Loading field map
Why do standard competitive market models fail to explain how healthcare is bought and sold? Patients rarely know what treatment they need, physicians control both diagnosis and supply, and insurance insulates consumers from the true cost of care. These features create a deep tension: the same institutions that protect patients from financial risk also distort the incentives that make markets efficient. Since the 1960s, health economists have built a sequence of frameworks to analyze this tension, moving from a general diagnosis of market failure toward increasingly specific models of institutional behavior, market design, and empirical competition analysis.
Kenneth Arrow's 1963 article "Uncertainty and the Welfare Economics of Medical Care" established the starting point for all later work on healthcare markets. Arrow argued that healthcare is fundamentally different from the textbook competitive market because of pervasive uncertainty and asymmetric information. Patients cannot predict their need for care, cannot easily evaluate the quality of the care they receive, and must rely on physicians as both advisors and providers. These conditions violate the assumptions under which competitive markets guarantee efficient outcomes.
Arrow's framework did not simply catalog imperfections. It identified a specific welfare logic: when information is unevenly distributed and risk is unavoidable, nonmarket institutions such as professional ethics, nonprofit ownership, and licensure can improve social welfare. The framework remains active today as a baseline against which all later models measure themselves. Every subsequent framework in healthcare markets accepts Arrow's core insight that healthcare is not a standard commodity, but they differ sharply on what follows from that insight.
Arrow's diagnosis was powerful but abstract. It told economists that healthcare markets were different, but it did not provide a behavioral model of the key actors—hospitals and physicians. Two frameworks in the 1970s filled this gap by giving institutional content to Arrow's claims.
In 1970, Joseph Newhouse proposed an economic model of the nonprofit hospital that treated it as maximizing a combination of quantity and quality of care, subject to a break-even constraint. Unlike a profit-maximizing firm, a nonprofit hospital had no residual claimants; its managers could pursue objectives such as prestige, technological sophistication, or community service. This framework coexisted with Arrow's market failure analysis by providing a specific mechanism for why nonprofit ownership might arise in healthcare: if patients cannot judge quality, a nonprofit's lack of profit motive can serve as a trust signal.
The nonprofit behavior framework narrowed Arrow's broad institutional argument into a testable model. It predicted that nonprofit hospitals would compete on quality and technology rather than price, and that they would cross-subsidize unprofitable services. Later empirical work showed that the distinction between nonprofit and for-profit hospitals is often blurrier than the model assumes, but the framework remains influential for understanding hospital objectives, especially in policy debates about tax exemptions and community benefits.
Arrow had noted that physicians act as agents for their patients, but he did not fully model the conflict of interest inherent in that relationship. The physician agency framework, developed by Mark Pauly and others in the mid-1970s, formalized the problem: a physician is both the patient's advisor and the seller of services, creating an incentive to recommend more care than a fully informed patient would choose. This is the supplier-induced demand (SID) hypothesis.
The SID framework transformed Arrow's information asymmetry into a principal-agent model with a specific empirical prediction: when physician supply increases, physicians may respond by shifting the demand curve for their services outward rather than lowering prices. The framework generated decades of debate. Critics argued that professional ethics, reputation concerns, and patient resistance limit inducement; proponents pointed to evidence from fee-for-service settings where physicians have clear financial incentives to provide additional procedures. This living disagreement persists because the extent of inducement is difficult to measure cleanly—a challenge that later frameworks would take up with new empirical tools.
By the 1980s, healthcare costs in the United States were rising rapidly, and policymakers began looking for market-based solutions that could contain costs without sacrificing quality. The frameworks that emerged in this period accepted Arrow's diagnosis of market failure but argued that carefully designed competition could outperform direct regulation.
Alain Enthoven's managed competition framework, developed in the late 1980s, proposed a middle ground between a fully regulated system and an unregulated market. The key innovation was the "sponsor"—a purchasing cooperative or government entity that structures the market on behalf of a large group of subscribers. The sponsor establishes rules for plan participation, manages enrollment, creates price-elastic demand by limiting employer premium contributions, and adjusts payments to compensate for differences in enrollee risk.
