The College Degree Just Became a Bad Trade

You’ve saved $1.5 million. Now comes the real test.

Can it produce $10,000 a month, or will that pace drain your portfolio?

Most retirees do not get a clear answer until it is too late.

The issue is not just how much you have. It is whether your portfolio was built to pay you, not just grow.

That difference can determine whether your money lasts decades or starts breaking down early.

Sequence of returns, taxes on withdrawals, healthcare costs, and whether the 4% rule still applies all play a role.

Fiduciary advisors created a breakdown showing what drives sustainable income and why the same $1.5M can produce very different outcomes.

If you have $1M or more invested, do not guess.

See How Long My Money Lasts

Learn More

The realized return on a US bachelor's degree has repriced from a uniform safe asset into a dispersion product. Four forces converge on the 2026-27 school year to widen the gap between the credential's winners and its losers: return dispersion, the demographic cliff, the small-college closure plateau, and the July 1, 2026 loan caps.

The break has a date. On July 1, 2026, the federal government capped student lending for the first time in a generation. Under the One Big Beautiful Bill Act's loan provisions, Grad PLUS was eliminated for new borrowers and graduate borrowing was capped at $20,500 a year and $100,000 in aggregate, with professional programs at $50,000 and $200,000. Parent PLUS was capped at $20,000 a year and $65,000 per student, and a $257,500 lifetime federal limit now applies to a borrower's own loans, excluding Parent PLUS (US Department of Education; NASFAA).

For twenty years the credit architecture ran the other way. Since Grad PLUS was created in 2006, families could borrow essentially unlimited federal money against any sticker price. That is the financing that underwrote two decades of tuition growth: when the buyer can borrow any amount, the seller has little reason to hold price. Colleges raised list prices into an open credit line, and the federal balance sheet absorbed the gap.

The marginal dollar of pricing power just moved from the seller back to the buyer.

The timing is what makes the break structural. The caps land in the same window as three other forces. The first post-peak high-school cohort enters college with the 2026-27 school year, so the buyer pool starts shrinking on schedule. The Western Interstate Commission for Higher Education (WICHE), which produces the standard state-by-state projections of high-school graduates, puts the 2025 peak at 3.9 million and the decline at roughly 13% by 2041. The payoff data went public in April 2025, when the New York Fed said in plain institutional language that college does not appear to pay off for at least a quarter of graduates. And the debt behind the product is visibly stressed, with 10.3% of balances now 90-plus days delinquent as pandemic forbearance fully unwinds.

The setup matters more here than any single number. A pricing mechanism, a demographic denominator, a payoff disclosure, and a debt-service shock all turn at once, and they turn on a calendar the business cycle does not set.

The macro backdrop is secondary. BCR's weekly macro work currently reads the US economy as late in a reflationary upswing, with growth and prices both still firm. None of this thesis's four legs depends on how that cycle resolves. The thesis rides the demographic and policy calendar, and that calendar has already started.

The New York Fed's April 2025 two-part study (Abel and Deitz) puts the median return on a bachelor's degree at 12.5%, comfortably above any reasonable investment hurdle. The dispersion around that median carries the story. The 25th-percentile graduate earns a wage premium under $10,000 a year over a high-school graduate, an estimated 2.6% return in 2024. The authors conclude that for at least a quarter of graduates in recent decades, college does not appear to pay off. The return is exquisitely sensitive to inputs a 17-year-old barely controls. Against a four-year finish, paying average sticker price with no aid drops it to roughly 10%, higher-than-average living costs take it near 9%, a five-year finish takes it to roughly 9%, and a six-year finish to roughly 7%.

The same shape appears in the wage data. Measured in constant 2010 dollars, the median graduate's premium over a high-school graduate was about $19,800 fifteen years ago and about $20,000 last year, a change of 1.1%. The 25th-percentile graduate's premium broke from its trend in 2010 and has since eroded to roughly $5,000 a year. The spread between graduate winners and losers widened about 16%.

The product now sorts its own buyers. The median premium held its ground, the loser tail eroded, and the spread widened. A quarter of the customer base is paying a six-figure price for a return that underperforms a savings bond, and until April 2025 no official source had said so this plainly.

