Asymmetric Information

George Akerlof's paper The Market for Lemons, published in 1970 after being rejected by three top journals, has become one of the most-cited papers in economics and won Akerlof a 2001 Nobel shared with Michael Spence and Joseph Stiglitz. The argument was a single sharp claim: when buyers cannot tell good cars from bad and information about quality is asymmetric, the market unravels — sellers of good cars get only the average price, good cars go for less than they are worth, their sellers withdraw, average quality falls, and only lemons remain. The argument was deceptively simple and enormously consequential, spawning the information-economics research programme — Spence's analysis of signalling, the analysis of screening and moral hazard, and most of modern microeconomics' workable framework for real markets.

The unravelling has a precise engine, and it is worth following once. A buyer who cannot judge quality will pay only what the average car is worth. That price is a bargain for a bad car and an insult for a good one, so the good cars withdraw first — and their withdrawal lowers the average, which lowers the price, which drives out the next tier of good cars. The end state can be a market that barely exists, not because anyone cheated but because the honest seller had no way to prove he was honest. Put it that way and the shape of every remedy becomes predictable: each is a way of making quality visible. The informed party can spend something a bad type could not afford to imitate — a warranty is cheap for a reliable manufacturer and ruinous for an unreliable one, which is exactly why it carries information. This is signalling, and the cost is not waste; the cost is the message. The uninformed party can work the other side by offering a menu that forces a self-report: an insurer who offers a low deductible at a high premium and a high deductible at a low one learns something about you from which one you choose. That is screening. Where neither works, trust gets outsourced — repeated dealing and reputation substitute for inspection when a seller must face you again, and certification or mandatory disclosure substitutes an institution's judgement for your own. Each remedy then fails in its own characteristic way, which is where the live arguments sit. If education works partly as a signal of ability rather than a producer of it, the private return can exceed the social return, and society pays for a sorting mechanism it mistakes for teaching. Disclosure assumes the disclosure is read; the mortgage documents of 2006 were disclosed and unread. Reputation assumes reviews are honest, and reviews can be bought. The framework keeps generating work because information about information is itself asymmetric — solve the problem at one level and it reappears at the next.