Payment Models and the Logic of Incentives

You Get What You Pay For — and America Has Been Paying for the Wrong Things

Every payment model in healthcare is a theory of behavior. It encodes an assumption about what providers will do when money flows in a particular direction. Fee-for-service assumes that paying for activity produces necessary activity. Capitation assumes that paying for coverage produces efficient care. Value-based payment assumes that paying for outcomes produces better outcomes. Population-based funding assumes that paying for community health produces healthier communities.

None of these assumptions are entirely right. None are entirely wrong. Each payment model produces the behaviors it incentivizes — and also produces the distortions, gaming, and unintended consequences that incentive theory predicts. The history of healthcare payment in America is a history of discovering what each model optimizes for, what it fails to address, and what it accidentally destroys.

This document traces that history through the intellectual traditions that explain it. The arc bends from transaction to relationship to community — from paying for what a provider does, to paying for what a patient experiences, to paying for what a population becomes. Whether it arrives at that destination is the open question of this generation of health policy.

I. Fee-for-Service: The Volume Engine

The Logic

Fee-for-service (FFS) is the oldest and most intuitive payment model: the provider performs a service, the payer pays for it. Each encounter, procedure, test, and visit generates a claim. Revenue is a direct function of activity volume. The model is simple, transparent, and aligned with the provider’s economic interest in delivering care.

The Intellectual Framework

The economics of fee-for-service are straightforward principal-agent theory. Kenneth Arrow’s 1963 paper identified the core problem: the asymmetry of information between physician (agent) and patient (principal) means the agent largely determines the quantity and nature of services consumed. When the agent is paid per service, the information asymmetry creates a structural incentive to oversupply — to order the additional test, perform the additional procedure, schedule the additional visit. This is not fraud. It is the rational behavior of economic actors within the incentive structure they face.

Mark Pauly’s foundational work on moral hazard in health insurance (1968, “The Economics of Moral Hazard,” American Economic Review) extended Arrow’s insight to the demand side: when patients are insured, the marginal cost of consuming healthcare falls to zero or near-zero at the point of service. Patients have no economic reason to decline recommended care. The combination of supply-side incentive to oversupply and demand-side insensitivity to cost produces what Victor Fuchs and others called the “medical arms race” — a self-reinforcing cycle of capacity expansion, technology adoption, and utilization growth that characterized American healthcare from the 1960s through the 1990s.

The Rural Distortion

Fee-for-service in rural settings produces a specific pathology. The volume-revenue relationship that sustains urban and suburban providers requires patient volume that rural communities cannot generate. A specialist in a metropolitan area can see 25 patients per day and generate revenue sufficient to cover overhead, staff, and profit. The same specialist in a rural community of 5,000 people may see 8 patients per day. The economics do not work — not because the provider is inefficient, but because the population base is insufficient to sustain volume-dependent revenue.

This is why rural providers have always required supplemental payment: Medicare’s Critical Access Hospital designation (cost-based reimbursement rather than DRG prospective payment), FQHC Prospective Payment System rates (which exceed standard Medicaid FFS rates), and various state-level supplemental payments. These are patches on a payment model that structurally disadvantages low-volume providers. They acknowledge, implicitly, that fee-for-service is a population-density-dependent model being applied to settings where population density is the defining constraint.

The Legacy

Fee-for-service built the American healthcare system as it exists. It built the hospitals, trained the physicians, developed the technology, and created the specialty-dominated delivery model that produces extraordinary acute care and mediocre population health. The model’s legacy is physical infrastructure (hospitals, clinics, imaging centers) organized around revenue-generating encounters — a built environment that subsequent payment models must either repurpose or work around.

II. Prospective Payment: The Efficiency Mandate

The Logic

Medicare’s shift to prospective payment in 1983 — paying hospitals a fixed amount per diagnosis (DRG) rather than reimbursing their actual costs — was the first systematic attempt to break the volume-revenue link. The logic is straightforward: if the hospital receives the same payment regardless of how many days the patient stays or how many tests are ordered, the hospital has an incentive to minimize resource use per case. Efficiency, not volume, becomes the revenue strategy.

