CEO perceived personality and corporate risk disclosure in prospectus: A multimodal machine learning analysis

This study examines how CEO personality traits influence IPO risk disclosure quality. We employ a novel multimodal deep learning approach to measure CEO personality traits using audio-visual data from IPO roadshow presentations. Our methodology integrates visual and audio data to capture both verbal and non-verbal behavioral cues, providing a more comprehensive assessment of personality traits than traditional single-modal approaches. Using a manually collected dataset of 866 IPO firms from China's ChiNext and STAR Markets between 2019 and 2024, we find that openness, conscientiousness, and extraversion are positively associated with risk disclosure quality, while agreeableness and neuroticism show negative associations. These relationships are moderated by underwriter reputation, particularly strengthening the positive effects of openness and conscientiousness. This study contributes to the literature by introducing a novel methodological approach to measuring CEO personality traits through multimodal deep learning analysis, extending our understanding of IPO risk disclosure determinants, and demonstrating the effectiveness of multimodal data analysis in organizational research.