Impact of adopting care robots on labor outcomes in long-term care: an observational study using a lottery-based natural experiment

Background Population aging increases long-term care (LTC) demand; however, worker shortage persists. Evaluating the impact of care robot adoption is essential for evidence-based policymaking. We used a natural experiment created by a lottery-based subsidy allocation to evaluate the effect of care robot adoption on labor outcomes in LTC facilities. Methods In this observational study, prefectural lottery records from 2021 were linked to Japan’s national LTC service disclosure system to construct a facility-level panel (2020–2023) of residential LTC providers in a metropolitan prefecture. The exposure was eligibility for a subsidy to adopt care robots, randomly assigned through lottery. Labor outcomes included the number of care workers, weekly hours worked, proportion of part-time staff, and number of night-shift care workers. Using 2020 as the pre-exposure baseline, we estimated within-facility changes and compared them between lottery-selected and non-selected facilities. Results Of the 200 facilities, 72 were lottery-selected, and 128 were not. Baseline key labor indicators were similar across the groups. Following care robot adoption, lottery-selected facilities reduced the number of care workers compared to non-selected facilities in 2021 (difference in change: −1.11; 95% CI − 2.20 to − 0.01). This difference became less pronounced in 2022 and 2023. Weekly hours worked increased slightly more in the exposure group, but the estimates were small and imprecise. The proportion of part-time staff decreased more in the exposure group over time, although the difference remained modest. Night-shift staffing levels remained stable in the exposure group and slightly increased in the control group, resulting in small between-group differences. The patterns suggest modest, directionally consistent labor adjustments after adoption. Conclusions Lottery-based eligibility for robot funding, used as a proxy for adoption, was associated with a short-term reduction in care worker headcount, suggesting the potential of substitution of robots for some care workers, whereas other labor changes were modest and imprecisely estimated. To precisely identify the mechanisms and heterogeneity, studies should incorporate granular information on the care robots.