Market Failure

Why this matters in health economics

A competitive market, left alone, is supposed to allocate resources efficiently: prices carry information, buyers and sellers know what they are trading, and no one outside the transaction is harmed or helped. Health care violates each of these assumptions systematically, not occasionally. Illness is uncertain and often catastrophic; patients cannot judge the care they buy; the person who advises treatment often profits from it; one person's vaccination protects their neighbour; and much of the bill is paid by a third party. The result is not a market that is slightly inefficient but one that, without intervention, would leave the sick uninsured, the poor untreated, and prices disconnected from value.

This is the intellectual foundation on which the rest of health economics is built. When Kenneth Arrow published his 1963 analysis of medical care, he was not making a political argument for or against public provision; he was showing that the specific ways health care departs from the competitive ideal are predictable, and that the institutions surrounding medicine — professional licensing, non-profit hospitals, insurance, trust-based ethics — can be read as responses to those departures. For a director, this matters because it reframes intervention. Regulation, subsidy, and public provision are not ideological impositions on an otherwise self-correcting market; they are attempts to repair failures the market cannot fix itself.

Getting the diagnosis right is what saves money and lives. A payer that treats over-provision as a pricing problem when it is really an agency problem will keep cutting fees while volumes climb. A government that mandates insurance without addressing adverse selection will watch its voluntary pool collapse. The failures interact, and the wrong remedy for one can deepen another. This chapter catalogues the failures and their mechanisms; the policy and regulatory responses to them are the subject of Chapter 3.2 — Health Policy, and the way systems are financed to contain them is the subject of Chapter 3.1 — Health Systems.

Core concepts

Market failure is any situation in which a market, left to itself, allocates resources inefficiently — producing too much of some things, too little of others, or excluding people who should be served. Health care is the textbook case, and Arrow's paper is the founding diagnosis; the discipline of health economics grew from it (the wider map is drawn in Chapter 1.1 — Introduction to Health Economics).

The first source of failure is uncertainty. The demand for care is irregular and unpredictable — you cannot plan the year you will need surgery — and both the incidence of illness and the effectiveness of treatment are uncertain. This uncertainty is what creates the demand for insurance (the demand side is Chapter 1.2 — Demand for Health and Healthcare), and insurance in turn introduces a fresh set of failures that this chapter owns.

The second is information asymmetry: the two sides of a health transaction rarely know the same things. A patient cannot readily judge whether a recommended scan is necessary, whether the surgery went well, or whether the drug was the best option. The clinician knows far more. This gap turns the clinical relationship into a principal–agent problem: the patient (principal) delegates decisions to the clinician (agent), trusting the agent to act in the patient's interest rather than their own. When agency is imperfect and the agent is also the seller, the door opens to supplier-induced demand — care generated by the provider's interest rather than the patient's need.

Insurance, the natural response to uncertainty, breeds two failures of its own. Adverse selection arises because individuals know more about their own health risk than the insurer does; those who expect to claim are keenest to buy, which raises premiums, which drives out the healthy, which raises premiums again — the "death spiral" that can unravel a voluntary market entirely. Moral hazard is the change in behaviour once insured: because someone else pays at the point of use, patients and providers consume and supply more care than they would if facing the full price. The RAND Health Insurance Experiment — a large randomized study run in the United States in the 1970s and 1980s — remains the landmark evidence that cost-sharing reduces the quantity of care used, mostly without measurable harm to average health, though with some harm to the poorest and sickest.

The third family of failures concerns spillovers. An externality is a cost or benefit that falls on someone outside the transaction. Vaccination is the classic positive externality: being immunized protects others by reducing transmission, so individuals acting on private benefit alone will under-vaccinate relative to the social optimum. Some health goods are public goods in the strict economic sense — non-rival and non-excludable — such as disease surveillance, clean-air regulation, and the knowledge produced by research. And some resources are commons that markets deplete: antimicrobial resistance is a slow-motion tragedy of the commons in which each prescriber's rational use erodes a shared resource — the continued effectiveness of antibiotics — for everyone.

Finally, health care markets are prone to concentration and cost pressure. Hospitals often behave as local monopolies; specialized services may be natural monopolies where one provider is efficient but unconstrained by competition. And the Baumol effect — cost disease — explains why labour-intensive services such as health and education grow structurally more expensive over time: wages must keep pace with the wider economy even where productivity gains are hard to achieve, so the relative price of care rises even when nothing is being done wrong.

