Why markets sometimes fail even without externalities or public goods — hidden information, predictable bias, and the design of better choices
Markets work well when buyers and sellers have accurate, complete information. When they do, prices reflect true value, products match their descriptions, and both parties benefit from exchange. But real markets are often clouded by unequal information, hidden characteristics, hidden actions, and the predictable limits of human reasoning.
This chapter introduces two sets of ideas that extend and deepen market analysis. The first set concerns information problems: situations where one party to an exchange knows something important that the other does not. The second set concerns behavioral economics: the study of how real people actually make decisions, including the systematic and predictable ways those decisions depart from the idealized model of perfectly rational, fully informed choice.
Together, these ideas help explain puzzles that standard supply-and-demand analysis cannot fully resolve: Why do used cars sell for much less than new cars of identical quality? Why do health insurers worry that the people most eager to buy coverage are the ones most expensive to cover? Why do employees save more for retirement when their employer chooses enrollment as the default? Why do people hold onto losing investments far longer than they should?
Understanding information problems and behavioral economics does not require abandoning economic reasoning. It requires extending that reasoning to account for the world as it actually is. The goal is not to replace markets with something else, but to understand when and why markets produce disappointing outcomes — and how better institutions, policies, and design choices can improve them.
Think of a time you bought something used, signed up for insurance, or accepted a "default" setting on an app or account without thinking about it. Did you know everything the seller knew? Did you act exactly as a perfectly rational calculator would? By the end of this chapter, you will have the vocabulary to analyze both situations with precision.
Economic analysis usually begins with the assumption that buyers and sellers are well-informed. A consumer knows what she is buying. A firm knows what it is selling. Prices communicate scarcity and value. Voluntary exchange leaves both parties better off.
This informational assumption does real work. It is part of why markets often function well. When information is reasonably complete and accessible, competition drives prices toward value, firms have incentives to produce quality products, and consumers can make sensible comparisons.
But information is itself scarce. Gathering it takes time and effort. Some information is held privately by one party and cannot easily be verified by the other. Some information is genuinely uncertain. And even when information is available, people do not always process it correctly.
Symmetric information exists when both parties to a transaction have access to roughly the same relevant information. Two farmers trading grain for tools, both knowing the current market price, is an example.
Asymmetric information exists when one party has significantly better information than the other. A seller often knows more about the quality of what she is selling than the buyer does. A borrower often knows more about her own creditworthiness than the lender does. A patient often knows more about his own health behaviors than his insurer does.
Asymmetric information is extremely common. It arises in used-car markets, insurance markets, labor markets, credit markets, healthcare markets, and countless other settings. When it is severe, it can prevent mutually beneficial trades, drive out quality products, or lead to outcomes that neither party intended.
Economist George Akerlof illustrated asymmetric information with a famous example involving used cars. In his framework, imagine a used-car market with two types of cars: high-quality cars (peaches) and low-quality cars (lemons). Sellers know which type they have. Buyers cannot tell the difference before purchase.
If buyers cannot distinguish a peach from a lemon, they will offer a price based on their estimate of the average quality in the market. This average price is too low for sellers of high-quality cars — who know their car is worth more — so those sellers may withdraw from the market. As high-quality sellers exit, average quality in the market falls. Buyers update their estimate downward and offer even less. More high-quality sellers exit.
In the extreme case, the market can unravel entirely, leaving only lemons. This is called adverse selection: when asymmetric information causes the less desirable type to be disproportionately represented in a market. Mutually beneficial trades — the sale of good used cars at prices buyers would willingly pay if they could verify quality — fail to occur.
This is a market failure caused not by externalities or public goods, but by information. It explains why used cars sell at a substantial discount relative to identical new cars, why warranties and reputation matter so much in markets with quality uncertainty, and why institutions that reduce information asymmetry — inspections, certifications, reviews, guarantees — create real economic value.
Asymmetric information problems take two broad forms, depending on whether the hidden information concerns a characteristic that exists before a transaction or an action that occurs after one.
Hidden characteristics are qualities or attributes that one party knows but the other cannot easily observe before agreeing to a transaction. A job applicant knows her own abilities better than her employer does. A car seller knows whether his car has been in accidents. A borrower knows whether she is likely to repay her loan.
Hidden actions are behaviors that one party takes after a transaction is completed, which the other party cannot easily monitor. An employee may work less diligently once she is hired than she promised during the interview. A homeowner with flood insurance may take fewer precautions against flooding than he would without it.
