The Label Cycle: When Sustainability Classification Systems Erode Their Own Informativeness

Sustainability ratings, ESG scores, and thematic indices increasingly guide the allocation of fund assets, yet the systems producing these classifications have received limited theoretical attention. This paper develops a conceptual framework, the label cycle, that treats sustainability classification as a feedback system linking producers who design classification rules, users who allocate capital on the resulting labels, and targets that adjust disclosures and conduct toward classification boundaries. Evidence from equity style indices shows that labels can move prices once investors trade on them, and sustainability labels add the response of rated firms to the rules that classify them. This response erodes a label’s informativeness only where firms can change the attributes a rule reads without a commensurate change in the construct the label measures and differ in their capacity to do so. A causal loop diagram and a stylized model ground three propositions on how rule mechanicality and transparency shape boundary management, how reliance raises comovement and, under these conditions, deepens the loss of informativeness, and when revision restores it. The Korea Value-Up Index and fund reclassification under the EU Sustainable Finance Disclosure Regulation illustrate individual mechanisms, and each proposition is paired with a research design and refuting evidence.