Description:
We follow a multivariate statistical approach to address whether stocks similar in terms of their ESG characteristics form a specific group similar in financial behavior. We employed the ESG ratings of seven major data providers and their artificial structural combination to determine the individual strength and the effect of shared information in the aggregated (ESG) and dimensional (E-S-G) perspective. The empirical classification of SP 1200 stocks regarding their ESG and several risk-return-liquidity variables shows no general persuasive effect for seven consecutive years, including the period of the COVID-19 pandemic. Nevertheless, we confirm that the unidimensional scores, particularly Governance, are more powerful than multidimensional indicators. Moreover, we show that the artificial structural scores considerably improve the quality of the model, implying the appropriateness of their use within financial decision-making. These results are supported by a robustness check on different time settings exploring ESG as a leading, concurrent, and lagged indicator