Both forms of asymmetric information create problems for markets and institutions. The problems have different names, different mechanisms, and different solutions.
| Source of Asymmetry | Why It Arises |
|---|---|
| Quality difficult to observe before purchase | Experience goods and credence goods (e.g., used cars, legal services) |
| Self-knowledge outsiders cannot verify | Health status, creditworthiness, work ethic |
| Costly or impossible monitoring | Employee effort, insured behavior, borrower use of loan funds |
| Specialized expertise | Doctors, lawyers, mechanics, financial advisors |
| Future behavior tied to unrealized circumstances | Whether a borrower will lose her job; whether a driver will take more risks |
Understanding that information asymmetry is normal, not exceptional, is important. It helps explain why so many economic institutions — professional licensing, warranties, credit scores, collateral requirements, performance reviews, insurance co-payments, referral systems — exist precisely to manage information problems.
For each: classify as a hidden characteristic or a hidden action, and explain your reasoning.
Adverse selection occurs when asymmetric information about hidden characteristics causes a market to attract disproportionately the participants who are least desirable from the other side's perspective.
The mechanism works through price. When prices must reflect average quality across a pool of participants — because the distinguishing characteristics cannot be directly observed — the price is too high for low-risk or high-quality participants and too low for high-risk or low-quality participants. The result is that low-risk and high-quality participants exit the market or choose not to participate, leaving behind a pool that is worse than average.
Health insurance provides a clear illustration. Suppose an insurer offers a single premium to a diverse population. People who know they are healthy and low-risk may find the premium too expensive relative to their expected benefit and decline to buy. People who know they are sick or high-risk find the premium attractive and eagerly enroll.
The insurer's pool of enrollees ends up sicker and more expensive than the general population. The insurer must raise premiums to cover costs. This drives away more of the healthiest remaining enrollees. Premiums rise again. In severe cases, the market for insurance can collapse or become accessible only to those with high expected costs — exactly the opposite of what insurance is supposed to accomplish.
This dynamic helps explain why health insurance markets often require government intervention, mandates, community rating rules, or risk-pooling mechanisms. These are responses to a real market failure rooted in information asymmetry, not arbitrary interference with voluntary exchange.
Lenders cannot directly observe how likely each borrower is to repay. If they charge a single interest rate, it may be too high for the safest borrowers, who will seek credit elsewhere or forgo borrowing. The remaining applicants will be disproportionately risky. Credit markets develop institutions — credit scores, collateral requirements, co-signers, loan covenants — specifically to reduce adverse selection.
Employers face a similar problem in hiring. Job applicants know their own abilities, work habits, and outside options better than employers do. If employers offer a single wage, highly productive workers who can command higher wages elsewhere will seek them, while the remaining applicants will be disproportionately those who cannot. Employers respond with screening, credentials, trials, and probationary periods.
Moral hazard occurs when a party to a transaction changes its behavior after the transaction in ways that are harmful to the other party — changes that occur because the first party no longer bears the full consequences of its actions.
The term "moral hazard" originated in the insurance industry, where it described the observation that insured parties sometimes take more risks than uninsured parties because they do not bear the full cost of loss. But moral hazard applies broadly wherever the consequences of an action fall partly on someone other than the person taking the action.
Moral hazard arises after a transaction — after a contract is signed, after insurance is purchased, after a loan is made. This distinguishes it from adverse selection, which arises from hidden characteristics that exist before the transaction.
A homeowner with fire insurance may be less careful about leaving candles unattended than she would be without insurance. A driver with comprehensive auto coverage may park in riskier locations than he would without it. In each case, insurance reduces the cost of a bad outcome to the insured, which reduces the incentive to avoid the outcome.
| Tool | How It Manages Moral Hazard |
|---|---|
| Deductibles | The insured pays the first portion of any loss, preserving some incentive to avoid loss |
| Co-payments and co-insurance | The insured pays a share of each claim |
| Policy limits | Coverage stops at a maximum amount |
| Exclusions | Excessive risk-taking behaviors are excluded from coverage |
| Experience rating | Premiums rise after claims, linking future cost to past behavior |
Employment contracts create moral hazard because employers cannot continuously monitor every action of every employee. Once hired, an employee may exert less effort than she promised. Employers manage this through performance reviews, output-based pay, piece rates, profit sharing, and monitoring technologies.
A bank that knows it will be bailed out if it fails has less incentive to avoid excessive risk — moral hazard at the institutional level. Corporate executives whose compensation is heavily weighted toward stock options may take on greater risk than shareholders would prefer, because the upside accrues to the executive while the downside falls on shareholders and creditors.
