CEO facial masculinity and real earnings management: a machine learning approach

Purpose Using a machine learning approach, this paper analyzes the relationship between chief executive officer (CEO) facial masculinity and real earnings management (REM) in the Asian context. Design/methodology/approach The analyses were based on a dataset of 348 listed firms from India, Japan, South Korea, Malaysia, Indonesia, Thailand, and Singapore from 2015 to 2024. The authors employed a machine learning framework using the Light Gradient Boosting Machine (LightGBM) model to predict REM. CEO facial masculinity was measured using the facial width-to-height ratio (fWHR), obtained from photographs of the CEOs collected from publicly available sources. To improve the interpretability of the model, SHapley Additive Explanations (SHAP) were applied to determine the relative contribution and directional influence of CEO- and firm-level characteristics. Findings The findings indicate that the LightGBM model provides satisfactory predictive performance across the three REM channels, with the strongest performance observed for abnormal discretionary expenses (ABDISEXP). SHAP analysis reveals that CEO facial masculinity is a meaningful factor in explaining REM, while firm-level financial characteristics remain important determinants. Compared with conventional CEO demographic characteristics, such as age and tenure, fWHR demonstrates relatively stronger predictive importance in identifying income-increasing REM. Country-level SHAP results further suggest that the influence of CEO facial masculinity and other explanatory variables varies across institutional settings. Research limitations/implications This research offers valuable insights into the field of study, particularly regarding the biological characteristics of CEOs in relation to earnings management. The findings offer new evidence that CEO facial masculinity is a significant factor in earnings management and highlight the importance of integrating executive physical characteristics into the study of managerial decision-making and corporate governance. Originality/value While much of the prior literature has primarily focused on the managerial aspects of CEOs, this study focuses on their biological characteristics. To the best of the authors' knowledge, this is the first research attempt to investigate this issue using a machine learning approach. The study provides novel evidence that behavioral traits captured through facial characteristics offer incremental explanatory power beyond traditional financial and demographic determinants of earnings management.