Moral hazard is a problem in economics in which a person or organization has incentives to take actions whose costs or risks are partly borne by another party. It commonly arises when insurance or guarantees reduce the consequences of losses, or when one party cannot adequately observe or contract over another’s behavior. The concept concerns how arrangements change incentives: an insured property owner, for example, may take fewer precautions because the insurer bears part of any resulting loss. (imf.org)
Mechanism and scope
Moral hazard combines a conflict of interests with limits on observation or enforcement. In contract theory, it is commonly modeled as a principal–agent problem: a principal delegates a task to an agent whose actions affect both parties’ interests. The principal may observe the outcome without observing the effort, care, or risk-taking that produced it. This is a form of information asymmetry. Even an observable action can generate contractual difficulties if a court cannot verify it. (nobelprize.org)
For example, company owners may observe profits but not every managerial decision. Profits also depend on market conditions and other influences outside management’s control. Consequently, poor results do not necessarily establish low effort, while favorable results do not necessarily establish good decisions. The contractual problem is to reward desirable behavior using imperfect evidence about it. (nobelprize.org)
Distinction from adverse selection
Moral hazard differs from adverse selection. In the standard distinction, adverse selection concerns private information about characteristics relevant to a transaction, such as a borrower’s underlying creditworthiness. Moral hazard concerns actions taken under an arrangement, such as how carefully the borrower manages an investment after receiving funds. The distinction is often described as hidden characteristics versus hidden actions, rather than simply as information before versus after a contract. (imf.org)
Both can operate simultaneously. People choosing more generous coverage may have greater underlying risks, may respond more strongly to coverage, or both. Therefore, a correlation between insurance coverage and claims does not by itself identify moral hazard. Research also examines “selection on moral hazard,” in which people choose coverage partly according to their anticipated behavioral response to it. (nber.org)
Formal analysis and intellectual development
A standard model represents an observable outcome as
where is the agent’s action and is a random variable representing outside influences. Each action induces a probability distribution over outcomes. A contract specifies payment conditional on verifiable outcomes rather than directly on the hidden action. (nobelprize.org)
The principal’s optimization problem includes two important restrictions. A participation constraint requires the contract to be acceptable to the agent. An incentive-compatibility constraint requires the intended action to be optimal for the agent under that contract. Incentive payments can encourage effort, but they also expose the agent to uncertainty. Where the agent is risk-averse, this creates a trade-off between incentives and risk sharing. (nobelprize.org)
The modern theoretical treatment developed through work by Kenneth Arrow, James Mirrlees, and other economists. Bengt Holmström’s 1979 analysis established influential results about observable performance measures, including the informativeness principle: compensation should incorporate information useful for inferring the agent’s actions. Holmström and Oliver Hart received the 2016 Nobel Memorial Prize in Economic Sciences for contributions to contract theory. (nobelprize.org)
Insurance and employment
In insurance, protection can weaken incentives to prevent losses. A homeowner who bears only part of a theft loss may invest less in security than an otherwise comparable uninsured homeowner. Deductibles, under which the policyholder bears an initial portion of a covered loss, retain some exposure to consequences. Risk-based premiums can also connect the price of protection more closely to observable risk. (imf.org)
In employment, performance-related compensation can align an employee’s interests with an employer’s. However, stronger incentives are not invariably better. If output quantity is easy to measure while quality is difficult to measure, rewarding quantity heavily may redirect effort away from quality. Multitask models explain why relatively weak explicit performance incentives can sometimes be appropriate, and why job design and promotion systems also matter. (nobelprize.org)
Banking and financial safety nets
Deposit insurance protects depositors and can reduce the likelihood of a bank run. It can also weaken depositors’ incentives to scrutinize a bank and reduce the funding consequences of greater risk-taking. The resulting moral hazard concerns the separation between private incentives and losses borne by the insurance arrangement. (federalreserve.gov)
Bank capital and liquidity requirements are mechanisms for limiting these risks. Their analysis involves costs as well as benefits: requirements can constrain banks’ liquidity provision or alter investment while reducing excessive credit risk and liquidity risk. Expectations of crisis assistance can likewise affect incentives when investors believe public intervention will absorb part of future losses. This is a potential consequence of a safety net, not proof that every intervention produces excessive risk-taking. (federalreserve.gov)
Empirical identification
Estimating moral hazard requires separating behavioral responses from differences between participants. Causal inference methods include randomized assignment, changes in available contracts, and other sources of variation in coverage or prices. Research on health insurance, for example, finds that out-of-pocket prices affect healthcare spending. Identifying such responses is distinct from determining the value of the additional services consumed. Models can complement experimental evidence when researchers examine contracts not directly observed in their data. (nber.org)