Artificial intelligence and firms’ labor income share: measurement and analysis based on LLM

We apply large language models and machine learning techniques to construct firm-level AI indicators, thereby overcoming the limitations of traditional word-frequency methods in capturing textual semantics. Using financial data on China’s A-share listed companies from 2007 to 2023, we then empirically examine the effect of AI application on firms’ labor income share. The results show that AI application increases firms’ labor income shares through a scale expansion mechanism induced by productivity effects and a wage-enhancing mechanism driven by augmentation effects. Further analysis indicates that, relative to general-purpose foundational AI technologies such as machine learning, the adoption of task-specific, application-oriented AI technologies – such as robotics, natural language processing and computer vision, planning and decision-making systems – has a more pronounced positive effect on firms’ labor income shares. Moreover, the positive effect of AI application is more pronounced in firms with stronger digital foundations, in technology-intensive firms, in small firms and in private firms.