Skip to content

Screening

Prime #
506
Origin domain
Economics & Finance
Also from
Operations Research, Psychology
Aliases
Self Selection Mechanism, Menu Design, Sorting Mechanism
Related primes
Signaling, Information Asymmetry, Adverse Selection, Moral Hazard, Mechanism Design, Incentive Compatibility, Price Discrimination, Agency Problem

Core Idea

Screening names the abstraction that (1) when an uninformed party must interact with agents whose types they cannot directly observe, (2) the uninformed party can design a menu — a structured set of terms (price, quality, coverage level, deductible, contract duration) — (3) such that agents of different types find it in their own interest to self-select into different options, (4) with the chosen option revealing the agent's type and allowing the uninformed party to respond differentially — overcoming adverse selection without requiring direct verification. The term itself was introduced by Stiglitz (1975)[1] in the context of education and labor-market sorting.[1]

Screening is the response; adverse selection is the problem it responds to. An adverse-selection-plagued uniform market (a single product for a heterogeneous population) is the baseline against which a screening menu is evaluated.

How would you explain it like I'm…

Pick to tell

Imagine a lemonade stand selling two sizes: a small cup for a little money and a giant cup for a lot. Big-thirsty kids pick the giant cup; small-thirsty kids pick the small one. You didn't have to ask how thirsty they were — they told you by which one they bought. Screening means setting up choices so people show you who they are by what they pick.

Sort By Choice

Sometimes you need to know something about a person but you can't just ask and trust the answer — like how risky a driver someone is, or how much they really want a job. Screening is a clever trick: you offer a menu of different deals, each one designed so that different kinds of people will naturally pick different options. A safe driver picks a low-deductible insurance plan; a risky driver picks the high-deductible one. By their choice, they tell you which type they are, even though you couldn't see it directly.

Self-selecting menus

Screening is a strategy used by an uninformed party who must deal with people whose hidden type matters. Instead of trying to verify the type directly, which is often impossible, the uninformed party designs a menu of options so that different types naturally find different options most appealing. When agents choose, their choice reveals their type. An insurance company offers a low-deductible expensive policy and a high-deductible cheap policy; high-risk customers prefer the first, low-risk customers prefer the second, and the company can charge accordingly. Screening is the response to adverse selection: rather than offering one product to a mixed pool and getting stuck with the worst customers, the menu sorts them.

 

Screening is a mechanism-design response to information asymmetry, specifically adverse selection. When an uninformed party (the principal) must transact with agents whose private types they cannot observe (insurance risk, productivity, willingness to pay), the principal designs a menu of contracts, combinations of price, quantity, quality, deductible, coverage, or duration, structured so that each type finds it in its own interest to choose a different option. This self-selection (formally, incentive compatibility) makes the chosen contract reveal the agent's type, enabling differential treatment without direct verification. Screening contrasts with signaling, where the informed party moves first to reveal its type (e.g., costly education in Spence's model); in screening, the uninformed party moves first by offering the menu. The term was introduced by Stiglitz (1975) in the context of labor-market sorting. The benchmark against which a screening menu is evaluated is the uniform-product market that collapses under adverse selection.

Structural Signature

Screening has a tightly specified anatomy, inverse to signaling, presented in canonical form by Laffont and Martimort (2002)[2] in their adverse-selection chapter:

  • An uninformed designer (employer, insurer, lender, buyer, regulator) who cannot directly observe agent types.[2]
  • Informed agents with heterogeneous types (high/low risk, high/low ability, high/low quality, high/low willingness to pay) known to themselves.
  • A menu of options — different combinations of terms (price, quality, coverage level, deductible, workload, compensation, warranty, contract duration) offered by the designer.
  • Type-differentiated preferences — agents of different types have systematically different preferences across the menu, so each type's utility-maximizing choice differs.
  • Incentive-compatibility constraints — the menu is designed so each type's preferred option is the one the designer wants them to take, with no type preferring to "cheat" by selecting a different type's option.
  • Individual-rationality constraints — each type must prefer their designated option to the outside option of not participating at all.
  • A revelation outcome — the agent's chosen option reveals enough about their type for the designer to respond appropriately.

The design problem is to construct a menu that satisfies incentive compatibility and individual rationality for all types while maximizing the designer's objective (revenue, efficiency, risk-pool quality).

What It Is Not

Screening is not signaling. Signaling is sender-initiated costly communication by the informed party; screening is receiver-initiated menu design by the uninformed party. The two address the same information-asymmetry problem from opposite directions and are routinely presented together as a tight pair, with Spence (1973)[3] supplying the canonical sender-side counterpart to the screening literature.[3]

It is not direct verification or monitoring. Screening operates through incentive-compatible menu design; verification relies on direct inspection, audit, or testing. In many real contexts both are combined — insurers screen through deductible menus and verify through medical exams — but the abstractions are conceptually distinct.

It is not customer segmentation. Marketing segmentation offers different products to different groups based on the seller's assessment of their characteristics; screening uses self-selection by the informed party as the type-revealing mechanism. A seller offering different SKUs to rural vs. urban customers based on demographic data is segmenting; a seller offering a menu and letting customers choose is screening.

It is not always welfare-improving. Screening often involves deliberate downward distortion of the low-type's option — worse than what would be offered under full information — to reduce the information rents the high-type would otherwise extract. Screening is a second-best response to information asymmetry, not a first-best allocation.

It is not limited to binary type structures. The theory has been developed for continuous type distributions (Mirrlees 1971[4] for optimal income taxation; Mussa-Rosen 1978[5] for quality discrimination; Maskin-Riley 1984[6] for quantity discrimination), with rich equilibrium structures including bunching of types at certain menu points.

