Adverse selection is a process in economics in which information asymmetry causes the participants or goods entering a transaction to be systematically less favorable to the less-informed party. For example, people who privately know that they face high expected losses may be especially willing to buy insurance, while sellers of low-quality goods may be especially willing to accept a price based on average quality. The resulting selection can change prices, restrict exchange, or prevent mutually beneficial transactions. (nber.org)
Economic mechanism
Adverse selection arises when information relevant to a transaction is privately held and influences the decision to participate. The less-informed party cannot fully distinguish different risks or qualities and therefore offers terms that reflect a pooled assessment. Those terms attract some types more strongly than others, changing the composition of the pool. In insurance, for instance, a common premium can be relatively attractive to high-risk customers and relatively unattractive to low-risk customers. (nber.org)
The crucial feature is that price affects not only the quantity traded but also who trades. In a conventional supply-and-demand analysis, demand and production costs can often be treated separately. In a selection market, the characteristics determining demand may also determine the supplier’s costs. Expanding insurance enrollment, for example, changes the average expected claims of the insured population. (pmc.ncbi.nlm.nih.gov)
Adverse selection can therefore produce a market failure even when firms compete. Competition does not itself supply missing information, and an apparently attractive price may become unprofitable once the composition of participants is taken into account. Nevertheless, complete market collapse is only a possible outcome, not an inevitable consequence of asymmetric information. (nber.org)
The lemons problem
The classic product-market example is the market for used cars analyzed in George Akerlof’s 1970 paper, The Market for “Lemons”: Quality Uncertainty and the Market Mechanism. Sellers may know more about a vehicle’s condition than prospective buyers. If buyers cannot distinguish good cars from defective “lemons,” they offer prices reflecting expected quality among cars offered for sale. Such prices can discourage owners of good cars from selling, reducing the quality of the remaining supply. (nobelprize.org)
An illustrative calculation shows the feedback. Suppose buyers value good cars at $12,000 and poor cars at $6,000. If they initially expect equal numbers of each, their expected valuation is $9,000. Assume owners of good cars require at least $10,000, whereas owners of poor cars require $4,000. At $9,000, only poor-car owners sell. Buyers who recognize this selection revise their valuation to $6,000. Poor cars can still trade, but good cars do not—even though buyers value them above their owners’ reservation prices. These hypothetical figures illustrate the mechanism rather than describe an observed market.
The loss is not merely that buyers pay too much for defective products. It includes the disappearance of transactions that would have benefited both parties if quality could have been credibly established. Akerlof’s argument identifies conditions under which such withdrawal occurs; it does not establish that every used-car market must unravel. (nobelprize.org)
Distinction from moral hazard
Adverse selection concerns hidden characteristics that influence participation or contract choice. Moral hazard concerns behavior that changes in response to the incentives created by a contract. In insurance, an already high-risk individual choosing generous coverage illustrates selection; greater coverage changing an individual’s subsequent behavior illustrates moral hazard. Both can generate higher claims among people with more insurance. (nber.org)
The distinction is often described as “before” versus “after” contracting, but the underlying difference is between selection on privately known characteristics and a behavioral response to incentives. Consequently, an observed correlation between coverage and claims does not, by itself, identify which mechanism is responsible. (pmc.ncbi.nlm.nih.gov)
Principal applications
Insurance. When consumers have private information about expected losses, those expecting greater benefits from coverage may disproportionately enroll or choose more generous policies. If premiums subsequently rise to reflect the higher-cost pool, lower-risk participants may withdraw. This feedback is sometimes called a death spiral. Selection can also occur between competing plans rather than simply between insured and uninsured people. (nber.org)
Annuities. An annuity pays income conditional on survival. People expecting to live longer may place greater value on it, increasing the expected duration of payments. Thus, “unfavorable risk” is defined from the provider’s perspective: greater longevity can be unfavorable to an annuity provider even though it is favorable to the purchaser. Research on annuity markets has documented selection associated with longevity. (nber.org)
