The following is the full opening chapter of We Need To Talk About Higher Education by Leon Shivamber, published in July 2026. The book applies investor-grade financial analysis to the college decision across nearly 6,000 institutions, 1,679 four-year degree-granting schools, and 26,000 distinct degree combinations.
See purchase details on the book page →
It is difficult to get a man to understand something when his salary depends upon his not understanding it.
Upton Sinclair
Somewhere today, a family is signing loan documents for a college program with a 67 percent underemployment rate. Nobody in that room knows the number.
One Nobel laureate explained exactly why college costs keep rising while value has fallen. We have ignored him for more than two decades.
The damage is what his theory predicts.
In 2001, the Nobel Prize in Economics went to three economists who changed how we understand markets: George Akerlof, Michael Spence, and Joseph Stiglitz.1 Their work on asymmetric information explained how markets can break down. The core problem is simple: one party knows more than another. When buyers and sellers have different information, the market can fail.
Akerlof showed how information gaps cause market failures. In his “Market for Lemons” example of used cars, sellers know more about quality than buyers do. Since buyers can’t tell which is which, they won’t pay a high price. They only offer an “average” price. Owners of good cars won’t sell for that low average price, so they leave the market. Only “lemons” remain, and eventually the market fails because no one wants to buy junk.
Stiglitz explained how the less-informed party can extract the truth through strategic choices. An insurance company does not know if you are a safe or risky driver. So they offer a choice: Cheap monthly payments, but you have a high deductible and pay a lot if you crash. Or, expensive monthly payments, but they cover everything if you crash.
The result? Safe drivers (who don’t expect to crash) will pick the first. Risky drivers will pick the second. By making a choice, customers “reveal” who they really are.
Spence demonstrated how the more-informed party can prove quality through costly signals.
It is Spence’s work that should terrify anyone paying for college right now.
In his Nobel lecture, Spence took on a question that still puzzles families today. Why would employers pay more for a credential that does not make the worker more productive?2
His answer, in the simplest version of his model: even a credential that taught nothing could still pay, because it sorts workers by the signal the diploma sends. He stressed that this was a simplification, not a claim that learning does not matter.2
This concept, signaling, has become foundational to modern economics.
It explains why a diploma from an elite institution opens doors that identical skills from a cheaper school do not. In that model, the degree is not valued for the knowledge transferred.
It is valued for what it supposedly signals about who you are: intelligent, disciplined, capable of completing complex tasks.2
The evidence is now clear: he was right. And we have spent those decades ignoring him while millions of families make financial decisions that do not pay off.
The Math Nobody Shows You
Most families do not know this when evaluating college: the “college premium,” that famous statistic showing college graduates earn more over a lifetime, hides a structural failure in how we allocate education resources.
When you attend college, you are investing. You spend four years and roughly $200,000 at a public college’s net price to about $400,000 at a private college’s full published price, in total economic cost, counting both direct costs and the wages you did not earn while in college.3
The expectation? Higher earnings will justify that investment.
But the averages hide what is underneath.
Look at the data for nearly 4,000 colleges. A decade after enrolling, most former students at 26 percent of those institutions earn less than $32,000 a year. That is what a typical high school graduate earns.4
These are not outliers in struggling schools. They are substantial programs at recognizable institutions.
My analysis takes this investigation further. Looking specifically at 1,679 four-year institutions serving about 11.5 million students, I find something stark. Nearly 9 percent of these college students, about 1 million people, attend institutions where the median graduate earns too little to justify attending even at zero cost. Nearly 30 percent of these institutions, 500 of the 1,679, fall into that group.5
Even if college were free, attending these institutions would still be a poor investment due to opportunity costs alone. Four years of forgone wages, four years of delayed career progression, four years of not building work experience. The median graduate from these institutions would have been financially better off starting work at 18. That is a comparison of medians, not a forecast for any one student. Chapter 31 takes up the difference.
The headline “college premium” you hear about combines excellent outcomes with terrible ones.
It’s like saying “the average American is financially comfortable” when wealth concentration means half the country lives paycheck to paycheck.
The number is technically accurate. It is also deeply misleading.
Signaling theory explains the dysfunction.
Employers cannot directly observe your productivity, work ethic, or intelligence before hiring you. They need a proxy. For decades, that proxy has been your diploma. Employers assumed the credential signals something meaningful about your capabilities.2
The math broke when the signal weakened and the costs exploded.