Managed competition was a market design framework, not a descriptive model of existing markets. It absorbed the Arrowian concern about information asymmetry by giving the sponsor the role of evaluating plans and managing risk selection. It also built on the nonprofit behavior framework by recognizing that many health plans were nonprofit and that their objectives mattered for market performance. The framework directly influenced the design of Medicare Advantage and the Affordable Care Act exchanges, making it one of the most policy-relevant frameworks in the subfield.
Managed competition assumed that a sponsor could manage risk selection effectively. But what if selection is so severe that competition itself becomes destructive? The health insurance competition and adverse selection framework, developed by David Cutler and Sarah Reber in the 1990s, analyzed exactly this problem. It showed that when insurers compete for enrollees, they have strong incentives to avoid sick patients by designing benefit packages that attract the healthy. This risk segmentation can unravel pooling and lead to market failure even when the sponsor is present.
This framework stands in a complementary but tense relationship with managed competition. Managed competition is a design proposal; adverse selection analysis is a diagnosis of the main obstacle to that design. The adverse selection framework revived Arrow's original concern about information asymmetry but gave it a precise mechanism: insurers know less about enrollees' health status than enrollees themselves do, and competition exploits rather than corrects this asymmetry. The framework remains central to evaluating insurance market reforms, including risk adjustment, reinsurance, and the individual mandate.
The frameworks up to this point were largely theoretical or reduced-form empirical. They identified incentives and predicted behavior, but they did not estimate market power or simulate the effects of mergers with the rigor available in industrial organization. The healthcare industrial organization (IO) framework, emerging around 1999, brought structural econometric models to healthcare markets.
Healthcare IO uses detailed data on hospital networks, insurer negotiations, and patient choice to estimate demand elasticities, marginal costs, and bargaining power. It builds on the earlier competition frameworks by providing tools to test their claims. For example, where the physician agency framework debated whether inducement exists, healthcare IO can estimate how physician market power affects prices and quantities. Where managed competition assumed sponsors could manage selection, healthcare IO models the strategic behavior of insurers and hospitals in response to payment rules.
This framework transformed the subfield by making empirical competition analysis the dominant method for evaluating mergers, antitrust policy, and payment reform. It coexists with the older frameworks rather than replacing them: Arrowian market failure remains the conceptual foundation, nonprofit behavior models still inform hospital objective functions, and adverse selection analysis is now estimated with structural models rather than simple correlations.
Today, all six frameworks remain active, but they occupy different roles. Arrowian market failure provides the overarching justification for studying healthcare markets as a special case. Nonprofit hospital behavior and physician agency offer behavioral models of the key institutions. Managed competition and adverse selection analysis frame the policy debate about how to structure insurance markets. Healthcare IO supplies the empirical toolkit for measuring competition and evaluating policy.
The major agreements across these frameworks are striking. All accept that information asymmetry is central to healthcare markets. All recognize that standard competitive models cannot be applied without modification. And all agree that institutions—ownership form, professional norms, regulation—matter for market performance.
The disagreements are equally important. The deepest divide is between those who believe that carefully designed competition can correct market failure and those who believe that market failure is so pervasive that nonmarket institutions must play a leading role. This tension runs through every framework: Arrow's original skepticism about for-profit healthcare, the nonprofit behavior model's trust-signaling logic, the managed competition sponsor's corrective role, and the adverse selection framework's demonstration that competition can make things worse. A second disagreement concerns the extent of supplier-induced demand—a debate that remains unresolved because the empirical evidence is sensitive to identification strategy. A third disagreement is methodological: reduced-form approaches that estimate average treatment effects coexist with structural models that recover deep parameters, and each camp questions the assumptions of the other.
Healthcare markets remain a laboratory for economic thinking about information, institutions, and competition. The frameworks developed over the past six decades have not converged on a single answer, but they have given economists a rich set of tools for asking better questions.