Moral Hazard

Moral Hazard in Health Insurance, Healthcare Use and Cost Sharing

In health economics, moral hazard most commonly describes behavioural responses to health insurance. When insurance reduces the price a patient pays for healthcare, the patient may use more care than they would if they faced the full cost. Moral hazard can also affect prevention, treatment choices, provider behaviour and decisions made by organisations protected from financial loss. The term does not imply dishonesty or immoral conduct. It describes how incentives change when actions are imperfectly observable and their consequences are shared with someone else.

How Moral Hazard Arises

Moral hazard is an incentive problem created by the way risk is shared. Kenneth Arrow's 1963 paper on the welfare economics of medical care noted that the cost of medical care depends on the choice of doctor and the willingness to use services as well as on the illness itself, and that widespread insurance increases the demand for care. Mark Pauly's 1968 comment argued that seeking more care when insured is rational economic behaviour rather than a moral failing, and this reading has shaped the health economics literature since. Moral hazard generally requires three conditions to hold together.

• one party is protected from at least part of a cost or risk;
• that party can take an action that affects the probability or size of the resulting cost; and
• the other party cannot perfectly observe, verify or contract over that action.

For example, health insurance transfers part of the financial cost of treatment from the patient to an insurer or public payer. Because the patient no longer faces the full price of care, insurance can change decisions about whether, when and how much healthcare to use.

This is an incentive effect rather than proof that an individual is deliberately exploiting the system. Behaviour can change simply because the price faced by the decision-maker has changed.

Ex Ante and Ex Post Moral Hazard

Moral hazard in health insurance is commonly divided into two forms, distinguished by whether the behavioural response occurs before or after illness. Zweifel and Manning's review in the Handbook of Health Economics also identifies a dynamic form, in which insurance increases the demand for expensive new medical technologies. Their review found the case for ex post moral hazard stronger than the evidence for ex ante or dynamic moral hazard, which remains limited.

Ex Ante Moral Hazard

Ex ante moral hazard occurs before illness or injury. Insurance may reduce the incentive to take preventive action because the insured person will not bear the full financial consequences of a future health event.

Possible examples include reduced investment in prevention, lower adherence to risk-reducing behaviour or greater participation in hazardous activities. Evidence of ex ante moral hazard varies considerably by behaviour and setting, and insurance can also increase prevention when preventive services are covered.

Ex Post Moral Hazard

Ex post moral hazard occurs after illness, injury or another insured event has occurred. Because insurance lowers the price paid at the point of use, the insured person may consume more healthcare than they would without coverage.

The additional use may include valuable treatment that was previously unaffordable as well as care whose expected benefit is small relative to its cost. Increased utilisation should therefore not automatically be classified as waste.

A Simple Economic Representation

Suppose a patient chooses healthcare quantity $q$. Let $B(q)$ represent the benefit obtained from care and $p$ represent the resource cost of each unit.

Without insurance, the patient may face the full price $p$ and choose the quantity of care at which marginal benefit equals that price.

$$MB(q)=p$$

where $MB(q)$ is the marginal benefit of an additional unit of care.

If insurance requires the patient to pay only a proportion $c$ of the price, where $c$ is at least zero and less than one, the price the patient faces per unit falls to a fraction of the resource cost.

$$p_{\text{patient}}=cp$$

where $p_{\text{patient}}$ is the out-of-pocket price per unit and $c$ is the coinsurance rate.

The patient then chooses the quantity of care at which marginal benefit equals the out-of-pocket price.

$$MB(q)=cp$$

Here the patient responds to the out-of-pocket price $cp$ rather than the resource cost $p$. Because $cp<p$, the insured quantity may be greater than the quantity selected at the full price. The difference represents the utilisation response to insurance. It does not establish that every additional service is unnecessary or socially inefficient.

Moral Hazard and the Value of Insurance

Insurance creates an important trade-off. It protects people against uncertain and potentially unaffordable costs, but it can also weaken the connection between healthcare use and the resource cost of providing that care.

Full insurance maximises financial protection but may produce stronger utilisation incentives. Greater cost sharing can reduce use but also exposes patients to more financial risk and may discourage high-value as well as low-value care.

The central policy problem is therefore to balance several objectives rather than to eliminate moral hazard altogether. The objectives usually weighed are listed below.

• protection against financial risk;
• access to beneficial healthcare;
• incentives for appropriate prevention and utilisation;
• administrative and monitoring costs; and
• the efficient use of limited healthcare resources.

Welfare Effects

A conventional analysis identifies a welfare loss when insurance causes consumption of care for which the patient's marginal benefit is below the full social cost. This can occur because the patient responds to the out-of-pocket price rather than the total cost of treatment. Pauly used this reasoning to conclude that, even when everyone is risk averse, some uncertain medical expenses would not be insured in an optimal arrangement.

