Abstract: Research on the impact of second-generation characteristics on family business innovation is limited by several factors. First, existing studies often focus on the interaction effects of a single or a few second-generation characteristic variables due to methodological constraints, lacking a systematic and comprehensive analysis of multidimensional second-generation variables. Second, most research employs an explanatory perspective to infer causal relationships, while giving limited attention to systematic quantitative analyses from a predictive perspective. This study, adopting a predictive approach and employing machine learning methods, uses data from 2,369 publicly listed Chinese family firms to comprehensively examine the predictive impact of multidimensional second-generation characteristics on family business innovation. It further identifies the key second-generation characteristic factors that significantly influence the prediction of family business innovation levels and elucidates their predictive mechanisms. The findings indicate that: (1) among personal characteristics, the duration of second-generation involvement is a strong predictor of R&D investment, while second-generation age is a stronger predictor of R&D output; (2) in terms of succession methods, the second-generation’s shareholding ratio is a strong predictor of family business innovation levels; (3) the relationships between the second-generation shareholding ratio, age, involvement duration, and family business innovation levels exhibit non-linear characteristics. The conclusions of this study provide valuable insights for family businesses during the intergenerational succession period on how to enhance innovation capabilities through the role of the “successor”.

Key words: intergenerational succession, second-generation characteristics, succession methods, family business innovation, machine learning