Signaling vs. Screening: Are You Hiring Talent — or People Who Are Good at Looking Talented?

Two finalists. One has a top-university degree, a stint at a famous consultancy, and a CV that reads like a press release. The other has a regional college diploma, five years at a company no one on the panel has heard of, and a portfolio of problems she actually solved. The panel leans towards the first candidate before anyone has asked a real question. The uncomfortable question is not who is better. It is whether your process could even tell.

Two theories, one hiring problem

Hiring is a problem of hidden information: the candidate knows more about their ability than you do. Economics offers two classic solutions. In Michael Spence's signaling model (1973), the informed party — the candidate — acts first, investing in a costly signal such as education. The signal works not necessarily because the education builds job skills, but because it is harder to obtain for less able people. In Joseph Stiglitz's screening model (1975), the uninformed party — the employer — acts first, designing a mechanism that makes candidates reveal their type. Spence and Stiglitz shared the 2001 Nobel Prize in economics for this work (with George Akerlof).

Why signals mislead

Signals tell you someone was selected by some earlier gatekeeper — an admissions office, a previous employer's recruiter. They carry that gatekeeper's criteria and biases, which may have little to do with your job. They also reward access: families that can pay for test prep, internships that are unpaid, networks that open prestigious doors.

The evidence on prediction is sobering. In 2022, Sackett, Zhang, Berry and Lievens re-examined the classic meta-analytic estimates of how well selection methods predict job performance and corrected for over-adjustment in earlier work. Structured interviews emerged as one of the strongest predictors. Job-knowledge tests and work samples performed well. Years of job experience and years of education — the backbone of most CV screens — were among the weaker predictors.

When signals are rational

Signals are not useless. When you have 2,000 applicants and one recruiter, a cheap signal is better than a coin flip. Licences matter for doctors and pilots. The mistake is treating signals as proof rather than as a first hypothesis — and letting them override better evidence later in the process. The prestige of a candidate's last employer should never outweigh how they performed on a job-relevant task in front of you.

Building a screening-first process

Limits and caveats

• Screens can be biased too if they are badly designed; validate them against job performance.
• Signaling theory does not claim education has no value — only that part of its labour-market value comes from sorting.
• Prediction figures are averages across many studies; your specific tools need local validation.

Two theories, one hiring problem

Michael Spence's job-market signaling model (1973) explains why candidates invest in costly signals such as degrees: the signal is valuable because it is harder to obtain for people with less of the underlying quality. Joseph Stiglitz's screening theory (1975) looks at the same problem from the employer's side: the employer designs tests, probation periods, or contracts that make candidates reveal information about themselves. Spence and Stiglitz shared the 2001 Nobel Prize in economics with George Akerlof for work on markets with asymmetric information.

Why signals mislead in modern hiring

• Signals are unequally available. A degree from a well-known university reflects ability, but also family income, geography, and luck.
• Signals inflate. When many candidates hold a credential, it separates people less, and employers ask for more — a process researchers call credential inflation.
• Signals can be manufactured. AI tools now help candidates write polished CVs and cover letters in minutes, which weakens writing quality as a signal.
• Signals age. A brand-name employer from ten years ago tells you little about current skill.

Research on selection methods supports shifting weight from signals to screens. The updated meta-analysis by Sackett, Zhang, Berry and Lievens (2022) ranked structured interviews among the strongest predictors of job performance, with job knowledge tests and work samples also performing well. Years of education and years of experience were far weaker predictors than earlier estimates suggested.

Designing better screens

Screens are not neutral either

Screening can also go wrong. Unpaid take-home assignments that take ten hours screen for free time, not ability, and discourage carers and people already in demanding jobs. Brain-teaser questions screen for a type of confidence, not job skill. A good screen is short, clearly job-related, respectful of the candidate's time, and scored consistently.

• Each job has a written list of critical tasks.
• At least one selection step directly samples the work.
• Degree requirements are justified by the job, not habit.
• Take-home tasks are time-limited or paid.
• Hiring decisions record evidence, not impressions.

Worked case: the degree requirement nobody could explain

A logistics company requires a university degree for its operations coordinator role. When HR asks why, the answers are 'it shows commitment', 'they write better' and 'we have always done it'. When HR compares current coordinators with and without degrees — some were grandfathered in — they cannot find a clear difference in quality, error rate or promotion. The degree was working as a signal of something the company never defined.

The company replaces the degree requirement with a two-part screen: a short written exercise (a realistic shipping delay email to a customer) and a structured interview about prioritising problems. Applications rise, the candidate pool becomes broader, and hiring managers now say what they are actually checking for.

Turning a vague signal into a real screen

What each role can do

• Hiring managers: for every requirement in an advert, write one sentence on what it proves and how else you could check it.
• Recruiters: track which requirements actually screen people out, and whether those people would have done the job.
• HR leaders: review requirements that create adverse impact against protected groups — they need a clear job-related reason.
• Candidates: show evidence, not only credentials — work samples, portfolios and concrete results.

Common mistakes

• Replacing one weak signal (a degree) with another (a famous former employer).
• Adding so many assessment steps that strong candidates with options drop out.
• Using unvalidated personality tests as screens.