If marginal benefit falls linearly with quantity and each unit costs $p$ to provide, the loss from the additional care is a triangle between the resource cost and the marginal benefit curve, given by the expression below.

$$WL=\frac{1}{2}(p-cp)(q_1-q_0)$$

where $WL$ is the welfare loss, $q_0$ is the quantity chosen at the full price $p$ and $q_1$ is the quantity chosen at the insured price $cp$. Each additional unit between $q_0$ and $q_1$ costs $p$ to provide but is valued by the patient at between $cp$ and $p$, so the triangle is the deadweight loss of the price reduction. Feldman and Dowd later published a new estimate of the welfare loss of excess health insurance, building on this reasoning.

Nyman challenged the conventional analysis by arguing that insurance is valued partly as a means of gaining access to care that would otherwise be unaffordable. On this view, some of the additional use that the conventional analysis counts as moral hazard is care that insurance makes affordable, and the value of that care is a gain from insurance rather than a loss.

Additional healthcare use after gaining insurance is therefore not necessarily inefficient. Insurance can enable access to effective care that patients could not otherwise afford, improve health, reduce uncertainty and transfer resources to people when they are ill.

Assessing the welfare effect therefore requires more than measuring whether spending increased. Analysts must consider the health benefits of the additional care, financial protection, distributional consequences and any costs imposed on other members of the insurance pool or health system.

Behavioural Hazard and Demand-Based Welfare Estimates

The welfare triangle rests on the assumption that the demand curve measures the patient's marginal benefit, so that the out-of-pocket price equals the health benefit of the last unit consumed. Baicker, Mullainathan and Schwartzstein questioned this assumption in a 2015 paper that introduced the term "behavioral hazard" (behavioural hazard in UK spelling). They argued that people also misuse care through mistakes, so that much high-value care is underused even when patient costs are low and some useless care is bought even when patients face the full cost.

When behavioural hazard is present, welfare calculations that use only the demand curve can be wrong by orders of magnitude or even have the wrong sign. A fall in use after a copayment rise may then reflect forgone high-value care rather than the removal of low-value care. The authors derive optimal copayments that depend on health responses as well as demand responses, and present this as a theoretical foundation for value-based insurance design.

Demand-Side and Supply-Side Responses

Demand-side moral hazard concerns changes in patient behaviour, such as increased healthcare use when out-of-pocket prices fall. It is the form measured by most insurance experiments.

Healthcare expenditure may also change through decisions made by clinicians or providers. Arrow observed that a physician's certification of need acts as only an imperfect check on insured demand, because physicians may find it convenient or pleasing to patients to prescribe more expensive care. A provider may influence the amount or type of care delivered when patients and payers cannot fully assess clinical necessity. This is more precisely examined through concepts such as supplier-induced demand, agency problems and payment incentives rather than being attributed automatically to patient moral hazard.

Ellis and McGuire distinguish demand-side cost sharing, in which patients pay more through copayments or deductibles, from supply-side cost sharing, which changes the incentives of providers through the way they are paid. Insurance arrangements may create interacting incentives for patients, providers and payers. A complete analysis should identify which party controls the relevant decision and who bears its consequences.

Managing Moral Hazard

Insurers and health systems respond to moral hazard by changing the price patients face, the incentives providers face or the rules governing access to care. Each mechanism targets a different decision-maker. The common mechanisms are listed below.

• deductibles, copayments and coinsurance;
• coverage limits and exclusions;
• prior authorisation and utilisation review;
• clinical guidelines and prescribing controls;
• provider payment methods;
• managed-care arrangements;
• incentives for prevention and adherence; and
• value-based insurance design.

These mechanisms involve trade-offs. Cost sharing may reduce unnecessary care, but it can also reduce effective treatment, worsen equity and impose financial hardship. Administrative controls may improve targeting but create delays, complexity and additional costs.

Evidence from the RAND Health Insurance Experiment shows why the trade-offs matter. Cost sharing reduced the use of highly effective and rarely effective care to a similar degree, so it did not selectively remove low-value care, and it influenced care seeking more among people on low incomes than among those with greater means, deterring care for poor children in particular.

Value-based insurance design attempts to vary cost sharing according to the expected clinical value of care rather than applying the same charge to every service. It encourages the use of services whose clinical benefits exceed their costs and discourages the use of services whose benefits do not justify their costs.

Evidence and Measurement

Moral hazard cannot be measured simply by comparing insured and uninsured people because their health risks and preferences may differ. This creates a selection problem.

Stronger empirical designs use randomised insurance arrangements, natural experiments, policy changes or credible quasi-experimental methods to isolate the effect of coverage or cost sharing on behaviour. Outcomes studied in this work may cover the measures listed below.

• healthcare utilisation;
• total and out-of-pocket spending;
• use of preventive services;
• treatment adherence;
• health outcomes;
• financial protection; and
• responses across different income or health-risk groups.

