Under What Conditions Do Workers Share in the Benefits of Artificial Intelligence? A Study Based on the Labor Income Share, Labor Market Power, and Resource Configuration

Whether AI’s efficiency gains are shared with workers is a key question in research on corporate value creation and income distribution, yet existing studies mainly examine AI’s average effect on labor income, employment, or labor market power, without systematically considering how firm resource conditions jointly shape worker outcomes. Using a sample of Chinese A-share-listed manufacturing firms from 2017–2024, this paper combines panel regression, production function decomposition, and panel fuzzy-set qualitative comparative analysis (fsQCA) to examine how AI relates to the labor income share and labor market power (Markdown), and the resource configurations behind this relationship. Results show that: (1) AI intensity is positively associated with the labor income share (significant at the 1% level), while its negative association with Markdown is statistically weaker (significant only at the 10% level). AI intensity is also associated with employment expansion and a higher share of technical and R&D personnel, whereas the estimate for average employee compensation is not statistically significant; (2) the AI-related decline in the labor-side wedge stems mainly from a rising labor income share rather than falling labor output elasticity, while product market markup changes partly offset the overall Markdown change; (3) no single condition—AI, R&D, human capital, financing capacity, supply chain structure, or government support—is individually necessary for worker-friendly outcomes; high labor income share and low Markdown can each be reached through multiple resource configurations, showing equifinality and causal asymmetry. These findings suggest that worker-friendly outcomes are associated not only with higher AI intensity but also with distinct configurations of complementary innovation, human, financial, supply-chain, and institutional resources.