The Intellectual Framework

Prospective payment is an application of mechanism design theory — the branch of economics concerned with designing incentive structures that produce desired behaviors from self-interested agents. The DRG system was developed by Robert Fetter and John Thompson at Yale in the late 1970s, originally as a utilization review tool. Its adoption as a payment mechanism by HCFA (now CMS) in 1983 was the largest-scale implementation of mechanism design in healthcare history.

The theory predicted, correctly, that prospective payment would reduce lengths of stay, decrease ancillary service utilization, and improve hospital operating efficiency. It also predicted, correctly, that the system would create new forms of strategic behavior: upcoding (classifying patients into higher-paying DRGs), cream-skimming (preferring patients whose expected costs fall below the DRG payment), and cost-shifting (transferring unreimbursed costs to other payers, particularly commercial insurers).

Michael Porter and Elizabeth Olmsted Teisberg’s Redefining Health Care (2006) argued that the fundamental problem with prospective payment — shared with fee-for-service — is that it pays for the wrong unit. Fee-for-service pays per encounter. Prospective payment pays per episode. Neither pays for what actually matters to the patient: the outcome. A hospital that discharges a patient quickly (reducing cost per DRG) but produces a readmission within 30 days has optimized for the payment model while failing the patient. The unit of payment determines the unit of optimization, and neither model makes health the unit.

The Rural Distortion

The DRG system was calibrated on urban hospital cost data. Rural hospitals, with lower volumes, higher per-unit fixed costs, and sicker patient populations (older, more chronic conditions, fewer preventive care options), found that DRG payments systematically underpaid their actual costs. The Critical Access Hospital (CAH) program, created in 1997, exempted small rural hospitals from DRGs and returned them to cost-based reimbursement — an acknowledgment that prospective payment’s efficiency logic requires a scale that rural hospitals cannot achieve.

This created a two-tier system within Medicare: urban hospitals incentivized toward efficiency through DRGs, rural hospitals incentivized toward cost documentation through cost reports. The CAH designation preserved rural hospitals but also froze them in a reimbursement model that rewards cost generation rather than value creation. The organizational implications — documented in the companion piece on organizational economics — are significant: CAHs developed institutional competencies around cost reporting, not around the outcome measurement and care redesign that subsequent payment models would demand.

III. Managed Care and Capitation: The Coverage Contract

The Logic

Capitation inverts the fee-for-service incentive. Instead of paying per service (which rewards volume), the payer pays a fixed amount per member per month (PMPM), and the provider or plan manages all care within that budget. Revenue is fixed. Cost is variable. The provider’s economic interest shifts from maximizing services to minimizing unnecessary utilization while keeping the patient healthy enough to avoid expensive interventions.

The Intellectual Framework

Alain Enthoven’s theory of managed competition (1978, “Consumer-Choice Health Plan,” New England Journal of Medicine; expanded in Health Plan, 1980) provided the intellectual architecture. Enthoven, drawing on microeconomic theory and his experience with systems analysis at the Department of Defense, argued that healthcare markets fail not because competition is inherently unsuitable for healthcare, but because the existing market structure lacks the institutional framework to make competition work. Managed competition would create that framework: standardized benefit packages enabling apples-to-apples comparison, risk-adjusted capitation preventing cream-skimming, and informed consumer choice creating pressure for quality and efficiency.

The theory is elegant. The implementation was messier. The managed care revolution of the 1990s produced HMOs and managed care organizations that achieved cost reductions but generated intense consumer and provider backlash. The backlash was not irrational. Capitation creates incentives to underserve — to deny care, restrict referrals, and narrow networks — that are the mirror image of fee-for-service’s incentive to overserve. Joseph Newhouse’s RAND Health Insurance Experiment (completed 1982, published in Free for All?, 1993) had demonstrated that cost-sharing reduces utilization, including necessary utilization, and that the reduction harms the health of the poorest and sickest. Capitation, taken to its logical extreme, produces the same result through supply-side restriction rather than demand-side cost-sharing.

