The Hidden Arithmetic of Climate Disclosure: How a Decade of CDP Climate Scores Masks a Large Within‐Firm Rise and an Equally Large Composition Drag

ABSTRACT Using the full population of letter‐graded organisations in the licensed CDP climate score files from 2016 to 2025 (37,450 firm‐years across 8959 firms), this study examines how CDP‐assessed climate disclosure quality evolves over time. We decompose the aggregate trend into within‐firm improvement and composition effects, separate the composition effect into entry and exit contributions and analyse the role of disclosure detail using a Textual Detail Index supported by multiple construct‐validity checks. The modest aggregate increase of +0.42 points masks substantial within‐firm improvement of +1.12 points (bootstrap 95% CI 1.03–1.22) among 1470 stayers on the 1–8 graded scale, which is largely offset by a composition effect of −0.70 driven mainly by lower‐scoring new entrants (−0.90) and partly offset by the exit of lower‐scoring firms (+0.20). The pattern is essentially unchanged when the window ends in 2023, before CDP's 2024 scoring revision, and survives ordinal robustness checks based on grade transitions. The association between disclosure detail and scores is strong in the cross‐section ( β = 1.29) but attenuates considerably within firms ( β = 0.57), indicating that persistent reporting capability accounts for much of the observed variation; a blind structured audit of 60 disclosures supports the index's construct validity. Because CDP scores measure assessed disclosure quality rather than environmental outcomes, we interpret these gains as growth in reported climate‐management maturity. CDP scores remain a useful input for investors, analysts and researchers, but their aggregate trends can be systematically misleading when the reporting population is expanding, and cohort‐adjusted benchmarks offer a practical remedy.