Best practices

1. Diagnose the specific failure before you reach for a remedy. "The market isn't working" is not an analysis. Ask which failure you are facing — asymmetric information, adverse selection, moral hazard, an externality, or monopoly — because each has a different fix, and the wrong remedy can deepen the problem. A payer cutting fees to control spending has misread supplier-induced demand as a price problem; the volumes will simply rise to defend income.

2. Treat agency as the central relationship, and design to protect it. Because patients delegate decisions to clinicians, the integrity of that agency relationship determines whether care serves need or income. Separate, where you can, the person who recommends care from the person who profits from delivering it; use second opinions, referral guidelines, and clinical decision support as structural safeguards rather than exhortations to behave well.

3. Watch how payment shapes the agent's incentive. Fee-for-service rewards volume and invites induced demand; capitation rewards restraint and invites under-provision. There is no neutral payment method — every one pushes the agent somewhere — so choose the distortion you can best monitor and counterbalance it with quality measurement. Blended payment exists precisely because no single mechanism gets agency right (payment design is developed in Chapter 3.1 — Health Systems).

4. Assume adverse selection in any voluntary insurance pool, and price or pool against it. If enrolment is optional and buyers know their own risk, the healthy will leave and the pool will drift sick and expensive. Community rating, guaranteed issue, risk equalization between insurers, and — most powerfully — broad mandatory or automatic enrolment are the levers that keep a pool viable; adopting one without the others often fails.

5. Expect moral hazard, but distinguish its two kinds before you tax it. Some induced use is genuinely wasteful; some is valuable care that people would forgo if they paid full price. Cost-sharing dampens both, and the RAND evidence shows patients cut back on effective and ineffective care alike, with the poorest and sickest cutting most. Design cost-sharing so it bites on low-value care and exempts the chronically ill and the poor, rather than applying a flat deductible that deters the wrong people.

6. Correct externalities at the margin where behaviour actually changes. For positive externalities such as vaccination, subsidy, free provision, defaults, and mandates raise uptake towards the social optimum; leaving it to private choice guarantees under-provision. For negative externalities and commons problems, the lever is to make the shared cost visible to the decision-maker — stewardship rules, prescribing feedback, and pricing that reflects the damage done.

7. Treat antimicrobial resistance as a commons that ordinary markets will deplete. No individual prescriber, patient, or manufacturer bears the full cost of resistance, so all under-invest in preserving effectiveness and over-use existing drugs. Stewardship programmes ration current use; the harder failure is that the market will not fund new antibiotics whose value lies in being held in reserve, which is why "pull" incentives that pay for availability rather than volume are being trialled (the policy design belongs to Chapter 3.2 — Health Policy).

8. Name the monopoly and regulate the price it cannot be competed down to. Where one hospital serves a region or one manufacturer holds a patent, there is no rival to discipline price or quality, and entry is often blocked by scale, licensing, or intellectual property. Merger review, tariff regulation, reference pricing, and health technology assessment substitute for the competition that is absent; pretending a concentrated market is competitive is a recurring policy error.

9. Read rising costs through the Baumol lens before blaming inefficiency. When the relative price of care rises decade on decade, part of the cause is structural: care is labour-intensive, and wages track a whole economy whose productivity is rising faster than health care's can. This means some cost growth is neither waste nor mismanagement, and cost-control targets that assume otherwise will demoralize good services and still miss.

10. Use information remedies before heavier intervention, but do not overrate them. Quality reporting, price transparency, accreditation, and independent guidance reduce asymmetry and can shift behaviour. But information helps only where users can act on it; a frightened patient at the point of care cannot shop, so transparency complements regulation rather than replacing it.

11. Match the remedy's cost to the failure's size. Every intervention has its own failure — regulatory capture, administrative burden, blunted incentives, unintended gaming. A remedy that costs more than the failure it corrects is a net loss, so weigh the government failure against the market failure rather than assuming intervention is free (this trade-off is the heart of Chapter 3.2 — Health Policy).

12. Expect the failures to interact, and sequence your fixes. Mandating insurance without risk equalization, or transparency without capacity to switch, can worsen the very problem you meant to solve. Map the failures together, fix the one that unlocks the others first, and monitor for the new distortion each remedy introduces.