James Robinson’s “The End of Managed Care” (JAMA, 2001) documented the political and market dynamics that constrained managed care’s cost-control ambitions. Employers, the primary purchasers of commercial insurance, responded to employee dissatisfaction by demanding broader networks and fewer restrictions — effectively purchasing the appearance of managed care at fee-for-service prices. The managed care revolution achieved a one-time cost reduction but did not produce the durable restructuring of care delivery that Enthoven’s theory envisioned.

The Rural Distortion

Capitation assumes a population base large enough to pool risk. The actuarial mathematics of capitation — setting a PMPM rate that covers expected costs with acceptable variance — require thousands of enrollees. In rural communities, the eligible population may be too small for actuarial risk pooling to function. A single catastrophic case (a premature infant requiring NICU care, a major trauma requiring air transport) can consume a disproportionate share of a small capitated budget.

Moreover, capitation assumes provider choice — the ability of the plan to construct a network and the patient to select among providers. In rural communities, there may be one hospital, one primary care practice, and no specialists within 60 miles. The competitive dynamics that Enthoven’s managed competition requires — multiple plans competing on quality and efficiency — cannot operate in markets with one provider. Capitation in rural settings is not managed competition. It is a budget constraint imposed on a monopoly provider.

Medicaid managed care’s expansion into rural areas in the 2000s and 2010s produced exactly these tensions. Rural providers accepted Medicaid MCO contracts because Medicaid was a significant revenue source, but the MCO administrative requirements (prior authorization, utilization review, network credentialing) imposed costs that exceeded any efficiency gains from care management. The intermediary layer — the MCO standing between the state Medicaid dollar and the rural provider — extracted administrative fees without adding value proportional to its cost. This is the intermediary extraction problem that pervades American healthcare finance: every layer between the funding source and the care delivery point takes a cut.

IV. Value-Based Care: Paying for Outcomes

The Logic

Value-based payment (VBP) attempts to solve the problem that both fee-for-service and capitation share: neither pays for health. Fee-for-service pays for activity. Capitation pays for coverage. Value-based payment pays for outcomes — or more precisely, it adjusts payment based on measured quality and efficiency metrics, creating financial incentives for providers to improve outcomes while controlling costs.

The Intellectual Framework

The conceptual foundation is Donald Berwick, Thomas Nolan, and John Whittington’s “Triple Aim” (2008, Health Affairs): simultaneously improving the individual experience of care, improving the health of populations, and reducing the per capita cost of healthcare. The Triple Aim reframed the healthcare value proposition from volume or coverage to a three-dimensional optimization problem. Porter’s concept of “value” in healthcare — health outcomes achieved per dollar spent (New England Journal of Medicine, 2010) — provided the measurement framework: value is a ratio, and payment should reward organizations that improve the numerator (outcomes) relative to the denominator (cost).

The practical instruments of value-based payment include:

Pay-for-performance (P4P). Financial bonuses or penalties tied to quality metrics. Medicare’s Hospital Value-Based Purchasing Program (2012) and the Merit-based Incentive Payment System (MIPS) for physicians (2017) are the largest P4P programs. The evidence on P4P is mixed. Ashish Jha and colleagues at Harvard have documented that P4P tends to reward already-high-performing organizations and penalize those with the fewest resources — effectively redistributing payment from safety-net providers to well-resourced systems. The metrics themselves are contested: process measures (did the patient receive recommended care?) are measurable but weakly correlated with outcomes. Outcome measures (did the patient get better?) are meaningful but confounded by patient severity, social determinants, and factors outside provider control.

Bundled payments. A single payment for an entire episode of care (e.g., a joint replacement, from pre-surgical evaluation through 90 days of post-surgical recovery). The Bundled Payments for Care Improvement (BPCI) initiative and its successor BPCI Advanced have demonstrated savings for some episodes, particularly those with high variation in utilization. The intellectual logic is Porter’s: by making the episode (not the encounter or the DRG) the unit of payment, the model incentivizes coordination across the care continuum. The challenge is that episodes are defined administratively, not clinically. Where an episode begins and ends is a design choice that determines who bears risk and who captures savings.

