers.
Agency 2 (absolute unmanaged risk). Exposure score 60 points (cement is carbon-intensive). Management reduces 24 points. Unmanaged risk = 60 − 24 = 36, which is in the "high" category (30–40). The managed share is 24 ÷ 60 = 40%.
Result: Agency 1 says "average", Agency 2 says "high risk". Both are consistent; they measure different things. A fund selecting "BBB or better" could hold Kalkstein; a fund excluding "high risk" could not.
Common mistake
Practice activity
On paper: a fictional software company has three key issues: data privacy (weight 50%, score 8.0), human capital (weight 30%, score 5.0) and governance (weight 20%, score 6.0). The best software company scores 9.0 and the worst 3.0. Calculate the weighted score and the industry position (0 to 1). Then explain in one sentence why the same company might still get a "medium" absolute risk rating from a Sustainalytics-style provider.
Show solution
Weighted score: 0.5 × 8.0 + 0.3 × 5.0 + 0.2 × 6.0 = 4.0 + 1.5 + 1.2 = 6.7.
Industry position: (6.7 − 3.0) ÷ (9.0 − 3.0) = 3.7 ÷ 6.0 = 0.617, i.e. above the industry middle (roughly "A" territory on a seven-step scale).
Absolute view: even good management leaves some unmanaged privacy and human-capital risk, so an absolute score of, say, 18–22 points ("low" to "medium") is plausible while the relative rating is above average.
Limitations: real providers use many more indicators, controversy adjustments and analyst overrides, so this is only the skeleton of a rating.
Key takeaways
- A rating = material key issues × weights, then (for MSCI) industry-relative normalisation.
- MSCI is relative (AAA–CCC); Sustainalytics is absolute risk (lower is better); ISS uses A+ to D- and a "Prime" threshold.
- Agencies correlate only about 0.54 on average, mostly because they measure and scope issues differently.
- Use ratings as one provider's risk opinion, check sub-scores, and never mix scales.
Next step
Next you will apply explicit exclusion rules: controversial weapons, thermal-coal revenue thresholds and UN Global Compact violations, with the numbers that decide who stays in an index.
Self-check
Answer the questions, then check your answers. You can retry as often as you like.
Automatic self-check for your own revision. Your answers are checked in your browser and are not seen or assessed by us. It is not an exam and not a qualification.
Your progress
“Complete” only means you marked the lesson as done. It is not a test of your understanding. Progress and notes are saved only in this browser on this device — not in an account.