Figure 1. Annual return on a bachelor's degree, median vs 25th percentile. Source: NY Fed (Abel and Deitz), April 2025.

US high-school graduates peaked in 2025 at 3.9 million, fall to about 3.68 million by 2030, and decline roughly 13% to 3.4 million by 2041 (WICHE, 11th edition, December 2024). The first post-peak cohort enters college with the 2026-27 school year, and the decline arrives region-first, with Illinois, Michigan, New York, and Pennsylvania already past their peaks. The 2008 birth-rate collapse already happened, and these students are already born. The demographic leg is arithmetic.

On nine paths in ten, the traditional freshman pipeline sits below its 2025 peak in both 2030 and 2035. The central path has it roughly 14% smaller by 2035 and a quarter smaller by 2041. Holding the pipeline flat would require the immediate college-enrollment rate, the share of each year's high-school graduates who start college that fall, to climb to 65.6% by 2030 and 70.9% by 2041. The 2041 requirement sits above the roughly 70% high recorded in 2016. The 2030 requirement sits 3.6 points above the current rate, which has fallen to 62.0% (2022) and drifts down about half a point a year (National Center for Education Statistics).

That gap is the whole demand story: a smaller pool is buying, and a smaller share of that pool is buying, at the same time.

Figure 2. US high-school graduates, 2025 peak through 2041. Source: WICHE, 11th edition (December 2024).

Nonprofit college closure announcements ran 14 in 2023, 16 in 2024, and 16 in 2025 (Inside Higher Ed's tracking), with 8 more announced for 2026 by late April, the last date on which the two most-cited public trackers were updated (The College Investor; BestColleges). Counting all sectors, 2024 alone saw 28 closures (Federal Student Aid and SHEEO data via BestColleges). This is an elevated plateau holding at 14 to 16 a year, and it has held for three straight years before the demographic cliff has even reached campus.

Two facts about a college are enough to sort the ones that close from the ones that survive: how small it is, and how fast it is shrinking. Across 4,496 institutions in the Education Department's census between 2019 and 2023, the smallest and fastest-shrinking tenth, a median of 121 students with enrollment down 34%, exited at 22.5% over four years. Schools above 5,000 students exited at 0.9%. Applied to the 1,566 private nonprofit four-year colleges still open, that pattern implies about 21 exits a year, rising to about 28 under a further 10% enrollment shock.

Twenty-one a year runs above the 14 to 16 the trade press counts, and the gap is definitional: the federal data counts every institutional exit, including mergers and absorptions, while the press counts announced closures. Size and enrollment trend alone account for the closures the sector has recorded. The Philadelphia Fed's widely quoted figure of as many as 80 additional closures in a single year is a worst-case scenario tied to an abrupt 15% enrollment drop, well above the base case; a gradual 15% decline produces roughly 5 additional closures a year.

Figure 3. Four-year exit rate by institution size and enrollment trend. Source: BCR closure-hazard model (IPEDS 2019-2023 panel, 4,496 institutions).

Recent college graduates were unemployed at 5.7% in Q1 2026 against 4.2% for all workers on the NY Fed's comparable series. And 41.5% were underemployed, meaning working in a job that does not require a degree (NY Fed, The Labor Market for Recent College Graduates, May 2026). The gap is entrenched. Recent-grad unemployment has exceeded the all-workers rate every single month since January 2021: 63 months and counting, nearly six times the longest prior run in the 1990-2026 series (11 months in 2019-20). The gap has averaged 0.81 percentage points, twice the 0.41 average across all prior inversion months. Nothing in the post-1990 record comes close enough to call this one chance. There was no inversion at all in 2010-2013.

Since mid-2023, recent-grad unemployment rose about 1.6 points, roughly three times the increase in the overall rate, concentrated in entry-level and tech-adjacent hiring. Oxford Economics attributes the rise to a structural shift in tech-sector hiring plus strong graduate supply growth. Undergraduate computer-science enrollment fell 8.4% in spring 2026 (National Student Clearinghouse), which reads as students pricing in the durability of that shift.

The persistence is the signal. Five straight years of a negative entry-ticket premium is a structural fact, and press framing of a recent flip understates how long it has held.

Figure 4. Recent-graduate unemployment against all workers (Q1 2026), and the length of the current inversion against the longest prior run. Source: NY Fed, The Labor Market for Recent College Graduates (May 2026); BCR inversion model.