Shared savings. Provider organizations accept accountability for the total cost and quality of care for a defined population. If they reduce costs below a benchmark while meeting quality thresholds, they share the savings with the payer. The Medicare Shared Savings Program (MSSP) and Pioneer ACO model (later Next Generation ACO, now ACO REACH) are the primary vehicles. Elliott Fisher and colleagues at Dartmouth — whose research on geographic variation in Medicare spending (Annals of Internal Medicine, 2003) demonstrated that higher spending does not produce better outcomes — provided the empirical foundation for the shared savings model. If spending varies and higher spending doesn’t help, then reducing spending without harming outcomes should be possible, and providers who achieve it should be rewarded.

The Rural Distortion

Value-based payment models depend on measurement infrastructure — electronic health records, quality reporting systems, data analytics capability, care coordination staff — that rural providers disproportionately lack. The fixed costs of measurement and reporting are spread across fewer patients, making the per-patient cost of value-based care participation higher for low-volume rural providers than for large urban systems.

The statistical requirements are also problematic. Quality measurement requires sufficient sample size to distinguish signal from noise. A Critical Access Hospital performing 50 joint replacements per year cannot produce statistically meaningful outcome data at the individual facility level. Rural providers are measured on the same metrics as urban systems with ten times the volume, but their data is noisier, their confidence intervals wider, and their performance scores more susceptible to random variation. The result: rural providers are more likely to appear as outliers — both high and low — not because their care is more variable but because their sample sizes are smaller.

The deeper problem is that value-based payment assumes the care delivery system is the primary determinant of health outcomes. For rural populations, social determinants — poverty, food access, housing quality, transportation, educational attainment, environmental exposure — may account for more outcome variation than clinical care. Paying providers for outcomes that are substantially determined by factors outside their control is not value-based payment. It is misattributed accountability.

V. Accountable Care and the Population Turn

The Logic

Accountable Care Organizations (ACOs) represent the organizational vehicle for value-based payment at population scale. The concept: a group of providers accepts collective responsibility for the cost and quality of care for a defined population. The ACO is not a payer. It is not an insurer. It is a provider-led entity that coordinates care across settings and shares financial risk with the payer.

The Intellectual Framework

The ACO concept emerged from Fisher and colleagues’ research on “extended hospital medical staffs” — the networks of physicians who naturally cluster around hospitals and collectively account for the utilization patterns of a geographically defined population. Fisher, Staiger, Bynum, and Gottlieb (“Creating Accountable Care Organizations,” Health Affairs, 2007) proposed that these naturally occurring referral networks could be formalized as ACOs and held accountable for the cost and quality of care they collectively deliver.

The intellectual lineage runs through several traditions:

Systems thinking. W. Edwards Deming’s work on quality management — adopted in healthcare through the Institute for Healthcare Improvement (IHI) and Berwick’s leadership — argued that quality is a property of the system, not of individual actors. ACOs operationalize this insight: quality and cost outcomes emerge from how the system coordinates, not from how individual providers perform in isolation. This connects directly to the emergent systems perspective: population health outcomes are emergent properties of complex adaptive systems, not the sum of individual clinical decisions.

Integrated delivery systems. The Kaiser Permanente model — combining insurance, hospital, and physician functions in a single organization — has long been cited as evidence that integration produces better outcomes at lower cost. Stephen Shortell’s research on integrated delivery systems (beginning with Remaking Health Care in America, 1996, with colleagues) examined the organizational requirements for integration and found that structural integration alone is insufficient — cultural integration, information system integration, and aligned financial incentives are also necessary. Most ACOs attempt functional integration (coordinating across independently owned organizations) rather than structural integration (merging into a single organization), which is organizationally harder and slower.

Common-pool resource governance. Elinor Ostrom’s framework for commons governance, discussed at length in the companion document on organizational economics, is directly relevant here. An ACO managing population health is governing a commons: the shared resource of healthcare capacity within a community. Ostrom’s design principles — clear boundaries, proportional cost-benefit distribution, collective choice arrangements, monitoring, graduated sanctions — describe the governance requirements that ACOs must meet to be sustainable. Most ACOs meet few of them.

