| 摘要: | This dissertation comprises three thematically related papers that collectively examine how firm size, firm age, and artificial intelligence integration influence the operational efficiency of manufacturing firms in the context of Industry 4.0. Grounded in production economics, the dissertation uses data from Taiwan's 2016 Industrial and Commercial Census and applies a synergistic approach that combines stochastic frontier analysis (SFA) with an eXtreme Gradient Boosting (XGBoost) algorithm to evaluate and predict efficiency.
The findings reveal several critical insights. First, firm size shows a U-shaped or inverse U-shaped relationship with efficiency, depending on whether the firm is a small or medium-sized enterprise or a large enterprise. Second, firm age exhibits an inverted U-shaped pattern with respect to efficiency, indicating that younger firms initially lag in efficiency, gradually catch up during mid-term stages, and may experience diminishing gains in the long run. Third, AI integration exerts uneven effects across industries: Commerce, Construction, Transportation and Storage, and Finance and Insurance show stronger efficiency improvements, whereas Manufacturing does not demonstrate significant benefits.
Moreover, by combining SFA's ability to distinguish inefficiency from random disturbances with XGBoost's robust predictive capabilities, this dissertation identifies firm-level and industry-level conditions under which firms can achieve full efficiency. From a policy perspective, the findings underscore the necessity of designing differentiated strategies for firms varying in size and age, alongside fostering AI-friendly ecosystems across industries.
Methodologically, this dissertation introduces an innovative fusion of conventional econometric models and advanced machine learning techniques, illustrating the potential of a synergistic approach for research in operational efficiency and industrial economics. Finally, the dissertation addresses endogeneity concerns, data privacy restrictions, and future directions for moderation analysis, providing actionable insights for efficiency enhancement under Industry 4.0. |
