Data story: tracking the divergence in ESG ratings across providers and what it means for investors

Why It Matters

When MSCI rates Tesla as an ESG leader while Sustainalytics flags it as high risk, something structural is happening inside the ratings ecosystem. The average pairwise correlation between major ESG rating providers stands at just 0.53, according to a landmark MIT Sloan study by Berg, Kölbel, and Rigobon (2022), updated in 2025 to show only marginal improvement to 0.56 (Berg et al., 2025). Compare that to credit ratings, where the correlation between Moody's and S&P exceeds 0.99, and the scale of the problem becomes clear. For the more than $40 trillion in assets now managed under some form of ESG integration globally (Global Sustainable Investment Alliance, 2025), this divergence is not an academic curiosity. It shapes portfolio construction, drives capital allocation, and determines which companies face the cost of capital penalty that ESG underperformance is supposed to deliver. Understanding where, why, and how much ratings disagree is essential for any investor who relies on third-party ESG scores.

Key Concepts

ESG rating divergence refers to the systematic disagreement between rating providers when scoring the same company on environmental, social, and governance criteria. Unlike credit ratings, which converge around standardised financial metrics, ESG ratings rely on heterogeneous methodologies, weightings, and data sources that produce materially different outcomes.

Measurement divergence arises when providers use different indicators to assess the same category. One provider may measure carbon intensity using Scope 1 and 2 emissions while another includes Scope 3 estimates, producing fundamentally different scores for the same company.

Scope divergence occurs when providers define the boundaries of ESG differently. MSCI evaluates 35 key issues across 10 themes, while Sustainalytics assesses over 70 indicators grouped into material ESG issues and idiosyncratic risks. The number and type of topics assessed vary substantially.

Weight divergence reflects differing judgments about the relative importance of E, S, and G pillars. Refinitiv (now LSEG Data & Analytics) applies roughly equal weights across pillars, while MSCI dynamically weights issues by industry materiality. A company with strong environmental practices but weak governance can score highly with one provider and poorly with another.

Rater effect describes the tendency of individual providers to systematically rate certain types of companies higher or lower, independent of the underlying data. Research by Christensen, Serafeim, and Sikochi (2022) showed that the rater effect accounts for roughly 30% of total rating variance.

The Data

Overall correlation patterns

The most comprehensive dataset on ESG rating agreement comes from the updated Berg, Kölbel, and Rigobon analysis covering six major providers: MSCI, Sustainalytics, Refinitiv, S&P Global, ISS ESG, and Bloomberg. Pairwise correlations across these providers for the same universe of approximately 3,500 large-cap companies range from a low of 0.38 (ISS ESG vs. Bloomberg) to a high of 0.71 (S&P Global vs. Refinitiv), with the overall average at 0.56 as of mid-2025 (Berg et al., 2025).

To contextualise this: a correlation of 0.56 means that roughly 31% of the variance in one provider's ratings is explained by another's. For nearly 70% of the signal, the two providers are telling investors different stories.

Divergence by ESG pillar

The social pillar consistently shows the greatest disagreement. Across the six-provider sample, the average pairwise correlation for social scores is 0.42, compared with 0.53 for environmental scores and 0.61 for governance scores (Aggregate Confusion Project, MIT Sloan, 2025). This pattern reflects the inherent difficulty of measuring social outcomes: metrics like "employee satisfaction," "community impact," and "human rights due diligence" are far less standardised than greenhouse gas emissions or board independence ratios.

Environmental score convergence has improved modestly since 2022, driven by the adoption of common disclosure frameworks such as TCFD and the ISSB's IFRS S2 standard. Governance convergence benefits from decades of corporate governance codes and proxy advisory research that have established relatively stable measurement conventions.

Divergence by sector

Rating disagreement is not uniform across industries. A 2025 analysis by Morningstar Sustainalytics found that technology companies exhibit the highest cross-provider divergence, with an average standard deviation of 18.2 percentile ranks across six raters for the same company. Oil and gas companies, despite being subject to intense ESG scrutiny, show lower divergence (standard deviation of 12.4 percentile ranks) because environmental metrics like Scope 1 emissions are more easily measured and compared (Morningstar Sustainalytics, 2025).