Questions to discuss with your team

1. Where in our system does the person who recommends care also profit from delivering it, and what protects the patient there? This question surfaces supplier-induced demand, the most under-examined failure in most organizations because it implicates respected clinicians rather than obvious villains. Look for the places where referral, diagnosis, and delivery sit inside the same income stream — an imaging suite the referring physician owns a share of, a surgical target tied to a bonus, a service paid per procedure. The honest answer is not that colleagues are corrupt; it is that structure, not virtue, is what should protect patients, and that relying on professionalism alone is a design choice with predictable failure rates. Distinguish demand you can justify clinically from demand that tracks provider income too closely to be coincidence. A good discussion ends with two or three specific places where you would separate advice from sale, or add an independent check, and an honest admission of where your current data cannot tell need from inducement.

2. If enrolment in our coverage became voluntary tomorrow, who would leave first, and how long before the pool failed? Adverse selection is easiest to see as a thought experiment about who exits. The people who leave a voluntary pool are the healthy, the young, and the confident, because they are subsidizing everyone else and know it; the people who stay are those expecting to claim. Trace that dynamic through your own numbers: which segments are cross-subsidizing which, and how thin is the margin before rising premiums trigger the next wave of departures. The tension is that everything protecting the pool — mandates, automatic enrolment, community rating, risk equalization — limits individual choice or redistributes cost, so someone always objects. An honest answer names the specific mechanisms holding your pool together today and how fragile each is, rather than assuming solidarity is self-sustaining.

3. Which of our rising costs are genuine inefficiency, and which are structural cost disease we cannot manage away? This question forces an uncomfortable separation that budget conversations usually blur. Some cost growth is waste — duplicated tests, avoidable admissions, prices no one negotiated — and is fair game for efficiency programmes. But the Baumol effect means part of health care's cost growth is structural: a labour-intensive service in an economy whose wages keep rising will get relatively more expensive even when it is run well. Confusing the two is expensive in both directions — you either chase savings that are not there and burn out staff, or you excuse waste as inevitable. An honest answer attempts the split explicitly, accepts that some of the rise is not a failure to be fixed but a fact to be funded, and changes what you promise your finance director accordingly.

4. Where does our cost-sharing deter care we want people to use, and where does it merely dampen waste? Moral hazard is the reason cost-sharing exists, but the RAND evidence is uncomfortable: patients cut back on effective and ineffective care alike, and the poorest and sickest cut most. So the real question is not whether to charge at the point of use but where the charge lands. Trace your co-payments and deductibles through the specific patients they touch — does a flat charge fall on the diabetic filling a maintenance prescription as heavily as on someone requesting a discretionary scan? The tension is that a charge precise enough to bite only on low-value care is administratively harder than a blunt deductible that treats all use alike. An honest answer identifies at least one place where your current cost-sharing is deterring the wrong people, and accepts that exempting the chronically ill and the poor costs revenue you were relying on.

5. For the externalities and shared resources our decisions touch, who bears a cost they never signed up for? Externalities and commons problems are the failures organizations notice last, because the harm falls outside the transaction and off the balance sheet. Ask where your services generate benefits or costs for people who are not the patient — a vaccination programme that protects a whole community, or an antibiotic prescribing pattern that erodes the shared effectiveness of the drug for everyone. The awkward truth is that individually rational decisions — one more antibiotic to be safe, skipping a jab that mostly protects others — aggregate into a worse outcome no single actor intended. Distinguish the positive externalities you should be subsidizing or defaulting into from the commons you should be rationing and stewarding. An honest answer names one externality you currently leave to private choice and asks whether a subsidy, a default, or a stewardship rule would move behaviour towards what the community actually needs.

6. Where are we the only game in town, and does anything discipline our price and quality when competition cannot? Concentration is easy to miss from the inside, because a dominant provider rarely feels dominant to its own staff. Ask honestly where patients have no realistic alternative — the sole hospital for a region, the only supplier of a specialized service, the holder of a patent with no substitute — because that is where market discipline is absent and price and quality drift unchecked. The tension is that scale often brings genuine efficiency, so the goal is not to break up every concentration but to substitute a deliberate check for the competition that cannot exist. An honest answer names where you hold market power, resists the comfortable fiction that you are simply competing well, and identifies what — regulation, transparency, an external benchmark, or a purchaser with teeth — actually constrains you today.

In practice: a health economics example

The Ministry of Health of the fictional middle-income Republic of Maravia wants to expand health coverage. Half the population is enrolled through a payroll-financed scheme for formal-sector workers; the other half — informal traders, farmers, the self-employed — has no coverage. A reform team proposes a voluntary insurance product: informal workers may buy in at a flat annual premium, with a generous benefit package matching the formal scheme. The finance ministry likes it because participation is optional and the budget cost looks contained. An economist on the team asks the team to think about who would actually enrol.

