Our report on underemployment uses a new data source—Burning Glass Technologies’ database of resumes, combined with our database of job postings. With a new technique naturally comes questions about its methodology. We thought we’d address some of the common questions we’ve heard, but we encourage you to look at the methodology in The Permanent Detour report for full explanations.

Resume data provides a distinctive view of the job market, because it describes how careers evolve in the real world. Resumes make connections that other forms of labor data don’t (such as how jobs match up with degrees or other training). What is particularly important in this study is the ability of resume data to allow for longitudinal analysis, in other words, track how workers progress, or fail to progress, once launched on their careers. Taken in combination with data from job postings, we can analyze how the skills workers have line up with what employers want.

How do you decide what counts as underemployment?

In our research, we used our database of more than 800 million historical job postings to examine the qualifications employers request. If a majority of postings for a certain occupation requested a bachelor’s degree over the previous three-year period (2015-17), we counted it as a college-level job.

We then used our database of 4 million resumes of graduates with bachelor’s degrees. All personally identifiable information was stripped out of the resumes prior to analysis. We used the work history graduates cited on their resumes to determine what jobs they held. If they were in jobs that were not college-level jobs, they were counted as underemployed.

What about people with graduate degrees who are in jobs that don’t require that level of education?

We considered all people with bachelor’s degrees or higher. If someone has a Ph.D and is in a job that only requires a bachelor’s, we did not count them as underemployed.

How does this compare with other estimates of underemployment?

Our assessment of the number of underemployed recent college graduates (43%) is comparable, or even a little lower, than a Federal Reserve Bank of New York study earlier this year, which used a more traditional definition.

Our goal was to stay as close as possible to the real world of the labor market. The traditional definition of whether a job requires a bachelor’s degree is O*NET, the federal government’s set of official occupational classifications. O*NET bases its data on surveys of workers.

While our method changes some jobs classified as non-college by O*NET to college-level jobs, many others go in the other direction, changing from college jobs to non-college. In total, 45 occupations moved from non-college to college, and 18 shifted from college to non-college. (A complete list is in the report’s appendix). On the whole, we found the changes made assessments of specific occupations more precise, but cancelled each other out otherwise.

For example, many middle-skill supervisory jobs, classified as non-college under O*NET, are categorized as college-level with our definition, based on current job requirements. We don’t count such middle-skill supervisors with college degrees as underemployed, but O*NET would. Other examples would be Paralegals, Industrial Production Managers, and Technical Writers, which we also reclassified as college-level jobs. In this way, we lower the reported underemployment rate and provide a definition that more closely aligns with the public perception of the job market.

Because it focuses on mismatches for college graduates, our approach is also different from the “U-6” underemployment rate, as defined by the Bureau of Labor Statistics and used by many analysts. The U-6 definition is designed to estimate underutilization in the entire civilian labor force.

But is this a fair way of assessing jobs, compared to survey of workers? Employers can ask for whatever they want, but that doesn’t mean they will get it, or that the actual skills needed to do the job have changed.

Upcredentialing, or degree inflation, may not be fair to job seekers, but it is a widespread practice among employers. And by including resume analysis, we gain insight into whether the qualifications of workers in those positions match up with what employers want.

What about people who go to graduate school, and take a lower-level job to get by until they finish their advanced degree? Wouldn’t that inflate the number of underemployed in the study?

In those cases, the “getting by” jobs for bachelor’s graduates might be either college-level or non-college. Since we used resumes to track employment at the five- and 10-year points, if those students get their advanced degree and an appropriate job, that would be shown in our analysis.

A full methodology is available in The Permanent Detour report, conducted with Strada Institute for the Future of Work.

Get the full report on The Permanent Detour: Underemployment’s Long-Term Effects on the Careers of College Grads.

To stay up-to-date on the latest news in today’s labor market, follow us on social media and subscribe to our newsletter today.