Investor appetite for ESG data pushes competing raters to cover the same ground

The private incentive in this model is standalone usefulness, not contribution to the aggregate picture. Azarmsa and Shapiro build an investor who commits capital only after receiving sufficiently positive information on both environmental and social performance, which turns a single-category specialist into a complement whose work pays off only when someone else supplies the other half ^[4]. A provider that covers both categories can sell a rating that carries a decision by itself, and it does so knowing that the two providers together now produce less precise information than two specialists would ^[5]. The pull toward breadth is strongest exactly where investors weight ESG most heavily ^[3].

That inverts the usual reading of a low correlation between two scores. Divergence among major providers has been documented before, often taken as an argument for consistency and standardisation ^[11]; in this framework, the configuration that produces the widest disagreement is also the one that puts the most information in front of investors ^[7]. The tradeoff is comparability against depth. A common taxonomy would make two ratings agree more often, and by the paper's own logic it would achieve part of that agreement by deleting what the disagreement was carrying.

A rating gap is not automatically something to worry about, but that reading is incomplete. The authors cite Berg, Kolbel and Rigobon, who attribute 38% of the discrepancy in category ratings among major agencies to scope divergence, meaning differences in which subcategories each provider examines ^[8]. That leaves 62% of the discrepancy attributed elsewhere in the same decomposition ^[13], and Azarmsa and Shapiro are explicit that ratings also diverge because providers use different data or measurement procedures ^[10]. Read uncharitably, the paper's logic could hand raters a permanent excuse for never reconciling anything. The answer sits in the conditional: only scope-driven divergence earns the benign reading ^[9], the 38% is a market-wide average rather than a decomposition of any single company's gap, and nothing here tells a company which kind of divergence produced its own two scores.

What the analysis does not take up is the payment side. It describes an industry that collects, analyses and sells information about firms' ESG performance ^[14], and it derives provider behaviour from competition and investor demand ^[15], so reading it as a verdict on how raters treat the companies they rate goes past the evidence.

The comparison with credit ratings is where the practical difference sits. A credit rating resolves one dimension, credit risk, while an ESG rating aggregates categories that need not be related to one another ^[12]. Under this model, "which of our two ratings is correct" has no answer, while "which subcategories did each provider actually score" has one that a provider's own scope documentation can supply. That is a reporting-cycle question rather than a this-week question, and it changes what an issuer asks a rater for well before it changes anything an issuer publishes.