Signaling happens when the informed party takes a costly action to reveal private information. Screening happens when the uninformed party designs a mechanism to get the informed party to reveal that information. Both address adverse selection and moral hazard, but they differ in who moves first: with signaling, the informed party initiates; with screening, the uninformed party does.

For a signal to work, it must be costly enough that low-quality types won't mimic it. A firm that sells a reliable product can afford to offer an expensive warranty; a firm selling junk can't. That cost differential is what makes the signal credible. More formally, the signal must satisfy a single-crossing condition: the cost of acquiring the signal must be lower for high-quality types than for low-quality types, so that at some signal level, it's worthwhile for high types to invest but not for low types.

Screening works by offering a menu of contracts that induces self-selection. An insurance company doesn't know your risk level, so it offers plans with different deductible/premium combinations and lets your choice reveal your type. The key constraint is that each contract in the menu must be designed so that each type strictly prefers the option meant for them.

A few conditions matter for both strategies:

Signals must be observable and verifiable by the uninformed party (a diploma hanging on the wall, not a vague claim of competence)
Screening contracts must be incentive-compatible, meaning each type genuinely prefers the option designed for them, and they must satisfy participation constraints, meaning each type prefers accepting their contract to walking away entirely
The cost structure of the signal must vary across types, so imitation is prohibitively expensive for low-quality agents

Pooling vs. Separating Equilibria

These two equilibrium concepts show up constantly in signaling and screening models:

Separating equilibrium: Different types choose different actions, so the uninformed party can distinguish them. High-productivity workers get a degree; low-productivity workers don't. The uninformed party updates beliefs accordingly and offers type-specific terms (e.g., different wages).
Pooling equilibrium: All types choose the same action, so no information is revealed. Every job applicant gets the same credential regardless of ability, and the employer offers a single wage based on the average productivity of the pool.

Whether a market settles into a pooling or separating equilibrium depends on the cost structure of the signal and the proportion of each type in the population. If high types are rare, the pooling wage is dragged down toward the low-type productivity level, giving high types a stronger incentive to separate. Separating equilibria are generally more informative, but they can be fragile: if off-equilibrium beliefs shift, a pooling equilibrium may dominate.

There's also a semi-separating (partial pooling) equilibrium, where some but not all types are distinguished. This can occur when the cost gap between types isn't large enough for full separation.

Effectiveness of Signaling and Screening

Education as a Labor Market Signal

Michael Spence's (1973) signaling model is the classic example. The core idea: education may not make you more productive, but it reveals your productivity because high-ability workers find it cheaper (in terms of effort, time, or opportunity cost) to obtain a degree.

Here's how the model works:

Workers differ in productivity (say, type $\theta_H$ and type $\theta_L$ , with $\theta_H > \theta_L$ ), but employers can't observe productivity directly.
Workers choose an education level $e$ before entering the labor market. The cost of education is $c(e, \theta)$ , and crucially, $\frac{\partial c}{\partial e}$ is lower for high types: acquiring an additional unit of education costs them less.
Employers observe education levels and offer wages based on their beliefs about the education-productivity link.
In a separating equilibrium, there exists a threshold $e^*$ $e^{*}$ such that high-productivity workers choose $e \geq e^*$ $e \geq e^{*}$ and low-productivity workers choose $e < e^*$ $e < e^{*}$ . The threshold $e^*$ $e^{*}$ must satisfy two incentive-compatibility conditions:
- High types prefer getting $e^*$ and earning $w_H$ to skipping it and earning $w_L$ : $w_H - c(e^*, \theta_H) \geq w_L$
- Low types prefer skipping education to mimicking: $w_L \geq w_H - c(e^*, \theta_L)$

The signal works because the marginal cost of education is lower for high-ability types. If everyone found school equally easy, education would carry no information.

Note that there are typically multiple separating equilibria (any $e^*$ in the range satisfying both IC constraints works). The Cho-Kreps intuitive criterion is one refinement used to select among them, often picking the least-cost separating equilibrium.

Potential drawbacks include overinvestment in education (workers acquiring credentials beyond what's socially optimal) and credential inflation (as more people signal, the bar keeps rising). From a social welfare standpoint, if education is purely a signal and adds no human capital, every dollar spent on it is a deadweight loss incurred just to sort workers.

Mechanisms for Reducing Information Asymmetry, Frontiers | Systematic Organization of COVID-19 Data Supported by the Adverse Outcome Pathway ...

Product Quality Signals in Consumer Markets

Warranties and money-back guarantees signal that the manufacturer is confident in durability. A firm selling a fragile product would lose money offering a generous warranty, so the offer itself is informative. The expected cost of honoring the warranty is lower for high-quality firms, satisfying the single-crossing condition.
Brand reputation functions as a repeated-game signal: firms invest in quality over time because the long-run value of their brand exceeds the short-run gain from cutting corners. The threat of losing future profits disciplines current behavior.
Third-party certifications (Energy Star, organic labels) provide independent verification, which is especially useful for credence goods where consumers can't evaluate quality even after purchase.
Advertising expenditure can itself be a signal. The logic (following Nelson and Milgrom-Roberts): a firm that spends heavily on advertising must expect repeat purchases, which implies confidence in product quality. The content of the ad matters less than the observable cost.

Financial Market Screening Mechanisms

Financial markets rely heavily on screening because lenders and insurers are the uninformed parties.

