The Credential Trap - How the qualification system stopped selecting for talent
Part of a series exploring the systems and ideas that shape modern society.
Somewhere in the last few decades, a quiet deal was struck. Employers would use degrees as a proxy for capability. Universities would supply them. Governments would subsidise access. And individuals would spend years - and in many countries, significant sums of money - acquiring qualifications whose primary function increasingly became getting them past institutional filters.
The deal is starting to break down. Not loudly, not all at once, but in ways that are becoming hard to explain away.
Goldman Sachs no longer requires a degree for many roles. Neither does Apple, IBM, or a growing list of firms that once treated the bachelor’s credential as non-negotiable. Hiring platforms built around skills tests and portfolio assessment are expanding rapidly. And AI-powered tools can now evaluate specific competencies directly - faster, more cheaply, and often more accurately than a degree classification ever could.
The credential isn't disappearing, but its authority is quietly being questioned by the very institutions that once enforced it. That shift is worth examining carefully, because the system it threatens to disrupt has shaped who gets opportunity, and who doesn’t, for the better part of a century.
What Credentials Actually Do
Credentialism is the practice of using formal qualifications - degrees, licences, certifications - as the primary signal of whether someone is worth hiring, promoting, or trusting with responsibility.
The term sounds technical, but the reality is straightforward: most employers can't directly observe whether a candidate is capable before they hire them. Credentials solve that problem cheaply. A degree doesn’t prove you can do the job, but it does tell the employer that you passed a sorting process someone else designed and paid for. The risk, in a narrow sense, is transferred. And in many contexts, that works reasonably well. Complex societies require scalable ways to establish baseline competence and trust, particularly in professions where mistakes carry real consequences.
There are three ways to understand what credentials are actually doing. Economists tend to see them as signals: information devices that allow candidates to communicate things about themselves that employers can’t easily verify. Sociologists - particularly Randall Collins, who made the argument with some force in the 1970s - see them primarily as instruments of status closure: mechanisms by which established groups raise the cost of entry to preserve their position. Some critics argue the system increasingly resembles a signalling market whose costs have expanded faster than its informational value.
None of these readings are necessarily wrong. A credential can simultaneously function as a signal, a class filter, and an entry ticket whose price has long since outrun its value. The tension between those functions has been present from the start. What has changed is that it’s becoming structural.
How the System Built Itself
Formal credentialing didn’t emerge from a neutral assessment of how best to match people to roles. It emerged from professional self-interest, then spread by imitation.
In the nineteenth century, associations in medicine, law, and engineering began requiring formal qualifications for entry. The publicly stated rationale was consumer protection - a licensed doctor is safer than an unlicensed one. That is partly true. In genuinely high-risk professions, formal qualification systems often improved reliability and public trust. But licensing also controlled supply, raised wages for incumbents, and excluded competitors who lacked the required paperwork regardless of their actual competence. The protection ran in both directions: for patients, and for practitioners.
After the Second World War, mass university expansion created a large population of degree-holders. Employers, facing more applicants than they could carefully evaluate, began treating the bachelor’s degree as a baseline screen. This was partly rational - degree-holders had been filtered already, and employers were borrowing that filter for free. But the logic, once adopted, proved difficult to contain.
Credential requirements began to migrate. Roles that had never needed degrees started listing them. As degrees became more common, master’s qualifications emerged as the new differentiator. Licensing spread into occupations well beyond any obvious safety rationale. In the United States, the share of jobs requiring a state licence grew from around five percent in the 1950s to roughly twenty-five percent today. The work hadn’t changed. The gatekeeping had.
The system is self-reinforcing in a specific way: once credentials become standard in a sector, the individual cost of opting out is high even if the collective cost of maintaining them is higher. Nobody wants to be the uncredentialed candidate in a credentialed field. So the qualifications accumulate.
Who Pays, and Who Benefits
The incentive structure of credentialism is worth being precise about, because it explains why a system whose costs and informational value no longer appear as tightly aligned as they once did has proved so durable.
Employers benefit from credential requirements as a liability management tool as much as a selection tool. Hiring a degree-holder who fails is defensible. Hiring someone without the expected qualification who then fails is harder to justify internally. The credential doesn’t just filter candidates, it absorbs institutional risk. It also reduces evaluation costs. Verifying a degree is cheap. Assessing actual competence is not.
