Adverse selection
When the hidden-information party self-selects, the worst risks crowd in.
What it means
A market failure arising before a transaction when one side holds private information about quality or risk, so that the terms offered attract a disproportionately bad pool. Akerlof's 'market for lemons' showed that if buyers cannot distinguish good cars from bad, prices fall to reflect the average, driving good cars out until only lemons remain. In insurance, a flat premium lures the highest-risk customers and repels the low-risk, threatening a death spiral. It matters because it explains why markets unravel under asymmetric information and motivates the remedies of signaling, screening, warranties, and mandates.
How the unravelling works
The damage comes from iteration, not from a single mispricing. A seller prices at the average of the pool it faces. That price is a bargain to whoever holds the worst risk and poor value to whoever holds the best, so the best withdraw. The pool left behind is worse, so the next honest repricing is higher and pushes out the next tier. Every round is individually rational and collectively destructive. Cutler and Reber traced this in one natural experiment at Harvard, where a change in how staff paid for plans drew the healthy out of the most generous option until it closed, at a welfare cost of 2 to 4 percent of baseline spending. The unravelling is not even the worst case: in Rothschild and Stiglitz's 1976 model of competitive insurance with hidden types, a stable equilibrium need not exist at all.
What the evidence shows
The mechanism is easy to draw, surprisingly hard to find. The standard test asks whether, among people offered the same terms, those who buy more coverage go on to claim more. Moral hazard produces the same correlation, so a positive result does not identify selection. Chiappori and Salanie ran it on French auto insurance and found no correlation for young drivers. Bond's 1982 test of the lemons model in a real used-vehicle market, pickup trucks, found traded trucks needed no more repair than equivalent trucks bought new. Where selection appears it is often narrow: Finkelstein and Poterba found UK annuity buyers sorting on contract features, not on the size of the annuity. Einav and Finkelstein's verdict: adverse selection exists in some insurance markets and not in others.
When selection runs the other way
Willingness to pay for cover does not track risk alone. It also tracks risk aversion, and the two can point in opposite directions. A cautious person buys more insurance and also drives more slowly, sees a doctor earlier, takes fewer chances. When that correlation dominates, the people who crowd in are better risks than those who stay out, and the market shows advantageous selection: a price set at the average is too high rather than too low, and the pool improves as it grows. Einav and Finkelstein point to long-term care insurance, where Finkelstein and McGarry found the more cautious both buy more cover and enter nursing homes less, leaving the raw correlation between coverage and use at roughly nothing. The remedies invert: a mandate that rescues an adversely selected market drags a favourably selected one toward the average.
Using it in practice
The diagnostic is two questions. Does the other side know something that predicts your cost, and does that knowledge drive whether they accept your terms? Both must hold, and in many markets the second quietly fails. The fixes then break one link or the other. Screening offers a menu priced so the types sort themselves: a higher deductible, a longer lock-in, a smaller discount. Signaling lets the good side pay to prove it: warranties, inspection reports, probation periods. Mandates and automatic enrolment remove the choice that does the selecting, which is why group schemes survive where individual purchase does not. Hendren is the warning: insurers reject applicants exactly where private information is strongest, so the market you cannot find may already have unravelled.
Examples
Health-insurance plans priced for average risk attract mainly the sick, forcing premiums up and chasing the healthy away.
A lender raises rates to cover its defaults. The careful borrowers go elsewhere and the desperate ones stay, so the higher price selects for exactly the risk it was meant to price.
A firm offers the same voluntary redundancy terms to everyone. The staff who could walk into another job take the money; the ones with nowhere to go stay.
An unlimited data plan priced on average usage is a bargain for the heaviest streamers and poor value for light users. The light users leave, and the average the price was built on climbs.
A retailer offering free returns spreads the cost across every price tag. Shoppers who already know they return half of what they order gravitate to the free-returns seller; the ones who never return anything are funding them, and drift to cheaper sellers that price returns separately.
First described in George Akerlof (1970).
Key references
- Hendren, N. (2013). Private information and insurance rejections. Econometrica, 81(5), 1713-1762. doi.org/10.3982/ECTA10931
- Einav, L., & Finkelstein, A. (2011). Selection in insurance markets: theory and empirics in pictures. Journal of Economic Perspectives, 25(1), 115-138. doi.org/10.1257/jep.25.1.115
- Finkelstein, A., & Poterba, J. (2004). Adverse selection in insurance markets: policyholder evidence from the U.K. annuity market. Journal of Political Economy, 112(1), 183-208. doi.org/10.1086/379936
- Chiappori, P.-A., & Salanie, B. (2000). Testing for asymmetric information in insurance markets. Journal of Political Economy, 108(1), 56-78. doi.org/10.1086/262111
- 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. doi.org/10.1162/003355398555649
- Akerlof, G. A. (1970). The market for lemons: quality uncertainty and the market mechanism. The Quarterly Journal of Economics, 84(3), 488-500. doi.org/10.2307/1879431
- Finkelstein, A., & McGarry, K. (2006). Multiple dimensions of private information: evidence from the long-term care insurance market. American Economic Review, 96(4), 938-958. doi.org/10.1257/aer.96.4.938
- 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. doi.org/10.2307/1885326
- Bond, E. W. (1982). A direct test of the 'lemons' model: the market for used pickup trucks. American Economic Review, 72(4), 836-840.