The Metric Your Graduate Outcomes Miss: Underemployment and Job-Match Quality

Answer first: "Employment rate" is the headline most programmes report to accreditors and prospective students, and it is close to useless on its own. It counts whether graduates have a job, not whether they have a graduate-level job that uses what they studied. Across the EU, roughly one in four tertiary graduates aged 25-34 is overqualified for their work, and the wage penalty for that mismatch runs to about 24% (CEDEFOP). If your programme evaluation stops at "did they get hired," it is blind to the outcome that most affects graduates'' earnings, satisfaction, and long-term careers. This piece argues that job-match quality — vertical and horizontal — belongs in graduate-outcome evaluation, and is honest about why it is hard to attribute to any single course.

The headline that hides the problem

An employment rate treats a law graduate drafting contracts and a law graduate pulling pints as identical successes. Both are "employed." Only one is in the outcome the degree was supposed to produce. Underemployment — working below one''s qualification level or outside one''s field — is the gap the headline conceals, and it is not a rounding error.

CEDEFOP''s indicator, which counts young tertiary graduates working in occupations outside the managerial, professional and associate-professional categories, put EU over-qualification at 23.6% in 2021, with roughly 23% in 2023 — close to a third of graduates by broader qualification-mismatch measures (CEDEFOP, mitigating overqualification in the EU). This is a structural feature of European graduate labour markets, not a fringe case.

Two kinds of mismatch, both worth measuring

The research literature splits job-match quality into two dimensions (Journal for Labour Market Research, mapping the mismatch of degrees, 2021):

• Vertical mismatch (over-education): the graduate holds a higher qualification than the job requires. A master''s graduate in an administrative role requiring no degree is vertically mismatched.
• Horizontal mismatch: the graduate works at the right level but in a different field from the one they studied. A chemistry graduate in graduate-level marketing is horizontally mismatched — employed, credentialled appropriately, but not using the discipline.

The distinction matters for programme evaluation because the two point to different questions. Vertical mismatch asks whether the qualification is being used at all; horizontal mismatch asks whether the specific field connected to a career. A programme can look successful on employment and vertical match while quietly failing on horizontal relevance — or vice versa.

The cost is real and it scars

This is not merely a labelling exercise. Over-qualified EU graduates earn about 24% less per hour than well-matched peers with the same credential, and report lower job satisfaction (CEDEFOP ESJ insights). Worse, the effect is not always transient. The "trap hypothesis" in the mismatch literature holds that early mismatch can become a permanent scar, with wage penalties that grow rather than fade over a career for those who never escape it (Springer Nature, vertical and horizontal mismatch, 2024). A first graduate job below one''s level is not just a slow start; for many it is a durable ceiling. That is a graduate-outcome your evaluation should want to see coming.

"But a course cannot control the labour market" — the counterargument

A rigorous reader will object immediately, and correctly. Three objections deserve a straight answer.

"Mismatch is driven by macroeconomic conditions, not curriculum." Largely true. Regional labour-market slack, sectoral demand, and the supply of graduates in a field shape mismatch far more than any single module. A programme in a depressed region will show more underemployment regardless of quality. This is a genuine attribution problem, and it means underemployment data must be read as context-adjusted signal, not as a verdict on teaching — the same caution that applies to reading graduate earnings as a quality proxy.

"The data arrives too late to act on." Also fair. Match quality is observed months or years after graduation, well after the cohort that generated it has moved on — the graduate-outcome lag problem. Underemployment is a lagging indicator; using it to steer a live programme requires pairing it with leading indicators collected while students are still enrolled.

"You risk turning universities into job-placement services." A values objection worth taking seriously. Not every programme exists to minimise horizontal mismatch — a philosophy degree that produces a thoughtful civil servant has not failed because the graduate is not a professional philosopher. Match quality should inform evaluation, not dominate it, and it should be weighed against a programme''s stated purpose rather than a blanket employability target. Measuring underemployment does not oblige you to treat every mismatch as failure; it obliges you to know the number and decide what it means.

What to measure instead of "employed: yes/no"

Concretely, a graduate-outcome evaluation that takes match quality seriously asks graduates: is your role graduate-level? Does it use your field of study? Are the specific skills the programme developed being used? Would you describe yourself as working at the level you trained for? These are perception questions that a tick-box destination survey handles badly — they need explanation, nuance, and follow-up, which is where richer methods earn their place. They also connect naturally to the employer-perception gap: graduates and employers often disagree about which skills a good programme should build.

Where Koji fits

Course evaluation cannot measure employability directly — it can only measure the proximal signals that produce it. But those signals, and graduates'' lived experience of match quality, are exactly what conversational methods surface well. Koji for Education runs AI-moderated conversational interviews with graduates and alumni that go beyond "are you employed?" to probe how the role uses their field, which skills transferred and which did not, and where the programme left a gap — with the same standardised, bias-aware moderation for every respondent. Its thematic analysis turns hundreds of such interviews into structured, programme-level evidence an accreditation panel can actually use, and its closing-the-loop action tracking links what graduates report back to curriculum decisions. Combined with employer feedback through the same engine, this closes the horizontal-match question from both sides: do graduates use the field, and do employers value what the programme built?

None of this makes underemployment a programme''s fault, and Koji is careful to frame graduate-outcome data as context-dependent evidence rather than a scoreboard. But it does mean the quarter of graduates working below their level stop being invisible behind a 92% employment headline. The same conversational research engine powers longitudinal customer and market research on the main Koji platform, for teams tracking outcomes well beyond graduation.

Report an employment rate and you have answered the easy question. Report job-match quality — vertical and horizontal, read against context — and you have answered the one graduates actually live with.

How to measure mismatch honestly

Mismatch can be measured three ways, and they disagree often enough that it matters which you use. Objective (normative) measures compare a graduate''s qualification level against the level an occupation is officially coded to require — the basis of the CEDEFOP indicator. Statistical (realised-matches) measures define the required level as the average or modal qualification of people already in that occupation. Subjective (self-assessment) measures simply ask graduates whether their qualification and field are being used. Each has a known weakness: objective measures miss within-occupation variation, statistical measures bake in whatever mismatch is already common, and self-assessment risks graduates rationalising their situation in either direction.

For programme evaluation, the honest move is to lead with graduates'' own assessment — because a graduate who says their field is going unused is reporting something real about the programme''s connection to their work — while cross-checking it against occupational coding so you are not misled by an individual''s framing. What you should not do is quietly pick the measure that flatters the programme and report it as the mismatch rate. State which definition you used, because a "12% mismatch" under one definition and a "28%" under another are both defensible and describe the same cohort. Transparency about the measure is the difference between graduate-outcome evidence and graduate-outcome marketing.

Measure the graduate outcomes a destination survey misses — see Koji for Education.