Persistent Firm Performance and ESG Ratings: A Chronologically Separated Evaluation in the Korean Capital Market

This study examines whether a persistent firm-specific component of financial performance, estimated only from a 2011–2017 training window, remains informative out of sample and whether it is associated with later ESG ratings in Korean listed firms. The component (PFPC) is the firm fixed effect from a two-way fixed-effects model. In a chronologically separated 2019–2025 holdout, PFPC outperforms firm-mean benchmarks and, for Tobin’s Q, a training-fitted covariates model, but not that model for ROA, and it does not outperform the latest observed outcome for either measure; Tobin’s Q holdout accuracy decays to zero by 2022–2025. Using one observation per firm, mean ESG is negatively associated with ROA-based PFPC and positively associated with Tobin’s Q-based PFPC (one-standard-deviation differences of −0.21 and +0.48 standard deviations of PFPC). Two-stage firm-bootstrap intervals exclude zero for Tobin’s Q, whereas the ROA interval includes zero at its upper end and the ROA sign reverses under alternative winsorization. Because the data are current-vintage extracts, the exercise is a chronological holdout, not a real-time forecast. The evidence supports a bounded evaluation framework, not a universal predictive or causal ESG claim.