Adverse Selection

Adverse selection is a problem that arises before a contract is signed, when one party knows something relevant about itself that the other party cannot observe. Because the uninformed side must set a single price for everyone, the deal is most attractive to those whose hidden characteristics make them costly to serve. People who expect to make large insurance claims are keenest to buy cover, and borrowers who are least likely to repay are often the most willing to accept a high interest rate.

Adverse selection is one of the two main consequences of asymmetric information, alongside moral hazard, which concerns hidden behaviour after the contract. The idea is central to insurance, banking and labour economics, and George Akerlof's market for lemons is its best-known illustration. Left unchecked, it can shrink a market or destroy it, a form of market failure.

Imagine a gym that offers an all-you-can-eat buffet membership for a single monthly fee. The people who sign up first are the ones with the biggest appetites. If the gym raises the fee to cover them, light eaters decide it is not worth it and leave, and the average appetite of those who remain goes up again.

The gym is not being cheated; it simply cannot tell in advance who will eat the most. Adverse selection is this sorting effect: a single price draws in the customers who are most expensive to serve and pushes away the cheapest ones.

Suppose a health insurer serves 100 people: 50 low-risk people with expected annual medical costs of $2,000 and 50 high-risk people with expected costs of $8,000. If the insurer cannot tell them apart, it sets one break-even premium equal to the average expected cost:

• $P$ is the premium charged to everyone.
• $s_{L}$ and $s_{H}$ are the shares of low-risk and high-risk customers in the pool.
• $C_{L}$ and $C_{H}$ are the expected annual costs of each group.

With equal shares, $P = 0.5 \times 2{,}000 + 0.5 \times 8{,}000 = 5{,}000$ dollars. Many low-risk customers decide that paying $5,000 for $2,000 of expected care is a poor deal and drop out. If they all leave, the pool becomes entirely high-risk and the premium must rise to $8,000. Each increase drives out the healthiest remaining members, a process known as a death spiral. In the end, coverage may be available only to those who need it most, at a price that reflects their high costs, and low-risk people who would have valued insurance at a fair price go without.

• Health insurance. The US Affordable Care Act of 2010 required insurers to accept all applicants and limited pricing by health status, and paired this with an individual mandate to keep healthy people in the pool. The Tax Cuts and Jobs Act of 2017 reduced the federal mandate penalty to zero from 2019, and some states, including Massachusetts and California, operate their own mandates.
• Life insurance and annuities. People who expect to live longer are more likely to buy annuities, and studies of the UK annuity market have found evidence of this selection. Debates over the use of genetic test results in underwriting turn on the same issue; in the US the Genetic Information Nondiscrimination Act of 2008 bars their use in health insurance, and UK insurers follow a voluntary code agreed with the government.
• Credit. Joseph Stiglitz and Andrew Weiss argued in 1981 that raising interest rates can drive away safer borrowers, so banks may limit lending rather than raise prices. Predatory lending concerns arise partly because lenders target segments with weak information.
• Payments and merchant acquiring. Acquirers that price card acceptance too cheaply may attract a disproportionate share of high-risk merchants with heavy chargeback exposure.
• Securities and labour markets. Investors may suspect that firms selling new shares know the stock is overpriced, and employers who cut wages may lose their best staff first.

• Risk classification and underwriting. Insurers and lenders gather information, using medical questionnaires, driving records or a credit score, to price risk more accurately.
• Screening menus. Following Michael Rothschild and Joseph Stiglitz (1976), firms offer choices such as high or low deductibles that lead customers to reveal their risk type.
• Pooling and compulsion. Group insurance through employers, and mandatory schemes such as social insurance or deposit insurance, keep low-risk participants in the pool.
• Waiting periods and exclusions. Pre-existing condition clauses and waiting periods stop people buying cover only when they already know they need it, though many jurisdictions restrict them.
• Signalling. Borrowers who pledge collateral or accept a guarantor show they expect to repay.

Economists debate how severe adverse selection is in practice. Empirical studies in car and health insurance sometimes find weaker effects than theory predicts, partly because people with greater risk aversion both buy more insurance and behave more safely, which can offset selection. There is also a trade-off between efficiency and fairness: finer risk classification reduces adverse selection but can price out people with health conditions or poor credit histories, raising concerns about discrimination and practices such as redlining. How to balance these goals is a policy choice that differs across countries.

• The Nobel Prize: Economic Sciences 2001