Broad Use

In insurance, structuring deductibles, copayments, and coverage tiers lets insurers offer multiple policies so high-risk clients select high-coverage-low-deductible plans while low-risk clients select low-coverage-high-deductible plans. Rothschild and Stiglitz (1976)[7] formalized this.[7]

In labor-market hiring, multi-stage interviews, coding tests, case-study work products, take-home assignments, and graduated trial-period arrangements screen applicants beyond what résumés and credentials alone reveal.

In credit markets, loan menus offering various combinations of interest rate, collateral, reporting requirements, and covenants induce borrower self-selection — Stiglitz and Weiss (1981)[8] showed how interest-rate menus combined with credit rationing serve as a screening device.[8] Low-risk borrowers accept strict covenants and lower rates; high-risk borrowers accept higher rates for looser covenants — or the high-risk ones self-select out entirely.

In education, application processes with essays, interviews, portfolios, and recommendation letters screen for applicant characteristics — commitment, fit, ability — that test scores alone do not reveal.

In software pricing and SaaS product design, tiered plans (basic/pro/enterprise) induce customers to self-select into the tier matching their willingness to pay and feature needs — a close cousin to second-degree price discrimination.

In regulatory policy, optional compliance pathways (self-certification vs. third-party audit vs. full agency inspection) let firms select the compliance level matching their risk profile and operational capability, with the chosen pathway itself conveying information.

In academic admissions and grant allocation, qualifying exams, essays, research proposals, and interviews screen applicants whose underlying characteristics are difficult to observe directly.

In cooperative and association membership, application processes, probationary periods, and participation requirements screen for commitment and alignment in ways that simple dues-paying does not.

Clarity

The abstraction clarifies why many real-world contracts, menus, and application processes look more complex than they "need" to be — the baseline against which screening is evaluated is the lemons-style market collapse described by Akerlof (1970)[9], in which uniform pricing under information asymmetry drives the market to unravel.[9] Insurance with multiple coverage levels, loans with covenants, software with tiered feature sets, and universities with detailed admission procedures make more sense once the screening structure is recognized: the complexity is not bureaucratic accretion but the mechanism through which the uninformed party overcomes adverse selection.

The clarity extends to policy design. Means-tested social programs with deliberately awkward application processes are often criticized as bureaucratic; screening theory suggests some of the friction is a deliberate screening device — the public-economics term of art is an ordeal mechanism (Nichols and Zeckhauser 1982[10]) — that imposes a cost only genuinely needy applicants will bear, improving targeting through self-selection. Kleven and Kopczuk (2011)[11] formalized the resulting trade-off: more rigorous screening improves classification but raises application complexity, induces incomplete take-up, and creates a margin between Type I exclusion and Type II inclusion errors. The stronger distributional critique — that the same frictions fall disproportionately on less-resourced participants — is developed in the administrative burden literature (Moynihan, Herd, and Harvey 2015[12]; Herd and Moynihan 2025[13]) and in Sunstein's sludge audits framework[14]. Empirical work on welfare take-up (Currie 2006[15]; Bhargava and Manoli 2015[16]) shows that simplification and default enrollment materially raise participation, while field evidence on self-targeting cautions that the incidence of ordeals is not theoretically fixed (Alatas et al. 2016[17]). The current policy-design recommendation is not "never screen" but manage the trade-off explicitly: prefer administratively verifiable eligibility and default enrollment when possible, reduce learning and compliance costs, and test whether a given friction improves targeting net of exclusion costs.

Manages Complexity

Without screening, the uninformed party in markets with heterogeneous types faces a bad choice: ignore the heterogeneity and offer a uniform option (producing adverse selection), or attempt direct type verification for every applicant (often impractical or prohibitively costly). Screening collapses the complexity by delegating type-revelation to the informed party — the agents reveal their type through their choices, and the menu is designed to make that revelation incentive-compatible. Bolton and Dewatripont (2005)[18] develop this delegation logic in their canonical contract-theory treatment.[18]

This delegation scales remarkably well. An insurer with millions of policyholders need not individually assess each one's risk profile; policyholders self-sort through deductible and premium choices. A university with thousands of applicants need not interview each one thoroughly; the application essay, test scores, and extracurricular record induce self-selection. A cloud-computing provider with hundreds of thousands of enterprise customers need not negotiate each contract individually; the tier menu lets customers choose their own scale.

Screening has informational limits. It can reveal type only along the dimensions the menu distinguishes; types differing on unobserved dimensions still pool. And screening does not solve moral hazard — once types have selected their option, they may still misbehave — requiring additional monitoring or incentive-compatible features within each option.

Abstract Reasoning

Consider the canonical two-type screening problem, presented in standard textbook form by Mas-Colell, Whinston, and Green (1995)[19] in their adverse-selection chapter.[19] Let types be \(\theta \in \{\theta_L, \theta_H\}\) with probabilities \(p_L, p_H\). The principal offers a menu \(\{(q_L, t_L), (q_H, t_H)\}\) (quality-quantity \(q\), payment \(t\)). The principal maximizes expected profit \(\sum_i p_i (t_i - c(q_i))\) subject to:

  • Individual rationality (IR) for each type: \(u_i(q_i, t_i, \theta_i) \geq 0\).
  • Incentive compatibility (IC) for each type: \(u_i(q_i, t_i, \theta_i) \geq u_i(q_j, t_j, \theta_i)\) for all \(j \neq i\).

