Quantitative Analysis of Citi's ESG Reporting: LDA and TF-IDF Approaches

Abstract

The rapid development of society is accompanied by significant changes in the environment. ESG is still in its nascent stages, and ESG policies have yet to be perfected. In this study, we examined Citibank's 2019 annual report and analyzed ESG factors using the Latent Dirichlet Allocation (LDA) method. We collected data through word clouds and identified words with the highest frequency through frequency analysis. Subsequently, we determined the optimal topic number from the perplexity curve obtained via the LDA method. Visualization charts further helped us distinguish these four ESG topics. Our findings indicate that both environmental and social factors are crucial in ESG, with corporate governance playing the most significant role.