How Robust Is the ESG Pillar-Profitability Link in Central and Eastern European Banking? A Robustness Analysis of Systemically Important Institutions

This study applies its own proposed methodological standard to test a widely repeated claim in banking–ESG research: that disaggregating composite ESG scores into environmental, social, and governance pillars reveals a social-pillar profitability premium the composite score obscures. Using LSEG/Refinitiv data for banks across seven Central, Eastern, and Southeastern European economies (FY2022–FY2023), once the sample is restricted to institutions unambiguously classified as commercial banks and to fiscal years with a consistently scaled governance measure (18 institutions, N = 36), the social pillar shows no significant association with return on assets (β = −0.102, p = 0.523, entity-clustered), a result stable across all leave-one-institution-out re-estimations and a wild-cluster bootstrap. By contrast, the same specification on the broader, vendor-classified sample as originally extracted, including two non-bank entities and an inconsistent governance-scale year (20 institutions, N = 59), shows a large, significant premium (β = 0.718, p = 0.019) that vanishes once either correction is applied and reverses sign once combined. This divergence, within a single dataset, illustrates how pillar-level ESG-profitability findings in small banking panels can be artifacts of sample composition and measurement error. We propose the resulting four-step Pillar Robustness Checklist (PRC) as a minimal reporting standard for future pillar-level ESG–banking research.