The Degree Was Earned

High school graduation produces an income premium larger than the ninth, tenth, and eleventh grades combined. The twelfth-grade diploma generates a bigger earnings jump than three years of attendance before it.¹

The pattern repeats at the next level. Senior year of college, the year in which a student completes a bachelor's degree, pays over twice as much as freshman, sophomore, and junior years combined. The final semester contains nearly identical lectures, coursework, and professors to the semester before it. The income jump is not close.¹

If education were primarily about acquiring skills, each year would add roughly equal value. A student who completes seven semesters has acquired nearly all the knowledge the eight-semester graduate has. The gap in learning between the two is trivial. The gap in earnings is enormous.

An economist at George Mason University spent years examining this anomaly. He called it the sheepskin effect, after the material diplomas were once printed on. His estimate: roughly 80 percent of the return to education is signaling. The diploma does not primarily certify what was learned, but what was demonstrated. Intelligence enough to gain admission. Conscientiousness enough to stay enrolled. Conformity enough to follow rules for four years and finish.²

The employer pays for the signal. Not because employers are foolish, but because the signal works. Hiring is expensive. The real requirements for a job are dispersed, contextual, and difficult to articulate. The knowledge that determines whether someone can do a specific role well is the kind of knowledge that resists compression, the kind that lives in the judgment of the person who has done the work.³ A credential compresses a person into something legible the way a forester once compressed a forest into board-feet.⁴ The simplification destroys information. But it makes the decision tractable.

The degree was a reasonable proxy once. When 11 percent of Americans over 25 held a bachelor's degree, the signal carried information. The problem began when the proxy became the point. When everyone optimized for the map, and nobody checked the territory.⁵

The signal

Nobody conspired. Nobody needed to. The incentive did the work.

2.52

In the early 1960s, the average grade-point average at American four-year institutions was approximately 2.52 on a 4.0 scale. By 2013, it had risen to 3.15. The climb was steady, unbroken, and nationwide.⁶

The students did not become 25 percent more capable over fifty years. A separate measure confirmed this. The amount of time students spent studying fell from roughly 40 hours per week in 1961 to approximately 27 hours per week by 2003. The grades went up. The effort went down. Both trends moved in the same direction for the same reason.⁷

The student optimizes for GPA and graduation. Takes the easier section that satisfies the same requirement. Avoids the professor who teaches more but grades harder. The behaviour is rational within a system that rewards the number. The student who takes the harder course and learns more but earns a lower GPA is penalized by the signal.

The professor optimizes for evaluations and retention. Research has shown that faculty who assign higher grades receive significantly better teaching evaluations, independent of actual learning outcomes. The incentive runs in one direction. Departments that fail too many students lose enrollment. Enrollment funds the department.⁸

Bachelor's required

The employer optimizes for the credential. A study by Harvard Business School and the consulting firm Accenture found that 67 percent of production supervisor job postings required a bachelor's degree. Only 16 percent of incumbent production supervisors actually held one. The same gap appeared across dozens of middle-skill occupations. The degree requirement was not measuring what the job required. It was filtering the applicant pool at low cost.⁹

The institution optimizes for rankings. US News & World Report awards 20 percent of its score to peer assessment surveys and weights student outcomes at 52 percent, measured by graduation rates, earnings, and loan repayment. None of these directly measure whether students learned anything. The institution that produces the most learning and the institution that produces the highest ranking are not necessarily the same institution, because learning is not what the ranking rewards.¹⁰

Four actors. Each individually rational. Nobody deceiving anyone. The system's incentive structure aligns everyone toward the signal, automatically, without instruction. The grades improved. The study time fell. The degree requirements inflated. And the thing all of these numbers were supposed to represent became invisible.

The evidence

Thirty-six percent

In 2011, two sociologists published a study that tested what four years of college actually produced. They administered the Collegiate Learning Assessment to approximately 2,300 students at 24 institutions, first when the students entered as freshmen and again when they graduated. The test measured critical thinking, analytical reasoning, and written communication.¹¹

Thirty-six percent showed no statistically significant improvement after four years.

