How We Define “Skilled” Work

And the problems with our classification systems

What occupations are considered “skilled”? Maybe it’s the jobs that require a college degree. Or the ones with the highest wages. Perhaps it’s the level of judgement required.

Although “skilled” and “unskilled” remain common terms used to describe and analyze the workforce, there isn’t a consistent way to classify work into those categories. Whether or not it’s productive to define skills in this way, it’s useful to understand when we started categorizing jobs by skill level, what are the prevailing measures we use today, and the consequences of those measures.

Early Use of Skill Level in American Labor Analysis

Back in 1938, a statistician named Alba Edwards published a report that had been decades in the making. He first took over the occupational statistics division of the United States Census Bureau in 1910, but in this most recent document, his new framework for occupational classification took its most explicit and cohesive form.

Edwards’ report was called the Social-Economic Groupings of Gainful Workers of the United States. It was his attempt to segment the complex labor market into simple categories, based on employee skill level. According to his definition, occupations which involved a long period of apprenticeship or formal training belonged within the “skilled” or “semiskilled” categories. And those which, “require no special training, judgment, or manual dexterity, but supply mainly muscular strength for the performance of coarse, heavy work” were called “unskilled.” This included all laborer occupations — ranging from fishermen and oystermen, carriage drivers, and garbage laborers — and all work categorized as servant occupations — ranging from porters, elevator tenders, waiters, and servants.

At the time, African Americans, immigrants, and women were systematically excluded from most “skilled” trades, and therefore were concentrated in the so-called “unskilled” occupations. I’m writing more on this in an upcoming piece, but unfortunately it’s not the focus of this article. See page 18 of the document for a breakdown of social-economic groups by race and gender.

Edwards was not the first to refer to “skilled” and “unskilled” work, but his report certainly gave some legitimacy to using this language when analyzing occupations.

So in the year 2021, how far have we come from this simplistic way of evaluating work? The two prevailing methods of evaluation are the use of 1) education and experience level and 2) job content, both of which have notable limitations and biases when evaluating a job’s value.

Evaluating Skill by Level of Education and Experience

“Low-skill” is frequently used as short-hand for occupations in which the average employee has a high-school degree or less. In economic theory, the canonical model, first introduced by Katz and Murphy in 1992, is used to study wage outcomes by using high-school education as a proxy for “low skill,” and college education as a proxy for “high skill.” A similar convention is used by the U.S. Department of Labor’s occupational database called the Occupational Information Network (O*NET), which sorts occupations into job zones ranging from “little to no preparation required,” listing occupations including models, baristas, and agricultural equipment operators, to “extensive preparation needed,” listing occupations including chief executives, lawyers, and librarians.

In 2018, more than 50% of the American labor force were employed in “low-skilled” jobs, where the typical entry-level education requirement is a high-school degree or less. On average, these same workers earn an estimated $33,473 per year, slightly above the poverty line for a family of four.

The types of jobs available for non-college educated workers varies geographically but are concentrated in the service sector. Home health aides, fast food and counter workers, cooks, janitors, and laborers and freight, stock, and material movers are among the occupations with the highest levels of projected job growth — none of which typically require more than a high-school degree.

There are several limitations in using education and experience as a proxy for skill-level:

  1. Firms themselves change their degree requirements based on labor market conditions and preferences. Between 2007 and 2012, the percentage of job postings that included a bachelor’s degree requirement increased by 175% for dental laboratory technicians and 55% for medical equipment preparers, a phenomenon referred to as, “degree inflation.” Google, Apple, Oracle, and Facebook are among the technology companies that are removing degree requirements from job applications for high-wage positions. All of these changes take place without necessarily adjusting the job requirements or desired skills.
  2. Highly educated workers can still earn less. Although women are more likely to have a college degree, their median annual earnings are 20% lower than men’s earnings. England, Levanon, and Allison looked deeper into this discrepancy, finding that when an occupation becomes female-dominated, such as in nursing or design, it will pay less, even when controlling for education.
  3. Ranking skill by education can perpetuate racial and class disparities. Research from the Brookings Institute has reported that children from families of low socioeconomic status are 3 times less likely to pursue postsecondary education and 33% less likely to graduate. This means the credentials to qualify workers for “skilled” positions are more likely to be awarded to children of wealthier families. Asian and non-Hispanic Whites are more likely to hold a bachelor’s degree than African Americans and Hispanics. People of color pursuing a degree are also more likely to be targeted in the marketing efforts of for-profit colleges, with the majority of these students defaulting on their loans within 12 years.
Educational Attainment in the United States: 2015

