The Great Repricing of American Healthcare: How Managed Care Conglomerates Rewrote the Laws of Underwriting
Section 1: The Machine That Swallowed American Healthcare
Every spring, on an evening in early April, a federal agency in Baltimore publishes a document that decides how much money roughly a third of America's seniors will be worth to the companies that insure them. It is called the Rate Announcement, and the Centers for Medicare & Medicaid Services has been issuing it, in one form or another, since private plans were first invited to take on Medicare risk. Most years it lands like a utility tariff filing: technical, expected, absorbed.
The announcement CMS published in April 2024, covering contract year 2025, did not land that way. It confirmed that the agency would proceed with the second year of a three-year phase-in of a rewritten risk-adjustment model, known inside the industry as V28, which stripped thousands of diagnosis codes out of the formula that converts a patient's documented illnesses into a monthly payment.1 It confirmed tighter quality thresholds on the Star Ratings system that determines which plans collect a five percent bonus on top of their benchmark. And it did both at a moment when seniors, having deferred hips, knees, and cardiac procedures through the pandemic years, were consuming medical care at rates the industry's actuaries had not priced.
Managed care equities lost tens of billions of dollars of market value in the sessions that followed. Humana, the most concentrated bet on Medicare Advantage among the large public insurers, took the worst of it. Its problem was arithmetic. More than four-fifths of Humana's revenue comes from Medicare Advantage premiums and adjacent senior products.5 When the federal government adjusts the formula that generates that revenue, Humana has nothing else in the building to absorb the shock.
UnitedHealth Group, whose insurance arm is the largest Medicare Advantage carrier in the country, fell too β and then, over the following quarters, told a different story about why its earnings held together. That difference is the subject of this article.
What the industry actually is
Start with the size of the thing. National health expenditure in the United States runs at roughly $4.8 trillion, about 17.5 percent of gross domestic product, and federal statutory spending on Medicare and Medicaid together exceeds $1.6 trillion a year.1 Health insurance is the toll booth on nearly all of it. Direct written premium across commercial and government-sponsored health lines exceeds $1.25 trillion, and the concentration is extreme: seven publicly traded managed care organisations β UnitedHealth Group, Elevance Health, CVS Health's Aetna, The Cigna Group, Humana, Centene, and Molina Healthcare β together stand between roughly 180 million Americans and their doctors.3
That toll booth used to be a modest business. An indemnity insurer collected a premium, paid a claim, and kept a spread. What changed, over five decades, is that the government decided to hand its own entitlement liabilities to those same companies at a fixed price per head. Medicare Advantage now covers about 54.5 percent of eligible Medicare beneficiaries, roughly 33.8 million people, which makes privately administered managed care the majority operating system for American senior medicine rather than an alternative to it.2
Here is the first data exhibit, and the shape of it is the whole story.
Exhibit 1 β Medicare Advantage enrolment and penetration, United States, 2020β2026
| Year | MA enrolment (millions of beneficiaries) | Penetration (% of eligible Medicare beneficiaries) |
|---|---|---|
| 2020 | 24.1 | 39.0% |
| 2021 | 26.3 | 42.0% |
| 2022 | 28.4 | 45.0% |
| 2023 | 30.8 | 48.0% |
| 2024 | 32.8 | 51.5% |
| 2025 | 33.5 | 53.8% |
| 2026 (estimate) | 33.8 | 54.5% |
Definition: beneficiaries enrolled in Medicare Advantage plans, including special-needs plans, as a share of total Medicare-eligible beneficiaries. Geography: United States. Source: KFF Medicare Advantage enrolment tracking and CMS enrolment files.2 Evidence status: 2020β2025 observed; 2026 is an estimate as of August 4, 2026, not an observed full-year figure.
Read that aloud and the inflection is impossible to miss. Between 2020 and 2024, Medicare Advantage added roughly two to two and a half million members a year and gained three percentage points of penetration annually. Between 2024 and 2026 it added about a million members in total and gained three points across two full years. The growth engine did not break. It was throttled deliberately, by insurers who stopped bidding for members they could no longer profitably serve β trimming dental and over-the-counter allowances, narrowing provider networks, and withdrawing plans from counties where the benchmark no longer covered the cost of care.
The margin picture moved in the opposite direction. Blended medical loss ratios across the large public carriers β the share of every premium dollar paid out in medical claims β sat in the low-to-mid 80s before 2023 and now run between roughly 87.8 and 89.2 percent.45 For a business whose pre-tax underwriting margin lives in a band of about 1.5 to 3.8 percent, a 400 basis point move in claims cost is not a bad quarter. It is the entire margin, twice over.
The question this article tests
The reader I am writing for is a professional thematic equity investor β the long-only manager judged against a benchmark and the long/short manager judged on absolute return face genuinely different versions of this problem, and I will flag where the difference matters. Both are asking the same diagnostic question about 2024 through 2026: was this a cyclical mispricing that annual bid submissions will fix, or a permanent resetting of what a managed care licence is worth?
The evidence points to a split answer, and the split is the investable part. Elevated utilisation is cyclical; carriers can and did raise premiums and cut benefits into the 2026 and 2027 bids. The V28 risk model and the Inflation Reduction Act's redesign of the Part D drug benefit are structural; no amount of repricing recovers a payment formula that no longer pays for the same documented conditions. Whether the two can be disentangled quickly is genuinely disputed: the sell-side consensus treats Medicare Advantage margin recovery as a roughly twenty-four-month repricing exercise, while the carriers' own bid behaviour β exiting counties rather than repricing into them β suggests they do not believe they can raise price fast enough without losing the members whose scale pays for the fixed cost base.
Which returns us to the divergence in April 2024. Two companies took the same federal rate action. One had an answer that did not depend on the rate. To understand why that answer exists at all, you have to go upstream, to the belief that made this industry worth examining in the first place.
Subscribe on LinkedInSection 2: Upstream Belief: The Vertical Integration Imperative
The proposition that sent us here is a single falsifiable claim, and it is worth stating flatly before testing it:
The United States is completing an irreversible migration from fee-for-service indemnity underwriting to government-sponsored, risk-bearing managed care platforms β and within that structure, durable shareholder value accrues only to entities that control both the risk-bearing insurance licence and the clinical care-delivery engine behind it.
Notice what that claim does not say. It does not say managed care membership will keep compounding. It does not say insurance is a good business. It says the insurance licence has become a permission slip to operate something else, and the something else is where the returns live. If that is right, the correct unit of analysis is a capital-allocation platform, not an underwriter.
Three independent lines of evidence point at the same conclusion, and they come from different parts of the economy.
The fiscal evidence
The first is demographic arithmetic wearing a fiscal costume. Roughly ten thousand Americans turn sixty-five every day; the population aged sixty-five and over has grown from about 41 million in 2010 to more than 62 million.1 Every one of them is a claim on a federal entitlement whose cost the Treasury cannot control directly, because traditional fee-for-service Medicare pays whatever volume providers generate.
Capitation solves that problem for the sponsor. When CMS pays a private plan a fixed, risk-adjusted amount per member per month, the federal government converts an open-ended liability into a budgeted one and transfers the variance to a corporate balance sheet. State Medicaid agencies made the same trade earlier and more completely. This is why the federal and state posture toward managed care has been, on net, expansionary for two decades despite periodic political hostility: governments are buying budget certainty, and they are willing to pay a margin for it. The tightening of 2024 through 2026 is best read as the sponsor renegotiating the price of that certainty, not withdrawing the offer.
The employer evidence
The second line runs through the commercial market, and it points somewhere unexpected. More than 65 percent of covered American workers are now in self-funded arrangements, in which the employer bears the medical risk and pays the insurer an administrative fee for network access, claims processing, care management, and pharmacy benefit administration.3
If insurance underwriting were the valuable part of a health plan, employers moving risk onto their own books would have gutted the carriers. It did the opposite. Employers keep paying, because what they cannot replicate is the negotiated provider network, the pharmacy rebate pool, and the utilisation-management apparatus. The customer, given a free choice, bought the infrastructure and declined the insurance. That is a behavioural revealed preference, and it is the single cleanest piece of evidence for the upstream belief.
The physician evidence
The third signal comes from the supply side of medicine. Venture and private-equity funding for independent primary care roll-ups, abundant through 2021, contracted sharply as capital costs rose and CMS revised the risk model that those business plans had capitalised into their projections. Independent practices that had planned to sell to a financial sponsor increasingly sold, or affiliated, with an insurer's own health services division instead β Optum at UnitedHealth, Carelon at Elevance, Evernorth at Cigna.46
Take those three together and you get a system in which the government is delegating risk downward, employers are buying infrastructure rather than insurance, and physicians are being absorbed by the entities that pay them. Each fact is explicable on its own. Together they describe one architecture.
The transmission mechanism
The mechanism that turns this belief into industry economics is a regulatory accident with enormous consequences.
