Predicting corporate culture using machine learning

This study utilizes machine learning (ML) to examine factors that predict corporate culture. Integrating firm financial metrics and nonfinancial factors, we evaluate eight ML models to identify the most effective one for predicting culture and the key factors for cultural values predictions. We also assess the roles of top executives and governance characteristics. We find the kNN regressor performs best among the eight models, and firm age, CEO age, and social performance scores are the most important predictors for the kNN model. Additionally, corporate cash holdings and environmental performance scores are among the major factors in predicting corporate culture in other models, while the Chief Executive Officer (CEO) compensation and board diversity also matter, particularly for future cultural values. Analyzing the components of culture, integrity, teamwork, innovation, respect, and quality, we document broadly similar predictors across dimensions. Grouping firm characteristics into financial metrics, nonfinancial performance, CEO characteristics, and board attributes, we show that financial metrics contribute most to predicting culture, CEO characteristics rank second, and nonfinancial performance and board attributes are least influential. Our study advances data driven research on corporate culture and highlights the factors that predict culture quality.