Under standard assumptions (single-crossing, type-ordered preferences) the optimal menu exhibits a characteristic pattern: the high-type's option is at the efficient quality/quantity; the low-type's option is distorted downward relative to the full-information efficient level. This downward distortion is the cost the principal pays to reduce information rents the high type would otherwise extract by threatening to select the low-type option. The high type receives a positive information rent; the low type receives zero rent at their option.

The pattern generalizes with remarkable consistency. In optimal income taxation (Mirrlees 1971), the government screens high- from low-ability workers through progressive tax schedules. In optimal insurance contracts (Rothschild-Stiglitz 1976), the insurer screens high- from low-risk clients through coverage-deductible menus. In optimal quality discrimination (Mussa-Rosen 1978), the monopolist screens high- from low-valuation customers with the low-quality option distorted downward. All three share the structural skeleton.

The mathematics yields useful design anchors. The revelation principle applies: without loss of generality the principal can restrict attention to direct menus in which each type truthfully reports their type and the menu assigns the designated option. The no-distortion-at-the-top result — the highest type's option is at the efficient level — appears with substantial generality.

Knowledge Transfer

Structural role mappings, drawing on Salanié's (2005)[20] cross-domain treatment of screening mechanisms:

  • Uninformed designer ↔ insurer / employer / lender / platform / monopolist / tax authority
  • Informed agents ↔ policyholders / applicants / borrowers / customers / taxpayers
  • Menu options ↔ policy tiers / interview stages / loan products / pricing plans / tax brackets
  • Type-differentiated preferences ↔ single-crossing / type-ordered utilities
  • Incentive compatibility ↔ "each type's chosen option is the one the designer wants them to take"
  • Individual rationality ↔ participation constraint / outside-option threshold
  • Downward distortion of low-type option ↔ information-rent minimization cost
  • No distortion at the top ↔ highest type always gets the efficient allocation

Where the structure transfers: in online lending platforms, loan menus with various interest-rate / collateral / reporting combinations induce borrower self-selection; riskier borrowers accept higher rates for looser covenants, lower-risk borrowers accept covenants for lower rates, and very-high-risk borrowers are screened out entirely by credit-score floors.[20] In retail loyalty programs, premium tiers with higher fees and higher benefits induce high-engagement customers to self-select, revealing willingness-to-spend and enabling marketing customization. In software subscription pricing, usage-based vs. seat-based vs. enterprise-unlimited tiers induce customers to self-select based on expected usage patterns and organizational scale.

The transfer extends to less commercial domains. In health insurance, high-deductible health plans paired with HSAs attract healthier enrollees who expect low medical spending, while lower-deductible plans attract enrollees anticipating more care; ACA exchange design, risk-adjustment mechanisms, and metal-level tiers extensively address the resulting screening and adverse-selection dynamics. In academic journal submission, journal tiers with different prestige levels, review-time expectations, and article-processing fees induce authors to submit to the tier they believe their work can reach — a soft screening mechanism. In cooperative membership, tiered membership classes with different governance rights, patronage-rebate structures, and participation requirements induce members to self-select into the class matching their engagement level. In technical hiring, take-home coding challenges and multi-hour on-sites screen not only for ability but for commitment — low-commitment applicants drop out before completing the screens, revealing themselves as poor fits for high-investment roles.

Example

Formal / abstract

The foundational paper is Michael Rothschild and Joseph Stiglitz's 1976 Quarterly Journal of Economics paper Equilibrium in Competitive Insurance Markets: An Essay on the Economics of Imperfect Information[7]. They showed that in a competitive insurance market with adverse selection (high-risk and low-risk types unobservable to insurers), pooling equilibria cannot exist in pure strategies — insurers must offer a menu of contracts, with low-risk types accepting less-than-full insurance at lower premiums and high-risk types taking full insurance at higher premiums, through a separating equilibrium. This paper launched a vast literature on screening.

James Mirrlees's 1971 Review of Economic Studies paper An Exploration in the Theory of Optimum Income Taxation[4] developed the formal theory of screening applied to income taxation, where the government does not observe workers' abilities but does observe their incomes. The paper derived the optimal tax schedule as a screening menu and included the celebrated no-distortion-at-the-top result. Mirrlees shared the 1996 Nobel Prize in Economics partly for this work[21].

Mussa-Rosen (1978)[5] for quality-differentiated goods and Maskin-Riley (1984)[6] for quantity-differentiated goods applied the framework to monopoly pricing with heterogeneous consumers, producing menu-design solutions that remain standard.

Joseph Stiglitz's subsequent work applied screening across credit markets (Stiglitz-Weiss 1981)[8], labor markets, and financial markets; Akerlof's[9] and Spence's complementary work on adverse selection and signaling, combined with Stiglitz's screening analysis, earned the three the joint 2001 Nobel Prize in Economics for foundations of the economics of markets with asymmetric information[22]. Eric Maskin's later implementation-theory work generalized screening to multi-type, multi-agent settings; Maskin shared the 2007 Nobel with Hurwicz and Myerson for mechanism-design foundations[23].

The literature remains actively developed in behavioral screening (agents with non-standard preferences), dynamic screening (menus evolving over time), and algorithmic screening (machine learning designing or refining menus).

Applied / industry

A regional community development financial institution (CDFI) providing small-business loans to underserved entrepreneurs in a multi-state service area is redesigning its lending programs to better serve both borrowers and the institution's mission and solvency. The empirical literature on screening in real markets — for example Cawley and Philipson (1999)[24] testing for asymmetric information in life insurance — informs how the redesign is evaluated.[24] Historically the CDFI has offered a single uniform small-business loan product — fixed interest rate, fixed collateral requirement, fixed reporting cadence, fixed maximum amount — with loan officers making subjective go/no-go decisions on each application.

