Insurance, Disintermediation, and the Limits of Coverage

Why Covering People Is Not the Same as Making Them Healthy

The dominant policy framework for healthcare access in the United States since 1965 has been insurance expansion. The assumption: if people have insurance, they can access healthcare. If they can access healthcare, their health improves. The sixty-year project of American health policy — Medicare, Medicaid, employer-sponsored insurance, the ACA — has been organized around this assumption. Coverage is the proxy for access. Access is the proxy for health.

The assumption is partially true and deeply incomplete. Insurance coverage is necessary for healthcare access in a system where providers require payment. But coverage does not guarantee access (narrow networks, prior authorization, geographic distance), access does not guarantee utilization (wait times, transportation, trust, cultural competence), and utilization does not guarantee health (overtreatment, iatrogenic harm, treatment of symptoms rather than causes). The relationship between coverage and health is real but mediated by a chain of intermediary factors, each of which introduces friction, cost, and failure.

This document examines the insurance model as an intellectual and structural framework — its origins, its logic, its failures, and the market responses that are emerging as the model’s limitations become increasingly apparent. It is not about any particular product or company. It is about the structural dynamics that are reshaping how healthcare is financed and delivered in America.

I. The Insurance Bargain: Pooling Risk, Creating Intermediaries

The Logic of Health Insurance

Health insurance, in its theoretical form, is a risk-pooling mechanism. Kenneth Arrow’s 1963 framework identifies the fundamental problem: illness is uncertain, costly, and unevenly distributed. Individuals cannot predict when they will be sick or how much care will cost. Insurance pools the risk across a population: many contribute premiums, few draw large claims in any given year, and the pool absorbs the variance.

The welfare economics of insurance are well-established. Under conditions of risk aversion (which most people exhibit for healthcare costs), insurance improves welfare by converting uncertain, potentially catastrophic expenditure into predictable premium payments. The expected utility gain from risk pooling is the foundational justification for health insurance as a social institution.

The Intermediary Problem

The theoretical elegance of risk pooling encounters a structural problem in implementation: someone must manage the pool. The insurer — whether private or governmental — becomes an intermediary between the person who needs care and the provider who delivers it. This intermediary role creates functions that add cost: underwriting (assessing who to cover and at what price), claims processing (adjudicating whether a service is covered), utilization management (determining whether a service is necessary), network management (deciding which providers to include), and administration (managing enrollment, billing, compliance, and appeals).

These functions are not waste in the simple sense — they serve real purposes in managing a complex pooled-risk system. But they consume resources that do not directly produce health. The administrative overhead of private health insurance in the United States is estimated at 12–18% of premium revenue, compared to 2–3% for Medicare fee-for-service. Steffie Woolhandler, David Himmelstein, and colleagues at the City University of New York have published extensively on administrative cost comparison (most notably “Costs of Health Care Administration in the United States and Canada,” New England Journal of Medicine, 2003; updated in Annals of Internal Medicine, 2020). Their findings are consistent: the multi-payer, private insurance system produces administrative costs that are multiples of single-payer systems, without corresponding improvements in health outcomes.

The intermediary problem is not unique to private insurance. Medicaid managed care introduces MCOs as intermediaries between state Medicaid agencies and providers. Medicare Advantage introduces private insurers as intermediaries between CMS and beneficiaries. Each intermediary layer adds administrative cost and creates opportunities for value extraction — the difference between what the payer sends in and what the provider receives.

Mark Pauly and others have argued that some administrative costs represent value — better care coordination, fraud detection, quality measurement. The empirical question is whether the administrative spending produces value proportional to its cost. In aggregate, the American healthcare system spends approximately $1,055 per capita on administrative costs, compared to $550 in Canada and $300–400 in most European countries (Papanicolas, Woskie, and Jha, “Health Care Spending in the United States and Other High-Income Countries,” JAMA, 2018). The differential spending does not correlate with better outcomes by any standard measure.