The answer is a textbook death spiral. At a flat premium, the people for whom buying makes obvious sense are those already sick or expecting to be — the diabetic trader, the pregnant farmer, the family with a chronic condition. Healthy young workers, facing a premium that exceeds their expected claims, stay out. The enrolled pool is therefore sicker than the population, claims exceed premiums, and the actuaries recommend a premium rise. The rise drives out the next-healthiest tier, and the cycle repeats. The team models three enrolment waves and finds the premium doubling within four years while enrolment falls — the voluntary product cannibalizing itself through adverse selection.

The team also flags a second failure hiding in the benefit design. The scheme pays participating clinics fee-for-service with no referral gatekeeping, and clinics in the pilot district are privately owned by the doctors who staff them. In a comparable pilot elsewhere, imaging and minor-procedure volumes had run well above clinical need — supplier-induced demand, the agency relationship bending towards provider income. A voluntary scheme with a sick risk pool and volume-rewarding payment would combine the worst of adverse selection and induced demand: high claims per member and high services per claim.

The redesign attacks both failures at once. Enrolment shifts from voluntary to automatic for anyone registering for a trading licence or agricultural subsidy, with premiums subsidized on a sliding scale from general taxation so the healthy cannot self-select out and the poor are not priced out — broadening the pool is the only durable answer to adverse selection. Payment moves to a blended model: a capitation base to remove the volume incentive, a small quality-linked component, and referral guidelines with independent imaging to protect agency. The team is candid about the residual: capitation now risks under-provision, so they add monitoring of under-referral. The lesson the Ministry takes is Arrow's — the failures are structural, they interact, and the design must answer them together rather than one at a time. The regulatory and legislative machinery to enact this sits in Chapter 3.2 — Health Policy.

Four sector lenses

Startup

A digital health start-up usually meets market failure as asymmetric information it promises to reduce — a symptom-checker, a price-comparison tool, a quality dashboard. The opportunity is real but narrow: information helps only where users can act on it, and a patient mid-crisis cannot shop. Founders should also notice which failure their business model quietly relies on; a telehealth service paid per consultation faces the same induced-demand temptation as any fee-for-service provider, and investors and regulators will eventually ask. The credible pitch corrects a failure without creating a new one.

Small business

A small established provider — a GP partnership, a single-site clinic, a community pharmacy, a care home — lives on the wrong side of the agency relationship every day, and its integrity is the practice's main asset. Unlike a start-up chasing growth, such a business is a price-taker facing payers who set fees and rules it cannot rewrite, so its room for manoeuvre is in how it behaves, not what it can build. The steady temptation is the induced-demand one: a fee-for-service income stream quietly rewards the extra test or repeat visit, and a small owner-operator feels that pull directly in the monthly accounts. The durable answer is to make trustworthiness visible — transparent referral, honest recommendation, restraint that patients and commissioners can see — because in a local market reputation is the one form of quality signal a small provider can actually own.

Enterprise

A large insurer or hospital group lives inside these failures as its core operating reality. An insurer manages adverse selection through risk pooling, underwriting rules, and risk equalization, and manages moral hazard through cost-sharing and prior authorization — each lever with its own backlash. A hospital system that has grown to regional dominance is a monopoly whether or not it uses the word, and will face merger scrutiny and price regulation accordingly. At enterprise scale the question is less whether failures exist than whether the organization is exploiting them or governing them, because regulators and the public increasingly tell the difference.

Government

Government is the actor of last resort for failures no market participant can fix — externalities, public goods, commons, and the unravelling of voluntary pools. Its toolkit is the widest: mandates and subsidies for vaccination, stewardship and pull incentives for antimicrobials, competition law and tariff regulation for monopoly, mandatory pooling for insurance. Its discipline must be the widest too, because every intervention risks its own failure — capture, gaming, deadweight cost — and a remedy dearer than the disease is still a loss. The design of these interventions is the subject of Chapter 3.2 — Health Policy; the point here is that the failures are what make government's role in health unavoidable in every system on earth.

Common failure modes

• Misdiagnosing the failure. Treating over-provision as a pricing problem when it is an agency problem, or treating an unravelling pool as a marketing problem when it is adverse selection. Fix: name the specific failure before choosing the remedy, using the catalogue in this chapter.

• Assuming solidarity is self-sustaining. Building a voluntary pool and expecting the healthy to stay out of goodwill. Fix: pool broadly through mandates or automatic enrolment, and equalize risk between insurers.