The RAND Health Insurance Experiment randomised people to insurance plans with different levels of cost sharing. Manning and colleagues reported in 1987 that a catastrophic insurance plan reduced expenditures by 31 per cent relative to free care, and estimated a price elasticity of demand for medical care of approximately $-0.2$. Demand is therefore inelastic but responsive to price, which is the central empirical evidence for ex post moral hazard. Analysis of health outcomes in the same experiment found that free care improved blood pressure control for low-income adults with high blood pressure and improved vision for people with poor vision, while no significant effects were detected for the average participant on eight other measures of health status and health habits.

The Oregon Health Insurance Experiment used a lottery to allocate the chance to apply for Medicaid, creating a randomised comparison of coverage for low-income adults. In the first year, people selected by the lottery were about 25 percentage points more likely to be insured and had higher use of primary care, preventive care and hospital care, lower out-of-pocket spending and medical debt, and better self-reported physical and mental health. After about two years, coverage produced no significant improvement in measured physical health outcomes, but it raised diabetes detection and management, lowered rates of depression and reduced financial strain.

A rise in utilisation provides evidence of a behavioural response to insurance, but additional analysis is needed to determine whether the response improves or reduces welfare.

Relationship to Information Asymmetry

Moral hazard is associated with hidden action. It arises because one party cannot perfectly observe or control behaviour that occurs after protection or a contract is in place.

Adverse selection is associated with hidden information. It arises when individuals possess relevant information about their risk before contracting, affecting who purchases coverage or the terms on which exchange occurs.

Both can result from information asymmetry, but they describe different mechanisms and require different policy responses. Higher spending in a generous plan can reflect either mechanism, so empirical work has to separate the risks people brought into the plan from the extra use the plan caused.

Applications Beyond Patient Insurance

The same incentive structure can occur outside individual health-insurance decisions. Wherever one party is shielded from the costs of its own decisions and another party cannot fully monitor them, a similar analysis applies. Examples are given below.

• providers protected from the financial consequences of inefficient practice;
• organisations expecting rescue when financial losses occur;
• manufacturers facing limited liability for certain risks;
• purchasers reimbursed by a separate budget holder; and
• institutions whose decisions impose costs on a wider system.

In each case, analysis should identify the protected party, the action that can change, the party bearing the resulting cost and the information preventing complete contracting.

Moral Hazard in Insurance and Payment Design Decisions

Moral hazard is best understood as a consequence of incomplete contracts, imperfect observability and shared costs. It explains why protection against risk can change behaviour, but it does not establish that insurance is undesirable or that every behavioural response is inefficient.

The relevant health-economic question is whether an insurance or payment arrangement achieves an appropriate balance between financial protection, access, health outcomes, incentives and resource use.

Sources

• Arrow KJ. Uncertainty and the welfare economics of medical care. American Economic Review. 1963;53(5):941-973.
• Pauly MV. The economics of moral hazard: comment. American Economic Review. 1968;58(3):531-537.
• Zweifel P, Manning WG. Moral hazard and consumer incentives in health care. In: Culyer AJ, Newhouse JP, eds. Handbook of Health Economics. Vol 1. Amsterdam: Elsevier; 2000:409-459.
• Feldman R, Dowd B. A new estimate of the welfare loss of excess health insurance. American Economic Review. 1991;81(1):297-301.
• Nyman JA. The value of health insurance: the access motive. Journal of Health Economics. 1999;18(2):141-152.
• Baicker K, Mullainathan S, Schwartzstein J. Behavioral hazard in health insurance. Quarterly Journal of Economics. 2015;130(4):1623-1667.
• Ellis RP, McGuire TG. Supply-side and demand-side cost sharing in health care. Journal of Economic Perspectives. 1993;7(4):135-151.
• Chernew ME, Rosen AB, Fendrick AM. Value-based insurance design. Health Affairs. 2007;26(2):w195-w203.
• Manning WG, Newhouse JP, Duan N, Keeler EB, Leibowitz A, Marquis MS. Health insurance and the demand for medical care: evidence from a randomized experiment. American Economic Review. 1987;77(3):251-277.
• Brook RH, Ware JE, Rogers WH, et al. Does free care improve adults' health? Results from a randomized controlled trial. New England Journal of Medicine. 1983;309(23):1426-1434.
• Lohr KN, Brook RH, Kamberg CJ, et al. Use of medical care in the Rand Health Insurance Experiment: diagnosis- and service-specific analyses in a randomized controlled trial. Medical Care. 1986;24(9 Suppl):S1-S87.
• Finkelstein A, Taubman S, Wright B, et al. The Oregon Health Insurance Experiment: evidence from the first year. Quarterly Journal of Economics. 2012;127(3):1057-1106.
• Baicker K, Taubman SL, Allen HL, et al. The Oregon experiment: effects of Medicaid on clinical outcomes. New England Journal of Medicine. 2013;368(18):1713-1722.