How ESG dimensions shape corporate green innovation: Necessary-condition thresholds, causal asymmetry, and the resource curse

Whether ESG dimensions shape green innovation through consistent mechanisms remains contested, in part because ESG rating divergence can alter empirical conclusions. Using three independent ESG data sources (Huazheng, CNRDS, Runling Global) and a multi-method design (fixed effects, NCA, fsQCA) covering Chinese listed firms (2015-2023), we document a construct-asymmetry thesis: ESG dimensions do not have uniform roles across rating sources. In the Huazheng sample, Governance (G) and Environmental (E) dimensions are necessary conditions for substantive green innovation (d > 0.28); in two independent ratings, the necessity pattern shifts toward the Social (S) dimension, while G necessity disappears. This divergence is not an artifact of sample composition - it persists in a matched same-firm sample - but reflects systematic differences in how rating agencies operationalize governance and social responsibility. Despite this measurement sensitivity, failure configurations are more identifiable than success recipes - a pattern consistent with the panel and NCA evidence across sources - and the average facilitative effect of G on strategic innovation is directionally consistent. Our findings suggest that the ESG-green innovation relationship is more method- and construct-dependent than previously assumed, and that multi-source validation is indispensable for credible ESG-innovation research.