摘要
在生产网络中对新一代通用目的技术——人工智能的采用和扩散是实现中国技术进步和高质量发展的关键所在。但在通用目的技术扩散初期,劳动生产率增长将经历较长时间的低迷阶段,这被称为"索洛悖论"。本文在对比信息时代和人工智能时代典型事实的基础上,将人工智能在生产网络中采用和扩散的过程内生化,构建了一个通用目的技术扩散影响劳动生产率增长的动态模型,揭示了"索洛悖论"形成及其演化背后的经济机制,并进行了参数校准与仿真模拟。模型模拟结合基于中国产业面板数据的实证研究发现:无论是用研发投入与SG&A费用,还是以上市公司股票估值溢价所衡量的无形资本对生产率增长的影响都呈现出短期抑制作用,但在长期将有效提升生产率。此外,由于生产网络中的上游产业传导效应并不显著,提高下游产业传导效应和激发企业家精神是提高短期和长期劳动生产率,实现高质量发展的重要途径。
The adoption and diffusion of artificial intelligence,a new generation of general purpose technology(GPT),in the production network is the key to achieving technological progress and high-quality development in China.Historical experience shows that during the early stages of GPT diffusion,labor productivity growth goes through a long period of deceleration,a phenomenon known as the Solow paradox.In the era of artificial intelligence,this phenomenon is caused by the unsynchronized diffusion of artificial intelligence technologies in the production networks of different industries and firms.Firms using artificial intelligence in the early stages of technology diffusion need to invest more intangible capital,so that the more intangible capital they invest in building their future competitiveness,the more the losses in their current financial statements,resulting in the illusion of low labor productivity.However,the premium of stock valuation of listed firms can capture such intangible capital that will generate more revenues and profits in the future.Based on the comparison and summary the typical facts of GPT diffusion and labor productivity growth in the information age and artificial intelligence era,this paper endogenizes the process of production network evolution and artificial intelligence adoption and diffusion,constructs a dynamic model of endogenous production network in which GPT diffusion affects labor productivity growth in the different stages,and reveals the economic mechanism of formation and evolution of the Solow paradox.And the parameters of the model are calibrated using Chinese economic data to predict the long-term trend of artificial intelligence technology diffusion and labor productivity growth in China,to explore the possible effects of changes in the model structure parameters and their mechanisms of action.An empirical study was also conducted using Chinese panel data of 48 industries from 2010 to 2014 to prove the hypotheses of the model.The data is from World Input-Output Database and China Stock Market and Accounting Research Database.Through model simulations and empirical studies,it is found that a larger share of intangible capital in the artificial intelligence era will lead to a more significant decline in labor productivity at the beginning of technology introduction phase and a more serious Solow paradox.However,Solow paradox changes in different phases,appearing in the identification and introduction phase of artificial intelligence and disappearing in the production synergy and maturity phases.The impact of intangible capital on productivity growth,measured by both research and development expense plus selling,general and administrative expenses(SG&A),and the valuation premium of listed firms’stocks,shows a short-term decreasing effect,but in the long run it is effective in raising productivity.In addition,since the upstream industry transmission effect in the production network is not significant,improving the downstream industry transmission effect and stimulating entrepreneurship are important ways to improve labor productivity in both short and long run and achieve high-quality development.Based on the above findings,China should adopt economic policies to accelerate the diffusion and integration of artificial intelligence technologies in various industries,stimulate entrepreneurship,and accelerate the construction of new infrastructure in order to improve labor productivity and reduce the negative impact of Solow paradox.The innovation of this paper is as follows.Firstly,by generalizing the typical fact of GPT diffusion and labor productivity growth in the information age and artificial intelligence era,it finds the general rule of labor productivity change under the S-shaped diffusion process of GPT.Secondly,the model builds on the long-term mechanism of automation influencing technological progress from the model of Aghion et al.(2017)and incorporates the S-shaped GPT diffusion process arising from the micro-decisions of firms in the production network.The model endogenizes the process of GPT diffusion in the production network and portrays its impact on the short-term fluctuations and long-term growth of labor productivity,thus explaining the Solow paradox in the era of artificial intelligence.
作者
程文
CHENG Wen(Huazhong University of Science and Technology;Massachusetts Institute of Technology)
出处
《经济研究》
CSSCI
北大核心
2021年第10期22-38,共17页
Economic Research Journal
基金
中央高校基本科研业务费(项目名称:通用技术扩散历史视角下人工智能对生产率的影响研究)的资助。
关键词
人工智能
索洛悖论
高质量发展
通用目的技术
Artificial Intelligence
Solow Paradox
High-quality Development
General Purpose Technology
