Asymmetric information exists whenever one side of a trade knows materially more than the other. A used-car seller knows the car's history; the buyer sees polished paint and guesses. An insurance buyer knows their own health and habits better than the insurer. A patient knows their symptoms, but the doctor knows what they mean — and what the treatment is worth.

This is the unit's last failure, and it has a special home. In 1963, the economist Kenneth Arrow argued that medicine is unlike ordinary markets precisely because of uncertainty and unequal knowledge: the patient cannot evaluate the product, often cannot judge whether treatment worked, and must trust an expert whose interests may differ from theirs. Health economics begins there. So we learn the tool where it was born.

Asymmetric information comes in two shapes, and the whole topic turns on telling them apart. One is hidden type — a fact about who you are that exists before the deal (how risky a driver, how sick a patient, how good a car). The other is hidden action — a choice you make after the deal that the other side cannot fully watch. Before and after. Type and action. Get that distinction clean and the rest of the chapter follows.

Adverse selection is the hidden-type problem. The classic illustration is the used-car market. Sellers know whether their car is a gem or a lemon; buyers do not. So buyers offer an average price. That price is too low for owners of genuinely good cars, who walk away — leaving more lemons, which lowers the average, which lowers offers again. Good cars get driven out of the market by the bad. The buyer's inability to tell them apart, not any lack of desire, is what breaks it.

Health insurance has the same skeleton, and it is the AP-favorite case. An insurer cannot fully see who is healthy and who is sick, so it charges a premium based on the average. To a healthy person, that average looks like a bad deal, so some drop out. With the healthy gone, the remaining pool is sicker, so average cost rises, so the premium rises, so the next-healthiest tier drops out — and around it goes. Economists call the result a death spiral.

Moral hazard is the after-the-trade twin of adverse selection. Once someone is insured, the price they face at the point of care drops toward zero, and behavior shifts: more visits, more tests, less attention to prevention. The patient is not lying or cheating — they are responding rationally to a changed incentive, the same way the fisher and the free rider did in earlier chapters. When the cost of the next unit lands on someone else, people consume more of it.

The repair is to keep a sliver of the price in front of the patient. A deductible makes the patient pay the first portion; a copay keeps a small charge on each visit. These tools restore some price signal — but only some, and here is the tension that runs through every health system: push too hard and you discourage care people genuinely need, including the prevention that saves money later. Protect too completely and you invite waste. There is no setting of the dial that is right for everyone. We are about to see why that dial exists at all.

Start with what insurance is actually for. To an economist, insurance pools risk that is rare, ruinous, and hard to predict — the house fire, the car crash, the cancer. You pay a small certain premium to be protected from a large uncertain loss, and pooling many people makes that possible. By that logic you would never insure the predictable. You do not buy "grocery insurance," because groceries are a known, routine cost, not a risk; running them through an insurer would just add a middleman and a markup. The same should be true of a routine checkup or a generic prescription. Those are known costs, not catastrophes.

And yet insurance in many countries pays for exactly those routine things. Why did it sprawl so far past catastrophe? The center of the answer is buyer market power. A large insurer or employer brings millions of patients to the table, and uses that bulk to negotiate "in-network" prices well below the list or cash price a lone patient would face. So even for ordinary care, running the bill through insurance often gets you a lower price than paying cash — you are borrowing the big buyer's bargaining power. That perverse discount quietly pulls more and more spending under the insurance umbrella, far past the catastrophic core. (A historical accident reinforced it: during World War II, US wage controls pushed employers to compete with tax-free health benefits, welding insurance to the job.) Notice the shape of that argument — a powerful buyer bending prices — because it is the doorway to Unit 4.

So is insurance good? Yes, and that is the point: it is good and it is a trade-off. Pooling catastrophic risk is what makes Chapter 15's inescapable demand survivable — it is the reason a heart attack does not have to mean bankruptcy. But stretch insurance to cover the routine and you trade away the price signal on everything it touches. This is where the third party who pays comes into focus. Whether the payer is a private insurer or the government, the patient is now spending someone else's money — Friedman's bottom row from Chapter 1, where the link between cost and value is weakest. That is the same fact as moral hazard, seen from the spending side rather than the behavior side. The honest question every system has to answer is where to draw the line between what we insure and what we simply pay for.

Hidden information does not always win. Two repair strategies sort the parties out, depending on who moves. Signaling is the informed side revealing itself with something costly and hard to fake: a surgeon's board certification, a used-car dealer's warranty, a degree, a reputation built over years. The signal works precisely because a low-quality type would find it too expensive to imitate.

Screening is the mirror image: the uninformed side designs choices that make the hidden types sort themselves. An insurer offering a menu of plans — high deductible and low premium, or low deductible and high premium — lets healthy and sick buyers reveal their type by which they choose. The test does the work the missing information could not.

Underneath all of this in medicine sits the principal-agent problem. Your doctor is your agent, choosing on your behalf, but with their own incentives and far more knowledge. Pay them per procedure and the structure quietly nudges toward more procedures; pay them a flat fee per patient and it nudges toward fewer. The point is not that doctors are venal — it is that, as with every failure in this unit, the structure shapes the choice before character ever enters.

If there were a clean answer, the world would have copied it. Instead, every country builds a different machine, and each machine is best understood as a bet — a decision about which of this unit's failures it will fight hardest, and therefore which it is willing to live with. Read the table from the most market-driven systems at the top to the most fully socialized at the bottom, and watch the trade-off rotate.

Look down the "bet" column and the lesson of the unit stares back. The market-driven systems guard the price signal and pay for it in catastrophic exposure and inequity. The socialized systems guarantee access and pay for it in rationing and queues. Singapore is the interesting outlier precisely because it read this chapter: it insures the catastrophic, makes people feel the price of the routine, and leans on the state only where buyers are weakest. It is not a free lunch either — nothing here is — but it is a deliberate answer to the exact failures we have catalogued.

Step back from the countries and ask why none of them found a clean answer. The reason is the whole unit at once. A market delivers its famous results only when a list of quiet conditions holds, and across six chapters we pulled them out one at a time and watched a market stumble each time one failed. Healthcare is the chapter where they fail together, in one place, on one person. Read the table as a map of everything you have learned.

Six failures, one patient, all at once. That is why no single lever works: inescapable demand argues for shielding people from catastrophic bills, moral hazard argues for keeping a price in front of them, adverse selection argues for broad mandatory pools, positive externalities argue for subsidizing the spillovers, and the antibiotic commons argues for restraint. Each diagnosis is solid; each points at a different, sometimes opposite, instrument. If economists share a discipline here, it is a sorting rule rather than a single policy: insure the catastrophic, price the shoppable, subsidize the essential and the spillover-rich, and publicly fund the public goods. Match the tool to the failure — and name which failure you are fixing before you reach for a fix.