人工智能发展会扩大收入差距吗——理论假说与省级证据

Does the Development of the Artificial Intelligence Enlarge China’s Income Gap: Theoretical Hypothesis and Provincial Evidence

基于人工智能的技术偏向性视角,理论诠释了人工智能的数据偏向、资本偏向和技能偏向性对行业之间、要素之间和异质性劳动者之间收入差距的影响机理。实证研究表明,人工智能发展有利于缩小中国行业收入差距,但会显著降低劳动收入份额,并提高劳动者技能溢价水平,该结论通过了内生性和稳健性检验。人工智能发展对中国收入差距的影响具有区域和时期异质性,中西部人工智能发展对收入差距的影响效应与全国样本一致,但东部人工智能发展对行业收入差距、东部整体和服务业部门的劳动份额影响不显著,人工智能对中国收入差距扩大的影响主要发生在人工智能的高速发展阶段。进一步检验发现,劳动生产率在人工智能与收入差距之间起着显著的中介或遮掩作用。

Based on the technical-bias of artificial intelligence, the paper theoretically analyses the influencing mechanisms of the data-bias, capital-bias and skill-bias of artificial intelligence on the income gap between different industries, factors and heterogeneous labors respectively. The empirical results show that the development of artificial intelligence(AI) is conducive to narrowing the income gap among different industries. However, it significantly reduces the labor income share and increases the skill premium of laborers. The conclusions have passed endogenous and robust tests. The study also shows that the impact of AI development on the income gap has regional and period heterogeneity. The impact of AI on the income gap of the central and western China is consistent with that of the national sample, while the impacts of AI development in eastern China on the industrial income gap, the overall labor income share and the labor income share in service industry are not significant. The impact of AI on widening China’s income gap mainly occurs in the rapid development stage of AI. Further research shows that labor productivity plays a significant mediating or hiding role between AI and income gap.