Pattern/Economics/No. 0018
Adverse Selection
You turn down another low offer for your well-maintained car because buyers price it like one hiding an expensive fault.
- Evidence
- Well established
- Read
- 6 min
- Links
- 9 connections
01You've seen this when…
- in life
You sell a camera you have carefully maintained. Buyers cannot verify its condition, so their offers assume hidden damage. You decide to keep it instead.
- at work
A maintenance firm offers one annual fee for every machine. Owners know their repair histories; the firm does not. Owners of unreliable machines rush to sign, while owners of well-kept machines decline.
- out in the world
A voluntary health plan charges everyone in an age band the same premium. People expecting expensive treatment enroll eagerly. Some healthier people decide coverage costs too much and stay out.
02The idea
Low offers disappoint the owner of a good camera and give that owner a reason to keep it. Someone with a damaged camera has more reason to accept. Buyers’ uncertainty can therefore change what is available to buy.
Adverse selection is a sorting problem caused by hidden information before an exchange. One side knows something relevant about quality, risk, or likely cost that the other side cannot reliably observe. The offered terms appeal differently to different types, and the people or products that participate become a less favorable mix for the less-informed side.
This is a particular consequence of asymmetric information, not a synonym for it. Adverse selection requires unequal knowledge and the offered terms to combine in changing who participates.
In insurance, customers may know more about their risks. In used goods, sellers may know more about quality. Nobody needs to lie. Honest people can simply decline a deal that undervalues what they offer, leaving the market worse for everyone else.
03Why it happens
- Different participants privately know different things. A machine owner knows it has broken down four times this year. A maintenance provider sees only its model and age. Those differences matter to the cost of the contract.
- One offer covers several hidden types. Unable to distinguish them, the provider sets a fee based on an expected average. Reliable machines look expensive to insure; unreliable machines look cheap.
- People select themselves into the deal. Owners expecting large repair bills accept more often. The actual customer pool costs more than the provider expected. Selection follows expected repair costs and can occur even if nobody deliberately tries to exploit anyone.
- Repricing can worsen the mix. Raising the fee covers today’s costly pool but may push out more owners of reliable machines. Another increase follows. In severe cases, this feedback shrinks the market until useful exchanges disappear.
That last spiral is possible, not inevitable. Better information, different contracts, subsidies, or strong reasons to participate can interrupt it.
04A worked example
In the mid-1990s, Harvard University changed its employee health-insurance contributions so that employees faced more of the price differences between plans. Economists David Cutler and Sarah Reber studied what happened. The more generous plan became less attractive to relatively healthy employees, who could save money by choosing cheaper coverage.
What it looks like Competition encourages employees to shop around and choose plans that fit their needs, while expensive coverage loses customers.
What’s actually going on Lower-cost enrollees disproportionately leave the more generous plan. Covering the remaining membership costs more, and the resulting pressure on its premium encourages additional departures. The study documented substantial adverse selection following the contribution change.
The distinction matters: a plan’s average spending can rise because its membership changes even when it delivers care as efficiently as before.
What would have helped Comparing plans after accounting for their members’ health risks, which can explain premium differences that might otherwise be attributed to efficiency. Risk-adjusted payments between plans could also reduce the financial penalty for covering sicker people. These approaches address the mechanism. Whether any particular payment formula would have solved Harvard’s problem remains uncertain.
05How to spot it
06What to do about it
- Make relevant information verifiable. Independent inspections, service records, and credible quality tests can separate goods that buyers would otherwise lump together. A warranty can act as a signal when defects would be costly for the seller to cover. A promise without enforcement does little.
- Offer choices that reveal useful differences. This is screening: the less-informed side designs options and watches which people choose. Insurance menus with different premiums and deductibles are one example. Check incentive compatibility: each type must actually prefer the option intended for it. Preferences and ability to pay can complicate the result.
- Adjust comparisons for the participant mix. Before judging a plan, supplier, or service by average cost, examine whom it serves. In insurance, risk-adjusted payments can make covering higher-risk people less financially punishing, though imperfect adjustment leaves room for selection.
- Protect the pool when broad participation is the goal. Subsidies, broad enrollment, or coverage requirements can keep lower-risk people participating. These are market design choices with costs and trade-offs. They need attention to affordability, privacy, and access, not just the provider’s balance sheet.
- Track departures as carefully as arrivals. Record which types leave after a price or contract change. A healthy-looking revenue figure can hide the loss of customers who made the arrangement sustainable.
07When it isn’t adverse selection
Moral hazard concerns how incentives change behavior: coverage might make someone less careful. Adverse selection concerns which people buy coverage in the first place. Both can operate together, but they call for different remedies.
Selection bias is broader. It describes a sample that gives a distorted picture of a population. Adverse selection describes an economic mechanism that changes participation because of private information and offered terms.
Buying more insurance can reflect caution about risk. Especially cautious people may both buy generous coverage and avoid accidents. Risk aversion, income, and other differences can offset the selection pattern. Empirical findings vary across insurance markets.
Risk pooling can support a society’s deliberate choice to subsidize people with higher medical needs. Differences in members’ costs are part of that arrangement. The problem is an unstable participation pattern that undermines that goal.
08Roots
In his 1970 paper, George Akerlof asked readers to picture a used-car market. Owners knew whether their cars were dependable or troublesome; buyers could not confidently tell them apart. A price reflecting average quality could drive dependable cars out, lowering the average quality further. The vivid detail was the lemon: a car whose problems become apparent only after someone else owns it.
The problem and the term were older than Akerlof’s example. They came from insurance, where applicants could know more about their health or other risks than the company did. Insurers worried about selection against them: the people most eager to buy could be those most likely to claim. There is no single clean moment when the pattern was first noticed.
Akerlof’s contribution was to show how this familiar insurance problem could undermine markets more generally. Michael Rothschild and Joseph Stiglitz later examined how competing insurers might use different contracts to separate customers by risk. Together, these arguments helped turn hidden information from a practical nuisance into a central subject of economics.
09How solid is this?
The mechanism is firmly established in economic theory and documented empirically, including in health insurance. Its strength varies across markets; hidden information does not automatically produce a damaging spiral, and other customer differences can offset it.
10Connections
- Often confused with Selection Bias, Moral Hazard
- Countered byIncentive Compatibility, Market Design, Signaling, Screening
- Part of Asymmetric Information
- See alsoRisk Aversion, Principal-Agent Problem
11Origin and sources
The term predates modern information economics and comes from insurance. George Akerlof’s 1970 paper, The Market for “Lemons”: Quality Uncertainty and the Market Mechanism, supplied the influential general account.
- [1]Akerlof, G. A. (1970). The Market for "Lemons": Quality Uncertainty and the Market Mechanism. The Quarterly Journal of Economics, 84(3), 488–500.
- [2]Rothschild, M., & Stiglitz, J. (1976). Equilibrium in Competitive Insurance Markets: An Essay on the Economics of Imperfect Information. The Quarterly Journal of Economics, 90(4), 629–649.
- [3]Cutler, D. M., & Reber, S. J. (1998). Paying for Health Insurance: The Trade-Off between Competition and Adverse Selection. The Quarterly Journal of Economics, 113(2), 433–466.
- [4]Cohen, A., & Siegelman, P. (2010). Testing for Adverse Selection in Insurance Markets. The Journal of Risk and Insurance, 77(1), 39–84.
Suggest an edit· Updated 2026-10-02