Universities have obvious incentives to expand credential offerings and maintain the labour market value of their qualifications. A university that successfully positions its degree as the expected entry point to a profession has, in effect, secured a recurring revenue stream backed by regulatory norms it didn’t create but benefits from enormously.
Professional associations use licensing to sustain wage premiums for their members. This is framed as quality assurance. It functions, at least partly, as market restriction.
Candidates have no rational choice but to participate. Even if you understand that the credential’s informational value is low, the cost of being excluded from a credential-filtered market is high. You acquire the qualification because the system demands it, not because you believe it will make you better at the work.
The cumulative cost of this arrangement often falls unevenly. The financial burden of credential acquisition - tuition, foregone income, debt - is proportionally heaviest for people from lower-income backgrounds. Credentialism doesn’t just select for ability. It selects for the financial capacity to spend years in a sorting process. The meritocratic framing obscures a wealth filter that has been built into the architecture of opportunity.
The Gap That Opened Up
The central problem with credentialism, the one that makes its current moment genuinely interesting, is that the credential has drifted from the competence it was supposed to represent.
This happens through inflation. As more people acquire a qualification, its signal value declines, and the next level becomes necessary to differentiate. A high school diploma gave way to a bachelor’s degree. A bachelor’s degree is giving way to a master’s in more and more fields - not because the underlying work has become more complex, but because the credential arms race has moved the bar. The cost rises. The informational content doesn’t keep pace.
Employers have noticed. The quiet removal of degree requirements at firms like Goldman Sachs and Apple isn't primarily an act of social conscience, it is a response to a talent problem: credential thresholds were excluding capable candidates who hadn’t followed the expected pathway. Skills-based hiring, portfolio review, and structured assessment grew because the credential was becoming a less reliable proxy for the thing employers actually wanted to measure.
AI is now making that failure harder to paper over. When a candidate’s capability can be assessed directly - through technical tests, verifiable work samples, or automated competency evaluation - the credential’s function as an indirect proxy becomes difficult to justify. Why use a four-year degree as a signal for analytical ability when you can test for analytical ability in forty minutes? The informational asymmetry that made credentials useful is shrinking. The systems that built up around them haven’t caught up.
Credentials also perform social functions beyond pure skill verification: they standardise expectations, legitimise authority, and provide institutions with a common language for evaluating strangers. That is part of why the system survives even as its informational efficiency declines. It is not merely inertia or employer laziness - it is that credentials coordinate trust across large institutions in ways that are genuinely difficult to replace quickly.
This is the fault line: the credential has become a status mechanism sustaining costs and access barriers that no longer track reliably with the competence they claim to certify. Its durability now rests less on what it communicates and more on how difficult it is to coordinate a switch to something else when everyone is still transacting in the same currency.
Whether It Holds
The pressures are arriving from several directions simultaneously, which is unusual. Normally, institutional systems of this kind face challenges sequentially and absorb them. The current convergence is different.
Credential inflation is visible and documented. The cost of qualification pathways is under sustained scrutiny in most developed economies. Alternative signalling systems - competency certifications in technology, portfolio-based assessment, direct skills testing - have matured to the point where they carry genuine labour market weight in some sectors. And AI is systematically reducing the cost of direct evaluation, which undermines the economic case for using indirect proxies.
Against this, the credential system has real inertia. Professional licensing is politically difficult to reform because incumbents benefit directly from barriers to entry and are well-organised. Universities are embedded in the social infrastructure of most countries in ways that make rapid change unlikely. Credential expectations propagate through hiring systems and informal norms without anyone actively choosing to maintain them - they persist through coordination, not conviction.
The question is less whether credential systems provide value - many clearly do - than whether expanding qualification requirements still track reliably with the competencies they are meant to signal. That is a harder problem to ignore than it was a decade ago, and the tools to test it directly are improving quickly.
Coordination problems are sticky. Institutions that serve multiple interests simultaneously, even imperfectly, tend to outlast critics’ predictions. The credential system will not collapse. But its authority rested on a specific claim: that it told you something useful about what a person could do.
When a signal separates from the thing it was signalling, it doesn’t always recover.