This paper investigates the eﬀects of the diﬀerent levels of ESG scores, and their underlying data, on the ﬁnancial performance of S&P500 companies. The study is based on Thomson Reuters ESG data between 2005 and 2018. Initially a thorough sampling process is executed, including data imputation, to take care of the missing observations. The ﬁnal sample includes 45 of the initial 500 companies and their respective ESG-score, pillar-scores, sub-scores and 21 of their fundamental enviromental- and raw data variables. The analysis, based on univariate time series analysis and multidimensional panel data regression analysis, aim to investigate the magnitude of ESG determinants on log-return. By conducting a thorough modelling process, where log-returns are ﬁtted with each and every variable as external covariate, 345 models are identiﬁed. These models show that the magnitude of ESG determinants are indiﬀerent. This is conﬁrmed in extension by the results of the multidimensional regression analysis, which furthermore shows that the initially omitted inter-individual diﬀerences have no impact on the results. The robustness tests of both the univariate time series analysis and the multidimensional regression analysis is checked by conducting multiple tests, which include, tests of informationcriteria, imputation, inﬂuential observations and timeintervals. These measures are taken in order to validate the methological choices made in this paper. In short, by validating the robustness of the sampling and modelling process, this paper establishes that both ESG-scores and their underlying determinants have no signiﬁcant impact on returns of S&P500 companies.
|Educations||MSc in Business Administration and Mathematical Business Economics, (Graduate Programme) Final Thesis|
|Number of pages||148|