Accountable Communities of Health

Washington State’s Accountable Communities of Health (ACHs) and Oregon’s Coordinated Care Organizations (CCOs) represent the most ambitious attempts to extend accountability beyond the clinical delivery system to the broader determinants of health. ACHs are regional, multi-sector entities that bring together healthcare providers, social service organizations, public health agencies, tribal nations, and community members to collectively address health outcomes for a geographic population.

The intellectual foundation draws on David Kindig and Greg Stoddart’s influential definition of population health (“What is Population Health?”, American Journal of Public Health, 2003): “the health outcomes of a group of individuals, including the distribution of such outcomes within the group.” This definition deliberately includes distribution — equity is not an add-on to population health; it is constitutive. An ACH that improves average outcomes while widening disparities has not improved population health.

The ACH model represents a fundamental shift in the unit of accountability. Fee-for-service holds the individual provider accountable for the individual encounter. DRGs hold the hospital accountable for the episode. ACOs hold the provider network accountable for the population’s clinical outcomes. ACHs hold the community accountable for the population’s health — including the social, economic, and environmental determinants that clinical care cannot address alone.

This is a radical expansion of scope, and it raises governance challenges that the healthcare system has not solved. Who speaks for the community? Who decides how resources are allocated between clinical care and social services? Who bears financial risk for outcomes that depend on housing policy, education funding, and economic development? The ACH model implies that healthcare payment should fund non-healthcare services — that the Medicaid dollar or the grant dollar should flow to food banks, housing programs, and job training if those interventions produce better health outcomes per dollar than clinical care. The evidence base for this proposition is growing (the work of Sandro Galea, Michael Marmot, and the County Health Rankings program at the University of Wisconsin Population Health Institute), but the payment infrastructure to operationalize it is nascent.

VI. Population-Based Funding: The Community Contract

The Logic

Population-based funding takes the ACO and ACH logic to its conclusion: instead of paying providers for services, episodes, or outcomes, pay a defined amount per person per year for the health of a defined population. The funding is not tied to utilization, quality metrics, or specific services. It is a per-capita budget for health, allocated to an entity (or network of entities) that has accountability for how it is spent and what it produces.

The Intellectual Framework

The intellectual roots run deeper than healthcare. Population-based funding is, at its core, a social contract theory applied to health: the community pools resources (through taxation or premium contribution), and those resources are managed by an accountable entity for the collective benefit. This is Rousseau filtered through public finance.

In healthcare, the most complete intellectual treatment is by Robert Evans and colleagues in the Canadian health economics tradition. Evans’ Strained Mercy: The Economics of Canadian Health Care (1984) argued that healthcare financing should be understood as a social insurance system, not a market for services. The “funding the system” approach — global budgets for hospitals, negotiated fee schedules for physicians, population-based allocation of resources to regions — produces lower administrative costs, more equitable access, and comparable or better outcomes than the American multi-payer, service-based model.

The relevance to the American conversation is growing. The RHTP legislation includes provisions for multi-stream funding integration and community-based allocation mechanisms that resemble population-based funding more than traditional grant programs. Several state implementation plans contemplate using RHTP funds to support the kind of infrastructure that population-based models require: community health workers, mobile care delivery, telehealth networks, and social service integration. These are not clinical encounter generators. They are population health infrastructure — and they require a payment model that funds capacity and access rather than visits and procedures.

The De-Centering of Hospitals

Population-based funding implies a restructuring of care delivery that destabilizes the hospital as the organizing institution of American healthcare. If payment is allocated per capita rather than per encounter, the economic rationale for channeling care through the hospital — the most expensive setting — dissolves. The economic logic shifts toward the lowest-cost setting that produces acceptable outcomes: primary care offices, community health centers, mobile clinics, home-based care, telehealth, and community health worker programs.

This is not a new idea. Barbara Starfield’s lifelong body of work on primary care (Primary Care: Balancing Health Needs, Services, and Technology, 1998; “Is Primary Care Essential?”, The Lancet, 1994) demonstrated that health systems oriented around primary care produce better outcomes at lower cost than systems oriented around specialty and hospital care. The evidence is international and robust: countries with stronger primary care infrastructure have lower mortality, better equity, and lower costs, controlling for wealth and other confounders.