Credit scoring uses historical data (payment history, debt levels) to sort borrowers by risk. It's a screening tool because the lender designs the criteria, not the borrower.
Collateral requirements screen borrowers by willingness to put assets at stake. A borrower confident in repayment will accept collateral terms more readily than one who expects to default. This is a direct application of the Stiglitz-Weiss insight about credit rationing under asymmetric information.
Insurance deductibles induce self-selection through menu design (the Rothschild-Stiglitz model):
- High-deductible, low-premium plans attract low-risk individuals (they don't expect to file many claims)
- Low-deductible, high-premium plans attract high-risk individuals (they expect to use the coverage)

The effectiveness of all these mechanisms depends on how well they separate types through self-selection rather than relying on direct observation. When the type distribution is continuous rather than discrete, perfect separation becomes harder, and screening menus must be more finely tuned.

Applying Signaling and Screening

Automotive Industry Applications

The used car market is the textbook adverse selection setting (Akerlof's 1970 "lemons" problem), so signaling and screening are especially visible here.

Certified pre-owned (CPO) programs combine both mechanisms: the dealer signals quality through inspections and reconditioning, and the added warranty screens out lemons that wouldn't qualify for certification.
Vehicle history reports (Carfax, AutoCheck) signal a car's condition by making maintenance and accident history verifiable. They reduce information asymmetry by converting the seller's private information into public information.
Lemon laws are a government-imposed mechanism that sets minimum quality thresholds and gives buyers recourse, which discourages sellers from passing off defective vehicles. These function more as a regulatory floor than as screening per se, since they don't rely on self-selection.

Mechanisms for Reducing Information Asymmetry, Adverse Drug Reactions: The benefits of data mining | eLife

Employment Market Strategies

Employers face a screening problem: they can't directly observe a candidate's ability before hiring. Several mechanisms help:

Multi-stage interviews screen for different dimensions (technical skill, communication, cultural fit) at each round. Each stage raises the cost of continuing for candidates who know they're a poor match, encouraging self-selection out.
Probationary periods are a form of screening through observation, where the employer gathers information about actual on-the-job performance before committing to a long-term contract. This converts a hidden-information problem into a situation with direct monitoring.
Academic credentials and professional certifications (CPA, CFA) serve as signals of specialized knowledge and commitment to a field.
Employee referral programs use existing employees' private information as a screening device, since workers tend to refer candidates they believe will perform well. The referring employee's reputation is implicitly on the line, which aligns incentives.

Digital Marketplace Mechanisms

Online markets face acute information asymmetry because buyers and sellers rarely interact face-to-face.

User ratings and reviews signal seller reliability. Aggregate scores give a quick summary; detailed reviews provide specifics. These are interesting because the buyers generate the signal rather than the seller, which reduces the seller's ability to manipulate it (though fake reviews remain a problem).
Verified purchase badges increase the credibility of reviews by confirming the reviewer actually bought the product, addressing a second layer of asymmetric information (is the reviewer genuine?).
Platform guarantees (buyer protection, dispute resolution) signal the platform's trustworthiness and reduce the risk of transacting with unknown sellers. The platform can offer these guarantees credibly because it aggregates risk across many transactions.

Welfare Implications of Signaling vs. Screening

Efficiency and Resource Allocation

Signaling and screening can improve efficiency by reducing adverse selection and enabling better matching between agents. But they come with costs.

Signaling costs may be socially wasteful. In Spence's model, if education only signals ability without increasing it, the resources spent on schooling represent a deadweight loss. Workers invest in credentials not to become more productive, but to prove they already are. Compare this to a world with perfect information: the same sorting would occur at zero cost.
Screening can generate economic rents for high-quality agents. If a separating equilibrium allows high-type workers to command a wage premium, income inequality may increase. However, in the Rothschild-Stiglitz insurance model, the high-risk type gets their first-best contract while the low-risk type is distorted (offered less than full insurance). So screening costs often fall on the "good" types, not the "bad" ones.
Multiplicity of equilibria creates its own inefficiency. When multiple separating equilibria exist, the market may settle on one that involves excessive signaling (e.g., requiring a master's degree when a bachelor's would suffice for separation).

The key question is whether the information gains from signaling and screening outweigh the resources spent on them.

Market Structure and Segmentation

Signaling and screening often lead to market segmentation, which can improve welfare by matching consumers to appropriate products (e.g., different insurance tiers for different risk levels). This is welfare-improving when different types genuinely have different preferences or needs.
However, pooling equilibria may persist when the cost structure doesn't support separation. In healthcare markets, for instance, adverse selection can cause insurers to offer only high-premium plans, pricing out low-risk individuals. This is the classic "unraveling" result.
Government intervention sometimes helps:
- Minimum quality standards (food safety regulations) act as mandatory signals
- Subsidies for education or certification can make signaling accessible to more people
- Mandates (like compulsory health insurance) can solve pooling problems by forcing all types into the market, preventing the unraveling that occurs when low-risk types opt out

The welfare effects of signaling and screening are not evenly distributed.

Consumers benefit from better information, but producers bear signaling costs (though they may recoup these through premium pricing). In labor markets, workers bear the signaling cost (tuition), and the benefit accrues partly to employers who can sort more effectively.
Signaling can promote social mobility if the signal is accessible. Education works as an equalizer only when students from all socioeconomic backgrounds can afford to invest in it. When it isn't accessible, signaling reinforces existing stratification.
When signals are expensive, advantaged groups are better positioned to invest, which risks reinforcing existing inequalities rather than correcting them. This is sometimes called the "Matthew effect" in signaling: those who already have resources find it easier to acquire the signal.
There are also trade-offs between information revelation and privacy. Credit scoring, for example, improves lending decisions but raises concerns about data use and fairness. Screening mechanisms that rely on demographic correlates can introduce statistical discrimination even when individual-level information would tell a different story.

Designing markets that balance efficiency, equity, and access to signaling and screening mechanisms remains an ongoing policy challenge. The theoretical tools from this unit give you a framework for analyzing these trade-offs rigorously.