人工智能、技术进步与低技能就业——基于中国制造业企业的实证研究

doi:10.16381/j.cnki.issn1003-207x.2019.1251

摘要/Abstract

摘要： 人工智能重塑中国制造业核心竞争力的同时，也加剧了行业内低技能型劳动力被取代的危机。本文以技术进步影响劳动力技能结构的理论机制为分析框架，聚焦人工智能与制造业的融合对低技能就业比重的因果效应。基于2011-2017年制造业企业的面板数据，首先将人工智能与制造业的融合定义为"AI产业化"和"产业AI化"，基于定义、逐家判断人工智能企业名单并定位人工智能的初始时间，再采用双重差分法围绕人工智能对低技能就业比重的因果效应及效应的传导机制展开实证分析。实证结果表明：（1）制造业企业融合人工智能显著降低了低技能的就业比重，即在一定程度上制造业内低技能潜在的就业机会被人工智能所挤出；（2）人工智能对低技能就业比重的负向效应具有动态异质性，即企业融合人工智能的时间越长，低技能就业比重下降的越多；（3）人工智能虽然通过刺激当期资本积累和收入扩张补偿了部分低技能就业比重的减少，但同时降低了低技能劳动力的相对边际产出，促使企业最终减少低技能员工的就业比重。

关键词: 人工智能, 技术进步, 低技能就业, 双重差分, 中国制造业

Abstract: As Artificial Intelligence (AI) is remodelling the core competitiveness for Chinese manufacturing industry, there is an increasing concern about a substitution for low-skilled labor.This study provides evidence on how AI affects the patterns of skill demands, measured by the share of low-skilled staff in Chinese manufacturing firms. Employing panel data on Chinese firms listed on the SSE, SZSE, and NEEQ over 2011-2017, the introduction of AI is treated as a quasi-natural experiment and use a robust DID model to estimate the effects of AI on patterns in the relative demand for low-skilled labor. AI firms are defined by investigating the AI production and application scenarios in manufacturing and select AI firms based on policy support, industry application, R&D output, and market recognition. The timing of the initial introduction of AI is defined based on firm-level information. In addition, using a PSM model, the control group whose skill patterns follow the parallel trend assumption is developed.
The empirical analysis results demonstrate that AI has significant influences on skill demands. First, AI significantly decreases the share of employment of low-skilled labor, which means a small portion of the low-skilled potential employment has been substituted by AI. Second,the negative effect of AI on the share of employment of low-skilled labor has dynamic heterogeneity, and the effect tends to rise over time. Third, AI creates the employment of low-skilled labor through capital accumulation and income expansion while finally urges to decrease the low-skilled share through cutting down marginal output of low-skilled labor.
Finally, several possible directions for future studies can be extended in two aspects:First, due to the limitations of the data sources, the range of sample observations is only seven years, and data from a longer period can be used to verify the long-term casual effects in further research. Second, the mechanisms related to different forms of AI introduction and firm-level skill premium can be analyzed.

Key words: Artificial Intelligence, technological change, low-skilled labor, Difference-In-Difference, Chinese manufacturing industry

中图分类号:

谢萌萌, 夏炎, 潘教峰, 郭剑锋. 人工智能、技术进步与低技能就业——基于中国制造业企业的实证研究[J]. 中国管理科学, 2020, 28(12): 54-66.

XIE Meng-meng, XIA Yan, PAN Jiao-feng, GUO Jian-feng. Artificial Intelligence,Technological Change and Low-skill Employment. Empirical Evidence from Chinese Manufacturing Firms[J]. Chinese Journal of Management Science, 2020, 28(12): 54-66.

