By Jeff Strohl, Director
Last month, I announced my plan to retire after 18 years with the Georgetown University Center on Education and the Workforce (CEW). Serving as CEW’s director for the last few years has been a great privilege—particularly as it involves working alongside immensely talented colleagues, both at the center and beyond. CEW has launched a national search for my successor, and I look forward to seeing how a new director will shape the center’s work in the years to come. If you are interested in learning more about the position and application process, information is available on Georgetown’s website. I will continue to serve as director until my successor assumes the role. In the meantime, it’s business as usual, so let’s talk about underemployment.
Our recent report, Rethinking Underemployment: Are College Graduates Using Their Degrees?, sparked some interesting reactions on social media. The report explored different methods of estimating underemployment, finding that underemployment among workers with a bachelor’s degree can range from 25 percent to 52 percent, depending on the methodology used. One commenter pointed out how important it is to reach a consensus approach on measuring underemployment because crafting good policy requires actionable data. Another noted that even the lower-end estimate of 25 percent underemployment among new graduates is a problem, because it suggests that a quarter of new workers with a bachelor’s degree are not using their college-level skills on the job. These comments and many others we received raise the ever-important “What now?” question. Indeed, identifying a problem is often much easier than resolving it.
The problem of underemployment actually involves two issues. The primary issue is that underemployment is too high, whether measurements fall at the low end or the upper bound. Too many workers have more advanced skills than their jobs require, signifying that the nation’s economic engine is not firing on all cylinders. The secondary issue is that the policy community doesn’t agree on the extent of the problem. Estimates of underemployment range widely because we lack a standardized method of determining what skills students develop in the classroom and assessing how those skills align with specific job requirements.
In contrast to unemployment, which has one accepted metric, underemployment is measured using proxies, such as data on entry-level education requirements, wage premiums, skill requirements listed in job ads, or survey practitioner questions (e.g., “Do you use the skills you learned in college at your job?”). A further complication is that work experience builds new skills and hones older ones. This complication is one reason why, regardless of the method of measurement, underemployment decreases for workers with a bachelor’s degree between entry level and mid-career.
These measurement details might be in the weeds, but they matter. Without a widely accepted standard metric, we can’t hope to move beyond disagreements about the full extent of underemployment to fully address the primary issue: that underemployment is too high. We need an effort to create consensus around measurement of underemployment; the US Bureau of Labor Statistics, as our nation’s premier arbiter of employment data with the ability to engage experts across the country, is well-positioned to lead this effort. An attempt to build consensus around underemployment would help us determine whether there is a natural rate of underemployment—as there is for unemployment. (The natural rate of unemployment reflects frictions and other forces in the labor market that make zero percent unemployment unrealistic.) We need a comparable baseline to evaluate whether the underemployment rate at a given time is atypically high or reflects expected labor-market dynamics.
To help illustrate how measurement consensus could move us toward answering important—and actionable—questions, let’s assume the underemployment rate among recent bachelor’s degree recipients (ages 22–23) working full-time, full-year is around 25 percent and naturally declines with additional work experience. Let’s also assume we have field agreement that a natural rate of underemployment for these early-career workers is 15 percent. Instead of debating the extent of underemployment, we could move on to explore the factors contributing to underemployment:
- Are recent bachelor’s degree recipients graduating from programs aligned with labor-market needs or working in fields that align with their majors?
- Are bachelor’s degree curricula aligned with workplace needs?
- Does the measurement of underemployment account for employment discrimination?
- At what points in workers’ early-career trajectories are they more or less likely to move out of underemployment?
- Does bachelor’s degree program quality affect whether graduates are more or less likely to be underemployed?
- How do factors such as company size, location, and employer type affect underemployment?
- How do personal circumstances, such as caregiving needs or health issues, correlate with underemployment?
- Given frictions in the labor market and in people’s personal lives, what is the appropriate natural rate (or range) of underemployment?
Answering these questions will require additional investments in labor-market and education data. For instance, we need to better map how majors, degrees, and fields of study align with the labor market. Postsecondary institutions need strong labor-market analytics to inform their curricular offerings and map projected demand. Better data will also help us identify how the business cycle—including, as is presently the case, a frozen labor market—affects underemployment.
Improvements to available data would help us move forward on multiple fronts. Better data on early-career outcomes would help us understand whether young workers are finding in-field employment, which we consider a first measure of whether their skills and employment are matched. It would also help us better evaluate career pathways into the labor market and distinguish between the natural consequences of early-career job change and indicators of program quality. Likewise, better and more data analysis on rejected resumes would help us understand the role that skills mismatch plays in creating skills shortages. Robust longitudinal data mapping career pathways would help us identify points where workers lose momentum, thus allowing us to develop effective interventions—one reason why the suspension of existing federal longitudinal studies is so troubling. And data on field of study at all degree levels would help us understand the connections between academic programs and career outcomes. The US Census Bureau could collect such data by widening its existing American Community Survey questions on major field of study from the bachelor’s degree level to all degree levels.
With the information that good data provide and consensus about underemployment measurement, we can begin the critical project of reducing underemployment. The stakes are high: Long-term underemployment can lead to permanent economic scarring, significantly reducing lifetime earnings and career opportunities over the course of individuals’ working lives. While it’s impossible for every job to use the full range of graduates’ skills and some degree of mismatch is inevitable, it’s clear that we can do more to better align education pathways with labor-market demands. But first, we will need to collect better data on what skills workers actually need and use in the workplace, and we will need to better evaluate what types of educational opportunities are setting individuals up for long-term success in their careers.
Recent Blogs
What Early-Career Earnings and Unemployment Rates Tell Us About the Returns on Bachelor’s Degree Majors
CEW Leadership Announcement