The board notices two problems. First, stronger-prospect borrowers who could easily qualify for conventional bank lending find the CDFI's uniform product over-priced relative to their risk; they self-select out, leaving a loan portfolio skewed toward higher-risk applicants that strains the institution's capital. Second, marginal-prospect borrowers — those with genuine business prospects but weak documentation, limited collateral, or thin credit histories — cannot afford the CDFI's required collateral and reporting burden; they go to predatory lenders or forgo financing entirely. The uniform product is failing both ends.

A mission-and-strategy review consultancy reframes the problem through the screening lens. The CDFI is an uninformed party attempting to lend to a heterogeneous borrower population; borrowers know their own prospects, commitment, and risk far better than the CDFI can externally verify; the uniform product is forcing the entire population into a single poorly-fit option. The fix is a menu of loan products with terms that induce borrowers to self-select into the product matching their type, each structured to be financially viable for the CDFI given the borrower types who select into it.

The redesigned menu includes four products. Product A (Graduation Loan): competitive interest rate close to prevailing bank rates, modest collateral, light monthly reporting, longer amortization — priced to retain strong borrowers who would otherwise graduate to conventional banking but who benefit from the CDFI's technical-assistance services. Product B (Community Growth Loan): moderate rate, moderate collateral, moderate reporting, bundled business-coaching hours — attractive to middle-segment borrowers who want both capital and expertise. Product C (Stepping-Stone Microloan): small principal, higher rate, minimal collateral, intensive monthly check-ins, strong coaching bundle — attractive to earlier-stage or weaker-documentation borrowers who need intensive support and accept the monitoring in exchange for access. Product D (Character-Based Micro-Credit): very small principal, peer-group accountability structure (borrower circles modeled on Grameen-style group lending, which combines screening with moral-hazard control through peer monitoring), minimal formal credit review, direct mentor pairing — attractive to first-time borrowers with no credit history.

The consultancy stress-tests the menu for incentive compatibility. Would a strong borrower be tempted to take Product D (cheapest, lowest collateral) rather than Product A? No — the peer-accountability and mentor-pairing structure in D is unattractive to borrowers with clean financials who can pass A's underwriting. Would a weak borrower take Product A rather than C? They might try, but A's credit-history and collateral requirements screen them out automatically. The menu's incentive-compatibility rests on the joint structure: each product's non-price terms combine with the price to make self-selection work.

Three years after implementation the CDFI's loan portfolio mix has rebalanced. Loan volume has grown across all four products; default rates have declined in aggregate even as the CDFI serves a broader borrower population; stronger borrowers are staying with the CDFI through Product A rather than graduating away; mission metrics (borrower-business survival, job creation, community wealth-building) have improved across segments. The evaluation credits the screening-menu redesign explicitly: letting borrowers sort themselves into fit-to-circumstance products produced better outcomes than loan-officer-driven sorting of a uniform product. The outcomes are qualitative and post-hoc, not a controlled test; the screening framing is used to design and interpret the intervention rather than to causally isolate its contribution from confounds (improved loan management, macro conditions, coaching effects). This is screening in a mission-driven CDFI context — the theoretical framework translated into an operationally viable lending program serving dual financial and social objectives.

Structural Tensions and Failure Modes

  • T1: Revelation Power vs Downward Distortion Cost. [5] As Mussa and Rosen (1978)[5] established for quality-differentiated screening, the optimal menu requires distorting the low-type option below its full-information level.

    • Structural tension: To reduce the information rent the high-type would otherwise extract by mimicking the low-type's option, the optimal menu distorts the low-type's option below the full-information efficient level. This distortion is a real welfare cost — the low type receives a worse deal than they would under perfect observability — and it is the price the designer pays to get any separation at all. The mechanism trades allocative efficiency for information revelation.
    • Common failure mode: Designing menus that over-distort the low end in pursuit of rent extraction, hollowing out participation at the very segment screening was supposed to serve. Insurance deductibles that exceed household savings, loan terms that strip rather than finance, welfare applications whose documentation burden matches their payment — each satisfies incentive-compatibility in the model while failing individual-rationality for the low-type in practice.

  • T2: Incentive-Compatibility Within Menu vs Walk-Away to Outside Option. [25] Wilson (1977)[25] showed that participation constraints in adverse-selection insurance markets can dominate incentive-compatibility, with whole types refusing the menu rather than choosing among options.

    • Structural tension: IC makes each type's utility-maximizing choice the one the designer assigned to them, but it cannot prevent types from walking away entirely if every menu option is worse than the outside option. The screen is a conditional sort — among those who participate — not an unconditional one. Participation is the first thing that can fail.
    • Common failure mode: Welfare-eligibility menus with take-up rates in the single digits; marketplace insurance that sorts nicely among enrollees but misses the uninsured; talent pipelines that screen cleanly among applicants but lose whole populations who never apply. The menu looks well-designed internally while the population that would have benefited most has routed around it to predatory lenders, emergency rooms, or the gray market.

  • T3: Menu Richness vs Participant Cognitive Load. [26] Handel (2013)[26] documents that, when health-plan menus offer too many options, bounded-rational consumers exhibit inertia and choose dominated plans — a behavioral failure of menu design.

    • Structural tension: Richer menus extract more information by offering finer type-differentiated options; but richer menus also demand more cognitive work from participants to identify and select their utility-maximizing option. Bounded-rational agents exposed to a 7-tier menu may select the most salient option, the default option, or the option a friend chose, rather than the option that matches their type.
    • Common failure mode: SaaS and health-plan menus with enough tiers and feature-bundled options that empirical take-up clusters on the default and the second-cheapest regardless of actual fit. The menu designer sees "the market chose tier 2," interprets it as type-revelation, and builds on a sorting signal that is really a choice-architecture artifact.

