Classification of CSR Using Latent Dirichlet Allocation and Analysis of the Relationship Between CSR and Corporate Value

Abstract

Corporate social responsibility (CSR) is a business approach that aims to help address social or environmental problems. Many researchers conducted empirical research to identify the relationship between CSR activities and corporate value. Some researchers explain that CSR is positively correlated with corporate value, others explain that CSR is negatively correlated with corporate value. This disagreement among the researchers has arisen because CSR standards are ambiguous. Therefore, we use topic classification to create a CSR standard. We rank the CSR activities. Our approach involves two steps. First, a CSR standard is constructed using a topic model from CSR reports. Second, the CSR rankings are calculated by using a random forest to calculate the importance of features related to CSR activities. The results show a new CSR standard. Topics represents activities related to reducing CO\(_{2}\) emissions or diversity promotion, however it is not helpful to consider too many topics: with more topics, more unrelated topics appear. CSR rankings show that medical activities have the strongest relationship with corporate value.