The Affordable Care Act imposed statutory minimum medical loss ratios: an insurer must spend at least 85 cents of every large-group and Medicare premium dollar, and 80 cents of every individual and small-group dollar, on medical claims and quality improvement, or refund the difference.1 The intent was to cap insurer profiteering. The effect was to cap insurer gross margin on regulated premium β and to leave everything outside the regulated premium entirely untouched.
So consider what happens when a health plan pays a claim. If the claim goes to an independent hospital, the dollar leaves the enterprise and counts against the MLR ceiling. If the same clinical service is delivered by a surgery centre, home-health agency, infusion suite, or specialty pharmacy that the plan's parent company owns, the dollar still counts against the MLR β but it does not leave the enterprise. It reappears as revenue in an unregulated health services segment, where margin is limited by competition rather than by statute.
That is the whole game. The MLR floor did not reduce the profitability of health insurance conglomerates. It relocated it. Anywhere the regulated premium dollar can be routed into a wholly owned service business, the conglomerate captures a second bite that the pure underwriter never sees. Pharmacy benefit management is the largest such channel: the three largest PBMs β OptumRx, CVS Caremark, and Cigna's Evernorth β process more than 79 percent of United States prescription claims, and all three sit inside companies that also hold insurance licences.46
The same belief implicates several sibling industries directly β acute-care hospitals, specialty pharmacy and drug distribution, revenue-cycle and healthcare IT, and medical technology all sit downstream of the same routing decisions β though this article follows the money only as far as the payer's own perimeter.
What would prove this wrong
A belief without a kill switch is a slogan. Three observable events would falsify this one.
First, a federal statutory ban on health plans owning PBMs, specialty pharmacies, or physician practices. That would sever the routing mechanism at the source, and the conglomerate premium in the equity market would have no economic basis.
Second, multi-year net disenrolment from Medicare Advantage back into traditional fee-for-service Medicare. If seniors, faced with thinner benefits and narrower networks, walked away in sustained numbers, the delegation of federal risk would be reversing rather than deepening.
Third, federal legislation establishing a public option that caps commercial reimbursement near Medicare fee-for-service rates. That would collapse the spread between what commercial plans pay providers and what the government pays, and the administrative-services franchise built on network discounts would lose most of its reason to exist.
None of the three has happened as of August 4, 2026. All three are live enough to be worth monitoring, and the third is the one that reprices the whole sector rather than one segment of it. Before we can judge how close any of them are, though, we need to understand the machinery they would be aimed at β the actual mechanics of how a health insurer makes and loses money.
Section 3: The Mechanics of Underwriting and the Great Statutory Moat
There is a week each spring when the most consequential work in American health insurance happens in a conference room full of actuaries.
Medicare Advantage plans must file their bids with CMS by the first Monday in June for the plan year beginning the following January. In that filing, a carrier commits β county by county, plan by plan β to a premium, a benefit package, a provider network, and an implicit forecast of how sick its members will be and how much care they will consume eighteen months into the future. The bid is compared against a county benchmark that CMS sets. Bid below the benchmark and the plan gets to keep a share of the difference as a rebate, which it must spend on extra benefits: dental, vision, hearing, over-the-counter allowances, reduced Part B premiums. Bid above it and the member pays the difference, and in a market where zero-premium plans are the norm, that is usually the end of the plan.
Everything an investor needs to understand about this industry's economics is visible in that one filing.
Capitation, and why it inverts the incentive
In fee-for-service medicine, a provider bills for each service delivered; more volume means more revenue. Under capitation, the payer receives a fixed monthly amount per enrolled member regardless of what that member consumes. The insurer's revenue is set at the start of the year and its costs are discovered over the following twelve months.
A useful analogy: capitation turns a health plan into something closer to a fixed-price construction contractor than a toll collector. The price is agreed before the work is scoped, so all the value creation and destruction lives in cost control and scope estimation. The analogy has a limit worth naming β a contractor can walk away from a bad job, and a health plan cannot. Regulatory requirements on network adequacy, benefit continuity, and mid-year member protections mean an insurer that mis-prices a county is contractually trapped in it for the full plan year. That asymmetry is precisely why the 2024 utilisation surge hurt as much as it did: the revenue was already fixed when the costs arrived.
The medical loss ratio, and the ceiling nobody talks about
The medical loss ratio is medical claims paid divided by premium revenue. Insurers often report it as a health benefits ratio; the definitions differ at the margin in how quality-improvement spending and taxes are treated, which is one reason cross-company MLR comparisons need care rather than a simple ranking.
The statutory floors β 85 percent for Medicare and large group, 80 percent for individual and small group β are usually described as consumer protections, and they are.1 But run the arithmetic from the shareholder's side. If at most 15 cents of a Medicare premium dollar can be retained, and administrative costs consume roughly 7 to 12 cents of it depending on the carrier's efficiency, then the theoretical maximum underwriting margin is a low single-digit percentage. In practice the industry earns a pre-tax underwriting margin of roughly 1.5 to 3.8 percent.3
Grocery-store margins on a trillion-dollar revenue base is a viable business β but only if two things hold. Capital turnover must be high, and the float must earn something.
Float, and the balance sheet that funds it
Health plans collect premiums at the start of each month and pay claims thirty to ninety days later. The gap generates a permanent pool of other people's money β operating float β invested in short-duration fixed income. In a zero-rate world this contributes almost nothing. At the short rates prevailing across 2023 through 2026, it contributes meaningfully to pre-tax earnings across every carrier simultaneously, which is an underappreciated common factor: a sustained fall in short rates would compress earnings sector-wide without any change in medical cost trend.
The float is not free money, because state regulators require it to be backed. Risk-based capital rules, administered through the National Association of Insurance Commissioners framework and enforced by each state's insurance department, set a minimum statutory capital buffer relative to underwriting and investment risk.3 Large carriers typically operate their licensed subsidiaries at 250 to 350 percent of the authorised control level. Capital above that threshold can be dividended up to the holding company, where it funds buybacks, debt service, and acquisitions. Capital below it cannot move at all β which is why a bad underwriting year does not merely reduce earnings, it traps cash inside regulated entities exactly when the parent wants it.
The statutory moat
Put these constraints together and you get the most durable competitive feature of this industry, which is that it is nearly impossible to enter.
A new entrant needs a licence in every state where it operates, statutory capital sized to its projected premium, a provider network contracted at rates competitive with incumbents who negotiate on behalf of tens of millions of lives, an actuarial function capable of surviving CMS bid review, and enough member scale to spread fixed administrative and compliance cost. Capital alone does not buy the network discount, and the network discount is the product.
This is Hamilton Helmer's scale economies and cornered resource operating simultaneously, with regulation supplying the barrier that would otherwise have to be built. It also explains, more convincingly than any story about execution, why a decade of well-funded insurance technology start-ups produced so few durable underwriters.
Two companies illustrate the two legitimate ways to win inside those constraints.
Molina Healthcare wins on cost. It runs a selling, general, and administrative expense ratio of roughly 7.2 percent β among the leanest in the industry β serving Medicaid, dual-eligible, and marketplace populations across nineteen states.3 In a business where the loss ratio is largely dictated by the medical cost of a state-assigned population and the premium is set by a state agency, the administrative line is one of the few variables management genuinely controls. Molina's leadership on that specific parameter is real and measurable, and its durability rests on an operating model deliberately built without the corporate overhead of a diversified conglomerate. The trade-off is equally real: a lean administrative structure buys less clinical care management, which limits Molina's ability to bend the medical cost curve rather than merely process it efficiently.
Alignment Healthcare wins on quality. More than 90 percent of its Medicare Advantage members are enrolled in plans rated 4.5 Stars or higher, an unusually high concentration achieved through a proprietary clinical workflow platform the company calls AVA, which identifies rising-risk members for proactive intervention.8 The Star Ratings system is CMS's annual one-to-five quality score; plans at 4.0 Stars and above receive a five percent quality bonus payment on their benchmark.1 For a carrier with a five-state footprint concentrated in California, that bonus is the difference between a viable bid and an uncompetitive one. Alignment's lead here is genuine, parameter-specific, and β importantly β not evidence that it leads on anything else. It is a fraction of the scale of any national carrier, and its geographic concentration means a single adverse California rate cycle affects most of its book.
V28, Stars, and the disputed arithmetic
The two regulatory levers that reset the industry between 2024 and 2026 both operate on this bid.
The first is the risk-adjustment model. CMS pays plans more for sicker members, calibrated through hierarchical condition categories that map diagnosis codes to risk scores. The V28 model, phased in across three payment years, removed or reclassified thousands of ICD-10 codes that the agency judged were being documented aggressively without corresponding cost β reducing average measured risk scores across chronic disease categories by low single-digit percentages.1 Because a plan's revenue equals benchmark multiplied by risk score, a two to four percent reduction in measured risk flows almost entirely to the bottom line of a business earning two to four percent margins.