Consider two data points.
The Federal Reserve Bank of New York takes a snapshot of all recent college graduates aged 22 to 27. As of Q4 2025, about 42 percent of them are underemployed, working in jobs not requiring a bachelor’s degree.6
A separate study by Burning Glass and Strada follows individual four-year graduates over time, rather than measuring a current age cohort. It finds 52 percent are underemployed in their first job, and 45 percent are still underemployed a decade later.7
Different studies. Different methods. Same direction.
This is not a temporary setback. It is a structural feature of the market college graduates are entering. If the credential were a reliable proxy for productivity, employers would find a use for it. They do not.
Read that again. Nearly half of the graduates in that study are working in jobs that do not require their degree. Not temporarily. Not for a year or two. A decade later, they are still there.
These outcomes expose the flaw at the heart of the signaling premium.
The signal is not functioning as intended. Yet the cost of obtaining it continues to rise.
You might push back here. The wage premium still exists. College graduates still earn more than high school graduates, on average. If signaling is failing, why do the earnings data still show a premium?
The objection is accurate as stated. But the flaw lies in the word “average.” The premium that persists is real for graduates in nursing, engineering, accounting, and computer science. In these fields, employers have validated the signal. Graduates actually perform the work their degrees promise. But the premium does not reach everyone. At the net price families actually pay, the median graduate’s lifetime after-tax return on the investment is negative at 57.6 percent of four-year institutions, more than half.5
The headline “college premium” blends these strong and weak results into one number, and that number hides how many schools leave their median graduate behind.
This is what signaling theory predicts when the cost of the signal climbs but the productivity behind it does not. The winners keep winning. The signal retains nominal value. And the losses spread quietly across the schools and programs where credential supply has outrun employer demand. The headline looks fine. The spread hides the wreckage.
A 2025 analysis constructed from 2023 Federal Reserve Bank of New York primary data shows dramatic variation in underemployment by major. Table 1-A shows a sampling.8
| Major | Underemployment rate |
|---|---|
| Criminal Justice | 67.2% |
| Performing Arts | 62.3% |
| Liberal Arts | 56.5% |
| Computer Science | 16.5% |
| Chemical Engineering | 16.5% |
| Nursing | 9.7% |
| Source · “The Labor Market for Recent College Graduates,” Federal Reserve Bank of New York (2023 data)8 | |
When 67 percent of criminal justice graduates are underemployed, that is not an aberration. The market produced exactly this outcome: a credential supplied at several times the level employers actually need, and the students who bought it are paying for the miscalculation.
These are not small programs. Tens of thousands of students choose criminal justice and performing arts each year, accumulating debt for credentials employers demonstrably undervalue.
The market is sending a clear signal: We need nurses and engineers. We don’t need this many degrees in criminal justice or performing arts.
Yet students keep pursuing those paths.
Meet DeShawn. He’s 19, enrolled in a criminal justice program at a local college in North Carolina. His net price after grants is about $13,500 a year. He plans to become a probation officer, stable work, public service, the kind of career his high school counselor described as “solid.”
The program accepted him. The federal loan office approved him.
Nobody mentioned that in the Fed’s 2023 data, 67.2 percent of recent criminal justice graduates worked in jobs that did not require their degree. Nobody showed him that the median earnings for criminal justice graduates at his institution, four years after graduation, are $35,000. By age 26, an 18-year-old who went straight to work after high school would be earning about $42,000, some $7,000 more, with no debt.
Spence’s model fits DeShawn’s situation. The signal has a cost: four years, $54,000 in direct costs (below the $80,000 four-year median, consistent with his school’s $13,500 annual net price), plus the income he is not earning. When that cost is disconnected from the value employers attach to the signal, the student who buys it loses. The institution selling it moves on to the next enrollment cycle.
DeShawn didn’t fail the signal. The signal failed him.
Why? Because the conventional wisdom remains “college is always worth it,” and the diploma still signals “educated” even when employers do not want what you studied.
This result fits Spence’s model, in which a costly signal adds nothing for society as a whole.2 When a signal stops delivering value but persists through inertia and information gaps, almost everyone loses.
The Data Hidden in the Averages
I have spent considerable time analyzing College Scorecard data, Federal Reserve research, and Department of Education reports because this information should be easier to access than it is. That opacity hides the problem.
