Work

Measuring What Companies Say about Net Zero and Climate Transition: A Weighted-Lexicon Framework with ClimateBERT Corroboration

Year: 2026
Type: preprint
Abstract: What companies disclose about their net-zero and climate transition is not the same as what they achieve, yet existing text-analytic approaches face a trade-off between scalable but undifferentiated keyword counts and contextually sensitive but costly manual coding. This paper develops Net-Zero/Climate-Transition Disclosure (NZCTD), a transparent, auditable measure of the volume and specificity of net-zero and climate-transition language in corporate annual reports — not a measure of actual climate performance, emissions reduction, implementation, compliance, or sincerity. NZCTD is built from an eighteen-category, 273-pattern lexicon with evidentiary weights from 0 to 4, applied at the paragraph level with contextual safeguards, and corroborated by an independent, model-based ClimateBERT layer applied only to qualifying paragraphs. Applied to 468 Indian listed companies and 4,048 company-year observations spanning FY2013–FY2025, the measure has a mean score of 319.58 (median 140.0) with pronounced right-skewness and a mean rising from approximately 32.2 in FY2013 to approximately 909.7 in FY2025, an independently recomputed 28.25-fold increase. Weighted and unweighted scores correlate highly (Pearson r = 0.9855; Spearman rho = 0.9919), and principal component analysis indicates a coherent common structure alongside meaningful multidimensionality (PC1 explains 46.85% of variance; r = 0.8655 with the total score), while ClimateBERT confirmation ratios reach approximately 0.684 in later years. NZCTD offers a scalable, transparent framework for measuring climate-transition disclosure that preserves contextual insights and the disclosure-versus-performance distinction throughout.
Language: English
Open Access status: green
OpenAlex ID: W7233552512
Cites: 29
Cited by:
Related to: