Ethan, thanks for having me. I’ve spent the post-pandemic years helping presidents and boards redesign programs, budgets, and digital delivery under mounting financial and demographic strain. The big themes we’ll unpack today are unmistakable: closures cluster where tuition dependence is high and enrollment is sliding; two-year for-profits face distinctly harsher odds; discounting and adult-market strategies can either stabilize net tuition or accelerate decline; and risk assessment markedly improves when you track trajectories—recent enrollment and cash signals—rather than static snapshots. We’ll also dig into what our predictive work shows—why an XGBoost model reached 83 percent accuracy versus 77 percent for standard federal metrics, how a 15 percent enrollment shock could trigger about 80 additional closures in a year, and what smaller colleges around 1,389 FTE can do to leverage intimacy at scale. Throughout, I’ll translate those findings into playbooks—quarterly operating targets, revenue diversification roadmaps, and rapid-response protocols that can keep students learning and communities intact.
Northland College closed in 2025 after 133 years. What specific warning signs there mirror patterns you’ve seen elsewhere, and how early did they appear? Walk us through a timeline, with anecdotes from campus leaders and concrete metrics that tipped you off.
When I visited campuses with Northland-like profiles, the early signal was dependence on tuition paired with softening yield, two or three cycles before the end. Leaders would tell me, “Our spring melt is the worst in memory,” while the net tuition line barely budged despite heroic discounting. In our broader data, schools that later closed had median margins around 3 percent versus 9 percent for peers that stayed open, and tuition accounted for 86 percent of revenue two years prior—those ratios don’t flip overnight. About two years out, you often see the 58 percent median year-over-year enrollment drop for future closures; on the ground, it feels like empty residence halls and under-enrolled gateway courses. Months before the announcement, deferred maintenance is frozen, auxiliary revenues sag from thinner activity, and presidents confide they’re modeling teach-out options alongside next term’s schedule.
You found 21% of for-profits closed versus 7% of nonprofits and under 1% of publics. What business model features drove those gaps, and which numbers matter most? Share examples of pivots that worked and steps that failed.
The core difference is revenue mix. For-profits rely on tuition for about 93 percent of revenue; nonprofits sit closer to 37 percent tuition with meaningful cushions from grants (18 percent), gifts (12 percent), and auxiliaries (about 8 percent). That concentration makes for-profits exquisitely sensitive to enrollment shocks and policy shifts. Pivots that worked started by cutting the tuition share of revenue—adding grant-funded training or philanthropic micro-scholarships tied to completion—while keeping operating margins closer to the 9 percent profile common among open schools. Failures tended to double down on tuition-only growth in shrinking markets, expanding sites or programs without evidence of demand, which only accelerates the slide.
Two-year for-profits had 33% closure rates. What operational levers could have shifted that trajectory, and which levers tend to be illusions? Describe a step-by-step turnaround plan you’ve seen succeed, with enrollment and margin benchmarks.
The viable levers are speed and focus: compress time-to-launch for high-demand certificates, align schedules with working adults, and build partnerships that bring non-tuition revenue to the table. Illusions include chasing rapid geographic expansion or stacking low-yield programs just to “add SKUs.” A workable sequence I’ve used: Q1, freeze nonessential hiring and resection courses to hit minimums; Q2, stand up short programs co-funded by employer grants; Q3, shift advising to proactive outreach keyed to early momentum; Q4, renew only programs with clear placement. As a guardrail, I ask teams to move away from the 86 percent tuition profile and back toward diversified revenue, and to close the operating gap toward the 9 percent margin norm rather than hovering near 3 percent.
The “demographic cliff” projects 13% fewer 18-year-olds by 2041. How should presidents translate that into five-year budgets, staffing plans, and program portfolios? Offer a scenario analysis with specific enrollment, tuition, and cost assumptions.
I advise presidents to budget with scenarios tied to enrollment trajectories, not just headcount goals. Using our study’s framing, model a no-change case, a sudden 15 percent enrollment drop, and a gradual 15 percent decline over five years, assuming revenue and expenses scale with enrollment. Then layer fixed-cost realities—facilities and tenured lines—to avoid overestimating flexibility. In practice, that means: pre-authorize contingent reductions if yield misses midyear; prioritize programs with demonstrated demand from adults and employers; and avoid tuition-only solutions where discounting already pressures net tuition, especially with the 51 percent average discount rate at nonprofits in 2022.
Adult enrollment has fallen by nearly half since 2008. What has actually worked to re-engage learners 25 and older, and what failed fast? Share campaign-level metrics, program formats, and conversion funnels that moved the needle.
What worked was simplifying the funnel and aligning delivery to adult lives: eight-week online courses, clear prior-learning pathways, and employer-tethered cohorts where grants or gifts cover part of tuition. Campaigns that succeeded were message-light and deadline-driven—“start in four weeks, finish in less than a year”—with advisors calling prospects within hours, not days. What failed was assuming adults would return for traditional schedules or that discounting alone would spark demand; in a sector where auxiliaries dipped during Covid and state pressures capped tuition growth, the adult market needs flexibility more than coupons. The tell is conversion speed—when outreach is fast and program formats fit, adult enrollment stabilizes instead of mirroring the broader multi-year decline.