Behind the credential sits $1.66 trillion of student debt (NY Fed Household Debt and Credit, Q1 2026). With pandemic forbearance fully unwound, 10.3% of balances are 90-plus days delinquent, up from 9.6% in Q4 2025. Roughly 1 million borrowers defaulted in Q4 2025 and 2.6 million more in Q1 2026, and over 17% of borrowers have been 90-plus days late at least once since credit reporting resumed in Q1 2025.

Read this as household balance-sheet strain and a demand signal for the education product itself. That is the limit of what the data supports here. Aggregate household delinquency held near 4.8% in Q1 2026, with student loans the clear outlier category, and in BCR's own testing the level of consumer delinquency failed to lead any of the last three recessions. The stress sits inside one product's customer base, and it feeds straight back into how households price that product at the kitchen table.

Figure 5. Share of balances 90-plus days delinquent, student loans against all household debt, Q1 2026. Source: NY Fed Household Debt and Credit, Q1 2026 (released May 2026).

Stage 1. The birth-rate break of 2008 to 2023 reaches campus. The US birth rate fell in the Great Recession and never recovered, and the 2008 cohort turns 18 in 2026. High-school graduates peak in 2025 at 3.9 million and fall 13% by 2041 (WICHE). The freshman pipeline now shrinks on actuarial rails, region-first, with the Midwest and Northeast already declining. This stage runs on arithmetic. Each successive cohort is smaller, and every one of them is already born and already counted.

Stage 2. The tuition-dependence ratchet meets that shrinking pool. State appropriations fell $13.8 billion in inflation-adjusted terms, close to 20%, between 2008 and 2013, and never fully returned, leaving most colleges reliant on tuition revenue to cover operating costs. A shrinking pool forces price competition exactly where balance sheets are weakest: small, regional, non-selective private colleges that recruit within fifty miles of campus. Tuition discounting rises, meaning the share of the sticker price a college hands back as institutional aid, and net revenue per student falls even when the headline sticker price holds.

Stage 3. Fixed costs refuse to flex. Tenure contracts, dormitories, and physical campuses cannot be shed on a single year's notice, so a revenue shortfall lands on cash reserves. For a college enrolling 250 freshmen a year, a few dozen missing students is a double-digit revenue shock that reserves absorb for a while and then cannot. The weakest institutions close, which is why nonprofit closure announcements have held at an elevated 14 to 16 a year since 2023.

Stage 4. The credit architecture flips from accelerant to brake. From July 1, 2026, capped federal lending (Grad PLUS eliminated, Parent PLUS at $20,000 a year, a $257,500 lifetime limit on a borrower's own loans) means high-sticker programs with weak payoff data can no longer be financed with unlimited federal money. Private lenders underwrite the gap, and private underwriting looks at exactly the dispersion data from Exhibit 1. Demand at the negative-tail programs, meaning long completion times, low-premium majors, and high sticker prices, loses its funding source first.

Stage 5. The payoff data is public and the labor market confirms it. The NY Fed's quarter-of-graduates finding, the 63-month unemployment inversion, and 41.5% underemployment give every kitchen-table conversation a number to work with. Households substitute at the margin, and the substitution is already visible in composition: undergraduate certificates grew 10.2% in spring 2026 and health professions 6 to 7%, while computer science fell 8.4% (National Student Clearinghouse).

Stage 6. Dispersion widens while the aggregate market holds. Selective, endowed institutions and high-payoff majors keep pricing power. The bottom tier reprices, merges, or exits. Aggregate undergraduate enrollment is rising now, up 1.3% in spring 2026, so Stages 1 through 4 remain a forecast with a schedule while Stages 5 and 6 are already visible in the enrollment mix. What reads as a healthy market in the aggregate is a widening gap between the schools that can hold price and the schools that cannot.

Figure 6. The six-stage transmission: Stages 1 to 4 run on a published calendar; Stages 5 and 6 are already observable in the enrollment mix. Sources: WICHE 11th edition (December 2024); SHEEO; NY Fed (May 2026); NASFAA; National Student Clearinghouse (June 2026).

