Automation ran into a roadblock in the last day with the news that an Uber self-driving car struck and killed a woman. That's extreme, but a new report says that automation and robotics may be causing enough anxiety in another group -- workers -- to cause poorer general, physical, and mental health.

The anxiety is over job insecurity. Once that was the province of factory workers as specialized robots took over assembly responsibilities. That has changed. Software systems have increasingly threatened managerial and professional jobs.

Researchers from Villanova University and Ball State University examined individual and county-level health data as well as county-level business data. The sources taken together allowed the researchers to compare employment by job category with associated health statistics. By doing regression analysis on the correlations, here is what they found with a 10 percent increase in automation risk at the county level:

  • 2.38 percentage points lower general health
  • 0.8 percentage points lower physical health
  • 0.6 percentage points lower mental health

The numbers at first glance may seem low, but they represent an iterative effect. A 20 percent increase in automation risk would double the health impact; 30 percent increase would triple the impact; and so on.

The study did offer some caution in interpretation:

First, the study does not take into account individual-level morbidities or health outcomes. Also, the effects of automation risk may get stronger over time (i.e., time-trend effects), and its effects may vary across levels of analyses (i.e., country, region, firm, and employee). How individuals cope with fear and anxiety around automation, leverage personal and social resources, and rely on firm-related resources, training, and support to cope with automation risk may be of interest for future research. Second, due to data availability limitations, our study does not consider longitudinal nor multilevel confluence of these factors; however, these cross-level and between-time interactions would be of interest to future research. It is plausible that automation risk could more strongly influence psychological comorbidities than physiological ones, or that job insecurity associated with automation risk could result in greater job demands, lower pay, and/or layoffs, resulting in more immediate job-related stress and strain and long-term negative physical health outcomes. Third, individuals with lower socioeconomic status are more likely to be in occupations at greater risk of automation. Although we controlled for income in the analyses, future studies could consider the influence of socioeconomic status in explaining the job insecurity-health risk association.

All important considerations. But for managers and entrepreneurs, a wait for later studies that may or may not happen isn't a viable alternative. Automation is a growing trend, but what happens if you make your workforce sicker and less productive in the meantime?

There is the initial hit on profits, of course. In addition are other implications. One is that you scare off talent. Better employees who do feel like they have options are more likely to jump ship, taking with them skills, attitude, and institutional knowledge.

Additionally, you may have a problem with automation systems themselves. The software frequently involves machine learning, which means humans have to provide a model. If you lose the best employees, you have the worse as examples for the software. Also, conditions change over time. People can adapt as needed. Current and near-future generations of software need continued retraining to change. As you lose employees, you also miss a fundamental ingredient to making automation work over time.