Nonprofits’ tuition discount rate hit 51% in 2022. How do you judge when discounting boosts net tuition versus erodes it? Walk us through the data checks, yield curves, and price elasticity tests you recommend.
I start by plotting yield against award bands to spot inflection points where deeper discounts aren’t lifting yield. Then I examine net tuition per student alongside the discount rate—if net tuition flattens or falls while discounts rise, elasticity has turned against you. I also watch the tuition share of revenue; creeping toward the 86 percent profile is a red flag because it removes your cushions from grants, gifts, and auxiliaries. Finally, track cohort persistence—if higher discounts don’t improve second-term retention, you’re subsidizing churn rather than outcomes.
You report median operating margins of 9% at open schools versus 3% at schools that later close. Which line items tend to swing that spread, and in what order should leaders act? Give a playbook with quarterly targets.
Personnel dominates higher ed costs, and instruction typically accounts for about 26–30 percent of spend depending on sector. The spread shows up where staffing and sectioning drifted out of sync with enrollment. My playbook: Q1, align course sections to minimums and freeze noncritical admin hiring; Q2, renegotiate procurement and shift auxiliary operations to break-even; Q3, convert fixed costs where possible—shared services and digital course shells; Q4, reinvest savings into programs with proven enrollment momentum. The north star is moving toward the 9 percent margin profile and away from the chronic losses that precede closures.
Schools that closed saw median year-over-year enrollment drops of 58% two years prior. What real-time indicators let you see that slide early, and how do you stop it? Share leading metrics, weekly dashboards, and a rapid-response sequence.
Weekly dashboards should track inquiries, applications, FAFSAs submitted, admits, deposits, melt, and section fill rates. The earliest tell is deposit volatility and unfilled gateway sections, often weeks before census. The sequence is: stem melt with immediate outreach, resection to preserve course viability, and stand up short-start cohorts so students aren’t forced to wait a term. If tuition is already the dominant revenue line, protect net tuition per student—don’t paper over the slide with across-the-board discounts that won’t hold students to week six.
For-profits rely on tuition for 93% of revenue, while nonprofits lean on grants (18%), gifts (12%), and auxiliaries (~8%). How should a college rebalance revenue safely? Outline a 24-month roadmap with milestones and risk checks.
Months 0–6: inventory grant-eligible programs and price employer contracts that can launch quickly; set a target to shift a visible slice of revenue away from tuition dependence. Months 6–12: launch contracted cohorts and align development efforts to fund completion micro-grants that reduce attrition. Months 12–18: expand auxiliary offerings that are already profitable, mindful of the Covid-era dip we saw when activity fell. Months 18–24: institutionalize reporting—monthly revenue mix reviews—to ensure the tuition share trends closer to diversified profiles rather than drifting back toward 93 percent.
Publics get 22% from government versus 17% from net tuition. What lessons from that mix can nonprofits adapt without public subsidies? Offer concrete steps, from contract training to research partnerships, with target margins.
The takeaway is resilience through diversified, predictable streams. Nonprofits can emulate the stability of government support by building multi-year employer contracts, expanding grant-funded initiatives, and tuning auxiliaries to reliably contribute. Contract training and sponsored projects act like quasi-appropriations when structured over multiple years, while keeping tuition at sustainable net levels. Pair those with disciplined cost control so your overall margin heads toward the healthier 9 percent median among open institutions.
Your data set covers 8,633 institutions and 1,671 closures using IPEDS and PEPS. What data gaps most compromise forecasts, and how do you patch them? Describe the exact imputation or proxy choices and their error checks.
The biggest holes are debt, assets, and leverage—many institutions simply don’t report consistently. We mitigate by using related financials—cash on hand, operating margin, and revenue mix—as proxies, and by incorporating trajectory variables like recent and medium-term enrollment changes. Machine learning methods handle missingness better than traditional models, and we sanity-check imputations by comparing predicted risk against actual closures over 1996–2023. The aim is not perfect precision on any single ratio but reliable classification across thousands of schools.
Your XGBoost model predicted closures with 83% average accuracy, versus 77% for federal metrics. What features drove the lift, and how do you validate out-of-sample? Walk us through the training data, cross-validation, and confusion matrix takeaways.
The lift comes from blending distress indicators with trajectories. The Financial Responsibility Composite Score added about 4 percent predictive power, and recent enrollment changes contributed another 2.4 percent. We trained on private institutions from 2002 to 2023, validated against held-out years, and compared the confusion matrices across models; the XGBoost approach reduced false negatives—fewer at-risk schools missed—relative to the federal metrics. The acid test: among the top 100 highest-risk calls, 84 closed within three years versus 47 under the federal metrics model.
In the top-100 highest-risk schools, XGBoost flagged 84 that closed within three years. How should regulators and boards act on a signal that strong? Share a triage protocol, from Heightened Cash Monitoring to teach-out planning, with timelines.