  • T4: Information Extraction vs Equity of Friction. [27] Chiappori and Salanié (2000)[27] developed the empirical framework for testing whether observed choices in insurance markets reflect informative self-selection or other frictions, providing tools for evaluating when menu friction screens type versus excludes the unresourced.

    • Structural tension: Friction that sorts types — documentation burden, wait time, complexity of application, payment timing — also disproportionately blocks less-resourced participants who have less slack to absorb friction. A mechanism that looks to the designer like efficient self-selection looks to the excluded like exclusion. The same feature that gives screening its power is what makes it ethically fraught.
    • Common failure mode: Administrative-burden regimes rationalized as "screening for genuine need" that in practice screen out exhausted, non-English-speaking, low-literacy, or disability-affected populations who are among the neediest. The designer invokes signaling/screening theory as justification while the population that was supposed to be sorted in is sorted out by fatigue rather than by type.

  • T5: Static Menu vs Drifting Type Distribution. [28] Courty and Li (2000)[28] formalized dynamic screening, where types reveal progressively over time and menus must adapt to shifting distributions rather than calcifying around their initial design.

    • Structural tension: The optimal menu depends on the distribution of types in the participating population. When the distribution shifts — through adverse selection, demographic change, platform growth, regulatory change, external shocks — a menu that was optimal calcifies into something that now misprices risk, misallocates quantity, or produces perverse self-selection. Menus inherit the assumptions of their moment.
    • Common failure mode: Insurance pools designed at enactment that discover, five years later, that the underlying risk distribution has shifted underneath the menu — premium/deductible combinations that no longer separate the intended segments, adverse selection spirals that the original designers did not anticipate. Redesign happens reactively in response to visible trouble rather than proactively from monitoring distribution drift.

  • T6: Continuous Types and Bunching vs Discrete-Type Intuition.

    • Structural tension: The tidy two-type toy models (high/low) predict a clean two-option menu with no distortion at the top. Once types are continuous, optimal designs exhibit bunching — intervals of types that pool on the same menu option because individual differentiation would demand uneconomically fine menu gradations. The real mechanism loses separation precisely where the simple intuition insists it should hold.
    • Common failure mode: Treating the discrete-type result as the structural truth and missing the bunching that continuous-type analysis predicts. Menu designers produce fine-grained tiers believing they will separate; empirical outcomes show large clumps of participants pooling on a few focal options; the designer reads this as "simpler menu would do" rather than as the structural bunching the mathematics predicted.

Structural–Framed Character

Screening is a hybrid on the structural–framed spectrum. Part of it is a bare pattern that recurs unchanged in any setting: one party who cannot see another's hidden type offers a structured menu of options engineered so that different types find it in their own interest to pick different items, and the choice itself reveals the type. Part of it is a frame inherited from economics, where the menu is read as a contract and the parties as utility-maximizing agents.

The structural core is a self-selection mechanism that can be stated without any human institution: a designer, a set of unobservable types, and a menu whose incentives sort those types apart. That same anatomy turns up far from markets—a software system serving different tiers of access so that users sort themselves by behavior, a biological setup where costly options separate organisms by quality, a survey designed so honest and dishonest respondents answer differently. Recognizing it is mostly a matter of spotting an incentive-compatible sorting structure that is already in place. But its home vocabulary—price, deductible, coverage, contract duration, the uninformed designer optimizing against adverse selection—carries the assumptions of rational agents and welfare that the bare sorting pattern does not need. The pattern leans structural, with a light economic frame riding along, placing it mid-spectrum with a structural tilt.

Substrate Independence

Screening is a highly substrate-independent prime — composite 4 / 5 on the substrate-independence scale. Its structural signature is elegant and fully substrate-agnostic — an uninformed designer offers a menu so that informed agents self-select and reveal their hidden types — and it spans economics, operations, organizational psychology, and risk management. The examples cross insurance markets, hiring, and organizational role definition, all driven by the same choice-architecture mechanism. What keeps the transfer evidence a notch shy of the top is that applications often dress the idea in domain-specific language even though the underlying type-revealing mechanism is universal.

  • Composite substrate independence — 4 / 5
  • Domain breadth — 4 / 5
  • Structural abstraction — 5 / 5
  • Transfer evidence — 3 / 5

Relationships to Other Primes

One-hop neighborhood: parents above, mutual partners to the right, children below.Screeningcomposition: Information AsymmetryInformationAsymmetrysubsumption: Mechanism DesignMechanism Design

Parents (2) — more general patterns this builds on

  • Screening is a kind of Mechanism Design

    Screening is a specialization of mechanism design in which the designer is the uninformed party facing agents of unknown type, and the desired outcome is type-revelation through self-selection. It inherits mechanism design's general inversion — start with the target equilibrium, search for rules implementing it under private information and self-interest — and specializes by fixing the rule class to menus of contracts and the objective to separating types. The participants' best responses, calibrated by their hidden types, deliver the information the designer needs without direct verification.

  • Screening presupposes Information Asymmetry

    Screening presupposes information asymmetry because its entire structure is a response to an unobservable type held by the agent: the uninformed party designs a menu of contracts such that agents of different types self-select, revealing the hidden information through choice rather than through verification. Without the prior distributional fact that one side knows something material the other cannot observe, there would be no need for self-selection mechanics. Screening inherits information asymmetry's structure of unequal private knowledge and supplies one of the canonical mechanisms — menu-based separation — by which the uninformed side recovers what it cannot directly see.

Path to root: Screening → Mechanism Design

Neighborhood in Abstraction Space

Screening sits in a sparse region of abstraction space (87th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.