The numbers:
Scale of the problem:
- $1.7 trillion in outstanding student debt across approximately 43 million borrowers as of Q4 20259
- 52% of recent four-year college graduates are underemployed one year post-graduation7
- 45% still underemployed ten years later7
- In more than 25% of higher education institutions, most former students earn less than a typical high school graduate ten years after enrolling4
These numbers do not coexist by accident. They are the result of a system that lends without checking and certifies without disclosing. The 18-year-old and parents at the loan desk are the only parties in the room who bear the full risk of what follows.
Information asymmetry:
When you invest in stocks, the SEC requires detailed disclosure about historical performance, fees, and risks.
When you borrow for a home, lenders carefully underwrite both you and the property.
When companies go public, securities laws mandate extensive disclosure.
If this were a mortgage, it would be illegal. But because we call it education, federal loans have flowed to almost any student at almost any institution, regardless of how its students fare. A federal earnings test signed into law in 2025 applies an outcomes check to degree programs at public and nonprofit colleges for the first time, but it only asks whether a program’s graduates out-earn a typical high school graduate in the college’s state, and the first penalties cannot land before July 2028.10 The lender still does not check the risk for any individual borrower. The school keeps the tuition whether you graduate or not, and bears almost none of the downside. The entire burden falls on an 18-year-old and the family.
This is why economist Bryan Caplan argues in “The Case Against Education” (2018) that education works mainly as signaling rather than skill-building.11 It is why I dug into the institution-level data myself, to find where the failures concentrate.4 5
The Sheepskin Effect
Caplan’s case rests on a striking pattern in the data. He calls it the sheepskin effect, named for the parchment on which diplomas were once printed. The idea: an extra year of school appears valuable when you look at the wage premium in aggregate, but most of that value collapses into the diploma itself. Years short of the credential pay much less. The credential pays a lot.
To test this, Caplan ran a regression on the General Social Survey, the long-running survey of American workers that asks both years of education completed and highest degree earned. From 1972 through 2012, he found that an extra year of school appeared to raise earnings by 10.9 percent. After controlling for whether the worker actually finished a degree, the year-by-year payoff collapsed to 4.5 percent. The finishing line did the work. Crossing the bachelor’s threshold added 31.4 percent.12
His conclusion: a large share of the apparent education premium is not skill-building. It is signaling.
I went back to the source data and ran the same regression on Caplan’s exact period to confirm the methodology. My numbers land close to his on the years-of-education estimates and on the bachelor’s. The high school and graduate coefficients came out different, probably because the GSS file’s variable coding has changed in the years since. The thesis holds. Years pay little once you account for diplomas. The bachelor’s matters most.
Then I ran the same test on the 2014 to 2024 data to see what the sheepskin effect looks like in the current labor market.
| Education | Caplan original (1972 to 2012) | Replication (1972 to 2012) | Updated (2014 to 2024) |
|---|---|---|---|
| Years of education (years only) | +10.9% | +9.5% | +8.5% |
| Years of education (with diplomas) | +4.5% | +4.2% | +2.1% |
| High school diploma | +31.7% | +8.5% | +11.3% |
| Bachelor’s degree | +31.4% | +36.2% | +47.7% |
| Graduate degree | +18.2% | +72.5% | +68.7% |
Source · Caplan, The Case Against Education (2018), Chapter 4, Table 4.1 · Author’s replication using GSS 1972 to 2024 Cumulative File, Release 3 · All results control for age, age squared, race, and sex. Sample restricted to labor force participants.
The pattern Caplan saw has gotten stronger. In the current period, an extra year of school adds just 2.1 percent once you account for whether you finished. That is roughly half what it added in Caplan’s window. Meanwhile, finishing the bachelor’s is worth more than ever. The payoff has climbed from 36 percent in the replication of his window to 48 percent in the current window. The graduate degree continues to add about 70 percent.
This is exactly what Spence’s signaling theory predicted. The cost of the diploma keeps climbing. So does the premium attached to it. The 48 percent bachelor’s coefficient lands in the same range as the wage premium I use as a starting point in this book’s NPV model, drawn from New York Fed and Census data. Different data, different method, same direction. The credential pays a lot, and it pays more than it used to.
Call it a finishing-line economy. The data since show the finishing line has moved farther from the runner.
