Inside the college majors at elite universities where graduates earn more than $250,000—over 4 times their peers

Duke mathematics graduates had median annual earnings of $297,029 four years after graduation, while Carnegie Mellon computer science and electrical engineering graduates earned $268,121 and $250,168, respectively; the overall median across institutions was $64,500. The high figures cover small cohorts and graduates from 2017–19, and newer data shows 2024 unemployment rates of 7.8% for computer engineering and 7.0% for computer science graduates, versus 4.2% for recent graduates overall. Federal rules may cut student-loan access for programs that fail earnings benchmarks; about 95% of students are enrolled in programs likely to pass.

Analysis

The investable signal is not that elite STEM degrees guarantee exceptional returns; it is that hiring outcomes are diverging sharply by skill, institution, and cohort. The headline earnings data is backward-looking and based on small cohorts, so it is weak evidence for current labor demand or program-level ROI. Meanwhile, elevated unemployment among recent computer-science and engineering graduates suggests employers may be restricting entry-level hiring even while retaining demand for experienced technical talent. That mix can lower near-term wage pressure for technology employers but raises the risk that training pipelines, graduate outcomes, and future recruiting capacity deteriorate.

The federal earnings test is a slower, localized risk—not an immediate sector-wide shock given the broad share of students likely to pass. Exposure depends on which specific programs fail, enrollment, and reliance on federal loans; do not infer that an entire school or education company is impaired from a weak program outcome. The contrarian point is that the extreme salary figures are poor evidence for the average student's college ROI, but the policy change may also be less disruptive to education-sector revenue than the headlines imply. The key 1–3 month evidence is program-level eligibility and enrollment data; the 6–18 month risk is whether weaker graduate placement feeds into lower applications, financing access, or employer recruiting.

Key Decisions for Investors

• No immediate broad trade: the data is lagged, cohort sizes are small, and the article does not establish current revenue or earnings exposure for any listed company.
• Watch for program-level federal loan eligibility results and discloseable enrollment or funding exposure before considering shorts in career-focused education providers. Falsify the downside thesis if affected programs are immaterial to enrollment and revenue or remain eligible.
• Track entry-level technology hiring, graduate unemployment, and wage data over the next 1–3 months. Persistent weakness would support a relative preference for established technology employers over businesses dependent on expanding junior technical headcount; improving hiring would weaken that view.
• Treat selective STEM-program salary outcomes as a narrow signal, not a basis for buying education providers or student-credit names. Reassess only with current cohort placement rates, tuition/aid mix, and program-level loan eligibility data.