Higher education has already repriced once in living memory. The for-profit college collapse of 2010 to 2016 hit a sector that had over-expanded during the recession. It hit through the same three forces now facing the bottom tier of nonprofits: payoff transparency through the gainful-employment rules, which forced programs to publish what their graduates earned and owed; tightened access to federal credit; and a reputational break. For-profit enrollment roughly halved from its 2010 peak, and the sector's largest players shrank or failed outright. Corinthian Colleges closed in 2015, stranding roughly 16,000 students at its final closure, and ITT Tech closed in 2016, displacing an estimated 35,000 more. Over 1996 to 2023, nearly a third of the roughly 3,700 two-year for-profit colleges closed, against about 7% of four-year nonprofits over the same period (Kelchen, Ritter, Webber, Federal Reserve Bank of Philadelphia).

The differences set the pace. Nonprofits carry endowments, alumni bases, and political protection, so their adjustment runs slower and flows through mergers as much as through outright closure. The demand shock this time is demographic and permanent, where the for-profit shock was regulatory and reversible: a disclosure rule can be repealed, and a birth cohort cannot be re-created. And the for-profit sector concentrated in a handful of large chains whose failures made national headlines, where the nonprofit bottom tier is fragmented across hundreds of small campuses whose closures register one town at a time.

The precedent fixes direction and mechanism, and leaves the speed open. A tuition-dependent, federally financed education segment has repriced before when payoff data went public and credit tightened, and the demographic leg is the new variable pointing the same way.

Table 1. The two repricings compared.

This section is educational. It describes how asset classes have historically behaved in comparable environments, and it carries no recommendation, price target, or view on any specific security.

Equities. The historical pattern inside education-linked equities in dispersion environments has been internal separation across sub-sectors. Skill-certification, trade-education, and workforce-training providers sit on the growth side of the substitution now visible in the data, with certificates up 10.2% year over year in spring 2026. On the exposed side sit traditional-enrollment-dependent models, the firms that run universities' online degree programs and whose revenue tracks graduate lending, and student-housing operators concentrated around non-selective private colleges. In past environments this has read as a split outcome at the sub-sector level, one group gaining while another loses, with the index-level signal muted. The macro overlay is modest: a degree-holding cohort aged 25 to 40 resuming debt service is a slow drag on discretionary categories.

Rates and fixed income. The cleanest historical expression has been higher-education municipal credit. Small tuition-dependent private and regional-public issuers have historically faced downgrades and widening spreads, meaning the extra yield investors demand over safe government bonds, when enrollment prints turn. Flagship publics and endowed privates have historically kept paying on schedule through the same episodes, so the yield gap between the sector's strongest and weakest borrowers widens even as the sector as a whole does not cheapen. Rating actions have historically followed enrollment prints by 6 to 18 months, and the first post-cliff enrollment data lands in October 2026, the window in which any repricing would first surface.

Credit. Consumer credit quality in the 25-to-40 degree-holding cohort has historically deteriorated at the margin in this kind of debt-service shock while staying benign in aggregate. The 2.6 million Q1 2026 defaults feed credit-score cliffs, wage garnishment, and reduced mortgage eligibility for a cohort already delaying household formation. Read this as a composition signal and a demand read on the education product, which is the limit of what the data supports. Aggregate household delinquency held near 4.8% in Q1 2026, with student loans the clear outlier category.

Commodities and FX. Neither leg carries a clean channel. Commodities reach this thesis only through a marginal regional-construction drag as campus building stops in shrinking regions, and the leg is deliberately unweighted. FX has one narrow line: education is a US services export through international students, exposed to visa policy, a channel that runs separately from this thesis's mechanics. A shrinking domestic pool pushes tuition-dependent universities toward international recruitment, which trades demographic risk for visa risk.

Spring 2026 undergraduate enrollment grew 1.3%, the third consecutive year of recovery, led by community colleges (up 3.1%), certificates (up 10.2%), and adult learners. If participation rates keep climbing among high-school graduates, adults, and international students, the shrinking 18-year-old pool is offset and the bottom tier gets a slower, shallower adjustment. National Student Clearinghouse actuals have beaten the enrollment-doom narrative for three straight spring reports, a real base rate in the bull case's favor. The offset has never been tested against an actually shrinking pool; the post-peak cohort only arrives in fall 2026. Holding the traditional freshman pipeline flat requires the immediate-enrollment rate to reach 65.6% by 2030 and 70.9% by 2041, the latter above the roughly 70% high recorded in 2016, while the rate has since fallen to 62.0%. So the offset, if it comes, comes through adults, certificates, and international students, channels that do little for small residential private colleges. The probability below covers the aggregate-enrollment version of the claim; the traditional-pipeline version runs closer to one in ten.