The second is Star Ratings, where CMS changed how cut points are calculated, adopting an outlier-removal method known as the Tukey procedure that raised the score required to reach each rating tier. Several carriers, including Elevance and Humana, have challenged aspects of CMS's Star methodology in federal court, and the litigation remains unresolved as of August 4, 2026, leaving genuine uncertainty around 2027 quality bonus revenue.5 This dispute matters more than its legal obscurity suggests: the sector-wide share of Medicare Advantage members in 4.0-plus Star plans has fallen to roughly 62 percent from about 74 percent in 2023.2 Twelve points of Star share, applied to a five percent bonus on a multi-hundred-billion-dollar revenue base, is a large number moving on a methodological footnote.
The regulatory architecture that makes this industry impossible to attack from outside is the same architecture that caps what it can earn from within. That contradiction has only one resolution, and the industry found it two decades ago: go earn the money somewhere the statute does not reach.
Section 4: Anatomy of a Value Chain: How Regulated Dollars Become Unregulated Profits
Follow a thousand dollars of premium through the system and the architecture reveals itself.
Layer one: who actually pays
The money originates in four places. The federal government pays capitation to Medicare Advantage and Part D plans. State Medicaid agencies pay capitation to managed Medicaid organisations under contracts they re-procure every three to five years. Employers pay premiums for fully insured coverage or administrative fees plus claims funding for self-insured coverage. Individuals pay premiums on the ACA exchanges, most of them subsidised through federal premium tax credits.
Bargaining power at this layer is concentrated and it is public. CMS sets benchmark rates, risk models, and quality bonus criteria unilaterally, subject only to notice-and-comment procedure and occasional litigation. States run competitive procurements that can move billions of dollars of revenue between carriers on a single award decision. This is a market with a small number of very large buyers who can rewrite the terms of trade annually, which is the defining feature of the industry's risk profile.
Layer two: the licensed risk-bearers
Our thousand dollars now sits inside a licensed insurance subsidiary β UnitedHealthcare, Elevance's health benefits business, Aetna, Cigna Healthcare, Humana, Centene, Molina. Roughly 850 to 890 of it is legally committed to medical claims and quality improvement under the statutory floors. Perhaps 70 to 120 goes to administration. What remains is the underwriting margin.
This is the layer everyone means when they say "health insurance," and it is the layer with the least attractive economics in the entire chain: regulated gross margin, mandatory capital, annual repricing risk, and a single dominant customer that also writes the rules.
Layer three: the pharmacy chokepoint
Of our thousand dollars, a growing share β driven by specialty and now cardiometabolic drugs β is spent on pharmacy. That spending is administered by a pharmacy benefit manager, which decides which drugs sit on the formulary, at what tier, with what prior-authorisation requirement, and negotiates rebates from manufacturers in exchange for that placement.
OptumRx sits inside UnitedHealth Group. CVS Caremark sits inside CVS Health alongside Aetna. Evernorth, built around Express Scripts, sits inside The Cigna Group.46 Between them they process more than 79 percent of United States prescription claims. Each also operates a specialty pharmacy β Cigna's is Accredo β that dispenses the high-cost injectables and infused therapies where the dollars concentrate.
The leverage here runs in two directions at once, and it is worth being precise about who supplies whom.
Upstream, Eli Lilly and Novo Nordisk supply GLP-1 therapies β Zepbound, Wegovy, Ozempic β into these formularies. The pharmaceutical manufacturers hold patent monopolies, which is real power. The PBMs hold the gate: exclusion from a major formulary removes access to tens of millions of covered lives. The observable outcome is a contested equilibrium in which manufacturers win on list price and PBMs win on net price and volume control, with prior authorisation used as the throttle. Net GLP-1 cost currently runs at roughly $8.50 to $12.00 per member per month across commercial books, a line item that did not meaningfully exist five years ago.6
Downstream, the PBMs supply plan sponsors β including insurance subsidiaries that are corporate siblings, and including competitors. Centene, the largest managed Medicaid carrier, outsources its pharmacy benefit administration to Cigna's Evernorth under a multi-year contract rather than building the capability internally.6 That relationship is worth pausing on. A company with roughly 28 million members and $195 billion of revenue concluded it could not economically replicate what a rival's subsidiary does. Humana likewise sources specialty pharmacy distribution through Evernorth's Accredo.6 These are confirmed commercial relationships disclosed in filings, and they establish something important about switching friction: PBM contracts are multi-year, deeply integrated into claims adjudication and member-facing systems, and expensive to unwind. That is Helmer's switching costs, expressed as an eight-figure implementation project.
Layer four: the care-delivery layer
Some of our thousand dollars pays for primary care. Increasingly, it pays a primary care organisation that the insurer owns or has contracted on a capitated basis.
Optum Health employs or affiliates with more than 90,000 physicians, making UnitedHealth Group one of the largest employers of doctors in the United States.4 CVS Health owns Oak Street Health, a value-based primary care chain built for Medicare patients. Elevance's Carelon aggregates behavioural health, pharmacy services, and care management. Alongside these sit independent enablement platforms: Agilon Health, which takes capitated risk on senior populations in partnership with independent primary care groups, and Privia Health, which supplies technology, contracting, and administrative infrastructure to physician practices without itself absorbing full medical risk.
The difference between those last two is the difference between a business and a hazard, and Section 7 takes it up in detail.
Layer five: acute and specialty providers
Most of the remaining claim dollars leave the enterprise here β to hospitals, ambulatory surgery centres, and specialists. HCA Healthcare and Tenet Healthcare are the largest listed acute-care operators and sit squarely on the receiving end of managed care's utilisation controls, reporting shorter inpatient stays and rising prior-authorisation denial rates. They are outside this article's boundary because they hold no insurance licence, but they are the clearest external evidence of where power sits: a hospital with high fixed costs and a fully committed physical plant negotiates against a payer that can redirect volume.
Layer six: the infrastructure beneath
Finally, stop-loss reinsurers absorb catastrophic claim tails for self-funded employers; healthcare IT and revenue-cycle vendors run the plumbing; analytics and AI vendors supply fraud detection, risk-score capture, and utilisation review models. Optum Insight sells analytics services to health plans that compete with UnitedHealthcare, which is a small, strange, and revealing fact about how thoroughly one company has embedded itself in its rivals' operations.
The migration mechanism, stated plainly
Now put the two halves together.
When the licensed insurer pays a claim to an independent hospital, the money counts toward the statutory MLR and exits the corporate perimeter permanently. When it pays a claim to a wholly owned surgery centre, home-health agency, primary care clinic, or specialty pharmacy, the money still counts toward the MLR β satisfying the regulator β and simultaneously books as revenue in a health services segment where no statutory margin cap applies.
The consequence is that a vertically integrated conglomerate converts a regulated, capped, capital-intensive premium dollar into an unregulated, uncapped, capital-light service dollar, and does it entirely inside the rules. Consolidated return on invested capital in the services layer runs structurally above the insurance layer, which is why the segment mix, and not the membership count, has become the variable that explains valuation dispersion across this sector.
Two objections deserve a hearing. The first is that internal transfer prices could be set to manufacture the appearance of services profit β a concern regulators and the Federal Trade Commission have taken seriously in reviewing PBM rebate practices and intra-company transactions. The second is that owning care delivery imports the very cost inflation the insurer is trying to escape: a clinic with empty exam rooms is a fixed-cost problem, and the pandemic-era physician labour market made that cost inflation real. Both objections are live. Neither has yet overturned the arithmetic, because the routing advantage compounds across tens of millions of members while the operational drag is borne asset by asset.
Monopoly power in American medicine has migrated from owning hospital beds to controlling the gateway through which claims and prescriptions are routed. But that migration did not happen by design in a single boardroom. It happened across fifty years, in five distinct policy shocks, each of which taught the industry the same lesson slightly more expensively.
Section 5: The Five Inflexions: From Nixon's HMO Act to the Great Post-Pandemic Squeeze
On December 29, 1973, President Richard Nixon signed the Health Maintenance Organization Act. The bill was a response to medical inflation that had been running well ahead of general prices since Medicare and Medicaid were created eight years earlier, and its logic was the same logic that governs the industry today: pay a fixed amount in advance, and let the organisation receiving it worry about the cost.
The Act provided federal grants and loans to establish health maintenance organisations, overrode state laws that had restricted prepaid group practice, and required employers above a size threshold to offer a federally qualified HMO alongside their conventional plan where one was available. It did not immediately transform the market β enrolment grew slowly through the 1970s β but it established the legal foundation for prepaid managed care and, more importantly, the political precedent that the federal government would subsidise the transfer of medical risk into private hands.
