Concept/Economics/No. 0889
Screening
Screening is a way for a less-informed party to learn about others by designing tests or choices that reveal hidden traits. In economics, Joseph Stiglitz studied how employers and insurers use it to address gaps in information about workers or customers.
- Evidence
- Well established
- Read
- 6 min
- Links
- 8 connections
01You've seen this when…
- in life
You book a nonrefundable train ticket to save $30. The operator offers a flexible fare too, letting travelers sort themselves by how much they value changing plans.
- at work
Two finalists have polished résumés. The hiring manager gives each a paid exercise using a messy spreadsheet like the ones the job involves.
- out in the world
A city offers discounted transit passes after an income check. Applicants document a fact the transit agency cannot otherwise see.
02The idea
Someone on the other side of a deal knows something that matters to you. A job applicant knows more about their skills. An insurance buyer knows more about their health or driving habits. A customer knows how much flexibility they need.
That gap is asymmetric information. Screening gives the less-informed party a way to learn before making an offer or completing a deal.
There are two common routes. A test produces evidence: a work sample, an income check, an inspection. A menu lets people reveal information through their choices: different deductibles, cancellation terms, or service levels. Economists often focus on this second route, called self-selection.
The menu works when different people prefer different options because of the characteristic being screened. Someone who expects frequent claims has more reason to buy generous insurance. Someone with firm travel plans has more reason to accept a cancellation restriction.
A successful menu satisfies incentive compatibility: each type of person finds the option intended for them preferable to the other options. Designing those incentives is part of mechanism design.
Signaling starts with the better-informed party taking an action to convey information. Screening starts with the less-informed party arranging the test or choices. An applicant volunteering a credential and an employer requiring one can use the same credential from different directions.
03Why it matters
Information gaps change who participates in a market. If an insurer charges everyone the same premium, people expecting many claims may find the offer especially attractive. People expecting few claims may leave. This is adverse selection, and it can make the remaining pool more expensive to cover.
Screening can help keep a deal viable by matching terms to differences that matter. It also explains why everyday offers contain seemingly awkward conditions.
- Restrictions can reveal preferences. A cheaper fare with limited changes attracts travelers willing to sacrifice flexibility. Their acceptance gives the seller information about what they value.
- Tests can improve matching. A job-relevant exercise gives an employer evidence about how an applicant handles the work. Its value depends on how well performance predicts performance on the job.
Both routes have costs. A long application, difficult exercise, or restrictive contract changes who can participate. Those costs belong in the design decision alongside the information gained.
04A worked example
Consider an invented insurance market. Some applicants have a 2% chance of a single $5,000 loss this year; others have a 10% chance. Each applicant knows their own risk, while the insurer cannot tell them apart. To keep the arithmetic simple, assume buyers compare average annual costs.
The insurer offers two policies:
- Policy A costs $350 a year. The buyer pays the first $2,000 of any loss.
- Policy B costs $525 a year. The insurer covers the entire loss.
What it looks like A choice between a lower premium and more protection. For a low-risk buyer, Policy A has an average annual cost of $390: the $350 premium plus a 2% chance of paying $2,000. Policy B costs $525. For a high-risk buyer, those figures are $550 and $525.
What’s actually going on Low-risk buyers prefer A, and high-risk buyers prefer B. The deductible makes A more expensive in expectation for people who anticipate more claims. Under these assumptions, the choice reveals the buyer’s risk group without requiring a declaration of that risk.
What would have helped Checking the arithmetic for both groups before offering the menu. The insurer also needs to test whether the policies cover its costs and survive competing offers. Actual buyers differ in their tolerance for uncertain bills, so the clean separation in this illustration requires validation.
05Where people trip up
- Confusing performance on the screen with the desired trait. A timed interview may measure quick verbal responses alongside technical skill. A complicated benefits form may measure comfort with paperwork alongside eligibility. Check construct validity: does the test capture the characteristic the decision requires? Compare it with later outcomes whenever possible.
- Making participation costly in the wrong way. An unpaid assignment takes more from an applicant with caregiving duties or several jobs. Its opportunity cost varies across people. That can sort applicants by available time or money. Keep tests short, pay for substantial work, and remove requirements that add little evidence.
- Giving every type the same preferred option. A menu reveals little when everyone chooses one item. Work through what each group gains and gives up under every option. If a discount compensates even frequent claimants for accepting a deductible, the cheaper policy may attract both risk groups.
- Treating initial selection as the end of the problem. People can learn to game a test, and their circumstances can change. Behavior can also change after the deal: insurance coverage may affect precautions. That later incentive problem is moral hazard. Recheck the screen’s predictive value and consider how the contract shapes behavior after selection.
06Where it doesn’t identify a single trait
A generous-policy buyer may face high risk, strongly dislike uncertain bills, or have enough money to pay for convenience. The same choice can arise from several motives. Research across insurance markets finds that the relationship between coverage and risk varies; a simple pattern of riskier people buying more coverage is far from universal.
A menu can separate customers commercially while revealing little about the reason for their choices. Use the result as evidence whose strength depends on the assumptions behind it. When a relevant fact can be checked directly, a short verification step may provide clearer information than an elaborate menu.
07Roots
In his 1975 paper on education, Joseph Stiglitz examined a familiar object with an unfamiliar question: a diploma. Employers see the certificate while a worker’s productive ability remains partly hidden. How could schooling help sort workers, and what would that sorting do to wages and the distribution of income?
The question mattered because education can both develop skills and provide information about people who complete it. Stiglitz gave screening a formal economic treatment, examining how the information function could affect individual choices and the wider economy. Michael Spence’s earlier work on job-market signaling explored the better-informed applicant’s side of this problem.
In 1976, Michael Rothschild and Stiglitz moved the problem into insurance. Buyers could know their risk while insurers offered contracts without that knowledge. Different combinations of coverage and payment could separate risk groups. Competition added a complication: a rival’s new offer could upset that separation, and their model showed that a competitive equilibrium might fail to exist.
These papers helped establish information economics as a way to study markets where the terms of a deal also reveal information. Screening became a framework for understanding credentials, contract menus, and the conditions that make self-selection work.
08How solid is this?
Screening is an established mechanism in information economics, with formal results showing when tests or menus reveal hidden characteristics. Its success in practice depends on design and context; insurance research finds that coverage choices do not consistently identify risk alone.
09Connections
- Often confused with Moral Hazard, Signaling
- Helps counter Adverse Selection, Asymmetric Information, Principal-Agent Problem
- Part ofMechanism Design
- See alsoConstruct Validity, Incentive Compatibility
10Origin and sources
Joseph Stiglitz gave screening a formal economic treatment in 1975. Michael Rothschild and Stiglitz developed its application to competitive insurance markets in 1976.
- [1]Stiglitz, J. E. (1975). The Theory of "Screening," Education, and the Distribution of Income. The American Economic Review, 65(3), 283–300.
- [2]Rothschild, M., & Stiglitz, J. (1976). Equilibrium in Competitive Insurance Markets: An Essay on the Economics of Imperfect Information. The Quarterly Journal of Economics, 90(4), 629–649.
- [3]Spence, M. (1973). Job Market Signaling. The Quarterly Journal of Economics, 87(3), 355–374.
- [4]Cohen, A., & Siegelman, P. (2010). Testing for Adverse Selection in Insurance Markets. Journal of Risk and Insurance, 77(1), 39–84.
Suggest an edit· Updated 2026-10-02