Introduction and framing
Underemployment is one of the most cited and least carefully defined concepts in workforce policy. The headline numbers vary by a factor of three or more depending on which definition is used. A worker counted as underemployed under one framework is counted as fully employed under another. This page treats the disagreement as the starting point.
Three definitions dominate the labor market literature. Each measures something real. Each misses something important. Sorting out what they mean, and where they overlap, is the first step toward using the data well.
The three definitions
BLS Current Population Survey
Involuntary part-time work
Workers who hold part-time jobs and want full-time hours. This is the cleanest definition because it relies on a worker's stated preference rather than an analyst's judgment about whether a job matches the worker's skills. The Bureau of Labor Statistics tracks it monthly through the Current Population Survey and includes it as a component of the U-6 measure.
New York Fed Recent College Graduates
Skills underemployment
Workers in jobs that do not use their education or skills. This is what most policy analysts mean by underemployment, and it is the most relevant to higher education planning. There is no official measure. The Federal Reserve Bank of New York publishes the most widely cited research version, defining a college job as one where at least 50 percent of workers hold a bachelor's degree.
BLS Current Employment Statistics
Hours underemployment
Workers who want more hours than they currently receive, regardless of full-time or part-time status. This overlaps heavily with involuntary part-time work but extends to full-time workers whose hours have been cut. The BLS Current Employment Statistics program publishes average weekly hours by industry, which serves as the standard data foundation.
How the definitions relate
The three definitions measure overlapping but distinct populations. A barista with a philosophy degree working 40 hours a week is underemployed by the skills definition and not by the hours or part-time definitions. A nurse working 24 hours a week who wants 40 is underemployed by the part-time and hours definitions but not by the skills definition. The populations cannot be added together. They cannot be averaged. Each answers a different question.
Definitions are adjacent, not additive
Definitions use different denominators. Involuntary part-time work is a worker count, skills underemployment is a college-graduate share, and hours pressure is proxied with average weekly hours.
DataSource: BLS and NY FedAs of: 2026-03-31
- Involuntary part-time
- Workers in part-time jobs who want full-time work.
- Adjacent measure
- Skills underemployment
- College graduates in occupations classified as non-college jobs.
- Included in headline
- Hours pressure
- Average weekly hours proxy for pressure below desired work hours.
- Adjacent measure
The headline data
As of 2026-03-31, the underemployment rate for recent college graduates is 41.5 percent. The rate for all college graduates is 33.4 percent. The recent graduate rate has historically run several points above the all-graduate rate.
The recent graduate underemployment rate has hovered between 38 and 45 percent for most of the past two decades. It rose sharply during the 2008 recession, partially recovered, then climbed again in the early 2020s. The all-graduate rate has been more stable, suggesting that underemployment tends to resolve as workers age into the labor market, but only partially.
Recent graduates run above all graduates
DataSource: NY Fed Recent College GraduatesAs of: 2026-03-31
By major
Among recent graduates, the underemployment rate varies from below 20 percent for some engineering and computing majors to above 60 percent for several humanities and service-oriented majors. The major-level data is updated annually and currently covers 15 major categories.
The variation across majors is the most consequential single fact in the underemployment data. The choice of major matters more than most workforce-policy conversations acknowledge. At the same time, the rankings are not destiny. Some high-underemployment majors lead to good outcomes for workers who reach graduate-level credentials. Some low-underemployment majors compress earnings even for workers who avoid underemployment. The headline ranking is the starting point, not the conclusion.
