In July 2025, Congress tied federal student loans to a single question: do a program's graduates earn more than people who stopped at high school?¹ At the end of December 2025, the U.S. Department of Education released program-by-program data showing how colleges would fare.^2,3 For 38,301 undergraduate programs, it shows what graduates who received federal aid, and who were working and not in school, earned four years after they finished: about three million graduates in all.

The data were built to enforce a minimum standard, but they answer a more useful question as well. They show, field by field and college by college, what the first years of a career pay. This briefing uses them for that purpose.

A program loses access to federal Direct Loans if, in two of three years, the median earnings of its recent graduates fall short of a benchmark.¹ The graduates counted are those who received federal student aid and who, in the fourth tax year after finishing, are working and not enrolled in school. For an undergraduate program, the benchmark is the median earnings of working adults aged 25 to 34 who are not in school and whose education stopped at a high-school diploma. It is measured in the college's state, or nationwide if fewer than half of the college's students come from that state.^1,4

The law covers degree programs and graduate certificates. The Department's final rule, published on July 1, 2026, applies the same test to undergraduate certificate programs, and it takes effect on July 1, 2027.⁴ The first official results will be calculated in early 2027, from students who finished in 2020–21, and the earliest any program can lose federal loans is the school year that begins on July 1, 2028.^4,5 Pell Grants are at risk only in programs that have already lost loans, and only at colleges where, in two of three years, more than half of aid recipients or aid dollars are in such programs, with exemptions.⁴

The data in this briefing are the Department's preview. Its December release, which it calls the 2026 Program Performance Data, covers 209,321 programs at 5,096 colleges. Most have no earnings result, because they had no graduates in the measured classes or fewer than 16 with tax records. This briefing uses the 38,301 undergraduate programs that have one. The figures pool graduates from the 2017–18 and 2018–19 school years, with earnings measured in 2022 and 2023 and adjusted to 2024 dollars.² The Department calls the release informational. The official test will use finer field codes, and single-year classes where programs are large enough, so its results will differ.^2,6

On the preview data, failure is rare for degrees and common for short certificates. Of 24,936 bachelor's programs, 300 fail, or 1.2%. Of 7,642 associate programs, 504 fail, or 6.6%. Of 5,723 undergraduate certificate programs, 1,659 fail: 29%. Because the failing certificate programs tend to be large, they hold 37% of certificate graduates in the data.² The Department's own estimate in its final rule, which also covers graduate programs and counts only the roughly 61,900 programs it expects the rule to cover, is that 5.2% would fail, enrolling 4.2% of their students.⁴

A single field accounts for much of the certificate failure. Of 896 cosmetology certificate programs in the data, 830 fail, or 93%, the same share the Department estimates in its final rule. They make up half of all failing certificate programs.^2,4

Among degree programs, the visual and performing arts stand out. They account for 256 of the 804 failing associate and bachelor's programs, about a third. At the bachelor's level the share is higher: 184 of 300, or 61%. The rest are spread thinly across fields. Most are associate programs, in fields such as medical administration and assisting, teacher education and general studies.

Passing the test says only that a program's typical graduate out-earns a typical high-school graduate. It says nothing about by how much. Exhibit 3 sorts every bachelor's program by the size of that gap. About half of them, 51%, either fail or clear the bar by less than $25,000 a year. About one in six, 16%, clear it by $50,000 or more.²

The test, in other words, is a floor. For a family choosing among programs that all pass, the useful question is how far above the floor a program lands. Parts II and III show that the answer depends first on the field a student chooses, and then, in some fields, on the college.

Many families choose a college first and leave the major for later. The federal data suggest weighing them the other way around. Across the country's bachelor's programs, knowing a program's field tells you much more about what its graduates earn than knowing its college.

To measure this, we asked how well each factor, on its own, accounts for the differences in median pay among the 24,601 bachelor's programs in the 50 states and Washington, D.C. Knowing only a program's major accounts for 64% of those differences. Knowing only its college accounts for 30%. The state the college is in accounts for 4%. Together, major and college account for 82%.⁷

The two overlap, because colleges with strong outcomes tend to offer more of the well-paid majors. Counting only what each adds beyond the other, the major's contribution (52%) is nearly three times the college's (18%). The ranking holds when programs are weighted by their number of graduates (70% against 34%), and when programs in Puerto Rico and other territories are included (60% against 34%).⁷

These are descriptive patterns, but research that isolates cause points the same way. In Norway, a large study of students who just made or just missed admission cutoffs found that the field a student studied changed later earnings far more than attending a more selective university did.⁸ At UC Santa Cruz, students who barely qualified for the economics major earned about $22,000 (46%) more a year early in their careers than similar students who just missed the cutoff.⁹ Studies like these show that the major has its own effect on pay, separate from the kind of student who picks it.

