Finding ESG's Aptitude for Projecting Financial Value by Novel Machine Learning
by Daniel Pyeong Kang Kim
Abstract – In the last few years, Environmental, social and Corporate Governance (ESG) has established its place as a measure that discloses intangible assets or liabilities of a company. Prominent as it is, there is some skepticism regarding whether ESG serves as a suitable tool for analyzing the financial prospect of an investment. While many papers concerning this metric advocate the use of ESG with their study, some claim that there are better alternatives to ESG. Thus, this paper seeks to investigate the extent by which a company’s investment can be predicted with its ESG ranking, and the accuracy of ESG in doing so compared with that of other financial features. The main form of analysis used was Exploratory Data Analysis, which was employed to show any existing correlations between monetary traits. The data used for this analysis, namely the companies’ financial features, was extracted from company performances in the NYSE market for the last 10 years. With 21 financial features included in each company’s data, the study extracted or removed certain types of features according to their accuracy. To determine the accuracy and capability of several pipelines in classifying investments, the paper adopted eight Machine Learning Classifiers. Though these classifiers yielded similar accuracies amongst themselves, the pipelines showed a sharp distinction: algorithm classifiers containing ESG in the train process displayed a substantially higher accuracy than those without ESG. This paper demonstrates that ESG is a comprehensive, valid instrument for investors to evaluate the accurate investment worthiness of an entity.
Not yet available.