What it means
Adverse selection is a pricing problem created by asymmetric information, meaning one party holds facts the other cannot see or verify. When a seller sets a single price for a mixed group, the customers who know they are cheaper than average find that price poor value and drop out.
The ones who stay are disproportionately expensive, so the average cost of the remaining group rises. This matters because it can quietly wreck the economics of an otherwise sensible product.
Raising the price to cover the higher average cost pushes out the next tier of good risks, which raises the average again, a spiral that insurers call a death spiral. Businesses manage adverse selection by gathering better information, or by changing the terms so that good risks have a reason to stay.
Underwriting questions, health checks, credit scores, excess levels, waiting periods and group enrolment through an employer all serve this purpose. Each one either reveals hidden risk or makes the product deliberately less attractive to the people you least want.
The idea reaches well beyond insurance. Second-hand vehicle markets, freelance hiring, small business lending and flat-rate subscription pricing all face the same pattern, where the seller cannot easily tell a good counterparty from a bad one.
Warranties, trial periods, references and staged commitments are the everyday answers. It is worth separating adverse selection from moral hazard, which is a related but different problem.
Adverse selection is about hidden information before the deal is struck; moral hazard is about hidden behaviour after it, such as an insured driver taking more risk because losses are covered.
In practice
Real-world examples
Example 1
A travel insurer offers a flat annual policy with no medical questions. People with existing conditions and frequent long-haul trips sign up in large numbers, while healthy occasional travellers buy cheaper single-trip cover elsewhere. Claims run at twice the level assumed in pricing.
Example 2
A software company launches an unlimited support plan at $500 per month. Customers with tiny support needs stay on the pay-as-you-go option, so the plan fills up with the heaviest users and the support team is overwhelmed within a quarter.
Example 3
A dealer in used construction equipment prices every five-year-old excavator at the same figure. Owners of well-maintained machines sell privately for more, so the dealer's stock fills with the neglected ones and refurbishment costs run far above budget.
Think of it
Imagine setting up an all-you-can-eat buffet for a flat five pounds. You will not attract average diners; you will attract professional competitive eaters who will bankrupt your kitchen.
Formula
Calculation
There is no single standard formula, but the mechanism is easy to show with a pooled cost calculation:
Average cost per member = total expected claims / number of members
A health plan starts with 1,000 members: 800 low-risk members expected to cost $1,000 each per year, and 200 high-risk members expected to cost $6,000 each.
Total expected claims = (800 x $1,000) + (200 x $6,000) = $800,000 + $1,200,000 = $2,000,000
Average cost per member = $2,000,000 / 1,000 = $2,000
Premium set at cost plus 10% = $2,000 x 1.10 = $2,200
At $2,200 the low-risk members are paying more than double their own expected cost, so 600 of the 800 cancel. The pool is now 200 low-risk and 200 high-risk members.
Total expected claims = (200 x $1,000) + (200 x $6,000) = $200,000 + $1,200,000 = $1,400,000
Average cost per member = $1,400,000 / 400 = $3,500
Loss per member = $3,500 - $2,200 = $1,300
Total annual loss = $1,300 x 400 = $520,000
The plan priced correctly for the group it started with and still lost $520,000, because the mix of members changed in response to the price.Case study
Seen in the real world
The following is an illustrative, fictional example. Cedar Ridge Dental Cover, an invented regional plan, launched a no-questions-asked policy at $30 a month covering two check-ups and up to $1,500 of treatment a year. The pricing team had modelled the average adult, who uses roughly $260 of dental care annually.
Within eight months, average claims per member reached $470. Analysis showed that a large share of new members had joined within days of being told they needed treatment, then cancelled once the work was done. Members who simply wanted routine check-ups had mostly stayed with their employer scheme.
Cedar Ridge did not raise the headline price, which it feared would push out its remaining healthy members. Instead it introduced a 90-day waiting period for major treatment, a simple two-question application, and a discount for members who joined through an employer group. Average claims fell to $310 per member over the following year, and the plan returned to a small surplus.
Watch out
Common mistakes
- Treating adverse selection as the same thing as moral hazard. The first is about hidden information before the contract, the second is about changed behaviour after it, and they need different remedies.
- Responding to poor loss experience by simply raising the price. Without a change to screening or terms, a price rise drives out the best customers first and makes the average risk worse.
- Assuming it only applies to insurance. Any flat-rate offer to a mixed group, including subscriptions, warranties and unlimited service plans, is exposed to the same effect.
Questions
People also ask
Does adverse selection always mean someone is behaving dishonestly?
No, it usually just means customers are acting rationally on information they hold about themselves, which is why the fix is better design rather than blame.
How do group schemes reduce it?
Enrolling everyone at a workplace or in an association brings in good and bad risks together, so the pool reflects the population rather than only those who most want cover.
Can data analytics eliminate the problem?
Better data narrows the information gap and improves pricing, but it rarely closes it completely because customers still know things about their own plans and habits that no dataset captures.