That precedent has been re-ratified four times since, and each ratification followed the same pattern: a legislative expansion that created a profit pool, a period of aggressive capacity growth, and then a regulatory tightening that consolidated the field.
2003: the Medicare Modernization Act builds the growth engine
The Medicare Modernization Act, signed by President George W. Bush, created the Part D prescription drug benefit and restructured private Medicare plans into what became Medicare Advantage. Crucially, it set benchmark payments that in many counties exceeded what traditional fee-for-service Medicare spent on comparable beneficiaries, which gave private plans room to offer supplemental benefits β dental, vision, gym memberships, zero-dollar premiums β that traditional Medicare does not cover.
That design decision is the origin of everything that followed. Seniors did not choose Medicare Advantage because they preferred managed care; they chose it because it was visibly better value at the point of enrolment. Penetration compounded for two decades on that proposition, and Humana, which had spent the 1990s as a struggling HMO, rebuilt itself entirely around it.
2010: the Affordable Care Act sets the ceiling
The ACA did four things to industry economics simultaneously: it imposed the statutory medical loss ratio floors described earlier, expanded Medicaid eligibility in participating states, created individual exchanges with subsidised premiums, and prohibited underwriting on pre-existing conditions.1
The combined effect was to make health insurance a scale business with a capped margin, which is a formula that mathematically requires consolidation. If you cannot earn more per member, you must serve more members, spread fixed costs further, and find revenue outside the capped premium. Every subsequent strategic decision in this industry follows from that sentence.
Medicaid expansion also created something new: a large, federally funded, state-administered population that private carriers could bid to manage. Centene built a company on it.
2018β2020: the conglomerates assemble
Between 2018 and 2020, three transactions redefined the competitive map.
CVS Health acquired Aetna for approximately $69 billion, combining a retail pharmacy chain, the Caremark PBM, and a national health insurer under one roof. Cigna acquired Express Scripts for approximately $67 billion, creating the business now branded Evernorth.6 Centene acquired WellCare for approximately $17 billion, consolidating government-programme scale.
These were not diversification for its own sake. Read against the ACA's margin cap, they are the same move executed three ways: acquire the layer where the regulated premium dollar can be recognised as unregulated service revenue. UnitedHealth had made that move earlier and organically, which turns out to matter.
2024β2026: the sponsor takes it back
The current inflection has four components arriving together.
CMS phased in the V28 risk model, compressing measured risk scores. It tightened Star Rating cut points, compressing quality bonus revenue. The Inflation Reduction Act redesigned the Part D benefit, capping beneficiary out-of-pocket drug spending at $2,000 in 2025 and $2,100 in 2026 β a genuine improvement for seniors that transferred catastrophic-phase liability from the government and the member onto the plan.1 And post-pandemic utilisation normalised upward as deferred procedures returned, at the same moment GLP-1 prescribing scaled.
Any one of these is manageable. Arriving together, they produced the medical loss ratio expansion documented in the next section and the enrolment flattening shown in Exhibit 1.
There is a myth worth correcting here. The popular framing holds that insurers were caught by surprise by seniors suddenly getting sicker. The evidence points elsewhere. Utilisation surprise explains perhaps half the margin damage and is genuinely cyclical. The V28 phase-in and the Part D redesign were both announced years in advance, quantified in advance, and are permanent. What the industry got wrong was not the forecast; it was the bid. Carriers had spent a decade in a market where buying membership at thin margin was rational because scale and risk-score capture would rescue the economics later. When CMS removed the risk-score rescue, the entire vintage of aggressively priced plans became unprofitable at once.
That is a capital-cycle failure, not a forecasting failure β and it is the classic shape. Regulatory easing invites capacity expansion; capacity expansion competes margin away; regulatory tightening arrives with the capacity already built; the shakeout removes the players who financed growth without a buffer. The industry is in the shakeout phase now. Carriers exited counties, cut supplemental benefits, and in one case exited an entire line of business: Cigna sold its Medicare Advantage operations to Health Care Service Corporation across 2024 and 2025, converting a structurally exposed asset into capital at roughly the top of the regulatory cycle.6
Which raises the obvious question. If everyone faced the same five shocks, why did the outcomes diverge so violently?
Section 6: The Protagonists and the Parameter-Specific Leadership Matrix
There is no single leader in American health insurance, and anyone who names one is describing a different question than they think. The industry contains at least six distinct competitions, and different companies win each of them.
Parameter-specific leadership, United States managed care, as of August 4, 2026
| Competitive parameter | Leader | Closest rival | Evidence | Why the lead exists |
|---|---|---|---|---|
| Overall revenue and medical scale | UnitedHealth Group | CVS Health | UNH ~$448B revenue; ~52M medical members4 | Multi-decade head start in health services integration; scale compounds in network discounts |
| Medicare Advantage share | UnitedHealth Group | Humana | UNH ~28.5% MA share (~9.4M members); HUM ~19.5%2 | Broadest national provider network and county footprint; geographic spread dilutes local rate cuts |
| Medicaid and exchange scale | Centene | Elevance Health | CNC ~13M Medicaid members; ~4.2M ACA marketplace members3 | State-level regulatory capability and low-cost network contracting built over two decades of acquisitions |
| Commercial self-funded (ASO) | The Cigna Group | Elevance Health | CI >15M self-funded lives6 | Employer broker relationships plus Evernorth specialty pharmacy bundling |
| Medicare quality (Star Ratings) | Alignment Healthcare | Kaiser Permanente | ALHC >90% of MA members in 4.5+ Star plans8 | Proprietary clinical workflow platform driving proactive chronic care management |
| Administrative cost efficiency | Molina Healthcare | Centene | MOH SG&A ratio ~7.2%3 | Deliberately minimal corporate overhead; disciplined integration of distressed Medicaid plans |
Measurement note: shares are membership-based as of the most recent 2025β2026 disclosures and are not directly comparable across lines of business, since a Medicaid member and a commercial self-funded life carry very different revenue and margin per head.
Now the second data exhibit, which shows what the five shocks of Section 5 did to each of these competitors.
Exhibit 2 β Consolidated medical loss ratio, major public managed care organisations, 2022β2026 (% of premium revenue)
| Company | 2022 | 2023 | 2024 | 2025 | 2026 guidance / run-rate |
|---|---|---|---|---|---|
| UnitedHealth Group | 82.0% | 83.2% | 85.6% | 87.2% | 86.8%β87.4% |
| Elevance Health | 87.6% | 87.0% | 87.1% | 87.8% | 87.5%β88.0% |
| The Cigna Group | 81.7% | 81.3% | 82.2% | 86.5% | 82.0%β83.0% |
| Humana | 86.3% | 88.0% | 89.9% | 89.2% | 89.5%β90.2% |
| Centene | 87.7% | 87.7% | 88.3% | 88.5% | 88.0%β88.8% |
| Molina Healthcare | 88.0% | 88.1% | 88.4% | 88.2% | 88.0%β88.5% |
Definition: consolidated medical claims and quality-improvement expense as a percentage of premium revenue, as reported by each company; definitions differ modestly across carriers, so levels are less comparable than trends. Geography: United States. Source: company Form 10-K and 10-Q filings.456 Evidence status: 2022β2025 reported; 2026 is company guidance or run-rate, not an observed result.
Read the columns rather than the rows. UnitedHealth's loss ratio expanded roughly 520 basis points from 2022 to 2025 β the largest absolute move on the table β and yet its consolidated operating margin held near 7.8 percent, the highest among the large carriers.4 Humana's expanded roughly 290 basis points and its operating margin fell to roughly 2.1 percent.5 The loss ratio, taken alone, ranks these companies almost backwards.
The reason is segment mix. UnitedHealth derives roughly 42 percent of revenue and more than 45 percent of operating earnings from Optum, so premium-side deterioration hits a shrinking share of consolidated profit.4 Humana has almost nothing of the kind at comparable scale. Cigna's 2025 spike, and its guided reversion toward 82 to 83 percent, largely reflects the mechanics of exiting Medicare Advantage rather than a durable change in underwriting quality.6 This is the single most important interpretive point in the sector: the medical loss ratio measures the insurance segment, and for the conglomerates the insurance segment is no longer where the earnings are.
How UnitedHealth built the buffer
The decision that produced this outcome was made in 2011, when UnitedHealth consolidated its pharmacy, technology, and care-delivery businesses under a single brand. Stephen Hemsley, who ran the company from 2006 to 2017 and returned to the chief executive role in 2025, presided over the creation of Optum as a distinct operating engine rather than a collection of support functions.4
The mechanism is a flywheel with four turns. UnitedHealthcare's membership generates predictable patient volume. Optum Health acquires or affiliates the physicians who see those patients β now more than 90,000. Those physicians refer within Optum's ambulatory and home-based assets. OptumRx fills the prescriptions. Each turn converts a regulated claim dollar into services revenue, and each turn generates data that improves the risk models, network design, and utilisation management sold back through Optum Insight to competing plans.