Exhibit 5 ranks the larger fields by typical pay four years after a bachelor's degree. Engineering and computer science lead, at about $94,000, followed by engineering technologies ($85,000) and the health professions, a group dominated by nursing ($83,000). Business ranks sixth of 24 fields, at $71,000, above the typical bachelor's program ($63,000). At the bottom, graduates in education, English and psychology typically earn $50,000 to $51,000, and those in the visual and performing arts about $44,000.²

The gap is starker at the top of each field. In engineering, 84% of bachelor's programs pay graduates at least $50,000 more than the high-school benchmark. In computer science the figure is 57%, and in business 18%. In psychology it is under 1%, and in English no program reaches it.²

None of this is new to economists. Among men in the 2009 American Community Survey, the pay gap between electrical engineering and general education majors was nearly as large as the gap between college graduates and high-school graduates.¹⁰

The same pattern holds inside a single university. At Cornell, computer science graduates in the data earned a median of $223,309 four years out. Graduates in fine and studio arts earned $41,784, about one-fifth as much. Penn's spread is similar. At Tufts and Boston University, the best-paid major earns about three and a half to four times as much as the lowest.²

These four are among the widest, but wide gaps between majors are the norm. At the median university with 20 or more majors in the data, the best-paid major earns 2.7 times the lowest, and the major at the 90th percentile pays 1.87 times the major at the 10th. Compare that last figure with the same measure across colleges: in the typical major, the college at the 90th percentile pays 1.51 times the college at the 10th.⁷

The federal figures capture the start of a career, and early pay can mislead. David Deming and Kadeem Noray found that computer science, engineering and business graduates start with the biggest pay advantage and that it narrows with age. Compared with all other majors, computer science and engineering graduates earn about 45% more early in their careers and 33% more by age 50; business graduates' edge falls from about 38% to 20%. The premium for the life and physical sciences and the social sciences, which include economics, grows with age, “although most other majors never quite catch up.”¹¹

Two features of the data matter here. They leave out graduates still enrolled in their fourth year out, such as medical and doctoral students, which matters most in fields that feed those programs, such as biology and psychology.² And they cover only graduates who received federal aid. A Census Bureau study of colleges in a sample of states found that this makes little difference one year after graduation, but that by year ten the federal-aid group earns significantly less than all graduates, that the gap is larger where fewer graduates receive aid, and that a program's ranking can shift by about 10 percentile points either way.¹²

Part II showed that the major sets the range. Within a major, though, colleges differ, and by how much depends on the field. For some majors, pay differs little from college to college. For others it differs a great deal.

Take every college that offers a given major and line up their graduates' median pay. In economics, the college at the 90th percentile pays $116,443; the college at the 10th pays $63,341, a ratio of 1.84. In mechanical engineering the same two points are $101,347 and $85,351, a ratio of 1.19.^2,7

Of the 35 majors offered by at least 200 colleges, computer science has the widest spread (1.91), followed by the allied health diagnostic and treatment professions (1.85), economics (1.84), general computer and information sciences (1.81) and mathematics (1.75). Engineering has the narrowest: 1.19 in mechanical, 1.21 in civil and 1.24 in electrical engineering. Social work (1.35) and teacher education (1.37) are close behind. Finance, accounting and business administration sit in the middle, at about 1.6.^2,7

Nursing (1.41) sits below the middle, and part of its spread is regional: 11 of the 15 best-paid nursing programs in the data are in California. More generally, differences in pay levels between states do not explain the pattern. After removing each state's average pay level, the same five majors still vary most from college to college, and civil and mechanical engineering least.^2,7

The data cannot say why the spread differs, but it fits how some of these job markets work. Registered nurses must pass a national licensing exam, the NCLEX, written by the National Council of State Boards of Nursing for the state boards, and public-school teachers must be licensed or certified by their state. Engineering programs follow accredited curricula, although a license is not needed for entry-level engineering jobs.^13,14 In fields like these, pay is likely to depend more on the role and the region than on the name on the diploma.

At the other end, some employers recruit heavily at a chosen set of campuses. Our earlier briefings on investment banking and private equity, and on management consulting, describe how those firms concentrate entry-level hiring at “target” schools.^15,16 In earlier College Scorecard data of the same kind (2014–16 graduates, five years out), College Board researchers found that college selectivity tracks pay strongly in computer science, finance and economics, and shows “virtually no relationship” with pay in nursing, social work and teacher education.¹⁷

The pattern has exceptions. The allied health diagnostic and treatment professions, a code that groups many health specialties with different pay, vary as widely as economics, and accounting, business administration and communication show the same spread as finance. Licensing and recruiting are at most part of the story.