II. The Employer-Sponsored Model: An Accident of History

Origins

The American reliance on employer-sponsored insurance (ESI) is not the product of deliberate health policy design. It is an accident of wartime economic policy. During World War II, wage and price controls (the Stabilization Act of 1942) prevented employers from competing for scarce labor through wages. The War Labor Board ruled that employer-paid health benefits did not count as wages subject to controls. Employers began offering health insurance as a recruitment tool — a benefit made financially attractive by its tax-exempt status, which the IRS codified in 1943 and Congress confirmed in the Internal Revenue Code of 1954.

Jacob Hacker’s The Divided Welfare State (2002) documents how this wartime accident became the structural foundation of American health coverage. The tax exclusion for employer-sponsored insurance — the largest single tax expenditure in the federal budget, worth approximately $300 billion annually — creates a massive public subsidy for private coverage, flowing disproportionately to higher-income workers (who face higher marginal tax rates and therefore benefit more from the exclusion) and to workers at large firms (which are more likely to offer coverage).

The consequences for healthcare financing are profound. Employer-sponsored insurance created a system where coverage is tied to employment — a coupling that produces coverage gaps for the unemployed, the self-employed, part-time workers, gig workers, and retirees under 65. It fragmented the risk pool by employer size and industry, creating wide variation in benefit design, cost-sharing, and network access. And it created a system where the ultimate purchaser (the employer) is not the consumer of care (the employee), producing a principal-agent problem at the core of the coverage relationship.

The Self-Funded Employer

A critical structural development, largely invisible to public policy discussions, is the shift from fully insured to self-funded employer health plans. Under ERISA (the Employee Retirement Income Security Act of 1974), large employers can self-fund their health benefits — paying claims directly from corporate assets rather than purchasing insurance from a carrier. Self-funded plans are regulated by the federal Department of Labor, not by state insurance commissioners, which exempts them from state benefit mandates, premium taxes, and many consumer protection regulations.

Today, approximately 65% of covered workers are in self-funded plans (Kaiser Family Foundation, 2023). The employer bears the actuarial risk directly, typically purchasing stop-loss insurance for catastrophic claims and contracting with third-party administrators (TPAs) for claims processing and network access.

Self-funding changes the employer’s relationship to healthcare spending from premium (a fixed cost negotiated annually with an insurer) to claims (a variable cost driven by the actual healthcare utilization of the employee population). This creates a direct financial interest in managing the health of the workforce — and a direct interest in understanding, challenging, and controlling the cost of healthcare services consumed by employees.

The self-funded employer is a purchaser with unusual characteristics: it has the data (claims data revealing exactly what services are consumed, by whom, at what price), the financial incentive (every dollar of unnecessary spending comes directly from the corporate budget), and — for large employers — the market power to negotiate directly with providers, pharmacies, and specialty care organizations. This purchaser profile creates the conditions for disintermediation.

III. Where Insurance Fails: The Access-to-Health Gap

Coverage Without Access

The assumption that insurance produces access encounters systematic failures:

Network adequacy. Insurance coverage is only as useful as the provider network it connects the insured person to. Narrow networks — a cost-control strategy where insurers limit the number of contracted providers to negotiate lower rates — reduce the practical accessibility of care. In rural areas, network adequacy is a persistent problem: there may be no in-network specialist within a reasonable distance, and out-of-network cost-sharing makes specialist care effectively unaffordable even for the insured.

Prior authorization. Utilization management mechanisms — particularly prior authorization, where the insurer must approve a service before the provider delivers it — create administrative delays that function as access barriers. The American Medical Association has documented that 93% of physicians report care delays due to prior authorization, and 34% report serious adverse events related to authorization delays. The mechanism exists to prevent unnecessary utilization. Its side effect is preventing or delaying necessary utilization.

Cost-sharing. Even with insurance, the deductible, copayment, and coinsurance obligations can deter care-seeking. High-deductible health plans — now the most common plan design in employer-sponsored insurance — require the individual to pay $1,600–$3,200 (individual) or $3,200–$6,400 (family) before insurance coverage begins (2024 IRS thresholds). For households with limited liquid savings — which includes a majority of American households — a $3,000 deductible is a functional barrier to non-emergency care. The RAND Health Insurance Experiment demonstrated forty years ago that cost-sharing reduces utilization of both unnecessary and necessary care, with disproportionate harm to low-income and chronically ill populations. The lesson has been documented but not applied.