Evaluating Skill by Job Content

An alternative method to evaluating skill level involves breaking down the components of a job to determine its complexity and difficulty. In 2003, economists David Autor, Frank Levy, and Richard Murnane, concerned about which occupations were most vulnerable to automation or outsourcing, introduced a new model for analyzing occupations based on the composition of tasks. They classify jobs along 4 dimensions: routine tasks, nonroutine tasks, analytic and interactive tasks, and manual tasks. Routine is defined as tasks which are performed repeatedly with an unwavering procedure, analytic tasks require abstract problem solving, intuition, and creativity, and manual tasks often incorporate in-person interactions and physical labor.

Another method of skill-evaluation is used by The National Compensation Survey (NCS), a regular field study conducted by the Bureau of Labor Statistics, capturing the wage and benefit levels of particular occupations. To evaluate pay, they developed a point-factor leveling system along 4 dimensions — knowledge, job controls and complexity, contacts (nature and purpose) and physical environment — assigning those occupations to one of 15 levels.

Using this method, 60% of occupations are classified as a level 4 or less, earning no more than $15 an hour on average, and the least likely to provide health insurance, retirement benefits, and paid leave. Common Level 1 jobs included cashiers and stock clerks, level 5 includes licensed vocational nurses and customer service representatives, and level 12 jobs include software developers and post-secondary teachers.

There are several limitations in using job content as a proxy for skill-level:

  1. Researchers and organizations rely on potentially problematic sources of data when determining job content. Many researchers continue to rely on predecessor for the O*NET, a job catalog called The Dictionary of Occupational Titles (DOT). The DOT received criticism throughout the latter half of the 20th century, both for its emphasis on blue-collar jobs in an economic landscape shifting towards service and knowledge professions and for bias in its classification system. Male-dominated positions like Dog Pound Attendant receiving higher complexity scores than female-dominated positions like Nursery School teacher.
  2. There’s very little transparency in methodologies used to evaluate job content. In the modern-day NCS point-level system, the potential point values vary significantly between the major occupational groups, without much explanation. For example, to evaluate the knowledge level, points are designated to service occupations at a starting value of 50 points to a maximum of 1250 points, where communications and arts jobs have a starting value of 550 points, with a maximum of 1850 points.
  3. Although a job may require specific skills, they may not be accurately translated into scores. For instance, a report from the Paraprofessional Healthcare Institute (PHI) notes that the physical, social, and emotional labor required by direct-care workers is frequently minimized and disregarded. In 2012, a care-worker in New Zealand named Kristine Bartlett settled a pay equity court case after successfully arguing that caring for the elderly was as demanding as a prison-guard (her union noted that both jobs require, “dealing with challenging behaviors including sexual behaviors and/or aggression”). As a result of the case, 55,000 government-employed elderly care workers won pay raises.
Photo by Ehimetalor Akhere Unuabona on Unsplash

Why This Matters Now

Employers and institutions around the country and the world have latched on to the idea of “upskilling” and a “skill-based hiring approach.” On the one hand, this can be seen as a laudable effort to ensure employees are hired for their competencies, not their ability to afford a college education.

On the other hand, as we’ve seen, skills are not concrete objects that we can easily define. There is a lot of judgement involved in identifying relevant skills and prescribing their value. We don’t yet know the mistakes we will make and the biases we will perpetuate when selecting workers based on some notion of skill level.

There are no easy solutions to this problem, but it starts with recognizing the skills—the emotional labor of working with the elderly, the judgement needed to properly harvest various crops, the communication required in retail—needed to perform work that has been historically labeled as, “unskilled.”

I hope you enjoyed this article. If you’re looking for more stories and research on the topic of work and employment, consider subscribing to Workable, my free monthly newsletter on how work is changing around the world.

Obsessed with people & process; advocate for labor dignity, voice & ownership. Subscribe to Workable for monthly news on changing work:

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store