The RHTP legislation contemplates this shift. State implementation plans that emphasize mobile care delivery, community health workers, paramedicine programs, and telehealth are de-centering the hospital in favor of distributed care infrastructure that meets patients where they are. This is operationally and politically difficult — hospitals are the largest employers in many rural communities, and their closure or downsizing produces economic consequences that extend far beyond healthcare. The community economic impact of a rural hospital closure — documented by the National Rural Health Association and in the work of Mark Holmes and colleagues at the University of North Carolina’s Sheps Center — includes job loss, reduced local tax revenue, population out-migration, and a cascading effect on other businesses that depend on hospital employees as customers.

Population-based funding does not require hospital closure. But it does require that the hospital’s role change — from the center of the revenue model to one component of a distributed care delivery system. The organizational transformation this requires is enormous and connects directly to the structural inertia problems discussed in the companion document. Hospitals that were built, staffed, and governed to generate inpatient and procedural volume cannot simply become population health management hubs by administrative fiat.

VII. Multi-Stream Integration: The Braiding Problem

The Logic

No single payment model can fund the full scope of activities that population health requires. Primary care needs encounter-based or capitated revenue. Workforce development needs grant funding. Infrastructure needs capital investment. Social services need flexible dollars not constrained by clinical billing codes. Population health requires the braiding of multiple funding streams — Medicaid, Medicare, grant programs, commercial insurance, state general funds, tribal self-governance compacts, philanthropy — into a coherent resource pool that can be allocated to the highest-value activities regardless of funding source.

The Intellectual Framework

The braiding problem is, at its core, a transaction cost problem in the sense that Williamson described. Each funding stream comes with its own rules, reporting requirements, spending restrictions, and accountability mechanisms. The cost of managing multiple streams — maintaining separate accounting, meeting different reporting deadlines, reconciling conflicting definitions of eligible expenses — is a transaction cost that scales with the number of streams and the degree of divergence between their requirements.

The federal government has recognized this problem repeatedly without solving it. The Government Accountability Office has published multiple reports on the fragmentation of federal health and social service funding (GAO-11-318SP, 2011, is the landmark catalog). Lester Salamon’s work on “third-party government” (The Tools of Government, 2002) provides the theoretical framework: the American government increasingly accomplishes public purposes through third-party agents (nonprofits, state agencies, contractors) using a complex array of policy tools (grants, contracts, tax expenditures, regulations), and the complexity of the tool array creates coordination costs that reduce the effectiveness of each individual tool.

The practical challenge for organizations receiving RHTP funding is that braiding requires competencies that most have never needed: cross-stream budget management, blended cost allocation, coordinated reporting, and strategic resource deployment across funding streams with different timelines, restrictions, and accountability structures. The compliance burden of braiding exceeds the compliance burden of any individual stream, because the organization must simultaneously satisfy multiple sets of rules that may conflict.

The Rural Dimension

Rural health organizations face a particular braiding challenge because their funding streams are more diverse and individually smaller than those of large urban systems. A rural FQHC may receive HRSA Section 330 grant funding, Medicaid FFS and managed care payments, Medicare fee schedule payments, RHTP sub-grant funding, state behavioral health grants, SAMHSA substance use disorder funding, Ryan White HIV/AIDS funding, and charitable contributions — each with its own rules, reporting, and restrictions. The administrative capacity required to manage this portfolio is disproportionate to the organization’s size.

The RHTP provisions for multi-stream integration and community-based allocation are a partial response to this problem. But the integration they envision — coordinating federal, state, and local funding streams into a coherent community health investment — requires governance capacity that extends beyond any single organization. It requires the kind of polycentric governance that Ostrom described: multiple overlapping authorities coordinating around shared resources, with rules developed by the participants rather than imposed from above.

VIII. The Arc and the Unfinished Question

The payment model trajectory in American healthcare bends from transaction to relationship to community:

Fee-for-service pays for what the provider does. Prospective payment pays for what the patient has. Capitation pays for who the patient is. Value-based payment pays for what happens to the patient. ACOs pay for what happens to the population. Population-based funding pays for what the community becomes.

