A different way for a market to break
So far in this rung, the markets have broken because the *price* missed something real: pollution sat outside the bargain as an externality, and a lighthouse stayed unbuilt because nobody could be charged for a public good. In every case both buyer and seller could *see* the good plainly; the trouble was who paid for it. This guide is about a stranger and quieter failure — one where the good itself is hidden behind a fog, and the two sides of the deal simply do not know the same things. Economists call this [[asymmetric-information|asymmetric information]]: one party to a trade knows something the other does not, and that imbalance alone can make a perfectly willing buyer and seller fail to do business.
It helps to split the fog into two kinds, because they bite at different moments. Sometimes the hidden thing is a *quality you cannot observe before you sign* — is this used car a gem or a wreck? is this applicant truly healthy? That is the world of hidden information, and the failure it causes is called adverse selection. Other times the deal is signed cleanly, but afterwards one side takes a *hidden action* the other cannot watch — once you are insured, do you still bother to lock your bike? That is hidden action, and the failure it causes is called moral hazard. Same root, two branches; we take each in turn.
The market for lemons
The classic story comes from a 1970 paper by George Akerlof, who shared a Nobel Prize for it. In American slang a bad used car is a "lemon" and a good one is a "peach." The seller knows which one they are holding; the buyer, kicking the tyres, cannot tell. Suppose half the cars on offer are peaches worth 6,000 dollars to a buyer and half are lemons worth 2,000. Knowing the split but not which car is which, a buyer is willing to pay only the *average* — about 4,000. But here is the trap: no owner of a genuine 6,000-dollar peach will sell it for 4,000. So the peaches quietly withdraw, leaving only lemons on the lot.
Now watch the spiral. Once buyers realise the peaches have fled and only lemons remain, their willingness to pay drops to 2,000. That price drives out any merely-decent cars too, which pushes the price lower still, which drives out the next tier — and in the harshest version the good cars vanish entirely and the market shrinks toward nothing. This is [[eco-adverse-selection|adverse selection]]: the very buyers' caution that seems sensible ends up selecting *against* good quality, so the bad drives out the good. Notice what is lost. There were real, mutually beneficial trades — a buyer who would happily pay 5,500 for a peach, an owner happy to sell at 5,000 — that simply never happen. That unrealised gain is the deadweight loss of the information gap.
After the deal: moral hazard
Adverse selection strikes *before* you sign, when quality is hidden. [[eco-moral-hazard|Moral hazard]] strikes *after*, when behaviour is hidden. The name is older than the economics: insurers in the 1800s noticed that once a warehouse was insured against fire, its owner grew curiously careless about smoking near the stock. The point is not that people are wicked. It is that insurance, by design, shifts the cost of a bad outcome onto someone else — and the moment you no longer fully bear the cost of your own carelessness, your incentive to be careful quietly weakens. Fully insure my phone and I will treat it more roughly than if every crack came out of my own pocket.
The cure is rarely to abolish the cover — that would just throw the risk back onto people who wanted protection. Instead, contracts are designed to leave you with *some* skin in the game, so the hidden action still has a price you feel. A health plan charges a *deductible* (you pay the first 500 dollars) and *co-pays* (you split each bill); car insurance offers lower premiums for a clean record. Each device deliberately keeps a sliver of the cost on your shoulders, just enough to keep your incentives pointing the right way without erasing the protection you bought. Notice the elegant tension: perfect insurance would be the most reassuring and the most corrosive to care — so good contracts insure you *almost* fully, on purpose.
The principal and the agent
Moral hazard's most general home is the [[principal-agent-problem|principal-agent problem]], and once you see it you will spot it everywhere. A *principal* hires an *agent* to act on their behalf — but the agent has their own goals, and the principal cannot fully watch what the agent does. A shareholder (principal) hires a CEO (agent) who may prefer empire-building to profit; a landlord hires a property manager; a patient relies on a doctor; voters elect a politician. In every pair, the agent knows more about their own effort and choices than the principal can ever verify. The hidden action is simply *how hard, and how honestly, the agent actually works for you*.
The fix is to stop trying to *watch* effort and instead *reward outcomes you can see*, so the agent chooses hard work for their own sake. Tie a salesperson's pay to commission, give a CEO stock options, pay a contractor a bonus for finishing early. Each aligns the agent's incentives with the principal's by making the agent share in the result. But this is genuinely hard, not a tidy trick — economists still argue about it. Reward measured outcomes and people game the measure (a surgeon who is paid for survival rates may refuse the riskiest, sickest patients). Push too much risk onto an agent who cannot control luck and you must overpay them for bearing it. The principal-agent problem is less a puzzle with one answer than a permanent balancing act between motivating effort and sharing risk.
Fighting the fog: signalling and screening
Markets are not helpless against the fog. Two mirror-image strategies have evolved to close information gaps. Signalling is when the *informed* side takes a costly action to credibly reveal its hidden quality. A used-car dealer offers a warranty; a healthy firm pays a dividend; a job applicant earns a degree. The key word is *costly*: a signal only works if it would be too expensive for a low-quality type to fake. A genuine peach-seller can cheaply promise a warranty because the car rarely breaks; a lemon-seller offering the same warranty would be ruined by claims — so the warranty separates the two, and the buyer can finally tell them apart.
Screening is the same idea from the other chair: the *uninformed* side designs a test or a menu that makes the informed side sort itself. An insurer offers two policies — a cheap one with a high deductible, and a pricey one with full cover — knowing that safe drivers will pick the cheap deductible and risky ones the full cover, so the very choice reveals the hidden type. A lender runs a credit check; an employer sets a probation period. One honest caveat economists raise: a signal can be socially wasteful. Michael Spence won a Nobel for showing that schooling can work partly as a *pure signal* of pre-existing ability — if some of a degree's value is merely proving you were already capable, rather than teaching you anything, then society spends real resources just to sort people. That debate over how much of education is learning versus signalling is still very much alive.
ADVERSE SELECTION MORAL HAZARD (hidden quality) (hidden action) bites BEFORE the deal bites AFTER the deal | | fix from informed side: fix by sharing the cost: SIGNALLING deductibles, co-pays (warranty, degree) commission, stock options fix from uninformed side: = leave the agent SCREENING "skin in the game" (deductible menu, credit check)
Why information itself is a market failure
Step back and see why this belongs beside pollution and public goods as a genuine [[market-failure|market failure]]. The invisible hand's promise rests on a quiet assumption you met early in this ladder: that both sides of a trade know what they are buying and selling. When that assumption cracks, prices stop carrying honest information, mutually beneficial trades go unmade, and the outcome is inefficient even though nobody behaved badly. The gap is not a moral flaw in the traders; it is a missing market — the market for *trust*, or for *verified quality*, simply does not exist on its own.
And this is also where private signalling and public policy meet. Many institutions you take for granted exist precisely to fill the information gap that markets cannot fill alone: brand names that would be ruined by one bad batch, online review scores, professional licences for doctors and electricians, food-safety labels, mandatory disclosure of a car's accident history. Some of these are private; many are rules a government imposes. They are close cousins of the merit-good logic — the case for nudging or requiring goods people would under-consume when left fully to themselves. Be honest about the limit, though: regulators face their *own* information gap and can be captured or mistaken, so more disclosure is not always better and a stamp of approval can breed false confidence. Information failures are real, but the fixes are imperfect tools, not magic wands.