The effect of industrial robots’ adoption on urban income inequality in China

ABSTRACT

This paper empirically investigates the effect of the utilization of industrial robots on income inequality in urban China. By employing the data of imports of industrial robots and income inequality at the city level in China, we find that the adoption of robots has a positive effect on urban income inequality. This positive effect remains robust when we use instrumental variable estimation to tackle the potential endogeneity problem. In addition, this positive effect is more salient in cities with below average national income per capita. The results have important policy implications for achieving common prosperity in China.

KEYWORDS:

• Income inequality
• industrial robots
• common prosperity
• China

JEL CLASSIFICATION:

• D31
• D63
• O14
• O18

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Prettner and Strulik (2020) proposed a theoretical model of the race between education and automation. They find that transfers to low-skilled workers could dampen the economic growth rate due to its side effect on slowing down robot adoption and workers’ pursuing of education. As such, direct transfer to low-skilled workers, although could help alleviate income inequality, will have a negative effect on growth.

² This stands for the Harmonization System Code (HS-Code) for commodities. This proxy has been used in the literature to gauge the utilization of robots (Acemoglu and Restrepo 2021).

³ Because the surge of robot imports started since 2010 (Cheng et al. 2019; Fan, Hu, and Tang 2021), and at the same time import data after 2017 has yet to be released by the customs. As such, in this paper our time span is from 2010 to 2016.

⁴ Besides, domestic production of robot data is only provided in an aggregated form at the national or industry level.

⁵ This data has an obvious advantage, as customs data is administrative data, which is collected compulsorily by the government and provides more precise and detailed information compared to firm level survey data.

⁶ Fan, Hu, and Tang (2021) also used this data to proxy for robots adoption of firms in China. Nominal variables such as the import values of robots are deflated by provincial CPI to come up with real variables.

⁷ It may be a concern that if firms import through headquarters or trade firms as intermediaries, then robot imports may not reflect end users’ adoption of robots. This will cause problems if headquarters or trade firms are not located in the same city with the end users. The headquarter effect is not very severe, since the customs require the reporting firm to provide the information of where the robots will be used. For the trade firm effect, according to (Fan, Hu, and Tang 2021), the share of robots imported by trade intermediaries is less than 25% in the customs data. As such, the problem of importing through trade intermediaries is not severe either.

⁸ Because a few cities and provinces do not provide the five quintiles information of the disposable income of urban citizens, so we omit these cities and provinces in our dataset.

⁹ Based on the IFR statistics, the annual growth rate of newly installed robots is 37.73% from 2010 to 2017 in China (Wang and Dong 2020). The effect of robot adoption on income inequality seems to be quite small if we only consider the annual growth of robot adoption. If we consider the accumulated growth over 2010 to 2017, which is 3.0184, the estimated effect on income inequality will be a 6% increase of the income ratio between the highest and lowest income groups, which is not small.

¹⁰ For robustness check, we also performed a series of regressions in which we take log of the dependent variable of income inequality, and all of the control variables. The estimation results are broadly consistent with that of Table 2. These are not reported to conserve space, and is available upon request.

¹¹ This IV is similar as in (Bucher-Koenen and Lusardi 2011).

¹² However, governments should also be made aware of the negative side effects of such policies. As direct transfer to low-skilled labours will tend to discourage human capital accumulation and dampen robot adoption by firms, which will slow down the growth of the economy. Detailed discussion could be found in (Prettner and Strulik 2020).

Additional information

Funding

This work was supported by the Department of Education of Guangdong Province [2018WZDXM009]; National Natural Science Foundation of China [72073036]; Natural Science Foundation of Guangdong Province for Distinguished Young Scholars [2022A1515010238]; Ministry of Education of China, Humanities and social science research projects [20YJC790057].