- Studies of service sector robots in Japan suggest they have increased some employment opportunities, and helped to mitigate turnover problems among long-term care workers.
- Experts believe robots could alleviate challenges that face both the ageing population and those working in the care sector.
- However, more research is needed to fully understand the implications of artificial intelligence on wages, quality of service and other key factors.
Robots hold polar extremes in economic narrative and popular imagination. One narrative depicts a looming dystopian future with robots and other forms of automation increasingly replacing human workers, depressing wages (Brynjolfsson and McAfee 2014), feeding inequality, and contributing to further ‘deaths of despair’ (Case and Deaton 2020, Mulligan 2021). In counter-imaginations, robots embody innovative technology spurring productivity and freeing workers from repetitive, strenuous, monotonous work while helping to relieve labour shortages arising from ageing populations. Such demographic challenges are salient particularly in higher-income countries farther along in the demographic transition, such as the OECD nations, where populations in 18 out of the 36 countries are projected to decline by 2055. These nations face rising old-age dependency ratios, declining employment-to-population ratios, and challenges in providing services to the growing number of frail older adults.
Demography and the duelling narratives about robots
Indeed, demography can explain substantial differences in development and diffusion of robotics and automation technologies (Acemoglu and Restrepo 2018, Prettner and Bloom 2020). Even in the younger US, Varian (2020) posits that reduced labour supply from population ageing will offset the reduction in demand from automation for many years to come (after labour markets heal from the current pandemic). The rosier narrative about robotics may also encompass their use in hospitals, nursing homes, and other care settings as complements to telemedicine and physical distancing to protect frail populations during a pandemic like COVID-19 or in future seasonal influenza epidemics.
Several empirical studies have corroborated aspects of the first, negative view, including evidence that robots reduce manufacturing employment and wages (e.g. Acemoglu and Restrepo 2020, Dauth et al. 2017, Dixon et al. 2019, Bessen 2019). Yet evidence from the service sector remains scant, especially firm-level studies that go beyond anecdote to probe the impact of robots used in providing services that ageing populations increasingly need, like long-term care.
Have you read?
Learning from an early adopter
Japan’s experiences may be especially instructive, given its declining overall population, increasing proportion of seniors, and aversion to large-scale immigration, alongside technological prowess in many aspects of robotics and automation. Despite recognition that robots may be a poor substitute for many tasks demanding empathy and dexterity in the caring professions, Japan has been an early adopter of robots to address the shortage of care workers relative to growing demand for long-term care services, including assistance with basic activities of daily living such as eating, toileting, and bathing (see Figure 1). Official projections indicate a shortfall of 380,000 care workers by 2025 (MHLW 2017), in part because care workers often experience physical repercussions such as lower back pain, while receiving wages barely exceeding the minimum wage.
Figure 1 Demand for long-term care (LTC) in Japan
In response, Japan has actively promoted the development and use of robots in long-term care. The national government ‘Robot Plan’ aims to increase the share of people who want to use robots for providing care from 60% to 80%. Local governments also provide subsidies to adopt robots in nursing homes; prefectures typically subsidise 50% of the cost of robot adoption up to 100,000 yen (approximately US$1,000) per robot (METI 2015). As of FY2018, 36 of 47 prefectures offer such subsidies.
Figure 2 Subsidies for nursing care robots
The average nursing home employed 42 care workers, 8 nurses, and 80 total staff, the majority (66%) of whom are regular (generally full-time, monthly wage) employees. About one-third of the nursing homes reported that retention of staff is a problem.
The share of Japanese nursing homes that reported using any type of robot rose from 17.6% in 2016 to 26% in 2017. Monitoring robots, which help monitor whether individuals have got out of bed, fallen, or need assistance were the most common type, reported in use by 14.9% of nursing homes. Other common types were transfer aid robots (7.7%) to assist care workers with moving individuals; mobility robots (5.3%) to assist residents with movement, toileting, and bathing; and communication robots (2.8%) to provide comfort and interaction.
Examining the relationship between robot adoption and nursing home staffing, we find that robot-adopting nursing homes had between 3% and 8% more staff than their non-adopting counterparts. Nursing homes with robots also appeared to have higher management quality; they were 10% more likely to have a human resource manager and were more likely to report that they make effort to improve wages for retention of employees.
To identify the causal impact of robots on staffing, we use prefecture subsidies as an instrumental variable for the adoption of robots. The planned number of robots per nursing home in 2017 proved to be a significant predictor of robot adoption (Figure 3). Contrary to popular concern about job replacement, robot adoption bolstered employment, both among care workers and nurses.
Figure 3 Subsidies for nursing care robots predict robot adoption, 2017
Importantly, the increases in staffing occurred entirely among the non-regular employees. Robot adoption doubled the number of non-regular care workers and significantly increased the number of non-regular nurses. The estimates on regular employees were negative but statistically insignificant.
For Japan, the use of foreign employees is especially of interest as policies have long discouraged much immigration but have just recently begun to be relaxed, in part to relieve the forecasted shortfall of LTC workers. Nursing homes that adopt robots are more likely to have hired foreign workers, and to have plans to hire immigrants in the future, but we find that these associations are not causal.
Regarding wages, analyses suggest that robot adoption reduced the monthly wages of nurses, especially of regular nurses, by a modest but nontrivial amount. The estimates for care-worker wages are also negative but smaller, likely because care-worker wages are only slightly above the minimum wage, and this wage floor limits how much wages can decrease. The reduction in nurse monthly wages could reflect reduction in caregiver burden during night shifts, since monitoring robots are designed to substitute for tasks such as frequent night-time rounds to monitor residents’ wellbeing. It may also be due to more nurses shifting to part-time work, which has been encouraged by the Japanese government and the Ministry of Health, Labor and Welfare as a strategy to increase worker flexibility.
Finally, robot adoption reduced the likelihood that the nursing home considered retention problematic, which suggests that robots may indeed help reduce the burden on care workers and nurses.
Thus, one of the first studies of service sector robots suggests that robot adoption has increased employment opportunities for non-regular care workers, helped to mitigate the turnover problem that plagues nursing homes, and provided greater flexibility for workers (Eggleston et al. 2021). Such evidence suggests that the wave of technologies that inspires fear in many countries could help remedy the social and economic challenges posed by population aging in others. Moreover, governments will likely step in to moderate or regulate any potential adverse socio-economic impacts of new technologies (Lee et al. 2019). Since we are currently still in the early phase of robot diffusion in the service sector, researchers and policymakers need to continue to monitor and assess the extent to which robots complement or augment some types of labour while substituting for others, and the implications for wages, quality of services, productivity, and broader measures of social welfare.