Graduates of selective colleges earn more, but selective colleges also admit students who would probably earn more anywhere. Three lines of research help separate the two.

In a study of students who entered mostly highly selective U.S. colleges in 1976 and 1989, those who chose the more selective of the colleges that admitted them did not, on average, earn more than similar students who chose a less selective one. The exceptions matter: Black and Hispanic students, and students from lower-income families or whose parents had less education, earned more at the more selective college.¹⁸ Among Texas public universities, students admitted to the same set of schools earned fairly similar amounts wherever they enrolled, and a small early pay edge for the more selective school faded within a few years of graduation.¹⁹

The most selective private colleges are a partial exception. For students on the margin of admission, attending an Ivy-Plus college (the eight Ivies plus Stanford, MIT, Duke and Chicago) instead of a flagship public university raised the chance of reaching the top 1% of earners by about half, nearly doubled the chance of attending an elite graduate school and made working at a prestigious firm about two and a half times as likely. It did not significantly change where graduates typically land in the earnings distribution.²⁰

Read together, these studies suggest the college matters most at the very top, for reaching the highest-paid jobs. For typical outcomes, the available research, mostly from Norway and one U.S. university, suggests the choice of major does more than the choice of college.^8,9 That is also why the peer-group gaps in Part V likely overstate what the college itself adds.

More bachelor's graduates in the data studied business than any other field: 421,314 of them, or 22%, across 4,060 programs. Not one of those programs would fail the earnings test on the preview data.² By the federal measure, business is a safe degree. That does not make every business degree the same.

Exhibit 8 breaks business into its twelve larger majors, those offered by at least 60 colleges. Among them, the most quantitative pay the most. Typical pay four years out is $89,451 in management science and quantitative methods, $85,162 in management information systems, $83,501 in finance and $80,415 in managerial economics. Accounting follows at $76,123. The broad majors that most business students choose pay less: $69,592 in marketing, $68,285 in business administration and $65,373 in general business. Hospitality management is lowest, at $56,688. Two smaller majors pay more than any of these: construction management ($93,910) and insurance ($92,555).^2,7

The gap is wider at the top. In management science, 54% of programs pay graduates at least $50,000 more than the high-school benchmark. In finance the figure is 32%, and in accounting 23%. In business administration, by far the largest business major with 177,824 graduates in the data, it is 7%.²

Many students drawn to business weigh finance against economics. Their typical pay is almost identical: $83,501 for finance and $83,248 for economics. The difference is at the top. The economics program at the 90th percentile pays $116,443, against $101,476 in finance.^2,7

One likely reason is where each is taught. At many of the most selective colleges, economics is the closest thing to a business major, so economics carries more of their graduates. Part V shows how much those programs pay.

At the top of the business fields, a few colleges' graduates earn far more than the rest. Penn's finance graduates in the data earned a median of $202,069 four years out, 2.6 times the median finance program ($76,959). Washington University in St. Louis ($185,551), Notre Dame ($160,313), Georgetown ($152,744) and Boston College ($147,746) follow. Boston College's 250 finance graduates in the data are the largest group in the top ten.²

In business administration and general business, Carnegie Mellon leads at $160,783, ahead of Michigan ($144,654), Berkeley ($144,599), Virginia ($142,113) and NYU ($137,804). Several of the leaders are large public universities: Michigan's figure covers 551 graduates and North Carolina's 319.²

For a future business student, the data point to two choices worth weighing. The first is the track: among the larger business majors, those built on math, data and finance pay more than broad ones. The second, for students drawn to finance, is the college: the best-paid finance programs pay more than twice the median one, although finance's spread across colleges overall (1.6) is no wider than accounting's or business administration's.

This part looks at a group of colleges that many ambitious students aim for: the eleven colleges of the New England Small College Athletic Conference (Amherst, Bates, Bowdoin, Colby, Connecticut College, Hamilton, Middlebury, Trinity, Tufts, Wesleyan and Williams), the eight Ivy League universities, MIT and Boston University. The federal data let us compare them with every other college, major by major. They also come with a caution that matters most for exactly these colleges.

