Financial predictors of firms’ diversity scores: a machine learning approach

Purpose Departing from previous studies, this paper aims to explore the predictive roles of financial indicators on diversity. Design/methodology/approach Data on all companies that are publicly traded was acquired from the Refinitiv Eikon database. The final list, which comprises 873 worldwide business data from 2021, composed the dataset. We used fundamental forward selection techniques, multiple regression and best subset regression in R programming to look at the data and find the most critical factors. Findings We found support for the predictive roles of financial indicators on total diversity score and its three components in global companies. In addition, bagging and random forest algorithms were able to find a predictor role of total liability on the diversity pillar score and inclusion score. In contrast, the people development score was best estimated by R. The boosted regression algorithm was also able to find evidence of the predictor role of total liability for people development and inclusion score but not for diversity pillar score. Originality/value This study is one of the first to examine financial predictors of firms’ diversity scores using machine learning algorithms. The discussion section offers theoretical and practical implications and directions for further research.