# Adverse Selection **The model:** When the parties to a transaction have different information about what's being exchanged, the worse-off party tends to be drawn disproportionately into the market — and the resulting market deteriorates or collapses entirely. Akerlof's 1970 analysis used used cars as the example: sellers know whether their car is good or a "lemon"; buyers don't. Buyers offer prices that reflect the average quality. Sellers of good cars find the average price too low and don't sell. Only lemons stay in the market. Buyers anticipate this and offer even lower prices. The market spirals down to mostly lemons at low prices, with good cars unable to find buyers willing to pay what they're worth. Adverse selection appears wherever one party knows something material about the transaction the other party doesn't. Insurance markets have it (people who buy life insurance know more about their health than the insurer); labor markets have it (workers know more about their reliability than employers); credit markets have it (borrowers know more about their probability of default); political endorsement markets have it (political brokers know more about a politician's actual reliability than voters do). ## The Mechanism **Information asymmetry, before the contract.** Adverse selection arises *before* parties commit to a transaction. The asymmetric information leads to the wrong selection of partners. Distinct from [[Moral Hazard]], which is asymmetric information about *behavior after* a contract is signed. **The selection effect collapses the market.** Pricing for the average draws in the worse-than-average and drives out the better-than-average. The remaining pool is worse than average. Pricing for the new average draws in the worse-than-new-average and drives out the new-better. The cycle continues until either some external mechanism breaks it or the market disappears. **Several known partial fixes.** Markets with adverse selection have evolved various responses: - **Signaling:** the better-quality party finds a way to credibly demonstrate quality. Education degrees, brand reputation, warranties, certifications. - **Screening:** the less-informed party finds ways to verify quality. Inspections, drug tests, credit reports. - **Mandates:** mandatory participation forces both good and bad into the pool. Mandatory health insurance is the canonical case. - **Subsidies:** subsidize the desirable participants enough that they don't drop out. Each fix is partial; each has costs. The literature is enormous. ## KB Applications ### Health Insurance and the Mandate Question Health insurance is the textbook adverse-selection case. People who expect to have high medical costs (sicker, older, riskier) value insurance more than people who expect to have low costs. Without a mandate, the high-cost people pile in and the low-cost people drop out, leaving an insurance pool of sick people whose premiums must rise to cover their costs — driving out the next-marginally-healthy, and so on (the "death spiral"). The Affordable Care Act's individual mandate was an explicit Akerlof-style fix; its weakening (zero-dollar penalty since 2019) creates predictable adverse-selection problems that are still working through the market. The right-wing position on the ACA mandate has often denied the adverse-selection logic ("if people don't want insurance, they shouldn't have to buy it"). The economics literature is unambiguous that the mandate solves a real problem; whether to solve that problem is a value choice. Pretending the problem doesn't exist is empirically wrong. ### Selection in Right-Wing Media Audiences The right-wing media ecosystem has audience-side adverse selection: viewers who want extreme content stay; viewers who want measured content drift away. The remaining audience is more extreme than the original; the content adjusts to serve them, becoming yet more extreme; the next-marginal viewer drifts away. The dynamic is structurally identical to the lemons market: each iteration sorts the audience further, and the equilibrium is far from where the average audience would be if engagement weren't asymmetric. ### Election-Donor Selection Political fundraising has adverse-selection features. Candidates who would adopt mainstream positions can attract mainstream donors at modest amounts; candidates who adopt extreme positions can attract extreme donors at much larger amounts (intensity correlates with willingness to fund). Without external constraints (public financing, contribution limits), the selection effect favors candidates who appeal to the most-intense donors, regardless of broad voter preference. This is one structural reason candidate-extremism is observed even where median-voter analysis predicts moderation. ### Why Loyalty-Tested Cabinets Decay Quality Cabinet selection in second-Trump-era politics has been adverse-selection-heavy: candidates who would have been credentialed and capable choose not to apply (they recognize the loyalty cost and the reputational risk); candidates willing to swear loyalty are disproportionately less-credentialed and more loyalty-driven. The cabinet pool, therefore, deteriorates over time as the more-capable refuse and the less-capable accept. Standard adverse-selection prediction — and the empirical pattern is consistent. ### The Honest Steelman of Conservative Adverse-Selection Concerns Conservatives historically raised adverse-selection concerns about expanded government programs (the people most likely to use them are those least like the median voter). Sometimes the concern is correct: badly-designed welfare programs can produce selection effects. The honest engagement isn't to dismiss the concern but to ask whether the specific program design accounts for it. Modern social-policy design (income phase-outs, work requirements, time limits) is largely a response to selection concerns — sometimes overcorrecting. ## See Also - [[Moral Hazard]] — the post-contract sibling - [[Supply and Demand]] — adverse selection is what breaks supply-and-demand when information is asymmetric - [[George Akerlof]] — author of the foundational "Market for Lemons" paper - [[Incentive-Caused Bias]] — the cognitive engine