The federal data count only graduates who received federal aid, and only in majors with at least 16 of them working and not in school four years on. At wealthy colleges with generous grants, that is a minority of each class. Across the 21 colleges, the earnings data describe about one bachelor's graduate in four: roughly 7,100 of the 28,200 degrees awarded each year. Nationally, they describe 43% of graduates at private nonprofit colleges and 50% at public ones.²¹

Coverage is thinnest at the smallest colleges and the wealthiest universities. The data see about one graduate in ten at Bates, Colby and Princeton, and 12% to 15% at Harvard, Yale, Hamilton and Bowdoin. Bates appears with only 3 majors, Colby and Bowdoin with 4, and Hamilton with 5. Wesleyan files nearly all its graduates under one general code, so its majors cannot be compared.²¹

Who is counted differs, too. At Princeton, Harvard and Yale, most of the federally aided group are Pell Grant recipients, who come mainly from lower-income families. At BU, Tufts and Hamilton, most received federal aid but no Pell Grant. So part of any gap between these colleges reflects whose earnings are measured.²¹ A Census Bureau study also found that federal-aid figures drift further from those of all graduates where fewer graduates receive aid.¹²

The figures in this part are therefore the median pay, four years out, of federally aided graduates who were working and not in school, in the larger majors. They are not the earnings of a college's whole graduating class, and this briefing does not rank the colleges overall.

In every one of the 15 majors reported by at least six of the 21 colleges, the median peer-group program pays more than the median program nationally. We call the ratio of the two the peer-group premium. It is largest in computer science, where the peer median ($201,702) is 2.03 times the national median ($99,574), and in mathematics (1.84). History (1.66) and economics (1.64) come next. It is smallest in international relations (1.19), biology (1.16) and area studies (1.15).^2,7

Mechanical engineering, where pay varies little from college to college, is near the low end (1.23), and so is biology, where graduates still in medical or doctoral programs four years out are not counted. History is the exception among the humanities: its premium rivals economics, while English (1.27) trails. The data do not show what jobs graduates hold, so they cannot say why.

Economics is one of two majors, with political science, reported by 19 of the 21 colleges, the fullest comparisons available. Williams College's figure, $166,678, is the highest for any economics program in the data. Princeton ($160,763), Harvard ($155,592) and Dartmouth ($152,929) also exceed $150,000, although at Princeton and Harvard the data see only about one graduate in eight to ten. Two more NESCAC colleges, Amherst ($141,730) and Hamilton ($130,633), sit inside the Ivy range, though Hamilton's figure also rests on under 15% of its class.^2,21

As groups, the eight Ivy economics programs have a median of $140,436 and the ten NESCAC programs $122,772. Boston University's figure, $96,723, covers a larger share of its class (35.5%) than most of its peers' and is above nearly four in five economics programs nationally.^2,7,21

Thirteen of the 21 colleges report computer science under its own code. MIT ($225,141) and Cornell ($223,309) have the highest figures, and every Ivy shown clears $188,000. Williams ($209,574) again sits among the Ivies. Boston University's computer science figure, $136,667, is above 85% of programs nationally.^2,7

English sits in the middle of the range. Brown's English graduates in the data earned $73,497 four years out and Tufts's $69,475. Across the 15 colleges that report English, the median program pays 27% more than the national median of $48,920. The lowest figure in the group, Harvard's $50,428, rests on just 21 federally aided graduates over two years and says little about Harvard English majors as a whole.^2,7

The usual cautions apply with more force here. Each figure rests on 16 to 44 graduates, and English majors still in school four years out, such as doctoral students or those who started law school after a gap year, are not counted. Early pay in English is modest almost everywhere, and it varies less from college to college than in the recruited fields.

Three conclusions follow for families weighing these colleges. First, the gap between these colleges and the national median is largest in computer science (2.0) and mathematics (1.8), followed by history and economics (about 1.65). Second, in economics NESCAC's best programs match the Ivies, though the typical NESCAC program pays about $18,000 less than the typical Ivy program. Third, the gap is not equal across majors: in international relations, biology and area studies it is under 20%.

The federal earnings data were built to catch the worst programs. Read carefully, they also help families choose among good ones. Six rules follow from Parts I to V.

The factor most closely tied to pay in this data is the choice of field, and many students make it with little first-hand experience of the work. Business shows why that matters. It is the most common bachelor's field, and its larger majors range from hospitality management to management science, a gap of more than $30,000 in typical early pay.

The Boardroom gives high-school students a week inside real business decisions before they choose. Over five days in Boston they read a market, test the numbers, value a company, decide on a deal and defend the decision to a board of working executives.²⁴ A week will not choose a college for anyone, and this briefing does not measure the effect of programs like ours on pay. What a week can do is show a student whether finance, strategy or markets are where they want to spend four years, and give them a specific answer when an application asks why they want to study business.