Family — Strategic Mechanisms & Bounded Rationality (13 primes)

Nearest neighbors

Computed from structural-signature embeddings · 2026-05-29

Not to Be Confused With

Screening must be distinguished from Signaling, its closest and most important neighbor—they address the same adverse-selection problem from opposite sides. Screening is receiver-initiated: the uninformed party (lender, insurer, employer) designs a menu of incentive-compatible contract options and lets the informed party (borrower, customer, job candidate) self-select into the option that best matches their type. The lender says "here are four loan products with different terms; choose the one that fits"; the borrower's choice reveals their type. Signaling is sender-initiated: the informed party (borrower, job candidate) makes a costly, type-revealing investment that the uninformed party can observe. The borrower improves their credit history or the candidate earns an expensive degree; the signal's cost structure makes it uneconomical for weaker types to mimic, revealing the sender's type through their investment decision. Screening works by restricting the menu so that self-selection reveals type; signaling works by making type-revelation expensive enough that weak types won't bother. In practice, markets deploy both: lenders screen applicants via credit menus while borrowers signal via collateral pledges and covenants; employers screen via job-offer packages while job candidates signal via education and interview performance. The two are complements, not substitutes. Confusing them means misdiagnosing where information-asymmetry frictions come from and where to intervene. If adverse selection is driven primarily by information asymmetry, screening or signaling can address it; if it is driven by hidden information about unobservable characteristics, signaling is more powerful (the signal itself must reveal what cannot be directly observed). The distinction is essential for understanding market design.

Screening is also distinct from Price Discrimination, though the two can coexist. Price discrimination uses price differences to extract consumer surplus once types or willingness-to-pay is known or inferred; it presumes information about type is already available and leverages it for profit or revenue. Screening uses contract/menu design (which may include prices but also non-price terms: coverage levels, wait times, collateral requirements, service quality) to overcome information asymmetry—to reveal type in the first place. A screening equilibrium may or may not involve price differences; it must involve incentive-compatible self-selection. Price discrimination presumes types are known; screening solves how to discover types. An airline pricing economy fares cheaper than business fares; that is price discrimination because the airline already knows most business travelers will pay more (via self-selection into high-value routes or willingness to book last-minute). An insurance company using separate deductible/premium menus is screening because it doesn't know which applicants are high-risk and designs the menu so that their choices reveal their risk type. That menu may incidentally involve prices (premiums), but the screening mechanism is the menu structure—the fact that high-deductible low-premium appeals to low-risk types, high-premium low-deductible appeals to high-risk types. Price discrimination targets consumers whose types are already distinguishable; screening distinguishes types that initially cannot be observed.

Screening is also not Herding Behavior, the tendency of agents to imitate others' actions even when their private information suggests different choices. Herding is a decentralized coordination phenomenon driven by uncertainty and the inference people draw from observing others' behavior; it emerges without any deliberate institutional design. Screening is deliberate mechanism design by a principal (lender, insurer) to use incentive structures to overcome information asymmetry. In herding, individuals imitate each other; in screening, the principal structures the choice set so that individuals' utility-maximizing choices reveal their types. A bank run is herding (depositors see others withdraw, infer the bank is weak, withdraw themselves). Loan-product menus designed to screen borrower quality are screening (borrowers choose the product that matches their type, revealing it). Both can occur in the same market—a financial panic can herd everyone toward the safest institutions while those institutions simultaneously screen applicants via complex loan-menu terms—but they are structurally different phenomena. Screening is what the principal does; herding is what agents do when information is missing and they follow each other.

Screening is also distinct from the Price Mechanism, the broader coordination apparatus by which prices aggregate decentralized information and signal scarcity to independent decision-makers. The price mechanism presumes prices reflect and incorporate available information; it works best when information is roughly symmetric or publicly available. Screening operates because prices alone cannot aggregate or signal information about type—because the information is asymmetric and held by one party. A simple price (what a given borrower is charged in interest) cannot by itself reveal whether the borrower is high-risk or low-risk; the borrower knows their own risk profile, but the lender does not. The price mechanism would collapse into a separating equilibrium only if the price could be set based on type, but the lender cannot observe type without the borrower's cooperation. Screening solves this: by offering multiple options with different non-price and price terms, the lender creates a structure where the borrower's choice (informed by their knowledge of their own type) reveals that type. The price mechanism is the broad coordination framework; screening is a specific institutional design for revealing information within that framework when asymmetry exists.

Solution Archetypes

Solution archetypes in the catalog that build on this prime — directly (this prime is a source ingredient) or as a related prime.

Built directly on this prime (4)

Also a related prime in 13 archetypes

Notes

Pass B will articulate the tight-pair relationship with Signaling and the closer-than-it-seems relationship with Mechanism Design and Incentive Compatibility. Screening is a specific mechanism-design problem — menu design under adverse selection — and the incentive-compatibility constraints in screening are the same type imposed by mechanism design generally. Pass B should present screening as the complement of signaling, the receiver-side response to the same adverse-selection problem.

Pass B should address ethical and fairness considerations in screening design. The Clarity section frames the trade-off through the established ordeal mechanism / administrative burden / sludge literatures (Nichols-Zeckhauser 1982; Kleven-Kopczuk 2011; Moynihan, Herd, and Harvey 2015; Herd-Moynihan 2025; Sunstein 2022); Pass B should deepen the treatment by working through concrete screening contexts — means-tested welfare with onerous application processes, health insurance with deductibles exceeding household savings, and academic admissions with high application friction — and by presenting the policy-design recommendations (administrative verification of eligibility where feasible, default enrollment, routine sludge audits, and explicit evaluation of whether a screening friction improves targeting net of exclusion costs) as concrete design patterns rather than abstract caveats.