Credit. A bank raising its interest rate may alter the composition of applicants, not merely increase receipts from existing borrowers. In the Stiglitz–Weiss model, higher rates can disproportionately retain riskier borrowers and reduce the lender’s expected return. A lender may consequently ration credit rather than raise rates until all demand is satisfied. This is a model-dependent explanation of credit rationing, not a claim that every rejected loan reflects adverse selection. (mubsep.mubs.ac.ug)
Financial trading. In a financial market, less-informed liquidity providers may trade against counterparties with superior information about asset values. The risk of such transactions is an adverse-selection cost and contributes to explanations of bid–ask spreads and trading frictions. (nber.org)
Information and contract design
Two important responses are signaling and screening. Signaling occurs when the informed party takes an action that credibly communicates its characteristics. Screening occurs when the less-informed party structures choices so that different types reveal themselves through their decisions. Spence’s work on signaling and Stiglitz’s work on screening became central contributions to the economics of asymmetric information. (nber.org)
In contract theory and mechanism design, revealing private information generally requires incentive compatibility: participants must find it preferable to choose the option intended for their type rather than imitate another type. Intermediaries can design trading mechanisms that elicit private willingness to pay or opportunity costs, but doing so may constrain which exchanges can be implemented. (cambridge.org)
Insurance-market responses also include subsidies, participation requirements, and risk-adjusted payments between plans. Risk adjustment and reinsurance can change insurers’ incentives to attract lower-cost enrollees. Their effects depend on the market’s demand, cost structure, and available contracts; identifying selection does not establish that a particular intervention improves welfare. (nber.org)
Empirical evidence and limitations
A common empirical test asks whether consumers choosing more insurance subsequently have higher claims, conditional on information available to insurers. Results vary across markets and groups of policies. Separating selection from moral hazard requires additional evidence or research designs, because both mechanisms can produce the same coverage–claims relationship. (nber.org)
Private information is also multidimensional. Demand depends on risk preferences as well as expected losses. More cautious or more risk-averse people may both purchase more insurance and incur fewer claims, producing advantageous selection or offsetting adverse selection on another dimension. Therefore, the absence of a positive coverage–claims correlation does not necessarily demonstrate the absence of private information. (aeaweb.org)
The magnitude of selection is distinct from its welfare cost. In a simple competitive insurance model, premiums reflect average cost, while the efficiency of adding another customer depends on marginal cost and that customer’s valuation. Estimating lost gains from exchange therefore requires information about demand as well as costs. Administrative expenses, heterogeneous preferences, and restrictions on the contracts offered can substantially alter the results. (nber.org)
Historical development
Akerlof’s 1970 lemons paper helped establish how asymmetric information can undermine exchange. Subsequent work developed signaling, screening, and applications to insurance and credit. In 2001, Akerlof, A. Michael Spence, and Joseph E. Stiglitz jointly received the Nobel Memorial Prize in Economic Sciences for their analyses of markets with asymmetric information. The award recognized this broader research program, of which adverse selection is a central component. (nobelprize.org)
References
- Writing the “The Market for ‘Lemons'”: A Personal and Interpretive Essaynobelprize.org
- Testing for Adverse Selection in Insurance Marketsnber.org
- Selection in Insurance Markets: Theory and Empirics in Picturesnber.org
- Selection in Insurance Markets: Theory and Empirics in Picturesaeaweb.org
- Selection in Insurance Markets: Theory and Empirics in Picturespmc.ncbi.nlm.nih.gov
- Selection and Asymmetric Information in Insurance Marketsnber.org
- Adverse Selection in Health Insurancenber.org
- Disentangling Moral Hazard and Adverse Selection in Private Health Insurancenber.org
- Credit Rationing in Markets with Imperfect Informationmubsep.mubs.ac.ug
- Market-making with Search and Information Frictionsnber.org
- Adverse selection in product marketscambridge.org
- Joseph E. Stiglitz – Biographicalnobelprize.org