Periodic Reporting for period 1 - HealthDeterminants (Investigating the Impact of Risk of Automation on Health Outcomes of Employees: Evidence from Germany)
Periodo di rendicontazione: 2023-10-01 al 2025-09-30
Automation transforms work at a rapid pace, with gradually increasing shares of the workforce at risk of being replaced by machines. However, little is known about how this risk is affecting workers. In line with the original proposal, this project advances the growing body of research on the social outcomes of automation by examining how exposure to high automation risk in the workplace influences various health outcomes among workers in Germany. It leverages a country-specific measure of routine task intensity to capture automation exposure - an approach that has been largely underutilized in previous studies - and exploits the longitudinal design of the German Socio-Economic Panel (GSOEP) to address the following research questions:
Q1. To what extent does automation risk affect the health outcomes of German workers?
Q2. What are the potential transmission mechanisms (mediating factors) that explain the health disparities among German workers exposed to varying levels of automation risk?
Q3. Is selection into occupations with higher automation risk influenced by individuals' pre-existing health conditions, thereby suggesting reverse causality?
Q4. Beyond the scope of the original proposal, the project also explores whether exposure to automation risk in the workplace contributes to adverse health behaviors (i.e. drinking patterns), applied to the context of workers in low- and middle-income countries.
Q1. To what extent does automation risk affect the health outcomes of German workers?
Q2. What are the potential transmission mechanisms (mediating factors) that explain the health disparities among German workers exposed to varying levels of automation risk?
Q3. Is selection into occupations with higher automation risk influenced by individuals' pre-existing health conditions, thereby suggesting reverse causality?
Q4. Beyond the scope of the original proposal, the project also explores whether exposure to automation risk in the workplace contributes to adverse health behaviors (i.e. drinking patterns), applied to the context of workers in low- and middle-income countries.
Work Package 1 - From Efficiency to Illness: Do Highly Automatable Jobs Take a Toll on Health in Germany? (Q1 & Q3)
• Brief summary of the work performed:
We investigated the relationship between high automation risk defined at the occupational level and both subjective and objective health outcomes of workers in Germany. We conducted a heterogeneity analysis, stratifying our sample by several demographic and occupational characteristics. We explored how our findings are influenced by the measurement of automation risk by varying the threshold that distinguishes highly automatable occupations from others (test for the dose-response relationship). We also conducted several other robustness checks (including the test for reverse causality via estimating the cross-lagged panel models) to assess whether our estimates are sensitive to various modeling choices and change once potential statistical issues are addressed.
• Outcome and results:
Our findings suggest that employment in occupations exposed to high automation risk is negatively related to self-reported health and health satisfaction, and is positively related to sickness absence and, depending on the model specification, to anxiety among German workers. No statistically significant effect is found for healthcare use. In addition, our heterogeneity analysis demonstrates that the estimates tend to vary slightly across several demographic and occupational characteristics. Further, we provide evidence that the way automation risk is measured may affect the magnitude somewhat, and, in some cases, the significance of the observed effects (the dose-response relationship is confirmed). However, the overall impact of the measurement approach on the conclusions remains moderate. Additionally, we find no evidence supporting the reverse causality hypothesis that workers’ health status determines their selection into highly automatable occupations. Finally, we estimate several alternative model specifications as robustness checks, with the findings being largely consistent with our baseline results.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper.
Work Package 2 - The Hidden Costs of Technological Change: Investigating Pathways Through Which Highly Automatable Jobs Undermine Workers’ Health in Germany (Q2)
• Brief summary of the work performed:
We performed a mediation analysis using the Karlson-Holm-Breen (KHB) method to test both the direct and indirect effects of automation on various health outcomes (self-reported health, anxiety, and the physical and mental component summary scores of the SF-12 Health Survey) of workers in Germany. Building on findings from the European context, we incorporated both broad mediators - such as concerns about job insecurity and one’s future financial situation - and more specific automation-related indicators, including worries about coping with technological change, the risk of occupational obsolescence, and anticipated changes in health risks, work productivity, and qualification demands. We also decomposed the estimated effects on aggregated health outcomes - specifically the physical and mental component summary scores from the SF-12 health survey - into their underlying domains: physical functioning, role physical, bodily pain, and general health (for physical health); and vitality, social functioning, role emotional, and mental health (for mental health). This allowed us to identify the specific health components most affected by automation exposure. Finally, we conducted several robustness and sensitivity checks by applying specific sample selection criteria and estimating alternative model specifications to ensure the reliability of our baseline findings.