Underemployment by major
DataSource: NY Fed Recent College GraduatesAs of: 2026-03-31
Major comparison table
Source: NY Fed Recent College Graduates | As of: 2026-03-31
| Major | Rank | Underemployment | Median wage | Wage premium | Notes |
|---|---|---|---|---|---|
| Performing ArtsArts | 1 | 62.1 percent | $48,000 | $14,000 | NY Fed major category |
| Fine ArtsArts | 2 | 58.7 percent | $50,000 | $15,000 | NY Fed major category |
| English LanguageHumanities | 3 | 55.2 percent | $52,000 | $16,000 | NY Fed major category |
| CommunicationsHumanities | 4 | 51.9 percent | $54,000 | $17,000 | Common online program area |
| PsychologySocial Science | 5 | 48.8 percent | $51,000 | $15,000 | Common online program area |
| Criminal JusticeSocial Science | 6 | 47.6 percent | $52,000 | $15,000 | Common online program area |
| MarketingBusiness | 7 | 43.7 percent | $58,000 | $20,000 | Common online program area |
| EducationEducation | 8 | 40.5 percent | $48,000 | $12,000 | Common online program area |
| Business AdministrationBusiness | 9 | 39.9 percent | $60,000 | $21,000 | Common online program area |
| Information TechnologySTEM | 10 | 34.8 percent | $68,000 | $25,000 | Common online program area |
| AccountingBusiness | 11 | 33.9 percent | $64,000 | $23,000 | Common online program area |
| FinanceBusiness | 12 | 32.7 percent | $69,000 | $26,000 | Common online program area |
| NursingHealth | 13 | 23.4 percent | $70,000 | $25,000 | Common online program area |
| Computer ScienceSTEM | 14 | 21.6 percent | $82,000 | $35,000 | Common online program area |
| Chemical EngineeringEngineering | 15 | 17.2 percent | $78,000 | $30,000 | NY Fed major category |
Trajectory analysis
Workers who begin their careers underemployed do not all stay underemployed. The data on career trajectories is thinner than the cross-sectional data, but the available evidence suggests partial resolution over the first decade of work.
The gap between recent graduate underemployment and all-graduate underemployment is the best available trajectory signal. The fact that the all-graduate rate is meaningfully lower indicates that some workers escape underemployment as they age. The fact that it is not dramatically lower indicates that escape is incomplete. For policy purposes, the more useful question is which underemployed workers escape and which do not. The available data is not granular enough to answer this directly, but the major-level patterns offer suggestive evidence.
Underemployment falls with experience, but not to zero
DataSource: NY Fed Recent College GraduatesAs of: 2026-03-31
Major lookup tool
The ranking chart shows the landscape. The lookup tool below gives the full picture for any single major. Each major has a profile including current underemployment rate, historical trend, wage data, and comparison to similar majors.
Programs commonly offered by online universities
Online and competency-based universities tend to concentrate their program offerings in a specific subset of majors. The most common categories include business administration, information technology, nursing and other health professions, education, and criminal justice. The underemployment landscape for these majors deserves specific attention because they represent the practical choice set for many adult learners and working students.
The majors common to online universities cluster in the middle of the underemployment distribution. They are neither the highest-return majors, which tend to be engineering and quantitative fields, nor the highest-underemployment majors, which tend to be performing arts, humanities, and some social sciences. This middle position reflects the practical orientation of online programs toward fields with clear occupational pathways, but it also means that workers in these programs face nontrivial underemployment risk and should not assume that a degree in a practical field guarantees a college-level job.
Common online program areas
DataSource: NY Fed Recent College GraduatesAs of: 2026-03-31
Methodology and transparency
How the measure works
This section explains how the underemployment measure is built, why the threshold matters, and where the public data is strongest and weakest.
The NY Fed 50 percent threshold rule
The core method classifies an occupation as a college job when at least half of workers in that occupation hold or require a bachelor's degree.
The NY Fed underemployment rate starts with an occupation-level classification. An employed college graduate counts as underemployed when they work in an occupation classified as a non-college job. The NY Fed assigns that label using a 50 percent rule, informed by Department of Labor O*NET education data and Census occupation data.
The rule is transparent and repeatable, which is why it works well as a public benchmark. It also creates a hard boundary around a soft labor market fact. An occupation just above the line becomes a college job. An occupation just below the line becomes a non-college job. Small shifts in the occupational workforce can change the label without changing the lived experience of the job.
The main critique is not that the rule is careless. The critique is that the rule is necessarily arbitrary. A threshold at 45 percent or 55 percent would classify some occupations differently and would move the headline underemployment rate.
Why this matters at the margin
Occupations near the threshold can move between college and non-college status as credentials, licensing norms, and employer requirements change.
The margin matters because several large occupations combine skilled work with multiple entry pathways. Computer user support specialists are a clear example. O*NET-derived education distributions place the bachelor's degree share just below the 50 percent line, while BLS notes that many jobs require technical knowledge but not necessarily a college degree.
