FREOPP’s 2024 analysis of 11,600 US master’s degree programmes across 1,441 universities found that 40% of master’s degrees fail to produce a positive lifetime financial return, even after the credential is factored into projected earnings. The median master’s ROI is $83,000, but the distribution is bimodal: STEM, nursing, and licensed health-professional master’s degrees almost never produce negative ROI, while MBA, education, humanities, fine arts, and many social-sciences master’s degrees frequently do. The Federal Reserve’s 2023 SHED survey of 11,400 US adults corroborates the satisfaction picture: 68% of bachelor’s-or-higher holders feel the benefits of their education outweighed the costs, leaving roughly one-third who do not — a population that includes a disproportionate share of graduate-degree holders carrying $40,000+ in additional debt. The 40% action-regret framing is a financial-ROI proxy, not a direct survey measure, and is acknowledged as such in proxy_only: true.
On the inaction side, Pew Research Center’s May 2024 survey (n=5,109) documented a 25-percentage-point gap between bachelor’s-only and postgraduate degree holders on the question of whether their education was extremely or very useful for the skills and knowledge needed in their current job: 47% of bachelor’s-only holders agree, versus 72% of postgraduate holders. The Strada-Gallup Education Consumer Pulse independently shows that only 26% of working US adults with college experience strongly agree their education is relevant to their work and day-to-day life, with relevance the single strongest predictor of education-worth-it ratings across degree levels. Together these signals support an inaction-regret estimate of roughly 25% — the share of bachelor’s-only holders who rate their education’s job-relevance below the postgraduate baseline. Like the action side, this is a constructed proxy: no nationally representative survey directly asks “do you regret not getting a master’s degree.”
The action_dominates pattern (regret_delta 0.15) reflects two structural shifts. The Grad PLUS loan programme, introduced in 2006, allowed unlimited federal borrowing for graduate study and removed the lender’s underwriting check on programme value — the result has been faster tuition growth in graduate education than in undergraduate, often outpacing starting-salary gains. Combined with the time cost of a typical two-year master’s — bachelor’s-level earnings forgone while studying, which the FREOPP ROI figures already net out as an opportunity cost — and the field-dependent wage premium, the financial expected value of a master’s has compressed for the median graduate even as it has expanded for engineering and computer-science students. The decision is recoverable in a way that PhD enrolment is not — the time cost is shorter and the credential generally has value in industry — which suggests the action-side regret here is moderate rather than acute, and concentrated in specific fields rather than spread across all master’s graduates.