Pass B should also address dynamic screening (Courty-Li 2000[28]; Baron-Besanko 1984[29]) where menus evolve and agents may reveal type progressively, and continuous-type screening with its bunching and ironing phenomena (Mussa-Rosen 1978; Maskin-Riley 1984).

Review flags: tight_pair_with_signaling. The two abstractions must be presented together. The origin is unambiguously economics-finance (Rothschild-Stiglitz, Mirrlees, Mussa-Rosen, Maskin-Riley); operations research is a secondary domain where menu-design optimization and selection-procedure algorithms are studied; psychology_behavioral_sciences is relevant because real-world screening must account for bounded rationality, choice architecture effects, and non-standard preferences that depart from the classical model.

References

[1] Stiglitz, J. E. (1975). "The theory of 'screening,' education, and the distribution of income." American Economic Review, 65(3), 283–300. Foundational screening paper applying separation to education as a signaling device and screening mechanism in labor markets. ↩

[2] Laffont, J.-J., & Martimort, D. (2002). The Theory of Incentives: The Principal-Agent Model. Princeton University Press. Canonical text developing the systematic reframing of governance and policy problems as incentive design under private information—structuring rules so self-interested behavior produces collective objectives rather than relying on direction or compliance. ↩

[3] Spence, M. (1973). Job market signaling. The Quarterly Journal of Economics, 87(3), 355–374. Ports the information-asymmetry pattern from product markets into the labor market, showing that costly signals (education) can establish separating equilibria when employers cannot directly observe worker productivity — a canonical cross-substrate transfer of the asymmetry structure. ↩

[4] Mirrlees, James A. "An Exploration in the Theory of Optimum Income Taxation." Review of Economic Studies 38, no. 2 (April 1971): 175–208. DOI: 10.2307/2296779. ↩

[5] Mussa, Michael, and Sherwin Rosen. (1978). "Monopoly and Product Quality." Journal of Economic Theory 18(2): 301–317. DOI: 10.1016/0022-0531(78)90085-6. Continuous-type screening with quality-differentiated goods; introduced ironing and bunching techniques that became standard for second-degree price discrimination. ↩

[6] Maskin, Eric, and John Riley. "Monopoly with Incomplete Information." RAND Journal of Economics 15, no. 2 (Summer 1984): 171–196. DOI: 10.2307/2555674. JSTOR: 2555674. Continuous-type screening with quantity-differentiated goods; the companion to Mussa-Rosen for second-degree price discrimination. ↩

[7] 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. Canonical model of adverse selection (hidden type, pre-contract) and the screening response in insurance markets, where the uninformed insurer offers a contract menu inducing self-selection by risk type. ↩

[8] Stiglitz, Joseph E., and Andrew Weiss. "Credit Rationing in Markets with Imperfect Information." American Economic Review 71, no. 3 (1981): 393–410. ↩

[9] Akerlof, G. A. (1970). The market for "lemons": Quality uncertainty and the market mechanism. The Quarterly Journal of Economics, 84(3), 488–500. Founding formalization of information asymmetry: a seller-held quality fact unverifiable by buyers drives good products out of the market (the unraveling mechanism), with counteracting institutions such as guarantees, brand names, and reputation showing the distortion is a pressure rather than a deterministic outcome. ↩

[10] Nichols, Albert L., and Richard J. Zeckhauser. "Targeting Transfers through Restrictions on Recipients." American Economic Review 72, no. 2 (May 1982): 372–377. JSTOR: 1802361. Classical introduction of ordeal mechanisms — deliberately costly participation requirements as screening devices that improve targeting of transfers by inducing self-selection on unobservable need. ↩

[11] Kleven, Henrik Jacobsen, and Wojciech Kopczuk. "Transfer Program Complexity and the Take-Up of Social Benefits." American Economic Journal: Economic Policy 3, no. 1 (February 2011): 54–90. DOI: 10.1257/pol.3.1.54. Formalizes the targeting/complexity trade-off: more rigorous screening improves classification but raises application complexity, induces incomplete take-up, and creates a margin between Type I and Type II errors. ↩

[12] Moynihan, Donald P., Pamela Herd, and Hope Harvey. "Administrative Burden: Learning, Psychological, and Compliance Costs in Citizen-State Interactions." Journal of Public Administration Research and Theory 25, no. 1 (January 2015): 43–69. DOI: 10.1093/jopart/muu009. Defines administrative burden as learning, compliance, and psychological costs in citizen-state interactions and frames burdens as a deliberate policy instrument with distributive consequences. ↩

[13] Herd, Pamela, and Donald P. Moynihan. "Administrative Burdens in the Social Safety Net." Journal of Economic Perspectives 39, no. 1 (Winter 2025): 129–150. DOI: 10.1257/jep.20231394. Recent survey bringing the administrative-burden framework directly into mainstream public economics and arguing that burdens systematically reinforce inequality in access to safety-net programs. ↩

[14] Sunstein, Cass R. "Sludge Audits." Behavioural Public Policy 6, no. 4 (October 2022): 654–673. DOI: 10.1017/bpp.2019.32. Develops sludge as the term of art for excessive or unjustified frictions impeding access to goods and services, and proposes sludge audits as a policy-design tool for systematically identifying and reducing them. ↩

[15] Currie, Janet. "The Take-up of Social Benefits." In Public Policy and the Distribution of Income, edited by Alan J. Auerbach, David Card, and John M. Quigley, 80–148. New York: Russell Sage Foundation, 2006. (Related NBER Working Paper 10488, April 2004, nber.org/papers/w10488.) Survey establishing that administrative barriers and transaction costs materially reduce take-up of welfare benefits and that automatic or default enrollment tends to raise take-up. ↩