Adverse selection concerns hidden characteristics before a transaction (who you are); moral hazard concerns hidden actions after a transaction (what you do). Both arise from asymmetric information, but they occur at different points in the relationship and require different responses.
When one party has better information about quality than the other, and quality cannot be directly observed, markets face the adverse selection problem described above. But markets also develop responses to this problem: signaling (by the informed party) and screening (by the uninformed party).
Signaling occurs when the informed party takes an observable action specifically to credibly communicate hidden quality to the uninformed party. For a signal to work, it must be costly to fake. If both high-quality and low-quality types could costlessly produce the same signal, the signal would convey no information. Only signals that are cheap for high-quality types and expensive for low-quality types will successfully separate the two.
Economist Michael Spence analyzed education as a potential signaling device in labor markets. Even if education does not increase worker productivity directly, highly productive workers might still invest in education credentials because doing so signals their type to employers, provided acquiring the credential is less costly for high-ability workers than for low-ability workers. This does not mean education is purely signaling — most economists believe education increases human capital as well — but the signaling dimension helps explain some observed patterns in educational investment and returns.
| Signal | Why It Is Credible |
|---|---|
| Warranties | Expensive for a firm selling low-quality products, since it would face frequent costly claims |
| Job history and references | Verifiable track records are harder to fabricate convincingly |
| Collateral | Defaulting would cost the borrower the pledged asset |
| Advertising expenditure | Only rational if the product is good enough to generate repeat customers |
Screening occurs when the uninformed party designs mechanisms to induce the informed party to reveal hidden characteristics. Insurance companies screen by offering a menu of policies with different premium-deductible combinations. Risk-averse, high-risk individuals prefer comprehensive coverage at higher premiums; lower-risk individuals prefer high-deductible, lower-premium plans. By offering a range of options, the insurer induces customers to self-select into appropriate categories, partially revealing their private information through their choices. Employers screen through interviews, probationary periods, skills tests, and reference checks.
Signaling and screening reduce but do not eliminate information problems. Signals can be faked if the cost advantage of doing so is large enough. Screening mechanisms add administrative cost and may themselves create adverse selection within subgroups. Markets with severe information problems may remain inefficient even after extensive signaling and screening.
Standard economic models assume that people have well-defined, consistent preferences; know all relevant options and their consequences; correctly calculate which option maximizes their well-being; and act on that calculation without delay, bias, or error.
This model of the "rational economic actor" is a powerful simplification. It generates many useful predictions about how people respond to price changes, income changes, incentives, and constraints. Much of Parts II and III of this textbook rely on it.
But the model is a simplification. Real people have limited time and attention, imperfect memories, difficulty processing complex probabilities, emotional responses, social pressures, and cognitive shortcuts that serve them well in many situations but systematically mislead them in others.
Behavioral economics studies how real human decision-making departs from the idealized rational actor model, and what those departures imply for markets, institutions, and policy.
Economist Herbert Simon introduced the concept of bounded rationality: the idea that human rationality is real but limited. Instead of finding the mathematically optimal solution to every decision, people use heuristics — mental shortcuts or rules of thumb — that usually work reasonably well but can produce systematic errors. Bounded rationality does not mean people are irrational; it means people are rationally economizing on the costly resource of deliberation.
| Heuristic | Effect |
|---|---|
| Availability | People judge probability by how easily an example comes to mind — dramatic or recent events seem more likely than they statistically are |
| Representativeness | People judge probability by resemblance to a typical example, sometimes ignoring actual base rates |
| Anchoring | People rely too heavily on the first piece of information received when making estimates or judgments |
| Overconfidence | People systematically overestimate the accuracy of their own judgments and predictions |
Framing refers to the way options are described or presented. Rational choice theory predicts that logically equivalent descriptions of the same choice should lead to the same decision. Behavioral research shows that they often do not.
| Framing Type | Example |
|---|---|
| Gain vs. loss framing | "90% survival rate" is evaluated more favorably than the identical "10% mortality rate" |
| Price framing | "$10 off" vs. "10% off" (on a $100 item) — identical discounts, processed differently |
| Label framing | "75% lean" beef rated higher quality than equivalent "25% fat" beef |
Framing effects are not limited to simple consumer choices. They influence medical decisions, legal judgments, investment choices, and policy preferences. People's choices are partly constructed in the moment, shaped by how options are presented — not just by underlying, stable preferences.
