“头部化”抑或“均等化”:人工智能技术会改变企业规模分布吗?

"Cephalization"or"Homogenization":Will AI Technology Change the Size Distribution of Firms?

党的二十大报告明确提出“推动制造业高端化、智能化、绿色化发展”。虽然现有文献关注人工智能技术及其对产业集聚的影响,却普遍忽视了人工智能技术是否会影响企业成长平等性,即智能化情境中企业规模分布问题,这关乎社会公平与效率。为此,本文从技术偏向和企业规模分布均等性入手,首次考察人工智能技术是否会改变企业规模分布状态,结合2004~2019年天眼查企业微观数据及中国工业企业数据库,采用修正的Pareto指数测算地级市制造业企业规模分布情况,剖析人工智能技术对企业规模分布的作用,并检验不同技术属性企业和约束条件下的差异性影响。研究结果表明:人工智能技术会助推中小型企业规模均等化成长,多种稳健性和内生性处理后结论依旧稳健。这一表现依赖于企业技术属性和外部约束条件,即人工智能技术会使传统企业朝均等化方向发展,同时也会推动大型人工智能企业头部化;更高的智能化信息基础设施水平有利于企业均等化成长,而更高的劳动力成本则将推动企业头部化。机制分析显示,人工智能技术会通过技术进步和资源配置效率提升这两条路径诱发企业规模分布发生变化。以上结论为拓展技术进步方向和社会平等性问题的研究,以及企业平等地享受技术红利和实现“共同富裕”提供微观经验证据。

The report of the 20 th National Congress of the Communist Party of China proposed that"promote high-end,intelligent,and green development of the manufacturing industry."The existing literature focuses on AI technology and its impact on industrial agglomeration but generally neglects whether firms of different scales can obtain technological dividends equally when AI technology appears and whether it would cause inequality in firm growth.For example,how would AI technology affect the size distribution of firms in an intelligent environment,which is a matter of social equity and efficiency?Moreover,what types of companies will take advantage of AI technologies to scale up and become"supersized"and"star"?Is there any difference in the impact of AI technology on the size distribution of different types of firms?Using the microscopic data of firms from 2004 to 2019 from Tian-Yancha and the database of Chinese industrial firms,we examine whether AI technology will change the size distribution of firms from the perspective of technology bias and the parity of size distribution of firms and use the basic and modified Pareto index to measure the size distribution of manufacturing firms in prefecture-level cities.We then use Python to extract AI patents according to AI thematic keywords to measure the level of AI technology in prefecture-level cities.Next,we measure the impact of AI technology on the size distribution of firms and its difference across firms with different technology attributes and constraints.The findings reveal that AI technology promotes an increase in the Pareto index of the overall size distribution of manufacturing firms.AI technology impacts small and medium-sized firms more significantly than large-scale firms.Then,we conduct variable replacement,sample replacement,database replacement,instrumental variable tests,and endogeneity treatments such as PSM-DID,and the results are still stable.Further analysis reveals that this result depends on firm technology attributes and external constraints.Specifically,compared with non-high-tech firms,high-tech firms are more capable of integrating AI technologies.Thus,AI promotes the growth of small and medium-sized high-tech firms more significantly.Second,AI technologies not only move traditional firms toward parity but also push large AI firms to the top.Third,a higher level of intelligent information infrastructure will facilitate the homogenization of firms,whereas higher labor costs will drive the cephalization of firms.The analysis of mechanisms indicates that there are two mechanisms through which AI technology affects the size distribution of firms,specifically by facilitating technological progress and optimizing resource allocation efficiency.First,technological progress significantly inhibits the trend toward headiness in the size distribution of AI firms,but the impact on the size distribution of traditional firms is currently insignificant.Second,improved resource allocation efficiency will not only further promote the trend toward cephalization in the size distribution of AI firms but also has a weak impact on the size distribution of traditional firms.The contributions of this paper include the following.First,it is the first study to examine the selective bias of AI technology and the parity of the growth of firms of different sizes with micro-firm data from Tian-Yancha and analyze whether AI technology induces the cephalization of firms or contributes to the parity of firms.Second,it innovatively explores the constraints on the selective bias of AI technologies and examines the differentiated results of AI technology bias for different technology attributes,i.e.,whether they are high technology or AI firms.It also examines the impact of constraints,such as intelligent information infrastructure and labor costs,and dissects the transmission mechanisms through which AI technology affects the size distribution of firms.Third,it makes up for the lack of research on the size distribution of all manufacturing firms.This paper provides micro evidence to expand research on technology bias and social equality to provide firms with equal access to technology dividends and achieve"common prosperity."First,we should accelerate the pace of building intelligent digital information infrastructure.Second,we should accelerate the technological innovation and application of artificial intelligence and strengthen the market regulatory mechanism.Third,we should adjust human resources training programs.Fourth,we should strengthen production cooperation and technology exchange between firms to promote the compatible development of firms of different sizes.