County-level job automation risk and health: Evidence from the United States

doi:10.1016/j.socscimed.2018.02.025

Social Science & Medicine

Volume 202, April 2018, Pages 54-60

https://doi.org/10.1016/j.socscimed.2018.02.025 Get rights and content

Highlights

•
Exposure to automation risk could exacerbate health outcomes.
•
Using General Social Survey data, automation risk.→job insecurity→poorer health is supported
•
Using county-level data, automation risk lowers general, physical, and mental health.

Abstract

Rationale

Previous studies have observed a positive association between automation risk and employment loss. Based on the job insecurity-health risk hypothesis, greater exposure to automation risk could also be negatively associated with health outcomes.

Objective

The main objective of this paper is to investigate the county-level association between prevalence of workers in jobs exposed to automation risk and general, physical, and mental health outcomes.

Methods

As a preliminary assessment of the job insecurity-health risk hypothesis (automation risk → job insecurity → poorer health), a structural equation model was used based on individual-level data in the two cross-sectional waves (2012 and 2014) of General Social Survey (GSS). Next, using county-level data from County Health Rankings 2017, American Community Survey (ACS) 2015, and Statistics of US Businesses 2014, Two Stage Least Squares (2SLS) regression models were fitted to predict county-level health outcomes.

Results

Using the 2012 and 2014 waves of the GSS, employees in occupational classes at higher risk of automation reported more job insecurity, that, in turn, was associated with poorer health. The 2SLS estimates show that a 10% increase in automation risk at county-level is associated with 2.38, 0.8, and 0.6 percentage point lower general, physical, and mental health, respectively.

Conclusion

Evidence suggests that exposure to automation risk may be negatively associated with health outcomes, plausibly through perceptions of poorer job security. More research is needed on interventions aimed at mitigating negative influence of automation risk on health.

Introduction

In recent decades, workers in the United States have faced a growing sense of uncertainty about their future employment (Kalleberg and Marsden, 2013). Automation is an important aspect of the changing nature of work in the US, and it poses a significant threat to the job security of workers. While estimates of potential job losses due to automation vary for the US population, with high estimates topping 47% (Frey and Osborne, 2017), most would agree that automation risk is significant and growing. While most automation studies focus on labor market impacts (Autor, 2015; Ford, 2015; Frey and Osborne, 2017; Hicks and Devaraj, 2015), other potential impacts remain largely unexplored, including the effect of automation risk on health outcomes.

Our theoretical basis is the job insecurity-health risk hypothesis (De Witte et al., 2016), and our preliminary model tests whether higher automation risk results in worse health outcomes with perceived job insecurity as a mediator: automation risk → job insecurity → poorer health. Related to the automation risk fueling job insecurity (automation risk → job insecurity), automation fuels fear and anxiety of job loss (Reichert and Tauchmann, 2017). Expectations of reduced wages and higher unemployment from automation increase perceptions of job insecurity. Automation may compound anxiety over job insecurity (Chui et al., 2015; Frey and Osborne, 2017). Related to job insecurity leading to poorer health (job insecurity → poorer health), in a recent review of 57 longitudinal studies on job insecurity and health/well-being outcomes, De Witte et al. (2016) conclude that “job insecurity affects health and wellbeing on the long term, rather than the other way round” (pp 18). Other studies have found that anxiety associated with prospects of job loss deteriorates mental health and wellbeing (Khubchandani and Price, 2017; Reichert and Tauchmann, 2017), and significantly increases negative mental health indicators, including depression (Burgard et al., 2009; Meltzer et al., 2010; Strazdins et al., 2011) and exhaustion/burnout (De Witte et al., 2016). Additionally, automation and the corresponding expectation of reduced wages and higher unemployment increases perceptions of job insecurity, and fuels fear and anxiety of job loss (Reichert and Tauchmann, 2017). Overall, we expect that the concerns of exposure to automation may induce job insecurity, which in turn, negatively influences health outcomes.

