Can sustainability ratings agencies put markets at risk?
In « Sustainability Gatekeepers: ESG Ratings and Data Providers », Marco Dell’Erba and Michele Doronzo examine whether sustainable finance has effective gatekeepers akin to credit rating agencies.
They compare how ESG rating firms operate compared to traditional credit raters and propose policy fixes to improve transparency and trust. They show:
- Unlike credit ratings correlating at ~99%, ESG raters included in this study correlate only ~71% on average, which raises doubts and risk for investors.
- Index providers use varied methods, yet their risk-return profiles closely mirror mainstream market indices and delivered virtually identical Sharpe ratios to global benchmarks.
- In the U.S., a simple index that excluded slightly lower-rated stocks underperformed its parent index because it dropped high-performing tech stocks, while another outperformed slightly by maximising ESG scores under a tracking-error constraint.
- In emerging markets, both approaches beat the market, with the exclusion-based index outperforming most years individually (though with higher volatility).
- Many ESG ratings firms also sell consultancy or advisory services to the companies they rate, which can erode their independence, as an ESG rater paid to advise companies might inflate those companies’ ratings.
- Regulators and international bodies are evaluating oversight for ESG data providers and urge them to ensure their decisions are independent, free from political or economic pressures, and to avoid conflicts stemming from their business models.
- The paper notes innovation in more objective, outcome-based ESG metrics, to provide measures focused on tangible externalities rather than abstract scores.
This paper claims ESG ratings do matter, as they can materially affect investment performance and thus fall under fiduciary duty considerations, but are not consistently reliable.
However, the authors acknowledge that all data-driven models for sustainability carry the risk of either oversimplification or overfitting.
It is worth noting that this paper is largely a comprehensive review and policy discussion rather than a traditional quantitative study: some recommendations remain untested.