• Applying blunt cost-sharing. A flat deductible that deters the poor and the chronically ill as much as it deters waste. Fix: target cost-sharing at low-value care and exempt those the RAND evidence shows are harmed by it.

• Leaving externalities to private choice. Expecting optimal vaccination or antibiotic stewardship from individuals who bear only their private costs and benefits. Fix: subsidize, default, or mandate the positive externality; make the shared cost of the negative one visible to the decision-maker.

• Pretending a monopoly is competitive. Applying market rhetoric to a region with one hospital or a drug with one patent-holder. Fix: regulate price and quality and scrutinize mergers where competition cannot exist.

• Blaming Baumol cost growth on staff. Setting efficiency targets that treat structural cost disease as mismanagement. Fix: separate genuine waste from the structural rise, and fund the latter honestly.

• Ignoring interaction effects. Fixing one failure and worsening another — mandating coverage without risk equalization, or transparency without capacity to switch. Fix: map the failures together and sequence the remedies, monitoring for each new distortion.

Maturity model

Checklist

• For the problem in front of you, name the specific market failure — asymmetric information, adverse selection, moral hazard, externality, commons, or monopoly — before choosing a remedy.
• Identify where the recommender of care also profits from it, and state what structurally protects the patient there.
• Check every voluntary insurance pool for adverse selection: who would leave first, and what holds the pool together.
• Confirm cost-sharing bites on low-value care and exempts the poor and chronically ill.
• For each major positive externality (e.g. vaccination), confirm a subsidy, default, or mandate is pushing uptake towards the social optimum.
• Treat antimicrobial resistance as a commons: check that both stewardship of current drugs and incentives for reserve drugs are addressed.
• Name any monopoly explicitly and confirm price and quality are regulated where competition cannot exist.
• Separate structural (Baumol) cost growth from genuine inefficiency in your cost plans.
• Weigh each intervention's own failure risk against the market failure it corrects.
• Map how the failures interact and sequence remedies so fixing one does not worsen another.

Key sources

• Kenneth J. Arrow, "Uncertainty and the Welfare Economics of Medical Care" (American Economic Review, 1963) — the founding diagnosis of health care market failure.
• Cam Donaldson & Karen Gerard, Economics of Health Care Financing: The Visible Hand — market failure and the case for intervention in health care financing.
• Amy Finkelstein, "Moral Hazard in Health Insurance" (Marshall Lecture / Columbia University Press) — modern evidence on moral hazard and selection.
• RAND Health Insurance Experiment — the landmark randomised evidence on cost-sharing and moral hazard.
• Economics Network, Health Economics for Teachers — market failure module — https://economicsnetwork.ac.uk/health/teachers
• World Health Organization health financing publications — pooling, prepayment, and coverage design against selection and out-of-pocket failure.

References

1. Market failure — Wikipedia — https://en.wikipedia.org/wiki/Market_failure
2. Kenneth Arrow — Wikipedia — https://en.wikipedia.org/wiki/Kenneth_Arrow
3. Information asymmetry — Wikipedia — https://en.wikipedia.org/wiki/Information_asymmetry
4. Principal–agent problem — Wikipedia — https://en.wikipedia.org/wiki/Principal%E2%80%93agent_problem
5. Supplier-induced demand — Wikipedia — https://en.wikipedia.org/wiki/Supplier-induced_demand
6. Adverse selection — Wikipedia — https://en.wikipedia.org/wiki/Adverse_selection
7. Moral hazard — Wikipedia — https://en.wikipedia.org/wiki/Moral_hazard
8. RAND Health Insurance Experiment — Wikipedia — https://en.wikipedia.org/wiki/RAND_Health_Insurance_Experiment
9. Externality — Wikipedia — https://en.wikipedia.org/wiki/Externality
10. Public good (economics) — Wikipedia — https://en.wikipedia.org/wiki/Public_good_(economics)
11. Antimicrobial resistance — Wikipedia — https://en.wikipedia.org/wiki/Antimicrobial_resistance
12. Baumol effect — Wikipedia — https://en.wikipedia.org/wiki/Baumol_effect
13. Kenneth J. Arrow, "Uncertainty and the Welfare Economics of Medical Care" — American Economic Review, 53(5), 1963.
14. Economics Network — Health Economics for Teachers — https://economicsnetwork.ac.uk/health/teachers
15. Health economics — Wikipedia — https://en.wikipedia.org/wiki/Health_economics