Why has nobody copied it? Two reasons, both structural. Time is one: Optum was assembled over fifteen years of continuous acquisition at pre-2021 valuations, and rebuilding it today would require buying physician assets at prices set by a decade of competitive bidding. Antitrust is the other: the Federal Trade Commission and the Department of Justice now scrutinise payer acquisitions of provider assets and PBM rebate practices in a way they did not in 2013. The lead is durable partly because the door it came through has been narrowed. It is not permanently safe β an adverse structural ruling on payer-provider ownership is the one event that unwinds it, which is why it appears in this article's falsification criteria.
Cigna: winning by refusing to play
David Cordani has led Cigna since 2009, and his signature decision was declining to compete where the government sets the price.
Cigna concentrated on commercial employers, particularly self-funded arrangements, where the employer bears the medical risk and Cigna earns an administrative fee. Fee revenue does not carry underwriting variance, does not require the same statutory capital, and is not repriced by a federal agency each April. On top of that base, Evernorth monetises pharmacy benefit management and specialty distribution through Accredo, contributing roughly 70 percent of consolidated operating income against roughly 30 percent from the insurance business.6
Then Cigna sold its Medicare Advantage book to HCSC. Judged against the rate environment that followed, the timing looks excellent. Judged as a strategy, it leaves an obvious hole: Cigna has voluntarily exited the only demographically guaranteed growth market in American health insurance. A long-only investor benchmarked to healthcare indices is buying stability and low regulatory beta; a long/short investor should note that the same decision caps the upside if Medicare Advantage rates normalise and pure-play margins recover.
Elevance: the franchise with a licence problem it also benefits from
Elevance Health operates Blue Cross Blue Shield plans in fourteen states, serving roughly 47 million medical members on roughly $199 billion of revenue at an operating margin near 5.2 percent. The Blue licence is a genuine cornered resource β brand recognition and provider network depth accumulated over decades, in geographies where no competitor can use the mark. It is also a constraint: Elevance cannot simply enter New York or California under the Blue brand, so national expansion has to run through Carelon services or non-Blue products.
Carelon, contributing roughly 18 percent of revenue, is Elevance's answer to Optum, and it is roughly a decade behind. The company's exposure runs through commercial group repricing and, more acutely, Medicaid β where the post-pandemic redeterminations that removed millions from state rolls left behind a residual population that is, on average, sicker per member than the one that departed.
CVS Health: the most integrated and the most encumbered
CVS Health owns more of the chain than anyone: retail pharmacies, the Caremark PBM, Aetna's insurance licences, and Oak Street Health's primary care clinics. On roughly $365 billion of revenue it earns an operating margin near 3.8 percent β the lowest among the majors β and carries substantial debt from the Aetna and Oak Street acquisitions.
The company is the cleanest test of whether vertical integration creates value or merely aggregates it. Aetna's Medicare Advantage book suffered badly in the 2024β2026 squeeze, and the retail pharmacy footprint faces structural decline in front-of-store economics. That combination produces a valuation near 10.2 times forward earnings β the market extending limited credit to the integration thesis until Aetna's loss ratio proves it can normalise. The strongest bull argument is that all the pieces are present and only the insurance segment is broken; the strongest bear argument is that owning every layer of a low-margin chain compounds fixed cost rather than diluting risk.
Centene and Molina: the same customer, two philosophies
Both companies serve government programmes almost exclusively, and their strategies diverge completely.
Centene assembled its position through acquisition β Health Net in 2016, Fidelis Care in 2018, WellCare in 2020 β reaching roughly 28 million members and building Ambetter into the largest ACA marketplace franchise. Scale gives it state-by-state political capability and network leverage. It also gives it concentrated exposure to two policy variables it does not control: state Medicaid re-procurements that can strip a multi-billion-dollar contract on a scoring decision, and the enhanced federal ACA premium subsidies whose expiration would shrink marketplace enrolment substantially. At roughly 9.8 times forward earnings on $195 billion of revenue and a 2.4 percent operating margin, the market is applying a visible discount for that policy exposure rather than for operating quality.
Molina buys distressed and sub-scale Medicaid plans and integrates them onto a deliberately thin cost base, running roughly $40 billion of revenue at a 4.1 percent operating margin β better than Centene's despite a fraction of the scale. Its lead is execution discipline in a defined niche, and the honest limit on that lead is that it does not generalise: Molina has no PBM of consequence, no meaningful services segment, and therefore none of the buffer that protected UnitedHealth.
The insurtech cohort, five years after the hype
Three listed companies represent the technology-native approach, and they have arrived at three different places.
Oscar Health built a cloud-native claims and member-engagement platform and concentrated it on the individual ACA marketplace, where more than 90 percent of its revenue originates. Revenue reached roughly $11.7 billion in 2025 with 2026 guidance near $18.8 billion, alongside a turn to profitability.7 The growth is real and the platform advantage in member acquisition and servicing costs is credible. The concentration risk is equally real: Oscar's business is a single-line bet on the individual exchanges, and the enhanced federal subsidies that inflated exchange enrolment are the subject of an unresolved congressional debate. At roughly 22.5 times forward earnings, Oscar carries the sector's highest multiple on the sector's most policy-dependent revenue base.
Alignment Healthcare, discussed earlier for its Star Ratings concentration, grew revenue from roughly $3.95 billion in 2025 toward approximately $5.18 billion guided for 2026.8 It is the clearest evidence that a technology-forward clinical model can win on quality inside Medicare Advantage. Whether it can do so at national scale, in geographies where it lacks the California provider relationships it built first, is not yet demonstrated.
Clover Health operates Medicare Advantage plans in New Jersey, Georgia, and South Carolina, with roughly $1.5 billion of 2025 revenue and approximately $2.85 billion guided for 2026, alongside a physician-facing software product called Counterpart Assistant that surfaces risk and care-gap prompts at the point of care. The strategic question for Clover is whether the software is a genuine second business or an internal tool with an external label; its scale relative to national Medicare Advantage incumbents leaves little margin for that question to be answered slowly.
The organisations with no share price
Two of the most instructive competitors are not investable.
Kaiser Permanente operates roughly $110 billion of revenue and about 12.5 million members across a closed-loop system in which the health plan, the hospitals, and the physician groups are structurally aligned. It is the purest expression of the upstream belief in this entire article β full ownership of both the risk and the delivery β and it has existed for eighty years. Its constraint is the reason nobody has replicated it at national scale: enormous capital tied up in owned hospital real estate and a heavily unionised clinical workforce, which makes geographic expansion slow and capital-intensive in exactly the way Optum's asset-light physician affiliation model is not.
Health Care Service Corporation is the largest customer-owned mutual health insurer, running roughly $60 billion of revenue and about 18 million members as the Blue Cross licensee in Illinois, Texas, Oklahoma, New Mexico, and Montana. Its mutual structure removes quarterly earnings pressure entirely, which is precisely why it was willing to buy Cigna's Medicare Advantage business into a deteriorating rate environment.6 It also cannot issue equity, which caps its ability to compete for large acquisitions. Because HCSC does not file with the SEC on the same basis as public carriers, its margins and loss ratios are not directly comparable to Exhibit 2, and any ranking that includes it should say so.
Market share without service integration has become a liability in a tightening regulatory regime. Which means the central question for an investor is no longer who has the most members β it is whether a given company's cash flows actually come from where its story says they do.
Section 7: Exposure Proof, Keyword Traps, and the Agilon Cautionary Tale
In 2021, an investor could be shown a pitch offering pure exposure to value-based care without the drag of an insurance balance sheet. The proposition was elegant. Insurance companies are capital-intensive, regulated, and slow. Physicians actually control medical cost. So partner with independent primary care groups, take capitated risk on their senior patients, keep a share of the savings, and skip the licence, the statutory capital, and the state regulators entirely.
Agilon Health went public on that proposition. By 2024 and 2025 it was recording severe negative revisions to its medical margin.
The mechanism of the failure is worth understanding precisely, because it generalises. Agilon accepted full capitated risk for senior medical expense β the same risk a Medicare Advantage plan accepts β under contracts with UnitedHealthcare, Humana, and Aetna, among others.45 What it did not accept, and could not, were the things that make that risk survivable.
It had no premium float, so it could not earn investment income on collected-but-unpaid claims. It had no PBM, so it captured none of the pharmacy economics on the drugs its patients consumed. It had no diversified segment mix, so a bad utilisation year hit one hundred percent of its earnings. And critically, it had no seat at the bid: when CMS changed the risk model, the health plans repriced their bids in response, while Agilon's contracts had already fixed its economics against the old assumptions. It absorbed a regulatory change it had no mechanism to pass on.