Healthcare and financial services occupy the middle ground, with standard deviations of 15.1 and 14.3 percentile ranks respectively. The pattern suggests that sectors with more qualitative, forward-looking ESG issues (data privacy, AI ethics, digital inclusion) see greater rater disagreement than sectors where emissions and safety metrics dominate.

Divergence by region

European companies receive more consistent ESG ratings than their North American or Asian counterparts. The average cross-provider correlation for EU-listed companies is 0.62, compared with 0.51 for US-listed companies and 0.44 for companies listed in emerging markets (FTSE Russell, 2025). Mandatory sustainability reporting under the EU's Corporate Sustainability Reporting Directive (CSRD), which began phased implementation in 2024, provides a common data backbone that reduces measurement divergence.

In emerging markets, limited disclosure requirements, language barriers, and reliance on estimated rather than reported data amplify disagreement. CDP reported that only 38% of large companies in Asia-Pacific disclosed climate data in line with TCFD recommendations in 2025, compared with 72% in Europe (CDP, 2025).

Real-world divergence examples

Tesla remains the most-cited illustration. As of January 2026, MSCI rated Tesla "A" (leader in its industry group), while Sustainalytics assigned a "High Risk" score of 32.8 out of 100. The divergence stems from MSCI's heavy weighting of product-level environmental impact (electric vehicles) versus Sustainalytics' emphasis on corporate governance, labour practices, and controversy exposure.

Amazon shows a similar pattern. S&P Global placed Amazon in the top quartile of its industry for environmental management in 2025, while ISS ESG ranked it in the bottom third, primarily due to differing assessments of packaging waste, warehouse worker conditions, and Scope 3 logistics emissions (S&P Global, 2025; ISS ESG, 2025).

TotalEnergies scored in the 70th percentile with Refinitiv for its renewable energy transition strategy but sat in the 25th percentile with MSCI, which penalised ongoing fossil fuel production volumes more heavily than transition commitments. This case highlights how methodological choices about whether to weight current emissions or transition trajectories drive entirely different conclusions.

Investment performance implications

A 2025 study by the CFA Institute found that ESG-tilted portfolios constructed using different single-provider ratings produced annualised return differences of 1.8 to 3.2 percentage points over a five-year backtesting period (CFA Institute, 2025). In other words, the choice of rating provider alone can be as consequential as active stock selection. Portfolios built on composite scores aggregating multiple providers showed lower tracking error and more stable factor exposures.

Research from Dimensional Fund Advisors (2025) further demonstrated that stocks in the top quintile of one provider's ESG ranking had only a 43% probability of appearing in the top quintile of another provider's ranking. This means that more than half of the "best ESG companies" according to one rater are not considered best-in-class by a peer rater.

Key Takeaways

Divergence is structural, not incidental. The 0.56 average correlation across providers has improved only marginally from 0.53 in 2022 despite significant regulatory attention and industry consolidation. Expecting rapid convergence is unrealistic because the disagreement stems from fundamental methodological choices, not data gaps alone.

The social pillar is the weakest link. With an average correlation of just 0.42, social scores carry the highest noise and the lowest signal-to-noise ratio for investors. Until standardised social metrics emerge (potentially through the ISSB's future human capital and human rights standards), investors should treat social pillar scores with particular caution and supplement them with primary data.

Regulation is driving partial convergence. The CSRD in Europe and ISSB adoption globally are creating common disclosure baselines that improve environmental and governance score consistency. European companies already show 0.62 cross-provider correlation versus 0.51 for US peers. As ISSB adoption expands to 23 jurisdictions that have committed to incorporating the standards by 2026, further convergence on the E and G pillars is likely.

Single-provider reliance is a portfolio risk. The 1.8 to 3.2 percentage point return differential from provider choice alone demonstrates that relying on a single ESG rating is an uncompensated risk. Institutional investors managing $1 billion or more in ESG-integrated assets should aggregate scores from at least three providers and apply proprietary materiality overlays.