They enrolled. They attended. They completed the coursework. They passed the exams. They graduated. They did not measurably improve at thinking.

The students who improved the least were concentrated in programs that required the least reading and writing. Students who took courses with more than 40 pages of reading per week and more than 20 pages of writing per semester showed significantly higher gains. The curriculum that produces learning and the curriculum that produces the credential had diverged. The degree was earned. The learning was not.¹¹

The new bachelor's

In 1970, approximately 11 percent of Americans over 25 held a bachelor's degree. By 2024, the figure had reached roughly 38 percent.¹²

The bachelor's degree is what the high school diploma was. A signal that once carried information now carries less, because more people hold it. So the system produces a new signal. Master's degrees conferred annually in the United States rose from approximately 230,000 in 1970 to over 880,000 in 2022. Nearly four times as many, in a population that had not quadrupled.¹³

This is credential inflation. It follows the same logic as monetary inflation. When the currency loses value, the response is not to reduce the supply. The response is to print larger denominations. Each new credential layer, the master's, the MBA, the specialized certification, is a new map drawn over the same territory. None measure the territory better. Each one raises the cost of entry.

Fifteen

In Germany, roughly half of students enter the dual system at age fifteen or sixteen. They spend two to three and a half years learning a trade inside a company while attending a vocational school two days a week. The system is called the Duales Ausbildungssystem, and it has operated in recognizable form for over a century.¹⁴

A seventeen-year-old in Stuttgart works on the floor of a precision manufacturing plant. She earns a training wage. She learns from a Meister, a master craftsman who has held the certification for twenty years and who is, by German law, the only person authorized to train apprentices in that trade. She practices on the machines that will determine whether the product ships or does not. After her training, she sits a practical examination administered by the local chamber of industry. The exam tests whether she can do the thing. Not whether she can signal that she might be able to do the thing.¹⁵

Germany's youth unemployment rate in 2024 was 6.7 percent. The United States, where the university system routes roughly 70 percent of high school graduates toward some form of postsecondary education, recorded youth unemployment near 10 percent. The OECD average was 11.2 percent.¹⁶

The German system optimizes for capability. The American system optimizes for the signal of capability. The same age group, sorted by a different system, producing a different output. The same finding that appears wherever the design of the system determines what the system produces.¹⁷

Three layers

The degree system is not uniform. It operates at three levels, each signaling to a different audience, each with a different cost structure.

The first is the functional degree. Medicine. Nursing. Engineering. The board exam tests whether you can do the thing. The clinical rotation puts your hands on the patient. The licensing authority reviews your work, not your transcript. This is the counter-case. The map matches the territory because the measurement is upstream of the outcome. A surgical safety checklist of 19 items, taking two minutes, reduced mortality by 47 percent across eight hospitals on four continents.¹⁸ The functional degree operates on the same principle. It measures the process that produces the outcome.

The second is the generic degree. Business administration. Communications. General studies. The map is a proxy. The degree signals completion, not capability. Eighty percent signaling. The employer uses it as a filter. The student pays for the filter. The credential is the slotting fee.

The third is the degree whose primary economic value is neither capability nor conventional employability.

Programs where the underemployment rate exceeds 55 percent. Programs where average debt exceeds average early-career salary by a factor of three or more. The signal these degrees send is aimed at a peer group, not employers. The credential signals cultural membership.

The structural finding is who enrolls. At 38 elite colleges, more students come from the top 1 percent of the income distribution than from the entire bottom 60 percent. The institutions that produce the highest-status credentials draw from the narrowest economic band.²⁰

The person from a wealthy family acquires this degree and absorbs the cost. Parental support, family connections, social networks that convert the credential into opportunities the credential alone does not create. The cost is absorbed. The person who borrows $80,000 for the same degree without that safety net bears the full weight. The signal is identical. The absorption capacity is not.

