The effect of ESG scores on the performance of banks from OECD member countries included in the FTSE index

This study investigates the relationship between ESG scores and the performance of banks from OECD countries that are listed in the FTSE Emerging & Developing ESG Index. The sample covers annual data for 65 banks over the period 2015-2022, obtained from the Bloomberg Terminal. The banks' ESG scores serve as the key independent variable, while the dependent variables comprise ROA, Tobin's Q and net interest margin (NIM) at the bank level, alongside GDP at the country level. both static (fixed-effects) and dynamic (GMM) panel estimators are employed, and Granger causality tests are conducted. In the static panel model, ESG scores exhibit a negative association with ROA, Tobin's Q, and NIM, but a positive association with GDP. According to the results of the dynamic panel data analysis, while the initial relationship between ESG scores and both ROA and Tobin's Q is negative, it turns positive in ESG square scores. This indicates a U-shaped pattern in the relationship between ESG scores and the financial performance indicators ROA and Tobin's Q. The findings of the Granger causality tests conducted within the scope of the study reveal the existence of causal relationships among the variables. As a result of the empirical analyses, it is observed that when the cost factor associated with ESG is taken into account, the initially negative relationship gradually shifts to a positive one over time. In the dynamic specification, the contemporaneous coefficients on ESG remain negative for ROA and Tobin's Q, yet statistics indicate that the longer-run cumulative effect turns positive. Granger causality tests reveal bidirectional causality between ESG scores and all variables except ROA and Tobin's Q, for which no causality is detected. Given the limited country coverage and predominantly static approaches in prior research, this study offers an original contribution by adopting a more comprehensive, cross-country panel and combining static and dynamic methodologies.