Defaults are the options that occur automatically unless a person actively chooses otherwise. Standard rational-choice theory predicts that defaults should matter little: a person with a clear preference will simply opt out of a default she doesn't want. Behavioral research shows that defaults matter enormously in practice, through three mechanisms: inertia (changing requires an active decision, and decisions are often postponed indefinitely), implicit endorsement (people interpret the default as a recommendation), and loss aversion (departing from the default may feel like giving something up).
| Plan Type | Enrollment After 6 Months | Enrollment After 3 Years |
|---|---|---|
| Opt-in (active enrollment required) | 37% | 59% |
| Opt-out (automatic enrollment, must opt out) | 86% | 91% |
Illustrative values representative of findings in the behavioral economics literature on automatic enrollment. The gap is largest immediately after the enrollment opportunity and narrows somewhat over time, but it never disappears. The effect is consistently larger for lower-income workers, who had the lowest participation rates under opt-in systems.
Defaults are also powerful in organ donation (opt-out/presumed-consent systems show substantially higher donor rates than opt-in systems), energy plans (defaulting customers to green energy increases adoption significantly), and medication adherence (auto-refill subscriptions dramatically increase adherence to chronic medication regimens).
The power of defaults does not mean people are foolish. It means that the architecture of choices — how options are structured and presented — systematically shapes behavior even among thoughtful people.
Psychologists Daniel Kahneman and Amos Tversky demonstrated through extensive research that people do not treat gains and losses symmetrically. In standard economic models, losing $100 should reduce well-being by the same amount that gaining $100 increases it. In reality, losses loom roughly twice as large as equivalent gains in people's psychological experience. This phenomenon is called loss aversion.
A stylized value function plots utility against gains and losses relative to a reference point of zero. The function is concave in the gains domain and convex in the losses domain, and it is steeper in the loss domain than the gain domain — roughly twice as steep. This illustrates that a loss of a given size produces a larger reduction in well-being than a gain of the same size produces in increased well-being. The reference point (current endowment or status quo) determines what counts as a gain versus a loss.
Loss aversion helps explain several observed patterns. The endowment effect: once people own something, they value it more than they would if they did not own it. Status quo bias: people prefer their current situation over change, even when change would benefit them. Investment behavior: investors hold losing stocks too long and sell winning stocks too soon, because realizing a loss feels worse than maintaining an unrealized one. Wage rigidity: workers resist wage cuts more fiercely than they welcome equivalent wage increases, contributing to wage rigidity during downturns, which in turn affects unemployment.
Standard economic models assume people discount the future at a consistent rate. Behavioral research finds that people discount the near future far more steeply than the distant future — a pattern called present bias or hyperbolic discounting. A person with present bias might strongly prefer $100 today over $110 next week, yet feel near-indifferent between $100 in 52 weeks and $110 in 53 weeks — the same one-week delay, evaluated very differently depending on when it occurs.
Present bias helps explain self-control problems, undersaving, procrastination, and the appeal of addictive behavior: tasks or temptations with immediate costs or benefits are weighted far more heavily than logically equivalent costs or benefits in the future.
Changing a behavior — starting to exercise, stopping smoking, beginning to save — requires bearing immediate costs and forgoing immediate pleasures, while the benefits arrive later. Loss aversion makes the costs feel especially large. Present bias makes the benefits feel especially distant. Together, they help explain why self-control is genuinely difficult even for motivated, intelligent people who understand their own situation clearly.
Economists Richard Thaler and Cass Sunstein developed the concept of a nudge: an intervention that changes the architecture of choices to predictably alter behavior without restricting options or significantly changing financial incentives.
A nudge does not ban anything. It does not impose a tax or a subsidy. It does not mandate a choice. It exploits the behavioral patterns described in the previous sections — the power of defaults, framing effects, loss aversion, present bias — to steer people toward better outcomes while preserving freedom of choice. A well-designed nudge should be easy to opt out of. If it is difficult to opt out, it is closer to a mandate than a nudge.
| Nudge | Mechanism Exploited |
|---|---|
| Automatic enrollment in retirement savings | Default power, inertia, present bias |
| Simplified enrollment forms with pre-filled data | Reduces effort cost of an active decision |
| Calorie labeling at restaurants | Salience; effects on behavior are mixed |
| Energy usage comparison reports | Social norms and desire to conform |
| Eye-level placement of healthy foods | Convenience and default-like ease of choice |
| Commitment devices (locked savings accounts) | Pre-commitment to bind a future, more patient self |
A standard objection to nudges is that they are paternalistic: they assume the designer knows better than the individual what is good for her. The response associated with libertarian paternalism is that choice architects must set defaults and framing regardless — there is no neutral option. A cafeteria must arrange its food somehow; a retirement plan must have a default contribution rate. Given that choice architecture is unavoidable, the argument goes, it is better to design it thoughtfully, in ways that tend to serve the interests of the people making choices, while preserving their right to choose differently. This debate remains active in economics, philosophy, law, and public policy.