• Outcome and results:
Our findings suggest that, in the German context, high automation risk tends to negatively affect workers' health primarily through indirect pathways - specifically, via economic uncertainty channels - rather than through direct effects (with a few exceptions). However, the contribution of these channels varies by gender: among male respondents, concerns about job insecurity and future financial situation are weighted almost equally, whereas for female respondents, only job insecurity appears to play a significant role. The decomposition by physical and mental health domains closely mirrors our baseline results while offering deeper insights into the specific sources of the effects observed in the aggregated health indices. Furthermore, our analysis reveals no significant evidence that the relationship between high automation risk and workers’ health outcomes in Germany is mediated by specific automation-related factors - such as one’s ability to cope with technological change, fear of occupational obsolescence, or expectations regarding changes in health risks, qualification demands, and work productivity. Overall, the pathways through which high automation risk affects workers’ health tend to be highly gender- and context-dependent, shaped by shifts in the economic environment and working conditions.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper.
Work Package 3 - The Glass is Half Empty: The Role of Highly Automatable Jobs in Shaping Drinking Behaviors in Russia (Q4)
• Brief summary of the work performed:
We investigated the relationship between high automation risk, measured at the occupational level, and drinking behaviors among workers in Russia. We examined whether job insecurity, perceived labor market unviability, and dissatisfaction with growth opportunities at work serve as potential pathways linking high automation risk to drinking behaviors, building on insights from prior studies conducted in high-income countries. We tested whether additional physical (e.g. manual handling, awkward postures, vibration exposure) and psychosocial (e.g. high job demands, limited autonomy, low social support from supervisors or colleagues) workplace stressors may act as moderating factors, amplifying the psychological strain associated with automation-related displacement risk, and thereby increasing the likelihood that workers will adopt certain coping mechanisms, such as alcohol consumption. We conduct a series of robustness and sensitivity checks to assess the consistency of our estimates across various model specifications and sample selection criteria.
• Outcome and results:
Our findings indicate that employment in highly automatable occupations is positively related to all examined drinking outcomes among women. In contrast, no statistically significant associations are observed among men. Furthermore, we find that dissatisfaction with growth opportunities at work mediates the relationship between automation risk and both active drinker status and alcohol consumption in the female subsample suggests that skills utilization, professional fulfilment, and career development appear to be of growing importance to Russian women, signalling a significant societal move toward more egalitarian values. Finally, we show that high job exposures (physical and psychosocial) are significant determinants of drinking behaviors among men in Russia. Notably, high physical exposures moderate the relationship between high automation risk and alcohol use among female workers, underscoring the presence of gender-specific occupational and sectoral segregation in the labor market.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper
• Brief summary of the work performed:
We investigated the relationship between high automation risk defined at the occupational level and both subjective and objective health outcomes of workers in Germany. We conducted a heterogeneity analysis, stratifying our sample by several demographic and occupational characteristics. We explored how our findings are influenced by the measurement of automation risk by varying the threshold that distinguishes highly automatable occupations from others (test for the dose-response relationship). We also conducted several other robustness checks (including the test for reverse causality via estimating the cross-lagged panel models) to assess whether our estimates are sensitive to various modeling choices and change once potential statistical issues are addressed.
• Outcome and results:
Our findings suggest that employment in occupations exposed to high automation risk is negatively related to self-reported health and health satisfaction, and is positively related to sickness absence and, depending on the model specification, to anxiety among German workers. No statistically significant effect is found for healthcare use. In addition, our heterogeneity analysis demonstrates that the estimates tend to vary slightly across several demographic and occupational characteristics. Further, we provide evidence that the way automation risk is measured may affect the magnitude somewhat, and, in some cases, the significance of the observed effects (the dose-response relationship is confirmed). However, the overall impact of the measurement approach on the conclusions remains moderate. Additionally, we find no evidence supporting the reverse causality hypothesis that workers’ health status determines their selection into highly automatable occupations. Finally, we estimate several alternative model specifications as robustness checks, with the findings being largely consistent with our baseline results.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper.
