The nonlinear association between ESG disclosure and firm performance: Evidence from Vietnam’s listed energy firms

Type of the article: Research ArticleAbstractVietnam’s energy sector faces growing tensions between production expansion and environmental protection, making the economic consequences of ESG disclosure particularly important. Accordingly, this study examines the nonlinear associations between composite and pillar-level ESG disclosure on 30 energy firms listed on Vietnam’s stock exchanges during 2015–2024. Annual report data were extracted by optical character recognition, with a large language model supporting paragraph classification into predefined E, S, and G categories, then quantified into an ESG score and each pillar based on frequency score and coverage score on 300 firm-year observations. The feasible generalized least squares regression results, with cautious interpretation for the relatively small sample size, indicate that ESG disclosure exhibits a statistically significant, though not uniformly robust, inverse U-shaped association with return on assets (ROA) (β = 0.012; β² = –0.002), Tobin’s Q (β = 0.025; β² = –0.004), and return on sales (ROS) (β = 0.003; β² = –0.001) although the statistical evidence is stronger for ROA. The E pillar shows a significant nonlinear association with ROA (β = 0.006, β² = –0.001) and ROS (β = 0.003[HT1.1], β² = –0.001), but no statistically significant association with Tobin’s Q. Governance disclosure shows comparatively stronger associations, particularly with Tobin’s Q (β = 0.034; β² = –0.005), while social disclosure is significantly associated with ROA and ROS but not with Tobin’s Q. These findings are associated with stronger governance signaling for market valuation, suggesting that energy firms may benefit from weighing governance strength alongside environmental disclosure aligned with operational capacity.