Coverage Without Health

Even when insurance produces access and utilization, the relationship between healthcare utilization and health outcomes is weaker than the coverage model assumes.

Michael Marmot’s The Status Syndrome (2004) and the broader social determinants of health literature (the WHO Commission on Social Determinants of Health, chaired by Marmot, reported in 2008) demonstrate that the primary determinants of population health are not clinical care but social and economic conditions: income, education, employment, housing, social cohesion, and environmental quality. Clinical care accounts for an estimated 10–20% of variation in health outcomes (the County Health Rankings model, developed by Bridget Booske, Patrick Remington, and colleagues at the University of Wisconsin Population Health Institute, attributes 20% to clinical care, 30% to health behaviors, 40% to social and economic factors, and 10% to the physical environment).

This does not mean clinical care is unimportant. For individuals who are sick, clinical care is essential. But at the population level, expanding clinical care coverage produces diminishing health returns once a basic threshold of access is met. The marginal health dollar invested in housing stability, nutrition, or income support may produce more health than the marginal dollar invested in clinical services for a population that already has basic coverage.

The implications for the insurance model are fundamental: if insurance is a mechanism for financing clinical care, and clinical care is not the primary determinant of health, then insurance expansion alone cannot close health disparities. The model solves a financial access problem while leaving the structural determinants of poor health unchanged.

Sandro Galea’s Well: What We Need to Talk About When We Talk About Health (2019) articulates this most directly: America has a health system, not a healthcare system, and the health system includes every economic, social, and environmental factor that shapes the conditions in which people live, work, and age. Insurance is a component of the health system. It is not a substitute for the rest of it.

IV. The Disintermediation Impulse

The Market Response to Intermediary Failure

When intermediaries fail to deliver value proportional to their cost, markets produce disintermediation — the removal or bypassing of the intermediary. This is a general market phenomenon (Clay Shirky’s work on institutional disruption, Clayton Christensen’s theory of disruptive innovation) that is now manifesting in healthcare with increasing force.

The structural conditions for healthcare disintermediation include:

Price opacity becoming price transparency. The Hospital Price Transparency Rule (2021) and the Transparency in Coverage Rule (2022) require hospitals and insurers to publish negotiated rates. For the first time, self-funded employers and other purchasers can see the price variation that the intermediary system has obscured. The variation is enormous: the same procedure at hospitals in the same city can vary by 300–1,000%, with no correlation to quality outcomes. Zack Cooper, Stuart Craig, Martin Gaynor, and John Van Reenen’s research on hospital price variation (“The Price Ain’t Right?”, Quarterly Journal of Economics, 2019) documented that hospital prices are largely a function of market power, not cost or quality. Price transparency does not, by itself, reduce prices. But it creates the informational prerequisite for purchasers to route around high-cost intermediaries and negotiate directly.

Pharmacy benefit disruption. The pharmaceutical supply chain — manufacturer → wholesaler → pharmacy benefit manager (PBM) → pharmacy → patient — is a multilayer intermediary structure with opaque pricing at every node. The three largest PBMs (CVS Caremark, Express Scripts, OptumRx) manage approximately 80% of prescription drug claims. The FTC’s 2024 interim report on PBM practices documented that PBMs engage in spread pricing (charging plans more than they pay pharmacies and keeping the difference), rebate retention (negotiating manufacturer rebates and keeping a portion rather than passing them to plans), and formulary manipulation (favoring higher-list-price drugs that generate larger rebates over lower-cost alternatives).

The market response: direct-to-employer and direct-to-consumer pharmaceutical models that bypass the PBM intermediary. GLP-1 receptor agonists (semaglutide, tirzepatide) — the most commercially significant drug class in a generation — have become the proving ground. Manufacturers and specialty pharmacies are offering direct pricing to self-funded employers, bypassing the PBM rebate structure entirely. The self-funded employer, with claims data showing the cost and utilization of GLP-1s in its population, can calculate whether direct contracting produces savings relative to the PBM channel. For many, it does.