Before dismissing the premium entirely, we must face an uncomfortable reality. The median college graduate would have been better off financially by starting work at 18, about $17,500 behind at the median net price and about $82,000 behind at full cost of attendance. But that does not mean college is worthless. It means we have a massive sorting problem.
STEM programs, healthcare, accounting, and selective schools produce large returns. They often exceed $500,000 in Net Present Value (NPV). The model introduced later details this. But at more than half of four-year institutions, the median graduate’s investment does not justify the cost at the net price families pay.
College itself is not the issue. The failure is treating all colleges as equivalent. We steer students into expensive programs with weak outcomes while telling them “any degree is better than no degree.”
What the Theory Says to Do, and What We Did Instead
Most people haven’t read Spence’s work. They’ve just heard about “signaling.” But his findings are more nuanced and more useful than the popularized version.2
Spence demonstrated that when a signal becomes expensive but does not improve productivity, it creates an equilibrium where everyone overinvests. You have to get the degree, even if it does not make you better at the job, because employers expect it. It is a costly game reducing social welfare.2
The fix follows from the same logic. When an expensive signal stops working, you have to force the market to correct itself. Four moves follow:
- Uncover the truth: Make the outcomes of these degrees visible to everyone.
- Make people pay the real price: Stop subsidizing credentials that do not deliver, so students stop being directed into them.
- Let competition work: Allow the market to expose and penalize programs that do not deliver.
- Stop the waste: Use smart funding rules to actively discourage students from spending years and thousands of dollars on dead-end paths.
This set of changes is precisely what should happen in higher education. Instead, we have done the opposite. We have left the earnings and graduation data where families rarely look. We have insulated institutions from consequences. We have prevented price signals from working. We have subsidized expensive credentials regardless of whether they deliver value.
Imagine if financial markets worked like higher education. Brokers could sell products without disclosing failure rates. Risk-free lenders had zero incentive to verify quality. The entire risk fell on the buyer.
That is higher education. The reforms that would change it are the subject of Part III.
For now, the point is simpler: the data to evaluate college as an investment exists, is public, and is rarely consulted before the loan documents appear. The College Scorecard publishes earnings data. The Federal Reserve analyzes employment by major.
We could fix the information problem tomorrow.
We have not, because no one in the system has had a reason to. Transparency would show that some prestigious colleges deliver worse returns than public universities, and that donors allocate by prestige rather than impact.
Even that premium requires scrutiny. Does an elite university actually generate that extra wealth? Or does it simply admit highly capable students who were destined to be high earners regardless of where they enrolled? When you adjust for the baseline capability of the students, the so-called ’elite premium’ shrinks dramatically.
A Case Study in Capital Allocation
Marc Rowan, CEO of Apollo Global Management, chaired the Board of Advisors of the Wharton School at the University of Pennsylvania for nine years, through June 2026. He donated $50 million in 2018, the largest single gift in Wharton’s history at the time. He’s one of the most sophisticated investors in the world.13
Is that a good investment from a social impact perspective?
According to my ROI analysis of the College Scorecard data, a Penn graduate generates about $789,000 more in pre-tax lifetime earnings NPV than a high school graduate.13
At roughly $387,000 per student in nominal four-year cost of attendance, Rowan’s $50 million funds about 129 full scholarships. That adds about $102 million in social value. Measured against his after-tax cost of about $31.5 million, once the 37 percent charitable deduction is applied, that is a 223 percent return.
That is impressive. But consider the alternative.
The same $50 million at CUNY Baruch College, where the four-year nominal cost is about $62,000 per student, funds 811 full scholarships. Baruch graduates generate about $331,000 more in pre-tax lifetime earnings NPV than high school graduates.13 Total social impact: approximately $268 million. That is a 752 percent comparable social return (Table 1-C).13
Every donor dollar to Baruch produces 3.4 times the social return of the same dollar to Penn.
