In 1973, Michael Spence published a short paper in the Quarterly Journal of Economics with an unsettling implication that employers and universities have politely declined to absorb for half a century. In 'Job Market Signaling,' Spence modeled hiring as it actually is: a bet on an unobservable, made by an employer who cannot see productivity directly. Education, in the model, earns its wage premium not by creating capability but by revealing it, because completing the credential is cheaper for the capable than for the less capable. The signal can be perfectly informative while teaching nothing at all.
Reality is a blend; education both builds skill and signals it, and economists have argued about proportions ever since. But the signaling share, whatever its exact size, has a property that changes everything: it is positional. A signal's value lies in separating its holder from non-holders. If everyone acquires it, the separation vanishes, and with it the premium, while the tuition remains fully paid.
The race with no finish line
Positional goods obey a cruel arithmetic. When a degree distinguishes you, acquiring it is rational. When enough others follow, the degree stops distinguishing anyone, and the distinguishing move shifts up a level: honors, then a master's, then the brand of institution. Every rung is individually rational and collectively futile, like standing on tiptoe in a crowd. The crowd ends up on tiptoe, seeing exactly what it saw before, calves burning.
The empirical residue of this race is visible in hiring data. Harvard Business School's Dismissed by Degrees documented postings demanding bachelor's degrees for jobs whose current occupants overwhelmingly lack them, the 2015 flagship case being production supervisors: 67 percent of postings required the degree, 16 percent of incumbents held it. Nothing about supervising production had changed. What changed was the queue: with degrees abundant, employers could costlessly demand one, so they did, and the credential floor ratcheted up beneath jobs that had been performed fine without it. That is signaling logic operating in the wild, a race in which the entry fee rose while the track stayed the same length.
The private-versus-social ledger explains why the race persists. For the individual, the signal pays; the premium is real money, and unilaterally disarming is career malpractice. For society, the signaling share of education spending purchases sorting, not capacity: years and tuition burned to rearrange a queue. Spence's framework makes the uncomfortable accounting legible: when returns are positional, private returns exceed social returns, and a market left alone will overinvest in exactly the thing that helps each and, past a point, helps all not at all.
Exits from the trap
Traps built from individually rational moves cannot be exited by advising individuals to stop. The exits run through the demand side and the signal technology itself.
Employers hold the largest lever, because signals only pay what evaluators accept. Every employer that evaluates evidence of work directly (samples, trials, structured assessment of things actually done) reduces the wage attached to pure signal, and shifts the race's finish line toward capability, a good that is not positional: two capable people do not cancel each other out the way two identical diplomas do.
Second, cheaper signals. The degree's signaling power rests partly on its price; expensive-to-fake is the point. But cost-to-fake and cost-to-acquire need not travel together. A portfolio of verifiable work is hard to fake and cheap to build relative to four years of tuition. To the extent employers learn to read such evidence, the same information arrives at a fraction of the social bill.
Third, honesty in what is being purchased. Families and students deciding on credentials deserve the decomposition Spence made possible: how much of this purchase is learning, and how much is queue position? Queue position may still be worth buying. But it should be bought the way one buys any positional good, with open eyes and a wince, not with the warm belief that the spending itself creates the value.
The trap, at bottom, is a collective agreement to keep pretending the tiptoes make everyone taller. Agreements of that kind end the way they always end: not by everyone agreeing to stand flat-footed at once, but when enough evaluators, one hiring decision at a time, start looking at the work instead of the crowd.
Sources
- Spence (1973). Job Market Signaling (Quarterly Journal of Economics 87(3), 355-374). Founding model of education as a signal under unobservable productivity: credentials can carry a wage premium while adding no skill. https://academic.oup.com/qje/article-abstract/87/3/355/1909092
- Harvard Business School, Accenture & Grads of Life (2017). Dismissed by Degrees (Fuller & Raman). Degree demands rose for jobs incumbents perform without degrees (production supervisors: 67% of postings vs. 16% of holders, 2015), evidence of proxy escalation rather than changed work. https://www.hbs.edu/managing-the-future-of-work/Documents/dismissed-by-degrees.pdf