Jul 18, 2018 - Economy & Business

The quicksand of low-wage work

A forklift operator lifts crates full of quartz crystals in this photo from 1948.

A forklift operator in 1948. Image: Ed Clark/The LIFE Picture Collection/Getty Images

Low-wage workers get trapped in largely manual jobs because of the difficulty of acquiring better-paid skills, according to a new study from MIT.

Why it matters: Some argue that workers whose jobs become automated could learn new skills to stay employed. But the difficulty of jumping from physical work to a job that requires mainly social and cognitive skills may leave low-wage workers with no recourse when manual labor is turned over to robots.

How they figured it out: A team of MIT researchers mapped out the relative difficulty of jumping from one profession to another by assembling a comprehensive list of job skills. From there, they developed a spacial map that arranges the skills by how closely they’re related to one another, and another that does the same with job titles.

  • The skill network, shown below, reveals two clumps of abilities — complex problem-solving and interpersonal relationships on one side, and manual dexterity and stamina on the other — with few connections between them.
  • The job network has a similar shape. The clustered cognitive-heavy jobs earn higher wages, while manual jobs are lower-paying.
  • This result mirrored research from MIT’s Daron Acemoglu and David Autor, who divided work into cognitive and physical jobs.
A network with two distinct groups of nodes shows the bifurcation of skill types.
A spacial map from the MIT team reveals two largely distinct skill sets. Map: MIT

What’s new: The research unpacks job titles that have remained largely unchanged for decades even as the jobs they describe have shifted drastically.

  • The MIT study analyzes workplace skills, not jobs, allowing the researchers to understand how and why workers jump from one job to the next.
  • An interactive site explores the detailed skillsets that each job requires, lays out neighboring jobs, and shows how the distance between job types makes it hard to hop from one to the other.

What’s next: "We will eventually get to a point where we can model the impact of a specific new technology" on a job, said Morgan Frank, an MIT Ph.D. candidate and co-author of the study.

This has big implications for predicting the effects of automation on jobs:

  • If robots get really good at performing one skill, Skillscape could someday identify the job titles most at threat and the skills that would complement the technology.
  • The spacial skills map would also identify the work closest in skills to the job under threat — but the clumpy skillsets means it’s very difficult to jump from a job threatened by automation to one that isn’t.

Go deeper: When you're done playing with the results, read the MIT study in Science Advances

Go deeper