Michael Spence's 1973 job market signaling model posed a provocative question: what if a college education conveyed no productive skills — if employers paid college graduates more purely because degree attainment was a costly signal that correlated with innate ability, not because of anything learned? Spence showed this equilibrium is theoretically stable: if able workers find it easier to complete college than less able workers do, completing college is a credible signal of ability regardless of whether the curriculum teaches anything useful. He shared the 2001 Nobel Prize for showing how markets can extract information through costly signals — even when the signal itself creates no direct value.
In plain terms
Signaling is any action taken by an informed party to credibly communicate their type to an uninformed party. For a signal to work — to convey real information — it must satisfy one critical condition: it must be cheaper for high-quality types to send than for low-quality types. If both types can send the signal at the same cost, everyone sends it, and no information is conveyed.
Screening is the uninformed party's approach: designing mechanisms — menus, tests, contracts — that cause different types to self-reveal through their choices. Insurance companies screen by offering menu of policy options; applicants' choices reveal their private information about their own risk levels.
Both solve the adverse selection problem created by asymmetric information, but from different directions:
- Signaling: the informed party acts first to communicate
- Screening: the uninformed party acts first to elicit information
Signaling in detail
For a signal to be informative, it must be a separating equilibrium: low types find the signal not worth sending (too costly relative to the benefit); high types find it worth sending (cheaper for them and the payoff justifies the cost). Both types end up revealing themselves through their choices.
Educational credentials work as signals because, even if education imparted no skills, completing a demanding program would be easier for people with higher cognitive ability. The Bureau of Labor Statistics earnings data shows the persistent wage premium for degree holders — consistent with both a human capital interpretation (education raises productivity) and a signaling interpretation (degrees identify able workers). Both mechanisms likely operate simultaneously.
Product warranties are signals: only firms confident in product reliability can afford to offer comprehensive warranties. A manufacturer offering a 5-year warranty on its appliances is signaling quality — a low-quality manufacturer would incur massive warranty claims and couldn't profitably offer the same terms.
Screening in detail
Insurance companies use screening menus to elicit information about applicant risk. Offering policies with different deductibles at different premiums causes applicants to self-sort: high-risk individuals prefer low-deductible policies (they expect to file claims); low-risk individuals prefer high-deductible policies with lower premiums (they don't expect to file). The CFPB's research on financial product design documents how product menus screen consumers by risk tolerance and financial sophistication — information the provider cannot observe directly.
Why it matters
Signaling and screening explain why markets invest so heavily in credentials, certifications, warranties, audited financials, and disclosure requirements. These mechanisms exist to solve information problems that, left unsolved, would cause markets to produce the wrong mix of products, undervalue the best participants, and potentially unravel entirely. The design of information-revealing mechanisms — contract structure, disclosure requirements, certification systems — is one of the most practical applied areas of information economics.
