WANG Yongqin;DONG Wen
The rise of AI and robots has been changing the economy and society. While these technologies have boosted productivity and economic growth, they have increasingly replaced workers in every industry and have posed unprecedented challenges to labor markets. Then, as the world’s top user of robots, China has an important question to answer: how have robots affected China’s labor markets? Currently, the existing empirical studies primarily focus on developed countries, and there is scant evidence from developing countries. Although there are some theoretical discussions on the rise of AI and robots in China and its economic consequences, a systematic and in-depth empirical research is still yet to come. This paper is to fill this void. To our knowledge, this paper is the first to study the effect of robots on China’s labor market using micro-data and among the first to study the effects of robots worldwide. The paper used data on China’s listed manufacturing firms and robot usage data from the International Federation of Robotics (IFR). It employed the Bartik IV to establish causality. The paper further explores potential mechanisms. To measure the exposure of firms to robots, we constructed a variable, industrial robot penetration. It is a Bartik-like instrumental variable, defined as the interaction between the industry-level density of robots and the weight of the firm in the industry. The weight was defined as the ratio of the employment share of production workers of the firm to the median of the employment shares of all of the manufacturing firms in the baseline period. To address endogeneity problems, we instrumented China’s industrial robot penetration using an analogous measure constructed from the industry-level density of robots in the United States. This allowed us to causally estimate the impact of exogenous improvements in technology on China’s labor markets. A priori, there are three main effects of robots on labor markets, which have different directional implications for labor markets: the substitution effect, the productivity effect, and the job creation effect. The ultimate effects are an empirical question. The results showed that there was a substitution effect between robot penetration and labor demand: a 1% increase in penetration reduced labor demand by 0.18%. The results also showed that there was substantial heterogeneity among workers with different skills, resulting in a job polarization effect. The adoption of robots mostly reduced the demand for middle-skill workers (i.e., those with bachelor’s and associate’s degrees): the elasticity of substitution was ?0.27 and ?0.44, respectively, for these two groups. There was a crowding-in effect for low-skill (high school degree or lower education) workers, but no significant effect for high-skill (graduate) workers. In addition, robots have a little effect on wages. One possible explanation for this result was wage rigidity. In terms of channels, the substitution effect was more salient in more concentrated industries (market power channel), industries with more credit constraints (credit constraint channel), and private firms vis-à-vis state-owned firms (resource constraint channel). The paper further explored the network effect of robots on labor markets. The results showed that the adoption of robots will have an effect on labor demand of upstream and downstream firms through input-output linkages. These results have important implications for China’s labor market and policy making. In particular, relevant policies should be considered with competition policies, financial policies, and social insurance policies together. To be more specific, firstly, the findings suggested that there was a substitution effect between robot penetration and labor demand, especially for middle-skill workers. In this regard, improving the social security system, especially increasing the coverage of unemployment insurance and strengthening social security for informal sector workers, may help mitigate the negative shocks caused by robots while exploiting the overall productivity gains. Secondly, because AI and robots also create new jobs, job training matters for both efficient and inclusive economic growth in the era of AI. Finally, implementing fair competition policies and easing financial constraints on firms can also help China create more employment to promote high-quality development in the brave new world of AI.
AI and robots;labor market;job polarization
To explore the background and basis of the node document