While the core analytic focus of this paper is on the county-level association between the prevalence of workers exposed to automation risk and health outcomes, and given that job insecurity data is not available at the county-level, to provide baseline support for job insecurity-health risk hypothesis, a mediation effect (automation risk → job insecurity → poorer health) is tested using the two waves of General Social Survey (GSS, 2012 and 2014).

In the second step, based on the preliminary support for job insecurity-health risk hypothesis from GSS data, this paper investigates whether the higher prevalence of workers exposed to automation risk at the county-level is negatively associated with county-level health outcomes. The proposed county-level association could be explained as follows: the anticipation of job loss to automation threatens fundamental benefits provided by a job – the need for survival, social relatedness, and self-determination (Blustein, 2008); automation risk reduces wages, fuels job losses, exacerbates family stress, and promotes social withdrawal. In these indirect ways, the negative social and economic environment created by the reduced quality of life could drive county-level prevalence of negative health. The actual and felt threats from automation may not immediately manifest into morbidities, but the increasing prevalence of poorer self-reported health and feelings of deteriorating physical and mental health can have a direct and lasting impact on individuals, families, and communities. While we cannot fully unpack the black box between county-level automation risk and health, nevertheless, it is important for policymakers to understand the health effects of automation risk.

To our knowledge, this is the first study exploring the association between automation risk in a region and health. We find support for a negative association between county-level automation risk and general, physical, and mental health. A 10-percentage point increase in automation risk at the county-level worsens general, physical, and mental health by 2.38 percentage point, 0.8 percentage point, and 0.6 percentage point, respectively.

Section snippets

An exploratory test of job insecurity-health risk hypothesis

Based on the job insecurity-health risk hypothesis, we assess whether people in occupations exposed to higher automation risk report more job insecurity and whether that, in turn, is associated with negative health outcomes.

We use the General Social Survey 2012, 2014 cross-sections, and match the individual occupational membership codes to the Frey and Osborne (2017) study on the probability of occupational automation. The descriptions of the variables and definitions are listed in Online

Study sample

We merge three datasets – County Health Rankings (CHR) 2017, Frey and Osborne's (2017) occupational automation probabilities aggregated at the county-level using American Community Survey (ACS) 2015, and Statistics of US businesses 2014.

The CHR 2017 rankings are based on county-level data during 2013 and 2015, the period around Frey and Osborne's (2017) measure of occupational automation risk.

Using the occupation level probabilities identified in Frey and Osborne (2017), Devaraj et al. (2017)

Conclusion

This study finds a negative association between county-level automation risk and county-level general, physical, and mental health. Consistent with job insecurity-health risk hypothesis, and validating that our inferences are not subject to ecological fallacy resulting from focusing on county-level data, participants in GSS (2012 and 2014 cross-sectional waves) who were in occupations facing greater automation threat reported greater job insecurity, which was, in turn, associated with

Conflicts of interest

The study is not funded by any funding source and the authors have no potential conflicts of interest amongst them or with any other external organizations.

References (18)

S.A. Burgard et al.
Perceived job insecurity and worker health in the United States
Soc. Sci. Med.
(2009)
C.B. Frey et al.
The future of employment: how susceptible are jobs to computerisation?
Technol. Forecast. Soc. Change
(2017)
A.L. Kalleberg et al.
Changing work values in the United States, 1973–2006
Soc. Sci. Res.
(2013)
A.R. Reichert et al.
Workforce reduction, subjective job insecurity, and mental health
J. Econ. Behav. Organ.
(2017)
D.H. Autor
Why are there still so many jobs? The history and future of workplace automation
J. Econ. Perspect.
(2015)
D.H. Autor et al.
The China syndrome: local labor market effects of import competition in the United States
Am. Econ. Rev.
(2013)
D.L. Blustein
The role of work in psychological health and well-being: a conceptual, historical, and public policy perspective
Am. Psychol.
(2008)
R. Chetty et al.
Where is the land of opportunity? The geography of intergenerational mobility in the United States
Q. J. Econ.
(2014)
M. Chui et al.
Four fundamentals of workplace automation
McKinsey Q.
(2015)

There are more references available in the full text version of this article.