[16] Bhargava, Saurabh, and Dayanand Manoli. "Psychological Frictions and the Incomplete Take-Up of Social Benefits: Evidence from an IRS Field Experiment." American Economic Review 105, no. 11 (November 2015): 3489–3529. DOI: 10.1257/aer.20121493. Field experiment demonstrating that simplification and salience interventions can materially raise take-up of means-tested benefits, supporting the view that some friction is not informative screening but recoverable deadweight cost. ↩

[17] Alatas, Vivi, Abhijit Banerjee, Rema Hanna, Benjamin A. Olken, Ririn Purnamasari, and Matthew Wai-Poi. "Self-Targeting: Evidence from a Field Experiment in Indonesia." Journal of Political Economy 124, no. 2 (April 2016): 371–427. DOI: 10.1086/685299. Shows that self-targeting through ordeals can improve beneficiary composition but that marginal increases in ordeal cost may not further improve targeting; incidence of ordeals is theoretically ambiguous between richer and poorer applicants. ↩

[18] Bolton, Patrick, and Mathias Dewatripont. Contract Theory. MIT Press, 2005. Comprehensive contract-theory treatment; develops the delegation logic by which screening menus shift type-revelation from the principal to the agent, including detailed derivations of optimal menu design under adverse selection. ↩

[19] Mas-Colell, A., Whinston, M. D., & Green, J. R. (1995). Microeconomic Theory. Oxford University Press. Canonical graduate microeconomics textbook: develops the preference-based and choice-based approaches in parallel, takes the binary preference relation (with completeness and transitivity) as the primitive of consumer theory before introducing utility, and frames optimization as derived from a primitive preference ordering. ↩

[20] Salanié, Bernard. The Microeconomics of Market Failures. MIT Press, 2005. Comprehensive treatment of screening mechanisms, adverse selection, moral hazard, and their interactions in real-world markets. ↩

[21] Royal Swedish Academy of Sciences (1996). "The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 1996 — James A. Mirrlees and William Vickrey." Press release and scientific background. https://www.nobelprize.org/prizes/economic-sciences/1996/. ↩

[22] The Royal Swedish Academy of Sciences. "The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001." Awarded jointly to George A. Akerlof, A. Michael Spence, and Joseph E. Stiglitz "for their analyses of markets with asymmetric information." nobelprize.org/prizes/economic-sciences/2001/summary/. ↩

[23] Royal Swedish Academy of Sciences (2007). "The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2007 — Leonid Hurwicz, Eric S. Maskin, and Roger B. Myerson: for having laid the foundations of mechanism design theory." https://www.nobelprize.org/prizes/economic-sciences/2007/. ↩

[24] Cawley, John, and Tomas Philipson. "An Empirical Examination of Information Barriers to Trade in Insurance." American Economic Review 89, no. 4 (September 1999): 827–846. DOI: 10.1257/aer.89.4.827. Empirical investigation of asymmetric information in life insurance; provides one of the early tests of the predictions of screening models in real insurance markets. ↩

[25] Wilson, Charles. "A Model of Insurance Markets with Incomplete Information." Journal of Economic Theory 16, no. 2 (December 1977): 167–207. DOI: 10.1016/0022-0531(77)90004-7. Foundational analysis of equilibrium in insurance markets with adverse selection, including pooling vs. separating equilibria and the role of participation constraints when the outside option dominates the offered menu. ↩

[26] Handel, Benjamin R. "Adverse Selection and Inertia in Health Insurance Markets: When Nudging Hurts." American Economic Review 103, no. 7 (December 2013): 2643–2682. DOI: 10.1257/aer.103.7.2643. Documents that consumers facing rich health-plan menus exhibit large inertia and frequently choose dominated options, complicating the simple screening assumption that menu choices reveal type rather than choice-architecture artifacts. ↩

[27] Chiappori, Pierre-André, and Bernard Salanié. "Testing for Asymmetric Information in Insurance Markets." Journal of Political Economy, vol. 108, no. 1 (2000): 56–78. Empirical testing of asymmetric-information hypotheses underlying screening separating equilibria. ↩

[28] Courty, Pascal, and Hao Li. "Sequential Screening." Review of Economic Studies 67, no. 4 (October 2000): 697–717. DOI: 10.1111/1467-937X.00150. JSTOR: 2695945. Founding treatment of dynamic screening where types reveal progressively over time. ↩

[29] Baron, David P., and David Besanko. "Regulation and Information in a Continuing Relationship." Information Economics and Policy 1, no. 3 (1984): 267–302. DOI: 10.1016/0167-6245(84)90006-1. Multi-period regulatory screening under adverse selection; companion to Courty-Li for dynamic-screening foundations. ↩

[30] Riley, John G. "Informational Equilibrium." Econometrica 47, no. 2 (March 1979): 331–359. DOI: 10.2307/1912649. Framework for equilibrium analysis in signaling games; addresses equilibrium existence and uniqueness under information asymmetry.

[31] Crémer, Jacques. (1995). "Arm's length relationships." Quarterly Journal of Economics, vol. 110, no. 2: 275-295. Mechanism-design foundations of screening with multi-agent and continuous-type extensions; theoretical bridge to implementation.

[32] Crawford, Vincent P., and Joel Sobel. "Strategic Information Transmission." Econometrica, vol. 50, no. 6 (1982): 1431–1451. Strategic revelation and concealment in screening; foundations of partial information revelation in menu design.