Pell share, mobility rate, and social ROI all favor Baruch. Same national job market, different returns per donor dollar.
| Institution | Pell share | Mobility rate | Social ROI |
|---|---|---|---|
| Baruch | 57.4% | 12.9% | 752% |
| Penn | 17.3% | 2.1% | 223% |
| Baruch ÷ Penn | 3.3× | 6.1× | 3.4× |
Source · U.S. Dept. of Education College Scorecard · Opportunity Insights mobility data · Author’s analysis13
Data from Opportunity Insights, Harvard economist Raj Chetty’s landmark study of social mobility, show that 12.9 percent of Baruch’s students both grew up in the bottom 20 percent of the income distribution and reached the top 20 percent. At Penn, the figure is 2.1 percent.14
Not marginally better. Not twice as effective. Baruch’s mobility rate is six times Penn’s, largely because Baruch enrolls nearly eight times Penn’s share of students from the bottom fifth.14
If one of the most sophisticated capital allocators in the world makes this choice without seeing the efficiency gap, a family signing loan documents on a Tuesday afternoon has no chance of finding it on their own.
I’m not criticizing Rowan. Giving to an Ivy League endowment is a perfectly rational, low-risk way to purchase elite legacy and social capital. It makes sense for him. But we must stop pretending that these massive donations are driven by a desire for efficient social impact.
The opacity matters more than the motivation. Even the most sophisticated capital allocators cannot see the efficiency gap.
Marc Rowan made a $50 million gift to Wharton. DeShawn is on track to leave with about $34,000 in debt, the median for his program, while his parents spend another $20,000 for him to attend college. Both are capital allocation decisions about higher education. They never appear in the same conversation.
When was the last time you saw universities ranked by social return on investment? By NPV created per dollar of aid? By social mobility generated?
We rank by prestige. By selectivity. By amenities. Not by outcomes.
That is a market failure. And market failures create victims.
The Uncomfortable Reality About Signaling
Here’s what makes this conversation difficult: signaling is not inherently bad. And it is not going away.
Employers need a mechanism to identify capable workers from everyone else. A college diploma does communicate something real. If you complete a rigorous degree, you’ve demonstrated discipline, intellectual capacity, and the ability to finish what you start. That has genuine value.2
Signaling alone is not the issue. The danger is costly signaling that fails.
When the cost of the signal, four years of forgone wages plus tuition and living costs, becomes disconnected from the value it delivers, the system breaks down. The value delivered ranges from negative returns, where the graduate ends up worse off than skipping college, to $500,000 or more, depending on the program and school.
The fix that follows is elegant: keep signaling, but reduce its cost and increase transparency, so people can make informed decisions about whether the signal is worth purchasing.
That is precisely what market discipline accomplishes. It forces colleges to do one of three things:
- Improve outcomes (some forward-thinking universities are already doing this)
- Reduce costs (community colleges frequently deliver stronger ROI than four-year schools for this reason)
- Shut down or consolidate (the ones that do neither)
The market isn’t cruel. It’s clarifying. It rewards value and weeds out underperformance. Higher education desperately needs that clarity.
What This Means If You Are Making College Decisions
If you are making college decisions now, Spence’s work implies one obligation. Look past the university’s brand. Examine the specific program’s outcomes.
You can find median earnings by field of study on the College Scorecard. You can find underemployment rates by major through the Federal Reserve. This data is free and current. It is rarely consulted before the loan documents appear.
Most colleges fall into one of two camps: institutions where the signal still pays, and institutions where the math stopped working years ago. Knowing which is which is the only question that matters.
The Cost of Ignoring Spence
Over more than two decades of ignoring him, student debt grew from approximately $240 billion in 2001 to $1.7 trillion in 2025.9
Not a modest increase. Not double or triple. A roughly seven-fold increase.
Families are making financially ruinous decisions for credentials that do not deliver. Institutions almost never face consequences for failure. The wealth gap widens because only affluent families can absorb the cost of weak signals.
The tragedy is not that we do not know what to do. We do.
Spence laid out the theory.
The tragedy: we have chosen institutional convenience and opacity over student welfare.
We ignored him.
More than two decades later, the damage is overwhelming.
DeShawn is still enrolled.
Keep reading
This is Chapter 1 of We Need To Talk About Higher Education. The rest of the book builds the full financial model, names the schools and majors where the math works and where it fails, and lays out what families, students, and counselors should do about it.
Get the book → Read the argument in full → Run your own numbers →
Notes
- The Royal Swedish Academy of Sciences. “The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001.” Press Release, 10 October 2001. The prize was awarded to George A. Akerlof, A. Michael Spence, and Joseph E. Stiglitz “for their analyses of markets with asymmetric information.”