Primary care disintermediation. Direct Primary Care (DPC) — a model where patients pay a monthly membership fee directly to a primary care practice, bypassing insurance for routine care — is a growing but still marginal movement. The intellectual foundation draws on the concierge medicine model but at a lower price point, targeting working and middle-class patients rather than the wealthy. DPC removes the insurer from the primary care relationship, eliminating claims processing, prior authorization, and network management overhead. The trade-off: DPC patients still need insurance for hospital, specialist, and catastrophic care, creating a layered model (DPC for primary care + high-deductible plan for everything else) that reduces total cost for healthy populations but may leave chronically ill patients in a coverage gap.

Digital health and virtual care. Telehealth, remote monitoring, and digital therapeutics create care delivery channels that do not require the traditional insurance-provider-patient triangle. A mental health platform that offers therapy sessions for a flat monthly fee, a diabetes management program that provides CGM devices and coaching for a per-member-per-month rate, or an employer-contracted musculoskeletal program that provides physical therapy through an app — these are all disintermediation plays that remove the traditional insurer from at least some portion of care delivery.

V. The Self-Funded Employer as Agent of Change

Why Employers Are Driving Disintermediation

The self-funded employer occupies a unique structural position in American healthcare. It is the actual payer — not the insurer, not the government, but the entity that writes the checks for healthcare claims. And unlike individual consumers, who lack information, market power, and alternatives, large self-funded employers have all three.

Michael Porter’s work on value-based competition argues that the fundamental problem in healthcare is the absence of informed purchasers who can evaluate and reward value. Self-funded employers are the closest approximation to informed purchasers that the American system has produced. They have claims data. They have actuarial analysis. They have the financial incentive to reduce waste. And they are increasingly willing to act on that information.

The actions employers are taking include:

Centers of Excellence programs. Routing employees to high-quality, lower-cost providers for specific procedures (joint replacement, cardiac surgery, cancer treatment), often outside the employee’s home geography, with the employer covering travel and waiving cost-sharing. This is direct competition with the local provider market, enabled by employer claims data showing quality and cost variation.

Direct contracting with health systems. Bypassing the insurer entirely and negotiating bundled rates directly with provider organizations. Haven (the Amazon-Berkshire Hathaway-JPMorgan venture) failed, but the impulse it represented — large employers believing they could do better than the insurance intermediary — has not dissipated. Subsequent employer coalitions and direct-contracting models continue to emerge.

Condition-specific carve-outs. Contracting with specialized vendors for specific conditions (musculoskeletal, behavioral health, diabetes, fertility) outside the general insurance benefit. Each carve-out removes a category of spending from the insurer’s domain and places it with a specialized provider accountable directly to the employer for outcomes and cost.

Pharmacy carve-outs and pass-through PBMs. Replacing traditional PBMs with pass-through or transparent PBM models that charge an administrative fee rather than profiting from spread pricing and rebate retention. This is a direct response to the FTC findings and to employer awareness that the traditional PBM model extracts value without adding proportional benefit.

The Limits of Employer-Driven Reform

Employer-led disintermediation has structural limits. It works best for large employers (with the data, actuarial capacity, and market power to negotiate) and for conditions that are episodic, measurable, and high-cost (where the return on direct contracting is large enough to justify the administrative investment). It does not address chronic disease management for populations without employer coverage. It does not address the coverage needs of the unemployed, the underemployed, the retired-under-65, or the disabled. And it does not address the social determinants of health that drive the majority of health outcomes.

The employer-driven model is also inherently fragmented. Each employer makes independent purchasing decisions based on its own population’s needs, creating a patchwork of arrangements that lacks the standardization, risk pooling, and equity guarantees of a universal coverage system. The efficiency gains from disintermediation may accrue to the employed populations of large firms while leaving everyone else in the same — or a worse — insurance-mediated system.

