Comparative Analysis of CSR and ESG Actions in Greece: A Study Using Artificial Neural Networks and Machine-learning Techniques

Purpose:In Greece, large companies have started to focus more and more on corporate social responsibility (CSR) and ESG (environmental, social, and governance) activities, realising the importance of sustainability and social responsibility beyond traditional profits. Using machine-learning (ML) methods and artificial neural networks (ANNs) can enhance the process of measuring performance in these areas in several ways, including data analytics. This paper investigates and explores the correlation between CSR and ESG actions with financial and non-financial factors for the 100 largest companies operating in Greece. Methodology:The study runs from January 2019 until December 2021, and ANNs and ML techniques are employed. The comparison concerns both the control variables and the predictability of the methods. Findings:The main findings that emerged are the confirmation of the correlation between CSR and ESG actions and the financial performance and determinants of corporate responsibility of the companies in the sample. Moreover, good results were obtained for almost all of the techniques examined, but the superiority of deep learning models and gradient-boosted trees (GBTs) was found for the selected variables. Significance/Implications/Conclusions:The findings suggest that using ML techniques and neural networks to measure CSR actions can help companies evaluate their performance and make effective decisions to improve their sustainability. It can also be a valuable tool for institutional investors, banks, and regulators. Future Research:We believe that future research should focus on improving these models, exploring hybrid approaches that combine the strengths of different techniques, and expanding the range of variables considered.