- Spence, A. Michael. “Signaling in Retrospect and the Informational Structure of Markets.” Prize Lecture, December 8, 2001, pp. 411, 424, and 428 (PDF pp. 5, 18, and 22). Spence’s 2001 Nobel Prize Lecture explains how signaling equilibria develop when employers cannot observe worker productivity directly, and how costly signaling can lead to social inefficiency when signal costs exceed productivity benefits.
- The College Board. “Trends in College Pricing and Student Aid 2024.” October 2024, pp. 10, 18, 48. Reports average published tuition and fees of $11,610 for in-state public four-year institutions and $43,350 for private nonprofit four-year institutions for 2024-25, with an average net cost of attendance (after grant aid) of approximately $20,780 at public four-year schools. Note: Four years of forgone wages at approximately $30,000-$40,000 a year for high school graduates come to $120,000-$160,000, and more at the median time-to-degree of 52 months. Adding four years at the public four-year average net cost of attendance ($20,780) gives roughly $200,000. Adding four years at the private nonprofit four-year average student budget ($62,990 in 2024-25, or $251,960 over four years) gives roughly $370,000 to $410,000.
- Itzkowitz, Michael. “Ensuring a Living Wage Through Higher Education.” The HEA Group, February 2024. Analysis by HEA Group, using U.S. Department of Education College Scorecard data, which track nearly 5 million students across approximately 3,887 institutions, found that at 26.3% (1,022 of 3,887) of higher education institutions, the majority of former students earned less than $32,000 annually (the earnings of a typical high school graduate) ten years after initially enrolling.
- See Chapter 2 “Higher Education Needs a Transformation” (Figure 2-H) for the analysis of College Scorecard data on 1,679 four-year institutions serving 11.5 million students. At 500 of them, which enroll roughly 1.0 million students, the median graduate does not break even against a high school graduate even at zero cost. At 967 of them (the 467 that clear only when college is free plus the 500 that fail even then), or 57.6 percent, the median graduate falls behind at the median net price.
- Federal Reserve Bank of New York. “The Labor Market for Recent College Graduates.” Updated regularly.
- Hanson, Andrew, Salerno, Carlo, Sigelman, Matt, de Zeeuw, Mels, and Moret, Stephen. Strada Institute for the Future of Work and The Burning Glass Institute. “Talent Disrupted: College Graduates, Underemployment, and the Way Forward.” February 2024. See PDF at https://static1.squarespace.com/static/6197797102be715f55c0e0a1/t/65fb306bc81e0c239fb4f6a9/1710960749260/Talent+Disrupted+03052024.pdf. Reports 52% of four-year college graduates are underemployed in their first job one year after graduation, with 45% remaining underemployed ten years later (see p. 10).
- Federal Reserve Bank of New York. “The Labor Market for Recent College Graduates.” Underemployment rates by major (2023 data, recent graduates aged 22 to 27). Figures compiled from the highest- and lowest-underemployment tables in Hennerich, Heather. Federal Reserve Bank of St. Louis, “The Jobs and Degrees Underemployed College Graduates Have.” Open Vault, August 13, 2025. The article’s two tables list, by major: criminal justice (67.2%), performing arts (62.3%), liberal arts (56.5%), nursing (9.7%), computer science (16.5%), and chemical engineering (16.5%). Data are from 2023.
- Federal Reserve Bank of New York. “Quarterly Report on Household Debt and Credit, 2025:Q4.” Center for Microeconomic Data. February 2026. https://www.newyorkfed.org/microeconomics/hhdc.html. Accessed April 6, 2026. Reports outstanding student loan debt of $1.66 trillion as of Q4 2025. Additional federal sources report figures ranging from $1.65 to $1.81 trillion, depending on methodology. Author estimate extrapolated from FFEL/Direct Loan portfolio data; the NY Fed Household Debt and Credit Report series begins in Q1 2003 at $243B. Historical data from Stoll, Adam. Federal Student Loans: Program Data and Default Statistics, report, September 23, 2002; Washington D.C. (https://digital.library.unt.edu/ark:/67531/metadc810096/m1/5/ Accessed February 3, 2026), University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu crediting UNT Libraries Government Documents Department. Shows FFEL loans at approximately $179 billion and DL loans at $59 billion at end-FY2001. U.S. Department of Education, Office of Federal Student Aid. “Federal Student Loan Portfolio by Borrower.” Data Center, studentaid.gov, spreadsheet Accessed February 3, 2026. https://studentaid.gov/data-center/student/portfolio lists 42.7 million borrowers.
