People have been asking this question for decades, ever since neoliberalism began to eat away at the welfare state. The difference now is that the US government itself is answering affirmatively, without qualification.
Responding to pressure from the Trump Administration, the US Department of Education finalised regulations in June 2026 that would punish academic programs that don’t ensure a reasonable “financial return on investment”. This matters because the policy turns a debate about student debt into a test of which programs are worth supporting.
Under this new policy, undergraduate programs must show that graduates earn more than a typical worker aged twenty-five to thirty-four with only a high-school diploma. In contrast, graduate and professional programs are measured against workers with bachelor’s degrees.
A program that fails this “earnings premium” test in two of three consecutive award years will lose eligibility for federal Direct Loans for at least two years.
Higher education is expensive. Students can incur debts that follow them for decades, and institutions have sometimes sold credentials whose economic benefits bear little relation to their costs. The US government, which subsidises and guarantees much of this system, has an obvious interest in preventing students from being loaded with debt for programs that leave them financially worse off. That concern is real, but it is not the same as deciding what education is worth.
The problem begins when an argument about debt and consumer protection quietly becomes a theory of value.
Consider two people: a litigation attorney and an elementary-school teacher. The attorney earns several times what the teacher earns. We can therefore say, with some confidence, that the market price of the attorney’s labour is higher. But what else have we learned?
We have not established that litigation contributes several times as much to social welfare as teaching children to read. Nor would pointing to the difficulty, responsibility, or stress of the attorney’s work settle the question. If compensation reflected stress, danger, or sacrifice with anything approaching mathematical regularity, infantry soldiers would presumably be among the wealthiest people in the country.
Prices tell us something else. They record the results produced by institutions through which goods and labour are exchanged. The corporate client who purchases legal services possesses resources that the school district purchasing teachers’ labour does not. Occupational licensing limits the supply of attorneys. Public budgets constrain teacher salaries.
Price is real. That doesn’t make it a neutral instrument for revealing value.
The earnings test implemented by the new US Department of Education policy does not merely observe that graduates of one program earn less than graduates of another. It takes the wage structure produced by the existing political economy and feeds it back into the educational system as an administrative judgment. The result is to make existing inequality govern future educational access.
In other words, the market is being called as an expert witness in a case in which the market itself is the defendant.
The recursive character of this arrangement is easiest to see in occupations whose social necessity is widely acknowledged but whose compensation is comparatively poor. Suppose a society systematically underpays teachers, social workers, public defenders, or caregivers. Graduates entering those professions consequently record modest earnings. By this metric, the educational programs that produce them will then appear to generate insufficient economic returns.
As a consequence, federal loans will eventually be restricted, making entry into those professions more difficult. Yet no one has ever demonstrated that the work has little social value. The wage structure has simply been permitted to testify about its own correctness, and the cost is borne by the professions society still needs.
This is where the language of “return on investment” does considerably more ideological work than it initially appears to do.
An investment has an investor, a principal, a return, and a time horizon. Once education is understood primarily through this vocabulary, the student becomes a small proprietor purchasing an income-producing asset called a degree. The university supplies the asset. Future wages constitute the dividend.
The transaction’s success can therefore be measured by comparing the purchase price with the income stream it subsequently generates. The conceptual achievement here is to present this arrangement not as one historically specific way of organising education but as education’s natural form.
Michael Hardt and Antonio Negri’s Multitude, the sequel to their landmark book Empire, proves helpful here. Their account of contemporary production destabilises the boundary between economic production and forms of social life that older industrial models treated as external.
Whatever one thinks of their larger claims about “immaterial labour”, their essential insight presents an awkward problem for attempts to reduce education to individual human-capital investment.
As Hardt and Negri see it, knowledge, communication, language, affect, cooperation, and the production of relationships have become increasingly central to economic activity. The economy increasingly depends upon capacities whose production is social, dispersed, cumulative, and difficult to attribute to a single educational input.
Yet education financing increasingly demands that particular packets of knowledge demonstrate their value through the subsequent earnings of individual credential holders.
The contradiction is striking. Contemporary capitalism wants flexible people who can communicate, interpret, invent, adapt, combine forms of knowledge, and navigate environments that change faster than formal curricula can. But the accounting system through which education is evaluated wants to know what someone with a B.A. in philosophy does for a living and how much she made last year.
McKenzie Wark’s A Hacker Manifesto expands on this problem. For Wark, information occupies an anomalous position within capitalist property relations because it can move between material supports without being exhausted through use.
The “hacker” is therefore not simply a computer programmer but anyone engaged in producing new abstractions: scientists, musicians, artists, writers, programmers, philosophers.
A hack produces something that did not previously exist by discovering new relations among things already present. Knowledge, in this sense, is valuable partly because its uses cannot be specified in advance.
Wark draws an especially useful distinction between education and knowledge. Education organises knowledge institutionally, under conditions of scarcity, credentialing, property, and discipline. Knowledge nevertheless exceeds that organisation.
One can acquire an education as a credential, but to become knowledgeable requires a transformation whose consequences cannot be delimited by the credential itself. The most productive abstraction may be precisely the one that does not fit the institutional regime established to classify and commodify it.