The generalisable rule: taking underwriting risk without holding underwriting infrastructure is the most asymmetric position available in healthcare investing. The upside is a share of savings. The downside is the full variance of medical cost. Insurers accept that trade because float, scale, reserving discipline, statutory capital, and annual repricing are the compensating instruments. Strip those away and what remains is a leveraged bet on senior utilisation with no hedge.
Note the contrast with Privia Health, which serves a similar physician customer with a deliberately different model: technology, contracting, and administrative infrastructure sold to practices, with far less full-risk capitation on the balance sheet. Privia is a services business with services economics. Agilon marketed itself as a services business and carried insurance risk. The label was the same; the cash flow statement was not.
The keyword trap
The broader version of this error is thematic. An investor who screens for "value-based care" or "AI in healthcare" will surface companies whose language matches the theme and whose economics do not.
The early insurtech thesis held that superior algorithms could underwrite health risk more accurately than incumbents and thereby earn structurally better loss ratios. Three features of the industry made that difficult in ways software does not address. Statutory MLR floors cap the reward for underwriting skill β an insurer that achieves an 82 percent loss ratio in a market with an 85 percent floor rebates the difference. State network adequacy rules require contracted provider coverage that must be negotiated locally, hospital by hospital, at rates that depend on the volume you already have. And CMS risk-adjustment audits mean aggressive coding is a compliance liability rather than an edge.
The cohort thinned accordingly. Bright Health Group, which grew exchange membership rapidly and then encountered claims it had not reserved for, exited health plan underwriting altogether after severe losses and redirected itself toward clinical care management.3 Oscar Health survived and turned profitable, which is genuine evidence that the technology approach can work β but Oscar's advantage shows up in member acquisition cost and administrative efficiency rather than in beating the loss ratio, and its 2026 revenue trajectory owes as much to federal subsidy design as to its engineering.7
The exposure ledger
Sorting the universe by where cash actually originates produces four categories.
Diversified beneficiaries hold both the licence and the services engine. UnitedHealth, with Optum contributing more than 45 percent of operating earnings, is the archetype.4 Cigna, with Evernorth contributing roughly 70 percent of operating income, is the most extreme example β arguably a pharmacy services company that also sells insurance.6 Elevance, with Carelon at roughly 18 percent of revenue, holds a partial version.
Pure-play underwriters hold the licence and nothing else. Humana on Medicare Advantage, Centene and Molina on government programmes, Oscar on the individual exchanges, Alignment on Medicare Advantage. Their exposure to the theme is total and unbuffered, which cuts both ways: they carry the full downside of rate compression and the full upside of rate normalisation. For a long/short investor, this cohort is where operating leverage actually lives. For a benchmark-relative long-only investor, it is where tracking error concentrates.
Enablers supply capability without holding the licence β Privia in physician infrastructure, healthcare IT and analytics vendors, stop-loss reinsurers. Their exposure is to industry activity levels rather than to underwriting outcomes.
False positives carry thematic language and mismatched economics. Agilon is the documented case.
The attribution rule that separates these is simple to state and tedious to apply: before assigning a thematic multiple, verify whether the cash flow originates in underwriting premium, administrative fee, service capitation, or software licence β because each carries a different variance, a different capital requirement, and a different regulator.
The rejection tests
A senior investor would reject each of these names for a specific, respectable reason, and it is worth naming them without advocacy.
UnitedHealth: the entire thesis rests on the durability of payer-provider integration, which is the most exposed structure in the sector to a single adverse antitrust or legislative outcome. Cigna: the company has removed its exposure to the fastest-growing customer segment in American healthcare, and PBM economics are the specific target of bipartisan legislative attention. Humana: a genuine recovery requires both rate normalisation and Star Ratings restoration, two independent events on different clocks. Centene: two policy variables β state re-procurements and ACA subsidy extension β can each move earnings materially and neither is forecastable from company disclosure. Oscar: valuation already reflects the profitability turn, leaving limited room for error if exchange enrolment contracts. Alignment and Clover: sub-scale operators in a business where scale is the primary defence against exactly the shock the industry just experienced.
Each of those objections is answerable only by observing what happens next β which is why the honest form of this analysis is a set of scenarios with different causal machinery, rather than a point estimate with error bars.
Section 8: The Scenario Matrix: Three Causal Worlds for 2026β2028
Forecasting this industry by applying percentage haircuts to consensus estimates misses the point, because the variables do not move independently. Rate updates, utilisation, drug costs, and enrolment are linked through the bid. Three internally coherent worlds are more useful than one adjusted number.
Three causal worlds for United States managed care, 2026β2028
| Dimension | Bear: regulatory squeeze and utilisation spike | Base: disciplined repricing | Bull: technology productivity and policy relief |
|---|---|---|---|
| MA penetration by 2028 | 52.0% (net retrenchment) | 56.0% (low single-digit expansion) | 60.0%+ (accelerated adoption) |
| Industry blended MLR | 90.0%+ | 87.5%β88.2% | 85.5%β86.5% |
| CMS effective net rate updates | β1.0% to β2.5% sustained | +0.5% to +1.5% | +2.5% or better |
| Principal beneficiary | Commercial-weighted carriers with large self-funded books (Cigna) | Vertically integrated conglomerates (UnitedHealth, Elevance) | Technology-native operators (Oscar, Alignment) |
| Principal casualty | Pure-play MA carriers and unbuffered risk-takers (Humana, Agilon) | Regional plans without captive services | Fee-for-service hospital systems unable to manage capitated risk |
Evidence status: scenario construction, not observed data. Penetration and MLR figures are analytical constructions built from the drivers described below, not forecasts published by any named institution.
The bear mechanism
The bear world is not simply "rates go down." It is a specific sequence.
CMS sustains negative effective net rate updates while medical cost trend runs at or above five percent. GLP-1 utilisation broadens β through expanded clinical indications or coverage mandates that restrict prior authorisation β pushing pharmacy trend past what formulary management can absorb. Carriers respond the only way the bid permits: they cut supplemental benefits further and withdraw from more counties.
Then the feedback loop closes. Thinner benefits reduce the value proposition that drove Medicare Advantage adoption in the first place. Seniors in exited counties return to traditional fee-for-service Medicare. Penetration falls for the first time in two decades. Fixed administrative costs spread across a shrinking membership base, worsening the very margins the benefit cuts were meant to protect.
In this world, Cigna's decision to exit Medicare Advantage looks prescient rather than merely well-timed, and its self-funded commercial book β where the employer bears the medical risk β becomes the most defensive revenue stream in the sector. Humana and the unbuffered risk-takers face the worst version of their existing problem.
The base mechanism
The base world assumes the industry does what regulated oligopolies usually do when the sponsor cuts the price: it reprices, in disciplined fashion, because all the major participants face the same arithmetic simultaneously and none of them has an incentive to buy share into a loss.
Carriers trim the most discretionary supplemental benefits, narrow networks toward lower-cost providers, tighten utilisation management, and accept membership loss as the cost of margin recovery. Blended loss ratios settle in the 87.5 to 88.2 percent range β above the pre-2023 baseline, permanently, because V28 and the Part D redesign do not reverse β and underwriting margins normalise at a structurally lower level than the 2015β2022 era.
In that setting the winners are the companies for whom the insurance margin is a minority of consolidated earnings. UnitedHealth and, to a lesser degree, Elevance capture the difference between a repriced premium dollar and an internally delivered cost.
The bull mechanism
The bull world requires two independent things to go right.
The first is administrative automation delivering genuine operating leverage: AI-assisted prior authorisation, claims adjudication, and clinical documentation reducing SG&A faster than premium grows. This is plausible and partially observable in company disclosures, though it is also the claim most vulnerable to the regulatory risk discussed in the next section.
The second is a pharmacy cost reversal. Effective oral small-molecule GLP-1 therapies at substantially lower unit cost would convert the largest new line item in the pharmacy budget from a margin drag into an affordable preventive intervention β and, over a longer horizon, into a reduction in cardiometabolic claims.
In this world the operators with the lowest administrative cost per member and the most flexible technology stacks gain disproportionately, which is where Oscar and Alignment have their strongest case.
The variant view
Consensus, as reflected in sell-side modelling, treats Medicare Advantage margin recovery as a roughly twenty-four-month repricing exercise: cut benefits, raise bids, watch loss ratios revert toward the mid-80s.
The variant view this evidence supports is that recovery takes thirty-six to forty-eight months for pure-play carriers, for three reasons that compound. The V28 phase-in is multi-year and its full effect on measured risk scores is still working through the book. The Part D redesign permanently shifted catastrophic liability onto plans with no offsetting revenue mechanism. And Star Ratings operate on a two-year lag between measurement and payment, so a carrier that fixes its quality metrics in 2026 does not collect the bonus until 2028.