- National Association of Student Financial Aid Administrators (NASFAA). “Program Completer Earnings Accountability From the One Big Beautiful Bill Act: Brief for Campus Leadership.” https://www.nasfaa.org/uploads/documents/OB3_Accountability_Changes_Brief_for_Campus_Leadership.pdf. Accessed September 24, 2026. The One Big Beautiful Bill Act was signed into law on July 4, 2025. An undergraduate program passes if its graduates, four years after completion, earn more than the median working high school graduate aged 25 to 34 in the institution’s state. The framework takes effect July 1, 2027, with optional early implementation from July 1, 2026, and July 1, 2028 is the earliest a program could face penalties.
- Caplan, Bryan. “The Case Against Education: Why the Education System Is a Waste of Time and Money”. Princeton University Press, 2018. Argues that 50-80% of observed education wage premiums result from signaling and ability bias rather than human capital development, with extensive analysis of why education persists despite weak productivity benefits.
- Caplan, Bryan. “The Case Against Education: Why the Education System Is a Waste of Time and Money.” Princeton University Press, 2018. Chapter 4, Table 4.1, p. 99. Caplan’s original sheepskin regression on the General Social Survey, 1972 to 2012, with published coefficients of 10.9 percent (years only), 4.5 percent (years with diplomas), 31.7 percent (high school), 16.6 percent (junior college), 31.4 percent (bachelor’s), and 18.2 percent (graduate).Author’s replication using the General Social Survey 1972 to 2024 Cumulative File (Release 3), accessed via UC Berkeley SDA at sda.berkeley.edu. Sample restricted to labor force participants. Dependent variable is log respondent income (REALINC). Two specifications: years-of-education only, and years-of-education plus diploma dummies for high school, bachelor’s, and graduate. Controls: age, age squared, race, sex.Replication on Caplan’s 1972 to 2012 window: year-of-school 9.5 percent (years only) collapsing to 4.2 percent with diplomas; high school 8.5 percent, bachelor’s 36.2 percent, graduate 72.5 percent. Replication on the 2014 to 2024 window: year-of-school 8.5 percent collapsing to 2.1 percent with diplomas; high school 11.3 percent, bachelor’s 47.7 percent, graduate 68.7 percent. The replication of Caplan’s window comes close to his published coefficients on the years-of-education estimates and on the bachelor’s coefficient; divergence on the high school and graduate coefficients likely reflects evolution in GSS variable coding since 2018.
- Wharton University of Pennsylvania.“Wharton Receives $50 Million Gift from Marc J. Rowan and Carolyn Rowan for Teaching, Research, and Leadership.” October 2, 2018. Documents the $50M gift to the Wharton School (at the time the largest single gift in Wharton’s history) from Rowan, who then chaired the Wharton board the release calls its Board of Overseers, the board current reporting calls its Board of Advisors. On his nine-year term as chair, which began in 2017 and ended June 30, 2026: Chitirala, Isha, and Ni, Jasmine. “James Dinan to succeed Marc Rowan as chair of Wharton Board of Advisors.” The Daily Pennsylvanian, November 7, 2025. On the 2018 gift as the record at the time: Snyder, Susan. “Billionaire Marc Rowan quietly donated to Penn and still leads Wharton advisory board, while helping with Trump’s university compact.” The Philadelphia Inquirer, October 17, 2025, which reports the gift “was then the largest single gift Wharton had ever received.” See Chapter 12 “Why Big Donors Prefer Elite Schools” for comparative analysis using College Scorecard data showing that Penn graduates generate approximately $789,000 more in pre-tax lifetime earnings NPV than high school graduates, versus $331,000 more for Baruch graduates, but the per-dollar social return is 752% at Baruch versus 223% at Penn due to lower costs.
- Chetty, R., Friedman, J., Saez, E., Turner, N., and Yagan, D. “Mobility Report Cards: The Role of Colleges in Intergenerational Mobility.” The Equality of Opportunity Project. Opportunityinsights.org. Opportunity Insights at Harvard University, Accessed July 12, 2024. NBER Working Paper No. 23618, July 2017. The Equality of Opportunity Project. “Companion data.” The New York Times. “Baruch College.” ; Social mobility rates (bottom quintile to top quintile): Baruch: 13%.The New York Times. “University of Pennsylvania.” Penn: 2.1%.