Estimated probability counter-argument is correct: 35%

The rise in recent-grad unemployment may be a tech-sector correction plus a graduate supply glut that clears on its own. Oxford Economics attributes the rise primarily to a structural shift in tech-sector hiring amid strong labor-supply growth. Its separate global work finds economy-wide evidence of AI-driven displacement "patchy," with firms not replacing workers with AI at scale. On that read, if tech hiring recovers and the graduate supply wave crests, the entry-level market re-clears and the labor leg of the thesis fades. Two things cut against a fast, benign resolution. Oxford's own attribution is structural, a hiring-model shift plus supply, which implies slower normalization than a cyclical dip. And the base rate is unforgiving. Of the six completed inversion episodes since 1990, every true normalization, meaning the graduate rate falling back below the workforce rate, came from episodes of five months or less. The two longer endings came from the all-workers rate rising to meet graduates (2001, COVID), which ends the inversion without helping graduates. Nothing in the record shows a multi-year inversion resolving to graduates' benefit.

Estimated probability counter-argument is correct: 25%

States could restore appropriations, the Education Department's ongoing OBBBA rulemaking could soften the caps' bite, or political backlash could reverse the caps before the first capped cohort borrows. Any of these would relieve the credit brake in Stage 4 and slow the closure pace. The historical base rate is unfavorable. State funding never fully recovered after the 1990, 2001, or 2008 downturns, the documented ratchet that shifted cost from states to families and held there. The caps passed through budget reconciliation, the filibuster-proof budget process, which carries the kind of political commitment that is hard to unwind mid-cycle. States do rescue flagship publics, and flagships are already the strong tier, so a state rescue tends to widen the gap between strong and weak institutions and does little to close it.

Estimated probability counter-argument is correct: 15%

Status codes read against the thesis: G means a thesis-confirming signal is currently active, Y means neutral or not yet observable, and R means a thesis-challenging signal. Levels are current as of July 24, 2026; the NY Fed's Q2 2026 household debt release is due in early August.

Table 2. Monitoring framework.

If fall 2026 freshman enrollment falls more than 2% and Q2 delinquency pushes past 11%, the thesis accelerates on schedule. If freshman enrollment grows despite the smaller high-school class and the unemployment gap closes below zero for two consecutive quarters, reassess.

Methodology note: BCR's four internal models (closure-hazard, grad-unemployment inversion, wage-premium plateau, and enrollment-cliff decomposition) were built and run on public federal panels (IPEDS, NY Fed labor and wage series, WICHE projections, and NCES enrollment tables) in July 2026. Three returned CONFIRMED and one PARTIALLY SUPPORTED. The model figures cited here trace to those runs: the 22.5%-versus-0.9% closure gradient, the 63-month inversion result, and the 91% cliff probability. Supporting statistics held out of the body: the closure model ranks a closing college above a surviving one 78.5% of the time, and both of its inputs clear conventional significance. The 2010 break in the 25th-percentile wage premium carries roughly 2% odds of arising by chance. A block resampling of the pre-2021 unemployment record, which preserves month-to-month persistence, puts the odds of a 63-month inversion at effectively zero. Delinquency comparisons use 90-plus days past due as the standard, consistent with Federal Reserve reporting conventions. All levels are current as of the publication date; the NY Fed's Q2 2026 household debt release is scheduled for early August 2026 and will supersede the Q1 delinquency figures cited here. The 2016 immediate-enrollment rate is reported by NCES as roughly 70% and is not statistically distinguishable from the 2010 reading, so it is described here as a recorded high rather than a definitive all-time peak.

This report is for informational and educational purposes only. It does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. All asset class commentary reflects historical patterns and educational analysis, not personal investment advice. Past performance does not guarantee future results. Readers should consult a qualified financial advisor before making investment decisions.