That is a timing disagreement rather than a directional one, and timing disagreements are where thematic investors most often lose money on correct theses. An investor who is right that Medicare Advantage margins normalise, and wrong by two years about when, will have owned the operating leverage through the drawdown and sold it before the recovery.
The capital-cycle read reinforces the same caution. This industry is in shakeout, not deployment: capacity is being withdrawn, marginal operators are exiting, and returns on incremental capital in pure underwriting are below the cost of that capital. Shakeouts eventually produce excellent entry points for the survivors. They rarely produce them on the schedule the initial thesis assumed.
Which makes the question of what to watch, and in what order, more important than the question of what to conclude.
Section 9: The Crux KPIs, Falsifiers, and the Terminal Value Migration
Most healthcare investors monitor earnings. Earnings in this industry are a lagging summary of decisions made eighteen months earlier in a bid filing, under assumptions set by a federal announcement published the previous April. By the time a loss ratio appears in a quarterly report, the information that determined it has been public for a year.
Five observables sit upstream of that reported result. Each measures something causally prior to revenue, share, or margin; each has a named public source and a known publication cadence; and each discriminates between the bear and bull worlds described above.
Crux KPI monitoring set, as of August 4, 2026
| KPI | Latest reading | Why it leads | What it discriminates | Confirm / break | Source and cadence |
|---|---|---|---|---|---|
| CMS Medicare Advantage effective rate update | +3.70% headline effective growth1 | Sets the revenue ceiling on bids filed nine months before the plan year begins | Whether the sponsor is restoring or continuing to withdraw margin | Confirm: >+3.0% net of model changes. Break: <+0.5% | CMS Rate Announcement, annually in April |
| Blended loss ratio, top four carriers | 87.8%45 | Coincident measure of whether bid pricing matched realised cost | Repricing discipline versus runaway cost trend | Confirm: <86.5%. Break: >89.5% sustained three quarters | Company 10-Q and 10-K, quarterly |
| Share of MA members in 4.0+ Star plans | ~62%, down from ~74% in 20232 | Determines quality bonus revenue twelve to twenty-four months forward | Whether bonus revenue returns to the sector or stays impaired | Confirm: >75%. Break: <55% | CMS Star Ratings release, annually in October |
| GLP-1 net cost per member per month | ~$10.20 commercial average6 | Leads specialty pharmacy trend and reveals whether formulary control is holding | Whether PBM gatekeeping works against a demand shock | Confirm: <$11.00. Break: >$18.00 | PBM drug trend reports, semi-annual |
| Enhanced ACA premium subsidy status | Active, subject to congressional extension debate1 | Determines individual marketplace enrolment before any carrier reports it | Whether exchange-exposed carriers keep their revenue base | Confirm: extension enacted. Break: statutory expiry | Congressional action and CBO scoring, as legislated |
A definitional caution on the first line, because it is the most misread number in this sector. The headline effective growth rate CMS publishes is a gross benchmark input. The net change a plan actually experiences is that figure adjusted for the risk-model phase-in, the coding-intensity adjustment, and the plan's own Star Rating. That is why the same announcement can be reported accurately as a rate increase and experienced accurately as a cut. Any monitoring framework that tracks only the headline will miss the transmission entirely; the number that matters is the net effect after model and coding adjustments, which each carrier quantifies in its own guidance.
Where the indicators sit in the hierarchy
The rate update and the subsidy status are structural indicators β they change the size and shape of the market. Star share is an industry indicator that resolves annually and moves revenue mechanically. The loss ratio is a company indicator and the closest thing to a coincident scorecard. GLP-1 cost sits between industry and company, because formulary strategy differs meaningfully across the three large PBMs.
Theme kill criteria versus security kill criteria
These are different things, and conflating them is expensive.
The theme dies if federal law prohibits health plans from owning PBMs, specialty pharmacies, or medical practices; if Medicare Advantage experiences multi-year net disenrolment back to fee-for-service; or if a public option caps commercial reimbursement near Medicare rates. Any of the three invalidates the upstream belief itself.
A security thesis can die while the theme thrives. Humana can fail to restore Star Ratings while Medicare Advantage grows. Centene can lose a state re-procurement in a year when Medicaid managed care expands. Oscar can be right about exchange technology and wrong about the subsidy vote. The distinction matters most for the long/short investor, who can be correct on the industry and lose on the position; the long-only benchmark-relative investor faces the inverse risk, of being right on the securities while the sector's index weight compresses on policy headlines.
The factor exposures hiding inside a sector basket
Owning six managed care names is not diversification, and the reasons are specific.
Every one of them carries the same regulatory duration: the April Rate Announcement and the October Star Ratings release move the entire sector's market value on the same afternoons, regardless of individual operating quality. Every one of them holds a large short-duration fixed income portfolio, so a sustained decline in short rates reduces float income across all of them simultaneously. And every one of them experiences multiple compression during federal election cycles when single-payer and drug-pricing rhetoric intensifies β a repeatedly observed pattern that has little to do with realised policy and a great deal to do with headline risk.
There is one further hidden exposure worth naming. Because the three dominant PBMs sit inside three of the largest carriers, a legislative attack on PBM economics is simultaneously an attack on UnitedHealth, Cigna, and CVS β three companies an investor might reasonably believe are diversified against one another.
Three developments that could move the value pool
Restrictions on algorithmic prior authorisation. Insurers use automated models to triage medical necessity reviews and claim denials. Class-action litigation and state insurance department investigations have challenged those tools, and physician groups allege error rates on automated initial denials exceeding twenty percent β a figure insurers dispute, and which no neutral authority has established. The commercial mechanism is straightforward: if regulators require clinician review of automated denials, administrative cost rises and previously denied claims get paid, hitting both SG&A and the loss ratio. Beneficiaries would be providers, particularly hospital systems whose denial-appeal burden falls. Losers would be the carriers that have automated most aggressively, and the observable milestone is state-level legislation rather than federal, since insurance regulation is primarily a state function.
Oral small-molecule GLP-1s. Current therapies are injectable and expensive. Effective oral alternatives at lower unit cost would change the economics twice: immediately, by reducing pharmacy spend per treated member, and eventually, by making broad treatment of obesity affordable enough that the cardiovascular claims reduction β which actuarial data suggests lags treatment by five or more years, well beyond typical commercial member tenure β could actually accrue to the same payer. This is the sharpest live dispute in the sector: manufacturers argue GLP-1s pay for themselves; payers' own data shows immediate pharmacy cost increases with offsets arriving after the member has changed employers and plans. Both sides are describing the same clinical evidence over different time horizons. The distinction between announced research and scaled commercial adoption matters here; approval and price are separate events, and formulary placement is a third.
PBM unbundling. Bipartisan legislative proposals would separate PBMs from insurers or eliminate rebate retention. The direct effect would strip Evernorth, OptumRx, and Caremark of a core profit mechanism. The second-order effect is more interesting and less discussed: if the pharmacy routing channel closes, the conglomerates' incentive to route value through clinical delivery β Optum Health's clinics, Oak Street's centres, home-based care β intensifies rather than disappears. The profit pool would migrate within the vertical structure rather than out of it, unless the legislation also reached provider ownership. That distinction is the difference between an earnings event and a thesis-ending event.
Rotation logic, stated as analysis rather than instruction
The analytical relationship between the defensive and cyclical expressions in this sector is legible. Commercial-weighted, fee-based revenue outperforms while rates compress and utilisation runs hot. Pure-play Medicare Advantage operating leverage works in the other direction, and it inflects on evidence rather than on sentiment: sustained loss ratios below the mid-86s alongside a positive net rate update and improving Star share. Those are observable, dated, and public. I am describing the mechanism, not recommending a trade, a weight, or a moment.
Tracking this industry means watching regulatory feedback loops rather than earnings reports. By the time the earnings arrive, the value has already moved.
Section 10: Institutional Glossary and Thematic Investor Playbook
Everything above resolves into a small number of structural judgments, and it is worth assembling them in one place.
The first is about where cash originates. In a business where the statute caps what can be earned on premium, the proportion of consolidated earnings generated outside the regulated premium is the single most explanatory variable in the sector. It explains why UnitedHealth's loss ratio deteriorated more than Humana's while its margin held. It explains why Cigna's insurance segment is close to a rounding error in its operating income. It explains why Molina's operating discipline, genuine as it is, could not have protected it from a Medicare Advantage shock the way Optum protected UnitedHealthcare.
The second is about the nature of the roles available. Rather than assign weights β which would require knowing a mandate, a benchmark, a risk budget, and a price, none of which this analysis has β it is more useful to name what each expression actually is.