Behavioral economics began as an academic research program documenting departures from rational choice. Over the past two decades, it has moved into government policy, public health, environmental regulation, financial regulation, retirement plan design, tax administration, and consumer protection — as well as corporate marketing, product design, pricing strategy, and human resources management.
| Domain | Application |
|---|---|
| Retirement savings | Automatic enrollment and automatic escalation in employer retirement plans, exploiting default power and present bias |
| Tax compliance | Letters citing neighbors' compliance rates (social norms) or framing non-compliance as lost local services |
| Energy efficiency | Default enrollment in green energy programs and smart thermostat defaults |
| Health behaviors | Nudges for vaccination, medication adherence, and organ donor registration; effectiveness varies by context |
| Financial regulation | Clear disclosure formats, cooling-off periods, and restrictions on practices exploiting present bias or loss aversion |
Private firms were applying behavioral insights long before governments did — though often to influence consumers in ways that serve firm profits rather than consumer well-being. Subscription pricing with automatic renewal exploits inertia. "Original price" and "sale price" labeling exploits anchoring. Free trials exploit loss aversion, since consumers dislike giving up something they already have. Decoy pricing — placing a slightly inferior product at nearly the same price as a superior one — makes the superior option seem like a better deal. Interface design choices on checkout pages, and loss-framed bonus structures for employees, draw on the same behavioral mechanisms.
When firms exploit behavioral biases to increase profit, should this be understood as ordinary competitive strategy, or as a market failure that consumer protection policy should address? This question parallels the policy debate over paternalism and deserves serious consideration from both sides.
Behavioral economics is influential but not without critics. External validity: many behavioral findings come from laboratory experiments, and it is not always clear how reliably they translate to real markets with experienced participants and competitive pressure. Heterogeneity: behavioral biases vary substantially across individuals, cultures, and contexts, so nudges designed for an average response may poorly fit many specific individuals. Paternalism and autonomy: even well-intentioned nudges can reflect designers' values rather than the values of the people being nudged. The standard model's resilience: in many competitive markets with repeated interaction and strong stakes, departures from rational choice are smaller and less consequential than in laboratory settings.
The most balanced view is that behavioral economics extends economic analysis without replacing it. Standard economic reasoning — incentives, tradeoffs, supply and demand, marginal thinking — remains indispensable. Behavioral insights add realism about how people actually make decisions, which improves the analysis of institutions, policy, market design, and firm strategy.
A used-car market with two demand curves: the upper curve represents buyer willingness to pay if all cars are high-quality; the lower curve represents willingness to pay given the actual mix of quality. Supply is upward-sloping and concentrated among high-quality sellers at higher prices. As the price buyers are willing to pay falls, high-quality sellers exit, reducing average quality, which further reduces willingness to pay — a downward spiral. The surviving equilibrium may involve only low-quality goods at a low price, while beneficial trades in high-quality goods fail to occur.
Utility plotted against gains and losses relative to a reference point of zero. The function is concave for gains, convex for losses, and roughly twice as steep on the loss side. The reference point determines what counts as a gain versus a loss.
| Information Problem | Type | Example | Common Response |
|---|---|---|---|
| Adverse selection | Hidden characteristic | High-risk individuals dominate insurance pool | Risk pooling, mandates, screening |
| Moral hazard | Hidden action | Insured drives less carefully | Deductibles, co-payments, monitoring |
| Adverse selection | Hidden characteristic | Low-quality sellers crowd out high-quality sellers | Warranties, certification, reputation |
| Moral hazard | Hidden action | Borrower takes excessive risk with loan proceeds | Collateral, loan covenants, monitoring |
| Adverse selection | Hidden characteristic | Low-ability workers seek employment at any wage | Screening, credentials, probationary periods |
| Moral hazard | Hidden action | Employee shirks after being hired | Performance pay, monitoring, reviews |
Chapter 17 examined how hidden information and predictable departures from rationality can cause markets to underperform their idealized models. Chapter 18 builds on this — and on the externalities and public goods studied in Chapter 16 — by introducing the tools government uses to intervene in markets: taxes, subsidies, regulation, public spending, and redistribution. Students will analyze tax incidence, the difference between efficiency and equity, and the intended and unintended consequences of public policy, completing the picture of market failure and policy response that spans Part IV.