This is the fundamental tension in the disintermediation movement: it solves the intermediary problem for the populations that have the most market power (employed, commercially insured) while potentially exacerbating the access problem for the populations that have the least (Medicaid, uninsured, rural).

VI. The Insurance Model in Rural Healthcare

Why the Model Breaks Down

The insurance model’s failures are amplified in rural settings for structural reasons that compound each other:

Thin markets. Insurance depends on a sufficient number of covered lives to pool risk and a sufficient number of providers to create a network. Rural communities may have neither. A county with 10,000 residents and one hospital cannot sustain multiple competing insurance plans — there is no network to differentiate, no provider competition to leverage, and insufficient enrollment to spread risk.

Provider dependence on public payers. Rural providers typically have payer mixes dominated by Medicare and Medicaid, with commercial insurance representing 15–25% of revenue. The insurance model’s incentive structures — network negotiation, utilization management, value-based contracting — are designed for markets where commercial insurance is the dominant payer. In markets where public payers set rates administratively, the insurance model’s competitive mechanisms have no purchase.

Administrative burden disproportionality. The administrative requirements of participating in insurance networks — credentialing, claims submission, prior authorization, quality reporting, contract negotiation — impose fixed costs that are distributed across a smaller patient base in rural settings. A rural clinic spending 20 hours per week on insurance administration is spending the same absolute time as an urban practice, but that time represents a larger proportion of total staff capacity.

Disintermediation bypasses rural communities. The emerging disintermediation models — centers of excellence, direct employer contracting, specialty carve-outs — route patients to high-volume, high-quality providers, which are overwhelmingly urban. A center of excellence program for joint replacement sends the rural patient to a major academic medical center in the city, not to the local Critical Access Hospital. Each disintermediation play that routes volume away from rural providers further undermines the already-fragile rural provider revenue base.

The Alternative Models

The intellectual traditions that point toward alternatives to insurance-mediated healthcare in rural settings include:

Community-rated, community-governed health funds. Historical precedents include the mutual aid societies of the early twentieth century and the contemporary health care cooperatives. These models pool risk within a geographic community rather than within an employment group, with governance by community members rather than by insurance corporations. The Group Health Cooperative of Puget Sound (now part of Kaiser Permanente) was the most successful American example of this model — a consumer-governed health plan that operated for over 70 years.

Global budgets for rural providers. Maryland’s All-Payer Model (2014) and its successor, the Total Cost of Care Model, demonstrate that global budgets — fixed annual revenue for a hospital, adjusted for population and inflation — can stabilize rural hospital finances while shifting incentives from volume to value. The Pennsylvania Rural Health Model (2017) extended this concept to rural hospitals specifically, providing a guaranteed annual revenue floor that insulates hospitals from the volume fluctuations that threaten financial viability. Early results show improved financial stability without deterioration in quality metrics.

Tribal self-governance models. The Indian Health Service, tribal 638 compacts, and tribal self-governance demonstrate an alternative to insurance-mediated healthcare: direct federal funding for a sovereign government to provide healthcare to its population. The model bypasses insurance entirely — the tribal health program is both the funder (through federal appropriation) and the provider. The 638 model’s strengths (self-determination, cultural appropriateness, integrated service delivery) and weaknesses (chronic underfunding, geographic isolation, workforce challenges) offer lessons for non-tribal rural communities considering post-insurance models.

VII. The Structural Question

The insurance model in American healthcare is not collapsing. It remains the dominant framework for coverage and payment, and it will remain so for the foreseeable future. But it is cracking — at the edges, in specific populations, for specific conditions, and in the structural spaces where its costs visibly exceed its value.

The cracks are informative. They reveal where the intermediary adds cost without adding health. They reveal where the coverage-access-health chain breaks. They reveal where market actors — employers, providers, patients — are willing to bear the risk and complexity of going direct because the intermediary’s price exceeds its value.

The deeper structural question is whether these cracks are isolated — specific failures in specific market segments that can be patched — or systemic — evidence that the insurance intermediary model itself is approaching its limits as a framework for healthcare financing.