The integrated conglomerates (UnitedHealth, Cigna, and in partial form Elevance) are structural expressions of the upstream belief. Their business quality is high, their regulatory concentration risk is also high, and the two are the same fact viewed from different angles. The regulated operators (Elevance's Blue franchise, Molina's Medicaid book) are franchise businesses whose value rests on a licence or a cost position rather than on integration; they generate consistent free cash flow and offer limited upside to the theme itself. The unbuffered pure plays (Humana, Centene, Oscar, Alignment, Clover) carry the theme's full operating leverage in both directions, and their outcomes depend on policy events with known dates and unknown contents. The enablers supply capability without absorbing underwriting variance. And the false positives β Agilon being the documented case β carry the theme's risk with none of its instruments.
The third is about what an investor should refuse to conclude. A large end market does not establish per-share value. Roughly $4.8 trillion of national health expenditure has been available to this industry throughout the period in which pure-play underwriting margins collapsed. Membership growth does not establish it either: Medicare Advantage penetration rose every year through the worst margin compression in two decades. What establishes it is the durability of the mechanism by which a regulated dollar becomes an unregulated one, priced against what the market already assumes about that mechanism.
Which brings the analysis to price.
Exhibit 3 β Financial and valuation snapshot, major listed managed care organisations, August 4, 2026
| Company | Market cap / EV | Revenue (2025β2026 run-rate) | Operating margin | Forward P/E (2026E) |
|---|---|---|---|---|
| UnitedHealth Group (UNH) | ~$450B / ~$495B | ~$448.0B | ~7.8% | ~18.2x |
| The Cigna Group (CI) | ~$95B / ~$120B | ~$235.0B | ~4.8% | ~11.8x |
| Elevance Health (ELV) | ~$110B / ~$132B | ~$199.0B | ~5.2% | ~13.5x |
| CVS Health (CVS) | ~$100B / ~$175B | ~$365.0B | ~3.8% | ~10.2x |
| Centene (CNC) | ~$36B / ~$45B | ~$195.0B | ~2.4% | ~9.8x |
| Humana (HUM) | ~$35B / ~$48B | ~$118.0B | ~2.1% | ~15.5x |
| Molina Healthcare (MOH) | ~$22B / ~$24B | ~$40.0B | ~4.1% | ~12.4x |
| Oscar Health (OSCR) | ~$9.4B / ~$8.8B | ~$18.8B (2026E) | ~2.8% | ~22.5x |
Definition: consolidated revenue and operating margin as reported or guided; forward P/E on 2026 consensus estimates. Geography: United States. Evidence status: revenue and margin are company-reported or company-guided; multiples are market data as of August 4, 2026 and are estimates, not company disclosure. Source: company investor disclosures.4567 Direct links in this article's source register cover UnitedHealth, Cigna, Humana, Oscar, and Alignment; figures for Elevance, CVS Health, Centene, and Molina are drawn from those companies' own SEC filings and are not separately linked here.
Read the table for its internal contradictions rather than its rankings. Humana trades at roughly fifteen and a half times forward earnings on a 2.1 percent operating margin, while Centene trades at under ten times on a 2.4 percent margin. Both are pure-play government-programme operators with compressed profitability, and the market is applying a six-turn premium to one of them. That gap is a statement about recovery: Humana's earnings are depressed against a business the market expects to normalise, while Centene's are exposed to policy variables the market does not expect to resolve favourably. Whether that distinction survives contact with the 2027 bid cycle is the most concentrated single disagreement in the sector.
UnitedHealth's roughly eighteen times, on the highest margin in the group, is the price of the buffer. It requires the payer-provider structure to remain legal and the Optum earnings mix to remain durable. Oscar's roughly twenty-two and a half times requires the enhanced exchange subsidies to survive, growth to continue at something like the guided pace, and margins to expand from a thin base. Both are legible requirements, and both are testable against the KPIs above.
Does the upstream belief still hold?
The proposition we started with was that America is completing an irreversible migration to government-sponsored managed care, and that value accrues only to entities holding both the risk-bearing licence and the care-delivery engine.
Three years of the hardest regulatory environment in two decades have tested it, and the verdict is split in a specific and useful way.
The first half is holding and arguably strengthening. Medicare Advantage penetration crossed the majority threshold and stayed there through benefit cuts and county exits. Employers, given the chance to shed insurers entirely by self-funding, kept buying the infrastructure. Physicians kept consolidating into payer-affiliated organisations. The migration continued through conditions designed to slow it.
The second half is holding, and is more fragile than it looks. The earnings divergence between integrated conglomerates and pure-play underwriters through 2024 to 2026 is the clearest evidence anyone could ask for that the integration premium is real. But the mechanism generating it depends entirely on regulatory permission, and that permission is now under simultaneous pressure from antitrust review of payer-provider transactions, bipartisan PBM legislation, and state-level scrutiny of automated utilisation management. The belief is not fraying at the demand end. It is fraying, if anywhere, at the point where the industry's own success has made its architecture politically visible.
That is the tension a thematic investor is actually underwriting in American health insurance: a structural migration that keeps working, monetised through a mechanism that works precisely because a statute did not anticipate it. The migration is a demographic fact. The mechanism is a policy choice, and policy choices get revisited.
Glossary
Administrative Services Only (ASO) β An arrangement in which an employer funds its own employees' medical claims and pays an insurer a per-member fee to administer them. It matters because it strips underwriting risk from the insurer while preserving the network discount and fee revenue, and it now covers most insured American workers.
Capitation β A fixed monthly payment per enrolled member, paid by a sponsor to a health plan or provider group regardless of services consumed. It is the mechanism by which governments convert open-ended entitlement liabilities into budgeted expense, and the reason insurers bear medical cost variance.
Dual-Eligible Special Needs Plan (D-SNP) β A Medicare Advantage plan for people who qualify for both Medicare and Medicaid. These members are among the highest-cost and highest-revenue populations in managed care, and coordinating both funding streams is a distinct operating competence.
Hierarchical Condition Category (HCC) / V28 model β CMS's risk-adjustment methodology, which converts documented diagnoses into a risk score that multiplies a plan's benchmark payment. The V28 revision removed thousands of diagnosis codes, reducing measured risk and therefore revenue, and is the single largest structural driver of the 2024β2026 margin compression.
Medical Loss Ratio (MLR) β Medical claims and quality-improvement spending as a share of premium revenue. Statutory minimums of 85 percent for Medicare and large group and 80 percent for individual and small group cap insurer gross margin on regulated premium, which is the constraint that pushed profit into unregulated service segments.
Operating float β Premiums collected in advance of claims paid thirty to ninety days later, invested in short-duration fixed income. It is a meaningful earnings contributor at current short rates and a shared interest-rate exposure across every carrier in the sector.
Pharmacy Benefit Manager (PBM) β The intermediary that sets drug formularies, negotiates manufacturer rebates, operates mail and specialty pharmacies, and adjudicates prescription claims. The three largest process more than four-fifths of United States prescriptions and all three sit inside insurance conglomerates, making formulary placement the industry's most concentrated chokepoint.
Prior Authorisation (PA) β Advance plan approval required before certain procedures, tests, or prescriptions. It is the primary lever through which insurers control utilisation, increasingly automated, and consequently the primary target of litigation and state regulatory attention.
Risk-Based Capital (RBC) ratio β The statutory capital buffer a licensed insurer must hold relative to its risk exposure, enforced by state regulators. Capital above roughly 250 to 350 percent of the authorised control level can be dividended to the parent; capital below it is trapped inside the licensed entity precisely when the parent most wants it.
Star Ratings β CMS's annual one-to-five quality score for Medicare Advantage plans. Plans at 4.0 and above earn a five percent bonus on their benchmark, which makes a methodological change to the rating cut points a multi-billion-dollar revenue event operating on a two-year lag.
References
-
Announcement of Calendar Year 2025/2026 Medicare Advantage Capitation Rates and Part C and Part D Payment Policies β Centers for Medicare & Medicaid Services, April 2024, updated 2025 ↩↩↩↩↩↩↩↩↩↩↩
-
Medicare Advantage in 2025: Enrollment Update and Key Trends β KFF, 2025β2026 ↩↩↩↩↩
-
2024β2025 Accident and Health Market Share Report β National Association of Insurance Commissioners, 2025 ↩↩↩↩↩↩↩↩
-
Investor disclosures, 2025 Annual Report (Form 10-K) and Q2 2026 results β UnitedHealth Group, 2025β2026 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
2025 Annual Report (Form 10-K) and investor materials β Humana Inc., 2025β2026 ↩↩↩↩↩↩↩↩
-
Form 10-K and Evernorth Health Services operational review β The Cigna Group, 2025β2026 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
2025 Form 10-K and 2026 financial guidance β Oscar Health, Inc., 2026 ↩↩↩
-
2025 Annual Report (Form 10-K) and Q1 2026 earnings release β Alignment Healthcare, Inc., 2026 ↩↩↩
LinkedIn
WhatsApp
Reddit
Hacker News
Email
Kindle
RSS Feed
Spotify
Apple Podcasts
Amazon Music
Audible
YouTube