Treat it like a weather warning: move from monitoring to action quickly. Step one is enhanced oversight—Heightened Cash Monitoring-style constraints that ensure funds flow against verified enrollment. Step two is independent viability assessment—can the institution get back to the 9 percent margin profile, or is it drifting toward sustained losses? Step three is contingency: parallel teach-out planning so students aren’t stranded, with community leaders engaged early because closures reverberate beyond campus.
Adding trajectory variables boosted prediction: recent enrollment changes added 2.4% predictive power; the Financial Responsibility Composite Score added 4%. How should oversight fold trajectories into policy? Propose thresholds, look-back windows, and intervention steps.
Oversight should weight direction as much as level. Use a look-back window that captures recent changes—think multiple terms of enrollment movement—paired with composite financial scores. When trajectories breach set bands—like steep multi-term enrollment declines—trigger stepped interventions: deeper audits, spending controls, and teach-out readiness. It’s a shift from waiting for a single failing score to watching the slope of the curve.
Your scenarios suggest a sudden 15% enrollment drop could mean 80 extra closures in a year. How should regions prepare for that shock? Map out labor market mitigation, space reuse, and transfer pathways with measurable goals.
Regions should pre-build transfer compacts so credits move seamlessly if a local college falters. Workforce boards can line up grant-funded reskilling cohorts that absorb displaced staff and students, minimizing community whiplash. Facilities can be repurposed for high-demand training or community services, keeping activity alive in spaces that anchor local life. With a shock of that scale—about 80 additional closures in a single year—the planning has to be done before the storm hits.
The median closed school enrolled about 1,389 FTE. What unique risks and advantages do sub-2,000-student colleges face, and how can they leverage intimacy at scale? Share examples of niche growth with retention and referral metrics.
Small colleges feel unit-cost pressures fast, but their intimacy can be an asset. Lean advising models and cohort designs can translate into stronger early-term momentum, which stabilizes enrollment ahead of the census dates that make or break budgets. Referral loops from satisfied students matter more at this scale—every retained or referred student moves the needle. The key is resisting program sprawl; deepen in niches where your advising, faculty attention, and community ties create outsized stickiness.
Expenditures rose as a labor-heavy sector, with instruction around 26–30% of spend. Where do sustainable savings come from without harming student success? Detail staffing models, shared services, and procurement wins, with before-and-after KPIs.
Savings endure when they respect the student experience. Shared services for back-office work, standardized course shells to reduce prep redundancy, and procurement resets deliver impact without hollowing out instruction. Instruction remains the heart—roughly 26–30 percent of spend—so protect teaching quality while aligning sections to demand. Before-and-after, I watch course fill rates, student persistence, and operating margins; if those improve in tandem, you’ve cut fat, not bone.
Investment returns shape nonprofit revenues, while auxiliaries dipped during Covid. How should finance teams hedge against market swings and event shocks? Walk through reserve policies, liquidity ladders, and stress tests, with target days cash on hand.
Treat liquidity as an academic program’s best friend. Build reserve policies that don’t assume bull markets will bail you out, add liquidity ladders so cash is available when auxiliary revenues dip, and run stress tests tied to enrollment shocks like the 15 percent scenarios. The goal is to keep teaching uninterrupted, even when markets hiccup or dorms empty. When leaders sleep at night, it’s because they know how many payrolls they can cover without tapping volatile assets.
Current federal measures missed many at-risk schools, and many institutions lack debt and asset data. What simple, standardized disclosures would most improve surveillance? Propose a short-form dashboard and the quarterly cadence to publish it.
A short-form, quarterly dashboard would do wonders: operating margin, cash on hand, tuition share of revenue versus grants, gifts, and auxiliaries, and rolling enrollment changes. Add the Financial Responsibility Composite Score and a flag for multi-term declines. This set maps closely to features that improved predictive power and would surface risk earlier than a single annual score. Publish quarterly so trajectories are visible, not buried in year-old PDFs.
When, if ever, should communities let a failing college close rather than rescue it? Lay out the decision tree: educational outcomes, externalities, fiscal multipliers, and teach-out quality, with thresholds that tip the balance.
Start with student outcomes—if the institution can’t deliver value students, employers, or society recognize, propping it up may do harm. Next, weigh externalities: is the college a genuine anchor for local education, mobility, and culture, or are there nearby absorbers with capacity and better outcomes? Factor in fiscal multipliers—keeping an unviable campus on life support can crowd out healthier institutions. If teach-out pathways are strong and the financial path back to the 9 percent margin profile is implausible, an orderly closure may be the most responsible act.
Do you have any advice for our readers?
Make trajectories your habit. If you’re a leader, build weekly dashboards around enrollment momentum and liquidity; if you’re a trustee, ask how today’s discounts affect net tuition and second-term persistence; if you’re a community stakeholder, map transfer and workforce options now, not later. The data are clear: institutions that diversify revenue, protect margins, and act early when enrollment shifts do far better than those that wait for a single failing score. The “demographic cliff” isn’t destiny, but it does reward those who watch the slope and steer accordingly.