Cited by (30)

Robots, meaning, and self-determination
2024, Research Policy
This paper is the first to examine the impact of robotization on work meaningfulness, autonomy, competence, and relatedness, which are essential to motivation and well-being at work. Using surveys of workers and robotization data for 14 industries in 20 European countries spanning 2005–2021, we find a consistent negative impact of robotization on perceived work meaningfulness and autonomy. Using instrumental variables, we find that doubling robotization leads to a 0.9 % decrease in work meaningfulness and a 1 % decline in autonomy. To put this in perspective, if the robotization levels of the top 5 industry were to match those of the leading industry in terms of robot adoption in 2020 (equivalent to a 7.5-fold increase), it would result in a decline of 6.8 % in work meaningfulness and 7.5 % in autonomy. The link between robotization, competence, and relatedness is also negative but less robust. We also examine how tasks, skills, and socio-demographic characteristics moderate the main relationships. We find that workers with routine tasks experience an even greater negative effect of robotization in terms of declines in their autonomy, competence, and relatedness. However, we also discover that utilizing computers as tools for independent work can help workers maintain a sense of autonomy, competence, and relatedness in industries and job roles that adopt robots. Our results highlight that by deteriorating work meaningfulness and self-determination, robotization can impact work life above and beyond its consequences for employment and wages.
Are they coming for us? Industrial robots and the mental health of workers
2024, Research Policy
How does the increasing use of robots affect the mental health of workers? To investigate this question, we combine individual mental health data from the German Socioeconomic Panel with data on the stock of robots in 14 manufacturing sectors provided by the International Federation of Robotics for the period 2002–2018. Using mediation analysis and an instrumental variable approach, we find that higher robot intensity is associated with deteriorating mental health, an effect that is mainly driven by worries about job security and a lower sense of achievement on the job. A heterogeneity analysis reveals that higher robot intensity has particularly severe negative effects on the mental health of workers close to retirement, in low-skilled occupations and performing routine jobs. Women and men are affected similarly, as are workers of all educational levels. Our results indicate the presence of hidden (health) costs of automation that policymakers need to address.
Pension reforms, longer working horizons and depression. Does the risk of automation matter?
2023, Labour Economics
We investigate the effect of postponing minimum retirement age on middle-aged workers’ depression. Using pension reforms in several European countries and data from the SHARE survey, we find that depression increases with a longer work horizon, but only among workers in occupations with a relatively high risk of automation. We explain our results with the higher job insecurity associated with occupations that are more exposed to automation, and rule out alternatives, including pension wealth effects and the differential exposure of occupations to the business cycle.
People versus machines: The impact of being in an automatable job on Australian worker's mental health and life satisfaction
2022, Economics and Human Biology
Citation Excerpt :
There are two studies which have explored the relationship between job automation risk and mental health outcomes, both of which find a negative relationship. ( Patel et al., 2018) and Abeliansky and Beulmann (2019) for the US and Germany respectively. As such, there is opportunity for further research to explore whether more sizeable effects are found in occupations which are most easily codified.
This study explores the effect on mental health and life satisfaction of working in an automatable job. We utilise an Australian panel dataset (HILDA), and take a fixed effects linear regression approach, to relate a person being in automatable work to proxies of their wellbeing. Overall, we find evidence that automatable work has a small, detrimental impact on the mental health and life satisfaction of workers within some industries, particularly those with higher levels of job automation risk, such as manufacturing. Furthermore, we find no strong trends to suggest that any particular demographic group is disproportionately impacted across industries. These findings are robust to a variety of specifications. We also find evidence of adaptation to these effects after one-year tenure on the job, indicating a limited role for firm policy.
The impact of automation and artificial intelligence on worker well-being
2021, Technology in Society
Citation Excerpt :