Registered nurses show the opposite issue. O*NET-derived education distributions place the bachelor's degree share modestly above the line, while BLS describes three common entry paths: a bachelor's degree in nursing, an associate degree in nursing, or a diploma from an approved program. The work is skilled and licensed, but the bachelor's degree is not the only route into the occupation.
Technical sales roles add a third type of ambiguity. BLS classifies sales engineers as typically needing a bachelor's degree, yet it also notes that some candidates qualify through sales experience plus technical training. These examples show why threshold classifications are most defensible for broad trends and most contested for occupations near the boundary.
Alternative measures and how they compare
Different underemployment measures answer different questions about credentials, skills, wages, and job fit.
The NY Fed measure asks whether a college graduate works in an occupation that meets a college-job threshold. Burning Glass and Lightcast-style measures often use job postings, occupational skill requirements, and career pathway evidence to classify underemployment. These approaches can capture employer demand more directly, but they depend on postings data and proprietary taxonomies.
Fogg and Harrington's mal-employment framework focuses on whether college graduates work in occupations that use the knowledge and skills associated with college education. It is conceptually close to the NY Fed approach, but it comes from a broader labor market underutilization literature and uses the college labor market as its organizing frame.
The mismatch literature separates vertical mismatch from horizontal mismatch. Vertical mismatch asks whether the job uses the worker's education level. Horizontal mismatch asks whether the worker's field of study matches the occupation. A simple credential underemployment count asks only whether the job requires a bachelor's degree. That measure is easy to explain, but it can miss skill use, wages, licensing, and career pathways.
What recent means
The NY Fed recent graduate cohort covers ages 22 to 27, which makes the measure especially useful for early career transitions.
Recent graduates in the NY Fed data are college graduates ages 22 to 27. This age band captures the transition from school into the early labor market and makes the recent graduate rate distinct from the all-graduate rate.
The definition also excludes many adult learners and nontraditional students who complete degrees later in life. That exclusion matters for analysts studying online, part-time, and competency-based education, because those models often serve workers whose labor market history looks different from a traditional 22-year-old graduate.
For this page, the recent graduate rate should be read as an early career benchmark. It should not be treated as a complete measure of outcomes for all newly credentialed workers.
Sample size and statistical reliability
Major-level estimates are useful for comparison, but small majors can carry wider uncertainty than the ranking chart implies.
The headline underemployment rate pools many observations and is more stable than most major-level estimates. Major rankings are thinner. Some majors represent smaller samples, and those estimates can move more from survey noise, classification changes, or cohort composition.
This matters most when two majors sit close together in the ranking. A one or two point gap should not be read as a precise ordering unless the underlying samples are large enough to support that distinction. The ranking is better at separating broad clusters than at adjudicating tiny differences.
The page should therefore present major-level bars as estimates with methodological context. Users can still compare majors, but they should avoid treating every rank position as a statistically meaningful difference.
What this misses
Limits of the framework
The underemployment framework is powerful because it is narrow. The same narrowness creates blind spots that matter for workforce policy.
Workers without bachelor's degrees are outside the frame
The measure excludes workers without bachelor's degrees, even though they make up the majority of the workforce and may face the most acute underemployment. A worker with strong technical skills, a certificate, or years of experience does not enter this calculation unless they also hold a bachelor's degree.
Sub-baccalaureate credentials are invisible
Certificates, associate's degrees, industry certifications, apprenticeships, and microcredentials do not receive separate treatment in the NY Fed framework. That omission matters because many workforce development strategies rely on credentials below the bachelor's level.
The data is a snapshot, not a career path
The cross-sectional rate does not measure career mobility, skill development on the job, or the value of an underemployed first job as a stepping stone. Some workers use a non-college job to enter an industry and move up. Others remain stuck. The headline rate does not separate those pathways.
The bachelor's degree remains the unit of analysis
The framework assumes that the bachelor's degree is the relevant credential boundary. Competency-based education and skills-first hiring challenge that assumption directly. They ask whether the worker has the skill for the job, not whether the worker holds the traditional four-year credential.
Geographic variation is averaged out
National rates flatten local labor markets. The same underemployment rate can carry different implications in a high-cost coastal metro, a lower-cost interior city, or a region with a specialized industry cluster. Wages, commuting options, occupational mix, and housing costs all change the meaning of the same headline number.