Each step in this progression expands the unit of accountability, lengthens the time horizon, and increases the organizational and governance complexity required. Each step also moves further from the observable, measurable, attributable world of the clinical encounter toward the emergent, multidetermined, hard-to-attribute world of population health. And each step requires institutions — governance arrangements, information systems, payment mechanisms, accountability structures — that do not yet fully exist.

The payment model question is not which model is correct. It is whether the institutional capacity exists to operate the model that the evidence supports. The evidence consistently points toward primary-care-oriented, population-based, multi-stream-funded, community-governed health systems as producing the best outcomes at sustainable cost. The institutional capacity to operate such systems — particularly in rural communities with thin organizational infrastructure, limited governance capacity, and fragmented funding — is the binding constraint.

This is the gap between theory and implementation that characterizes the current moment. The payment models that would best serve rural communities are known. The organizational and institutional capacity to implement them is not present. Time-limited investment can build some of that capacity. Whether it can build enough — in the face of simultaneous revenue disruption, workforce shortages, and compliance burden — is the empirical question that the current policy moment forces into the open.

Intellectual Debts

This document draws on the following primary works:

• Arrow, K. (1963). “Uncertainty and the Welfare Economics of Medical Care.” American Economic Review 53(5). Healthcare market failures and information asymmetry.
• Pauly, M. (1968). “The Economics of Moral Hazard: Comment.” American Economic Review 58(3). Moral hazard in health insurance and demand-side incentive distortion.
• Enthoven, A. (1978). “Consumer-Choice Health Plan.” New England Journal of Medicine 298(12-13). The theory of managed competition.
• Enthoven, A. (1980). Health Plan: The Practical Solution to the Soaring Cost of Medical Care. Institutional framework for healthcare market competition.
• Evans, R. (1984). Strained Mercy: The Economics of Canadian Health Care. Healthcare as social insurance and population-based funding.
• Williamson, O. (1985). The Economic Institutions of Capitalism. Transaction cost economics applied to organizational boundaries and multi-stream coordination.
• Newhouse, J. (1993). Free for All? Lessons from the RAND Health Insurance Experiment. Empirical evidence on cost-sharing, utilization, and health outcomes.
• Starfield, B. (1994). “Is Primary Care Essential?” The Lancet 344(8930). The evidence base for primary-care-oriented health systems.
• Shortell, S., Gillies, R., Anderson, D., Erickson, K., & Mitchell, J. (1996). Remaking Health Care in America. Organizational requirements for integrated delivery systems.
• Starfield, B. (1998). Primary Care: Balancing Health Needs, Services, and Technology. The foundational argument for primary care as the organizing principle of health systems.
• Robinson, J. (2001). “The End of Managed Care.” JAMA 285(20). The political and market constraints on managed care.
• Salamon, L. (2002). The Tools of Government: A Guide to the New Governance. Third-party government and the complexity of policy tools.
• Fisher, E., Wennberg, D., Stukel, T., Gottlieb, D., Lucas, F., & Pinder, E. (2003). “The Implications of Regional Variations in Medicare Spending.” Annals of Internal Medicine 138(4). Geographic variation in spending without outcome variation.
• Kindig, D. & Stoddart, G. (2003). “What Is Population Health?” American Journal of Public Health 93(3). The foundational definition including distributional equity.
• Porter, M. & Teisberg, E. (2006). Redefining Health Care: Creating Value-Based Competition on Results. Value as health outcomes per dollar spent.
• Fisher, E., Staiger, D., Bynum, J., & Gottlieb, D. (2007). “Creating Accountable Care Organizations.” Health Affairs 26(1). The ACO concept derived from naturally occurring referral networks.
• Berwick, D., Nolan, T., & Whittington, J. (2008). “The Triple Aim: Care, Health, and Cost.” Health Affairs 27(3). The three-dimensional optimization framework for health system design.
• Porter, M. (2010). “What Is Value in Health Care?” New England Journal of Medicine 363(26). The value equation: outcomes relative to cost.
• GAO. (2011). Opportunities to Reduce Potential Duplication in Government Programs, Save Tax Dollars, and Enhance Revenue. GAO-11-318SP. The fragmentation of federal health and social service funding.