The US Department of Education’s new earnings test intensifies the very identification Wark wants us to question. It treats education as a bounded commodity whose consequences should be traceable through a determinate chain: a student purchases a credential; the credential corresponds to a field; the field corresponds to an occupation; the occupation produces an income; the income reveals the credential’s value.
From this perspective, knowledge that travels elsewhere, combines with other knowledge, changes its possessor in ways not captured occupationally, or produces consequences decades later will be difficult to perceive.
This does not mean that knowledge possesses some mystical value that should exempt educational institutions from scrutiny. It means almost the opposite. The difficulty is that knowledge produces consequences in too many places to be adequately represented by the price attached to one of them.
Gilles Deleuze and Félix Guattari give us another vocabulary for describing what happens when the state nevertheless attempts to make these heterogeneous effects administratively legible. A Thousand Plateaus, their sequel to Anti-Oedipus, repeatedly distinguishes between forms of organisation that allow movement across heterogeneous spaces and those that measure, classify, partition, and regulate them.
The distinction Deleuze and Guattari make between smooth and striated space should not be reduced to a moral opposition in which smooth is good and striated is bad; actual social arrangements continually combine the two. But striation offers an unusually suggestive description of what happens when qualitatively different activities are placed on a common grid so they can be compared and governed.
An earnings metric performs this operation precisely. Philosophy, nursing, chemistry, social work, engineering, music, education, and computer science are radically different practices involving different objects, temporalities, institutions, purposes, and relationships to social life. The metric need not understand those differences. Indeed, its administrative usefulness depends upon not understanding them. It requires only that graduates of programs in these fields be placed along the same quantitative axis.
This is an apparatus of capture in a fairly precise sense. The state does not abolish the market in favour of bureaucratic judgment, nor does it simply withdraw and permit “the market” to determine educational outcomes. Instead, it constructs a regulatory apparatus that takes market outcomes, processes them through administrative categories, and returns them to institutions as conditions of access to public resources. The result is not a neutral measure of value, but a recursive system that entrenches the very inequalities it claims to assess.
The distinction between state and market becomes difficult to sustain because each helps constitute the other. Market wages provide the data; the state transforms the data into thresholds; the thresholds alter educational institutions; those altered institutions help determine which workers and forms of knowledge subsequently enter the market.
The regulations therefore matter even if they do not endure. Universities operate on timelines much longer than presidential administrations. Hiring a tenure-track professor, constructing a department, admitting students to a doctoral program, or developing a new curriculum involves commitments whose consequences can extend for decades.
The US Department of Education’s new system, moreover, requires institutions to report data annually and review the earnings information generated from federal enrollment records. The machinery of measurement thus becomes part of ordinary institutional administration before the most severe sanctions are necessarily imposed.
Imagine that a subsequent administration repeals the earnings rule. A university cannot automatically return to the day before the rule existed. A faculty line may already have disappeared. A program may have stopped admitting students. Administrators may have shifted resources toward degrees whose earnings profiles seem safer. Students may have learned to regard certain disciplines as financially suspect. Departments may have rewritten curricula and promotional materials to emphasise occupational outcomes that regulators can see.
More importantly, administrators planning for twenty years cannot assume that repeal is permanent. A future administration could restore the policy or construct something resembling it. The rational institutional response is therefore anticipatory. Why rebuild a vulnerable program if another change in government could render it financially precarious four years later?
A regulation need not survive for the outcomes it is meant to produce to do so.
This institutional dimension matters because metrics do not merely measure institutions. Institutions adapt themselves to metrics. Once departments learn to explain themselves in terms of salary premiums, placement rates, occupational destinations, and loan eligibility, the categories initially imposed from outside become part of the institution’s self-understanding.
What was presented as descriptive, a neutral assessment of earnings potential for different educational paths, becomes prescriptive. The policy will end up producing the reality it claims merely to reflect.
The problem with the US Department of Education’s new policy is not that it cares about employment. Students care about employment, and they have every reason to. It makes sense for government to care whether educational institutions exploit students through programs carrying enormous costs and miserable prospects.
The policy goes wrong in how it converts one legitimate question—Will this program leave its graduates economically secure?—into a proxy for a much larger one: What kinds of knowledge should a society make materially possible?
Although those questions overlap, they are not the same. Once they are treated as interchangeable, however, the consequences extend far beyond the programs that immediately fail the test.
Universities begin learning which kinds of knowledge are safest to produce. Students learn which forms of curiosity carry regulatory risk. Administrators learn to translate intellectual activity into categories legible to the labour market.
The distinction between education and vocational investment becomes harder to articulate because the institutions themselves have been reorganised around a metric that presumes the distinction in advance.
Perhaps this is the deepest sense in which the US Department of Education’s new policy represents an apparatus of capture.
Nothing has to be banned. Nobody has to announce that philosophy, history, literature, music, anthropology, or theoretical physics are useless. Nobody even has to believe that they are.
The machinery requires only a number, a threshold, and a consequence. The spreadsheet will handle the philosophy.
And because its judgments arrive wearing the neutral language of earnings data rather than the embarrassing costume of ideology, we are encouraged to forget what has actually happened.
Photograph courtesy of Alisdare Hickson. Published under a Creative Commons license.