The last hypothesis focuses on anticipatory or forward-looking effects. On this view, workers aware of automation risk may experience fears of rapid job change or especially displacement [90,97], inducing feelings of job insecurity [79], increasing stress, and worsening health [85]. The link between job insecurity and negative well-being outcomes is not novel in the psychological literature: De Witte, Pienaar, and De Cuyper [86] found in a 30-year meta-analysis that job insecurity negatively impacts both worker psychological well-being and somatic health.
Discourse surrounding the future of work often treats technological substitution of workers as a cause for concern, but complementarity as a good. However, while automation and artificial intelligence may improve productivity or wages for those who remain employed, they may also have mixed or negative impacts on worker well-being. This study considers five hypothetical channels through which automation may impact worker well-being: influencing worker freedom, sense of meaning, cognitive load, external monitoring, and insecurity. We apply a measure of automation risk to a set of 402 occupations to assess whether automation predicts impacts on worker well-being along the dimensions of job satisfaction, stress, health, and insecurity. Findings based on a 2002–2018 dataset from the General Social Survey reveal that workers facing automation risk appear to experience less stress, but also worse health, and minimal or negative impacts on job satisfaction. These impacts are more concentrated on workers facing the highest levels of automation risk. This article encourages new research directions by revealing important heterogeneous effects of technological complementarity. We recommend that firms, policymakers, and researchers not conceive of technological complementarity as a uniform good, and instead direct more attention to mixed well-being impacts of automation and artificial intelligence on workers.
The future of labor unions in the age of automation and at the dawn of AI
2021, Technology in Society
Citation Excerpt :
Moreover, displaced workers may find that retraining, re-skilling, and re-educating themselves to enter new industries require time and money, which might prove too costly for them [51]. The risk of job loss may also cause increased physical and mental health issues [61], loss of self-esteem and loss of confidence in their professional aptitude [62], as well as degraded family relationships and even family breakups [63,64]. People who lose their job due to technological advancement might also experience difficulties in maintaining their social status and self-value as their assets and finances decline [51].
The COVID-19 crisis has accelerated an already-ongoing process of massive digitalization in economic production and services. AI and robotics are getting, for the first time, autonomous and self-learning, with human-like capabilities. The discussion about digitalization and the future of work has become even more imperative. So far, labor unions were the leading institutions representing employees. However, the rising possibility of human substitution by intelligent machines puts in question the feasibility of labor unions’ policies. This development undermines their traditional power sources, which depend on the membership of masses of paid workers and on their ability to stop production. In this context, this paper aims to discuss the challenges confronting unions in capitalist democracies. Most scholarly literature on labor relations has embraced the assumption that the digital revolution will eventually bring new, better jobs. We suggest considering an alternative scenario, namely, a digital revolution that causes mass replacement of human workers and structural, technological unemployment, which might expand our point of view, particularly for designing public policy. We suggest that unions now have two crucial roles. The first is to safeguard workers' rights and interests in the transition from an economy based on paid labor to an economy based on automated-autonomous production; and second, they should transform their primary calling from representing employees to representing the social rights of all citizens, and particularly the material interests of lay people.

View all citing articles on Scopus

View full text

Short communicationCounty-level job automation risk and health: Evidence from the United States

Highlights

Abstract

Rationale

Objective

Methods

Results

Conclusion

Introduction

Section snippets

An exploratory test of job insecurity-health risk hypothesis

Study sample

Conclusion

Conflicts of interest

Soc. Sci. Med.

Technol. Forecast. Soc. Change

Soc. Sci. Res.

J. Econ. Behav. Organ.

Why are there still so many jobs? The history and future of workplace automation

J. Econ. Perspect.

The China syndrome: local labor market effects of import competition in the United States

Am. Econ. Rev.

The role of work in psychological health and well-being: a conceptual, historical, and public policy perspective

Am. Psychol.

Where is the land of opportunity? The geography of intergenerational mobility in the United States

Q. J. Econ.

Four fundamentals of workplace automation

McKinsey Q.

Short communication
County-level job automation risk and health: Evidence from the United States