Rob Henderson arrived at Yale on the GI Bill after growing up in foster care and serving in the Air Force. He noticed that his classmates spoke a vocabulary that nobody in his hometown or the military used. The words did not primarily communicate meaning. They communicated membership. They proved you had been educated at a place like this one.²¹

He identified a pattern: beliefs and credentials that confer status on the affluent while imposing costs on those who adopt them without the same cushion. The luxury degree operates on the same mechanism. The credential confers cultural membership. The cost lands on whoever cannot afford the membership fee.²¹

The cost

The student pays. Not the employer. Not the institution. Not the politician who designed the loan program.

Average federal student loan debt per borrower: approximately $39,500. Total outstanding student debt in the United States: $1.84 trillion. The figure has roughly tripled since 2006.²²

In 2019, economists at the Federal Reserve Bank of New York published a study examining what happens when the government increases subsidized lending for higher education. For every dollar of subsidized federal loan made available, tuition rose by approximately 60 cents. The subsidy designed to make college affordable made it more expensive. The money flowed from the student through the institution into the cost structure the subsidy inflated.²³

The chess pieces moved. Tuition at four-year public institutions rose roughly 1,300 percent between 1980 and 2020, against approximately 275 percent general inflation over the same period. The gap is the cost of the subsidy flowing through the system and emerging as higher prices. The targets adapted. They always do.²⁴

Student loans cannot be discharged in bankruptcy. Congress made federal student loans non-dischargeable in 1976 and expanded the restriction to private student loans in 2005. The student borrows to pay for the signal. The signal depreciates as more people acquire it. The debt does not depreciate. The cost is permanent. The signal is temporary.²⁵

The person who designed the loan program bears no cost for being wrong. The institution that raised tuition bears no cost. The employer who requires the degree bears no cost. The student bears the entire cost. Knowledge in one place. Power in another. Consequences in a third. The same structure that appears every time the people who make the decisions and the people who live with the outcomes are not in the same room.²⁶

The counter-case

Not every degree is signaling. The functional degrees produce measurable capability. The board exam, the licensing test, the clinical rotation. The measurement is upstream of the outcome. You cannot game "did you confirm the patient's identity." Either you did or you did not. The medical board ensures the doctor can do the thing. The engineering exam ensures the bridge stands. The process metric improves the outcome. The proxy metric reshapes the territory.²⁷

The German dual system is the structural counter-case at national scale. A system designed for capability, not signaling, produces lower youth unemployment and higher skill match. Half the population routes into vocational apprenticeships. The other half attends university. The system sorts by aptitude and interest, not by the signal. The sorting produces outcomes that the American system, which sorts almost entirely by the credential, does not. The same finding: every system that produces depth has a traceable design.¹⁷

The argument is not that education is worthless. The argument is that the signal and the learning are not the same thing. And the entire system restructured itself around the signal.

The diploma

The degree was earned. That was never in question.

The student attended. Completed the coursework. Passed the exams. Walked across a stage in a rented gown and accepted a piece of paper that represented four years of time, an average of $39,500 in debt, and a signal that said the holder was capable.

The question is what "capable" measures. A set of skills that compounds over a career? Or a signal that depreciates as everyone else acquires it?

Thirty-six percent of graduates showed no significant improvement in critical thinking, analytical reasoning, or written communication after four years. Grades rose while study time fell. Tuition climbed 1,300 percent while the credential it purchased carried less information every decade. Master's degrees nearly quadrupled in half a century, not because the economy demanded four times as many, but because the bachelor's no longer separated anyone from anyone else. Each number improved. The thing each number was supposed to represent was never the thing being measured.

The diploma sits on the wall. The employer saw it. The loan officer approved on the basis of it. The ranking rewarded the institution that produced it. Everyone looked at the map.

Nobody checked the territory.

The earnings report was one visible number. The forecast was another. The price, the timeline, the shelf, the rating, the review. And now the diploma. Each one measured what it was designed to measure. Each one was calculated correctly. Each one was mistaken for the thing.

The degree was earned. The question is what was earned.