Technology and oil and gas sit at opposite ends of the divergence spectrum. Investors in tech should be especially sceptical of any single ESG score. Conversely, energy sector ESG ratings, while controversial, at least exhibit greater cross-provider agreement, making them more actionable as portfolio signals.

Action Checklist

• Use multi-provider composites. Subscribe to at least three ESG rating providers and build composite scores weighted by methodology transparency and data freshness.
• Decompose by pillar. Evaluate E, S, and G scores separately rather than relying on aggregated headline ratings, especially for the social pillar where divergence is highest.
• Prioritise disclosed over estimated data. Favour companies reporting under CSRD, ISSB, or CDP frameworks; their scores exhibit 15% to 20% less cross-provider variance than companies relying on estimated data.
• Apply sector-specific scrutiny. For technology and healthcare holdings, conduct proprietary ESG assessments rather than deferring to third-party scores.
• Monitor convergence trends. Track cross-provider correlation annually for your portfolio universe; improving convergence in specific sectors may signal more reliable score-based tilting opportunities.
• Engage rating providers directly. Request methodology documentation and flag material disagreements; several providers now offer bespoke data feeds that allow investors to re-weight indicators to match their own materiality frameworks.
• Integrate controversy screening. Overlay quantitative ESG scores with qualitative controversy monitoring from RepRisk, Truvalue Labs (Factset), or similar platforms to catch risks that static ratings miss.
• Disclose your ESG data sources. When reporting to clients or beneficiaries, specify which providers and methodologies inform your ESG integration process to avoid the appearance of false precision.

FAQ

Why do ESG ratings diverge so much more than credit ratings? Credit ratings assess a narrow, well-defined outcome: the probability of default on a financial obligation. Decades of standardisation, regulatory oversight (SEC recognition of NRSROs), and market feedback loops have driven convergence. ESG ratings attempt to capture a far broader set of outcomes across environmental, social, and governance dimensions, with no universally agreed definition of what "good ESG" means. Different providers make different but defensible choices about scope, measurement, and weighting, producing structural disagreement. As Berg, Kölbel, and Rigobon demonstrated, scope divergence and measurement divergence each contribute roughly 35% to total disagreement, with weight divergence and the rater effect explaining the remainder.

Will regulatory standardisation eliminate divergence? Partially, but not entirely. Mandatory disclosure standards like CSRD and ISSB improve the quality and comparability of underlying data, which reduces measurement divergence. European companies already show higher cross-provider correlation than peers in less regulated markets. However, weight divergence and scope divergence reflect genuine differences in how providers interpret materiality, and these are unlikely to converge unless regulators prescribe not just what companies must disclose but how rating agencies must process that information, which would raise significant concerns about intellectual freedom and innovation.

How should asset owners respond to rating divergence? The most robust approach is multi-provider aggregation combined with proprietary analysis. The CFA Institute recommends that institutional investors treat ESG ratings as one input among several, supplementing them with direct company engagement, sector-level materiality assessments, and controversy monitoring. Pension funds such as ABP in the Netherlands and CalPERS in California have built internal ESG scoring models that blend third-party data with proprietary research, reducing reliance on any single provider's methodology.

Does rating divergence mean ESG investing doesn't work? No, but it means that passive reliance on a single provider's scores is insufficient. Research consistently shows that companies with strong performance on material ESG issues outperform peers over the medium to long term (Khan, Serafeim, and Yoon, 2016; updated Serafeim, 2025). The challenge is correctly identifying which issues are material and measuring them accurately. Divergence does not negate the materiality thesis; it complicates the measurement task and places a premium on investor judgment.

Which ESG rating provider is the most accurate? There is no single "most accurate" provider because accuracy depends on the question being asked and the definition of ESG performance being used. S&P Global and MSCI tend to show the strongest correlation with subsequent stock price performance on governance-related issues, while Sustainalytics and ISS ESG have demonstrated predictive power for controversy events (Serafeim and Yoon, 2025). The optimal choice depends on the investor's specific use case: portfolio screening, engagement targeting, risk management, or regulatory compliance each favour different providers.