参考文献

[1] 陈彦斌,林晨,陈小亮. 人工智能、老龄化与经济增长[J]. 经济研究, 2019, 54(7):47-63.
[2] Acemoglu D, Restrepo P. Artificial intelligence, automation and work[R]. Working Paper,National Bureau of Economic Research, 2018.
[3] McKinsey China. A future that works:Automation, employment, and productivity[R]. Working Paper, 2017.
[4] Frey C B, Osborne M A. The future of employment:How susceptible are jobs to computerization?[R]. Discussion Paper,University of Oxford, 2013.
[5] Frey C B, Osborne M A. Technology at work v2.0:The future is not what it used to be[EB/OL]. (2016-01)[2016-01]. https://www.oxfordmartin.ox.ac.uk/downloads/reports/Citi_GPS_Technology_Work_2.pdf
[6] 普华永道. 人工智能和相关技术对中国就业的净影响[ER/OL].(2018-11-16)[2018-12]. https://www.pwccn.com/zh/services/consulting/publications/net-impact-of-ai-technologies-on-jobs-in-china.html
[7] Brynjolfsson E, Rock D, Syverson C. Artificial Intelligence and the modern productivity paradox:A clash of expectations and statistics[J]. Nber Working Papers, 2017.
[8] Autor D H, Katz L F, Krueger A B. Computing inequality:Have computers changed the labor market?[J].Social Science Electronic Publishing,1997,113(4):1169-1213.
[9] Acemoglu D. Why do new technologies complement skills? Directed technical change and wage inequality[J]. The Quarterly Journal of Economics, 1998, 113(4):1055-1089.
[10] 姚先国,周礼,来君. 技术进步、技能需求与就业结构——基于制造业微观数据的技能偏态假说检验[J]. 中国人口科学, 2005,(5).
[11] 宋冬林,王林辉,董直庆. 技能偏向型技术进步存在吗?——来自中国的经验证据[J].经济研究,2010,45(05):68-81.
[12] Korinek A, Stiglitz J E. Artificial intelligence and its implications for income distribution and unemployment[J]. Nber Working Papers, 2017.
[13] Acemoglu D, Restrepo P. Low-skill and high-skill automation[J]. Journal of Human Capital, 2018, 12(2):204-232.
[14] Acemoglu D. Labor- and capital-augmenting technical change[J]. Journal of the European Economic Association, 2003, 1(1):1-37.
[15] 张鹏飞. 人工智能与就业研究新进展[J]. 经济学家, 2018(08):27-33.
[16] 孙博, 刘善仕,姜军辉,等. 企业融资约束与创新绩效:人力资本社会网络的视角[J]. 中国管理科学, 2019, 27(4):179-189.
[17] HjortJ, Poulsen J. The arrival of fast Internet and employment in Africa[J]. American Economic Review, 2019, 109(3), 1032-79.
[18] Acemoglu D, Restrepo P. The wrong kind of AI? Artificial Intelligence and the future of labor demand[R]. Working Paper, National Bureau of Economic Research, 2019.
[19] Agrawal A, Gans J S, Goldfarb A. Artificial intelligence:the ambiguous labor market impact of automating prediction[J]. Journal of Economic Perspectives, 2019, 33(2):31-50.
[20] Bessen J. Toil and technology[J]. Finance and Development, 2015, 52(1):16-19.
[21] 杨蕙馨,李春梅.中国信息产业技术进步对劳动力就业及工资差距的影响[J]. 中国工业经济, 2013,(1):51-63.
[22] 宁光杰,林子亮.信息技术应用、企业组织变革与劳动力技能需求变化[J]. 经济研究,2014,(8):79-92.
[23] Acemoglu D, Restrepo P. Robots and jobs:Evidence from US labor markets[J]. Journal of Political Economy, 2019,DOI:10.1086/705716.
[24] 邵文波, 李坤望. 信息技术、团队合作与劳动力需求结构的差异性[J]. 世界经济, 2014,37(11):72-99.
[25] Autor D H, Katz L F, Kearney M S. The polarization of the U.S. labor market[J]. American Economic Review, 2006, 96(2), 189-194.
[26] Michaels G, Natraj A, Van Reenen J. Has ICT polarized skill demand? Evidence from eleven countries over 25 years[J]. CEPR Discussion Papers, 2010, 96(1), 60-77.
[27] Graetz G, Michaels G. Robots at work[J]. Review of Economics and Statistics, 2018, 100(5):753-768.
[28] 腾讯研究院. "人工智能+制造"产业发展研究报告[ER/OL]. (2018-06-20)[2018-06-21]. http://www.tisi.org/5031
[29] 解季非. 制造企业服务化路径选择研究[J]. 中国管理科学, 2018, 26(12):138-148.
[30] 清华大学. 2018人工智能发展报告[EB/OL].(2018-07)[2018-08].http://cistp.sppm.tsinghua.edu.cn/publish/cistp/1480/index.html
[31] 王亮. 微博媒体的信息传播分析[D]. 哈尔滨:哈尔滨工业大学, 2012.
[32] 郭研,郭迪,姜坤. 政府资助、项目筛选和企业的创新产出——来自科技型中小企业创新基金的证据[J]. 产业经济研究, 2015,(2):33-46.
[33] 潘文卿, 吴天颖. 中国技术进步偏向性的省际扩散效应:1996-2015[J]. 系统工程理论与实践, 2018,38(2):374-389.
[34] 黄凌云, 杨娜. 外资兼并对企业出口和技术进步的影响:基于倾向指数再加权估计法[J]. 中国管理科学, 2014,22(12):34-40.
[35] 徐春骐,周建,徐伟宣.外商直接投资与我国三次产业技术进步相关关系研究[J]. 中国管理科学, 2005,13(2):118-123.
[36] Levinsohn J, Petrin A. Estimating production functions using inputs to control for unobservables[J]. Review of Economic Studies, 2003, 70(2):317-341.
[37] Rosenbaum P R, Rubin D B. Assessing sensitivity to an unobserved binary covariate in an observational study with binary outcome[J]. Journal of the Royal Statistical Society. Series B:Methodological, 1983, 45(2):212-218.
[38] 简泽,段永瑞.企业异质性、竞争与全要素生产率的收敛[J].管理世界,2012,(8):15-29.
[39] 王立勇,高玉胭. 财政分权与产业结构升级——来自"省直管县"准自然实验的经验证据[J].财贸经济,2018,39(11):145-159.