The evidence points in both directions. For acute, episodic, high-cost care (surgery, cancer treatment, catastrophic events), insurance remains the most efficient risk-pooling mechanism available. No individual or employer can self-insure against a $2 million NICU stay. For chronic disease management, preventive care, primary care, behavioral health, and social service integration, the insurance model increasingly appears to be the wrong tool — an intermediary structure designed for unpredictable, high-cost events being applied to predictable, ongoing, relationship-dependent health needs.

The intellectual challenge is to distinguish which healthcare functions require insurance (risk pooling for catastrophic and unpredictable costs), which require direct payment (routine, predictable, relationship-based care), and which require public investment (population health infrastructure, workforce development, social determinants). The American system has historically channeled all three through a single intermediary structure — the insurance plan — and the result is a system that does none of the three as well as it could.

The disintermediation movement is, at its core, a sorting process — the market discovering which functions the intermediary adds value to and which functions it merely taxes. That sorting process is underway. It will not produce a single alternative model. It will produce a layered system: insurance for catastrophic risk, direct relationships for ongoing care, public investment for community health infrastructure. Whether that layered system is more equitable, more efficient, and more effective than the current intermediary-dominated model depends on how the sorting is governed — who decides, who benefits, and who bears the cost when the layers don’t align.

That governance question — not the market question and not the insurance question — is the one that will determine whether the disintermediation of American healthcare produces better health or merely different intermediaries.

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 the economic case for insurance.
• Pauly, M. (1968). “The Economics of Moral Hazard: Comment.” American Economic Review 58(3). Moral hazard and demand-side incentive distortion.
• Newhouse, J. (1993). Free for All? Lessons from the RAND Health Insurance Experiment. The empirical relationship between cost-sharing, utilization, and health outcomes.
• Hacker, J. (2002). The Divided Welfare State: The Battle over Public and Private Social Benefits in the United States. The origins and consequences of employer-sponsored insurance as public subsidy.
• Woolhandler, S., Campbell, T., & Himmelstein, D. (2003). “Costs of Health Care Administration in the United States and Canada.” New England Journal of Medicine 349(8). Comparative administrative costs across system designs.
• Marmot, M. (2004). The Status Syndrome: How Social Standing Affects Our Health and Longevity. Social determinants as primary drivers of health outcomes.
• Kindig, D. & Stoddart, G. (2003). “What Is Population Health?” American Journal of Public Health 93(3). Defining population health to include distributional equity.
• Porter, M. & Teisberg, E. (2006). Redefining Health Care: Creating Value-Based Competition on Results. The case for informed purchasers driving healthcare value.
• WHO Commission on Social Determinants of Health. (2008). Closing the Gap in a Generation. The global evidence base for social determinants of health.
• Christensen, C., Grossman, J., & Hwang, J. (2009). The Innovator’s Prescription: A Disruptive Solution for Health Care. Disruptive innovation theory applied to healthcare delivery.
• Booske, B., Athens, J., Chang, D., Remington, P., & Tran, V. (2010). County Health Rankings Working Paper. The model attributing population health outcomes to clinical care, health behaviors, social/economic factors, and physical environment.
• Papanicolas, I., Woskie, L., & Jha, A. (2018). “Health Care Spending in the United States and Other High-Income Countries.” JAMA 319(10). Comparative administrative spending across high-income nations.
• Cooper, Z., Craig, S., Gaynor, M., & Van Reenen, J. (2019). “The Price Ain’t Right? Hospital Prices and Health Spending on the Privately Insured.” Quarterly Journal of Economics 134(1). Hospital price variation driven by market power.
• Galea, S. (2019). Well: What We Need to Talk About When We Talk About Health. Healthcare versus health and the scope of the health system.
• Woolhandler, S. & Himmelstein, D. (2020). “Administrative Costs Update.” Annals of Internal Medicine. Updated comparative administrative cost analysis.
• Federal Trade Commission. (2024). Interim Report on Pharmacy Benefit Managers. Documentation of PBM pricing practices and market concentration.