Work Package 2 - The Hidden Costs of Technological Change: Investigating Pathways Through Which Highly Automatable Jobs Undermine Workers’ Health in Germany (Q2)
• Brief summary of the work performed:
We performed a mediation analysis using the Karlson-Holm-Breen (KHB) method to test both the direct and indirect effects of automation on various health outcomes (self-reported health, anxiety, and the physical and mental component summary scores of the SF-12 Health Survey) of workers in Germany. Building on findings from the European context, we incorporated both broad mediators - such as concerns about job insecurity and one’s future financial situation - and more specific automation-related indicators, including worries about coping with technological change, the risk of occupational obsolescence, and anticipated changes in health risks, work productivity, and qualification demands. We also decomposed the estimated effects on aggregated health outcomes - specifically the physical and mental component summary scores from the SF-12 health survey - into their underlying domains: physical functioning, role physical, bodily pain, and general health (for physical health); and vitality, social functioning, role emotional, and mental health (for mental health). This allowed us to identify the specific health components most affected by automation exposure. Finally, we conducted several robustness and sensitivity checks by applying specific sample selection criteria and estimating alternative model specifications to ensure the reliability of our baseline findings.
• Outcome and results:
Our findings suggest that, in the German context, high automation risk tends to negatively affect workers' health primarily through indirect pathways - specifically, via economic uncertainty channels - rather than through direct effects (with a few exceptions). However, the contribution of these channels varies by gender: among male respondents, concerns about job insecurity and future financial situation are weighted almost equally, whereas for female respondents, only job insecurity appears to play a significant role. The decomposition by physical and mental health domains closely mirrors our baseline results while offering deeper insights into the specific sources of the effects observed in the aggregated health indices. Furthermore, our analysis reveals no significant evidence that the relationship between high automation risk and workers’ health outcomes in Germany is mediated by specific automation-related factors - such as one’s ability to cope with technological change, fear of occupational obsolescence, or expectations regarding changes in health risks, qualification demands, and work productivity. Overall, the pathways through which high automation risk affects workers’ health tend to be highly gender- and context-dependent, shaped by shifts in the economic environment and working conditions.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper.
Work Package 3 - The Glass is Half Empty: The Role of Highly Automatable Jobs in Shaping Drinking Behaviors in Russia (Q4)
• Brief summary of the work performed:
We investigated the relationship between high automation risk, measured at the occupational level, and drinking behaviors among workers in Russia. We examined whether job insecurity, perceived labor market unviability, and dissatisfaction with growth opportunities at work serve as potential pathways linking high automation risk to drinking behaviors, building on insights from prior studies conducted in high-income countries. We tested whether additional physical (e.g. manual handling, awkward postures, vibration exposure) and psychosocial (e.g. high job demands, limited autonomy, low social support from supervisors or colleagues) workplace stressors may act as moderating factors, amplifying the psychological strain associated with automation-related displacement risk, and thereby increasing the likelihood that workers will adopt certain coping mechanisms, such as alcohol consumption. We conduct a series of robustness and sensitivity checks to assess the consistency of our estimates across various model specifications and sample selection criteria.
• Outcome and results:
Our findings indicate that employment in highly automatable occupations is positively related to all examined drinking outcomes among women. In contrast, no statistically significant associations are observed among men. Furthermore, we find that dissatisfaction with growth opportunities at work mediates the relationship between automation risk and both active drinker status and alcohol consumption in the female subsample suggests that skills utilization, professional fulfilment, and career development appear to be of growing importance to Russian women, signalling a significant societal move toward more egalitarian values. Finally, we show that high job exposures (physical and psychosocial) are significant determinants of drinking behaviors among men in Russia. Notably, high physical exposures moderate the relationship between high automation risk and alcohol use among female workers, underscoring the presence of gender-specific occupational and sectoral segregation in the labor market.
