Are They Coming for Us? Industrial Robots and the Mental Health of Workers

Abstract

We investigate how an increase in the robot intensity (the ratio of industrial robots over employment) affects the self-reported mental health of workers in Germany. To do so, we combine individual mental health data from the German Socioeconomic Panel with the deliveries of robots to 21 German manufacturing sectors provided by the International Federation of Robotics for the period 2002-2014 (every two years). Controlling for a range of individual and sectoral characteristics, and employing individual-, time-and sectoral fixed effects, we find that an increase in robot intensity of 10% is associated with an average decrease of 0.59% of the average mental health standard deviation. This suggests that in a fast automating sector (i.e. rubber and plastics), where the robot intensity increased by approximately 2000%, mental health would have decreased by 118% of one standard deviation. This effect seems to be driven by job security fears of individuals working in noninteractive jobs and the fear of a decline in an individual's economic situation. Moreover, further sample divisions into low, middle and high occupational groups shows that the negative effects are affecting mostly the middle-level occupational group. Splitting the sample according to different age groups shows that the mental health of younger workers is the most vulnerable to an increase in automation. Results are also robust to instrumenting the stock of robots, and to different changes in the sample.

Automation, workers’ skills and job satisfaction

“…To address endogeneity concerns, we follow recent research and use the lagged introduction of robots in local labor markets (industry-regions) as an instrument for individual workers' fear of replacement [11,12]. Existing studies on robot implications for labor markets where robot adoption is the main explanatory variable address endogeneity issues by incorporating spillover effects from robot adoption across industries in other countries as an instrument in a 2SLS setup [22,34,36].…”

“…Two recent papers explore the relationship between automation and workers' well-being. Abeliansky and Beulmann [11] present an empirical study on the impact of automation on the mental health of workers (which is one important dimension reflecting stress and weak job satisfaction). The analysis uses individual-level data from the German Socioeconomic Panel for the period 2002-2014 linked to industry-level data on use of industrial robots in 21 manufacturing sectors in Germany.…”

“…Robot stock for years 2018 and 2019 are extrapolated assuming a 9 percent annual growth in operational stock as projected by IFR [33]. IFR data have recently been used to analyze the impact of automation on employment and wages [22,34,35], as well as on workers' well-being [11,12]. We allocate robots in regional labor markets following extant research [22,34,36], assuming that robots are distributed across region and industries by their respective employment shares.…”

“…On the other hand, the corresponding marginal effect for the workers with a University degree is not statistically significant. Abeliansky and Beulmann [11] also carried out some regressions to study the relationships between automation and mental health for different educational groups (tertiary vs secondary education), but they did not find any significant difference among these groups of workers in Germany. As discussed in section 3,…”

“…In spite of the relevance of this topic, to the best of our knowledge only two papers have recently explored the relationship between automation and workers' well-being. Abeliansky and Beulmann [11] focuses on workers' mental health in Germany; and Schwabe [12] studies workers' life satisfaction in a sample of European countries. Neither of these studies, though, investigates explicitly the impacts of automation on job satisfaction.…”