• Achievement of scientific deliverables and milestones:
All related deliverables (i.e. literature review, STATA codes, descriptive statistics in the form of tables and graphs, summary of the data, methods and empirical evidence) are provided within or supplied with the pre-print version of the paper
The project generated meaningful impacts for both academia and society. From an academic perspective, it promoted gender-specific multidisciplinary research by integrating methodologies and modelling approaches from related social sciences (labor economics, health economics, sociology, demography) which include but are not limited to longitudinal data analysis, heterogeneity analysis, mediation analysis, IV regression models, cross-lagged models etc. It also showed how to effectively leverage diverse data sources (e.g. nationally representative survey data, administrative databases, expert estimations etc.) to address contemporary research challenges and to enrich the evidence for relatively novel topics. Furthermore, the project contributed to the ongoing discourse on methodological approaches to measuring automation risk in the workplace, while also assessing the reliability and robustness of these measures across diverse occupational contexts. From an economic and policy standpoint, the project advanced understanding of the social costs of automation by identifying the channels through which automation risk may adversely affect workers’ health and offering valuable insights to policymakers, companies, and employers on how to mitigate its potential negative spillover effects on the workforce. From a societal perspective, the project highlighted emerging public health risks associated with automation and underscored the potential burden this may place on healthcare systems, particularly in the absence of proactive policy measures targeted at the most vulnerable demographic groups.
The project can potentially contribute to the following EU headline priorities: “Economy that works for people” and “Europe for the digital age”.
Potential users of the project include but are not limited to:
1. Labor economists who deal with topics of labor market polarization, routine-biased technological change and the future of work, task-based composition of occupations, labor dynamics, job insecurity, gender inequality in the labor market etc.
2. Health economists who deal with sociodemographic and work-related determinants of public health, health dynamics of the population, subjective and objective measures of health, topics on well-being, gender differences in health outcomes etc.
3. Scientists from related disciplines (e.g. sociologists, demographers, epidemiologists) who are interested in labor market and public health research.
4. Policy-makers and consulting agencies working in the area of public health and labor market regulations.
5. Employers who aim at improving productivity, decreasing costs and attaining sustainable growth and development of their companies.
6. Current employees, recent graduates and those who decide on their future career trajectories.
The project results are summarized in the following pre-prints:
Vasiakina, M., & Dudel, C. (2025). From efficiency to illness: Do highly automatable jobs take a toll on health in Germany? https://doi.org/10.2139/ssrn.5209197(si apre in una nuova finestra).
Vasiakina, M., & Dudel, C. (2025). The Hidden Costs of Technological Change: Investigating Pathways Through Which Highly Automatable Jobs Undermine Workers’ Health in Germany. MPIDR Working Paper WP-2025-032. https://doi.org/10.4054/MPIDR-WP-2025-032(si apre in una nuova finestra).
Vasiakina, M., & Dudel, C. (2025). The glass is half empty: The role of highly automatable jobs in shaping drinking behaviors in Russia (MPIDR Working Paper WP 2025‑027). https://doi.org/10.4054/MPIDR‑WP‑2025‑027(si apre in una nuova finestra).
The project can potentially contribute to the following EU headline priorities: “Economy that works for people” and “Europe for the digital age”.
Potential users of the project include but are not limited to:
1. Labor economists who deal with topics of labor market polarization, routine-biased technological change and the future of work, task-based composition of occupations, labor dynamics, job insecurity, gender inequality in the labor market etc.
2. Health economists who deal with sociodemographic and work-related determinants of public health, health dynamics of the population, subjective and objective measures of health, topics on well-being, gender differences in health outcomes etc.
3. Scientists from related disciplines (e.g. sociologists, demographers, epidemiologists) who are interested in labor market and public health research.
4. Policy-makers and consulting agencies working in the area of public health and labor market regulations.
5. Employers who aim at improving productivity, decreasing costs and attaining sustainable growth and development of their companies.
6. Current employees, recent graduates and those who decide on their future career trajectories.
The project results are summarized in the following pre-prints:
Vasiakina, M., & Dudel, C. (2025). From efficiency to illness: Do highly automatable jobs take a toll on health in Germany? https://doi.org/10.2139/ssrn.5209197(si apre in una nuova finestra).
Vasiakina, M., & Dudel, C. (2025). The Hidden Costs of Technological Change: Investigating Pathways Through Which Highly Automatable Jobs Undermine Workers’ Health in Germany. MPIDR Working Paper WP-2025-032. https://doi.org/10.4054/MPIDR-WP-2025-032(si apre in una nuova finestra).
Vasiakina, M., & Dudel, C. (2025). The glass is half empty: The role of highly automatable jobs in shaping drinking behaviors in Russia (MPIDR Working Paper WP 2025‑027). https://doi.org/10.4054/MPIDR‑WP‑2025‑027(si apre in una nuova finestra).