The American
Uninsured

There is no contradiction in those two sentences. There is only the distance between the two ways America counts what its medical care costs: the way that produces a comforting number for a press release, and the way that produces a family sleeping in a Honda in a Walmart parking lot. This essay is about how to tell those two countings apart. It is also about a single method, durable and portable, for never again being fooled by the first kind. The method is simple enough to state in one sentence, and the rest of this is just proof that it works: when the government hands you a headline built on a survey or an index, do not argue with it, walk around it, find the number built from dollars that someone was legally required to report, and measure how far the two sit apart. The distance is the lie.

Healthcare is the cleanest place in American life to run that test, because the two numbers are not close. They are not even in the same universe.

IThe Number Built to Look Small

The headline is that 92 percent of Americans have health insurance and only 27.1 million people, about 8 percent, are uninsured. The Census Bureau publishes it¹ every September. Every official, of either party, who wants to say the system fundamentally works reaches for that 8 percent the way a drowning man reaches for a rope.

Now read the definition, because the trick is buried in a single word. You are counted as uninsured only if you had zero coverage for all 365 days of the year. One covered day, just one, and the survey files you under insured. Lose your employer plan in March when you lose the job, go bare through the summer, claw back onto something in December, and the United States government records you, officially, as a person with health insurance that year. The number does not measure who can get care. It does not measure who can afford care. It measures who was totally, continuously, uninterruptedly bare for an entire calendar year, which is, by deliberate design, the smallest group the question is capable of producing. It is the U-3 of healthcare: technically accurate, and built from the ground up to be small.

The Census cannot even hold its own story together across its own instruments. The Current Population Survey, which asks whether you had coverage at any point in the year, reported the rate held flat. The American Community Survey, the largest household survey in the entire country, which measures coverage at a single point in time, reported the rate rose, from 7.9 to 8.2 percent² in one year. Same bureau. Same calendar year. Two surveys, two answers, and the divergence is driven by people churning in and out of Medicaid as the pandemic-era continuous-enrollment rule expired and the states began throwing people off the rolls. The "92 percent covered" figure is not a fact about the condition of the country. It is an artifact of which survey a person decides to quote and exactly how the question was worded. Change the instrument, change the truth. That is not how facts behave. That is how marketing behaves.

So ask the harder question. Not do you possess an insurance card, which is what the Census effectively asks, but are you actually covered, and can you afford to walk into a doctor's office and use it. The Commonwealth Fund asked precisely that³ of working-age adults, and the answer is the real silhouette of the thing. Only 56 percent had adequate coverage for the full year. The remaining 44 percent fell into one of three holes: uninsured outright, insured but hit a coverage gap during the year, or underinsured, holding a card they could not afford to use because their out-of-pocket exposure consumed more than a tenth of their household income, or more than a twentieth if they were already poor. On a base of roughly 196 million working-age adults, that 44 percent works out to about 86 million people.

Hold those against each other. The distance between them, nearly 60 million Americans, is the entire population the survey was constructed not to see. They have cards. They have deductions on their pay stubs. They are, in the language of the press release, covered. And they cannot afford to get sick. Timmy's family lives in that 60-million gap, and they lived there the entire time, before the monkey bars and after, fully "insured" on the way into the parking lot.

IIThe Dollars That Settle It

When a survey and a pile of money disagree about the state of the world, believe the money. A survey asks human beings questions, and the answers can be shaped by how you ask, whom you ask, what you let them say, and how you define the words. Money is different. Money is what was actually sent, actually paid, actually written off, actually filed with a regulator under penalty of law. It does not adjust itself to produce a flattering headline. It is the closest thing the political economy offers to a confession.

So follow the money out of Timmy's broken arm, because this is where the abstraction becomes a family in a car. The emergency room stabilized the fracture, because a federal law requires emergency rooms to stabilize anyone who comes through the door regardless of ability to pay. That is the one piece of the system that behaved humanely, and it is also the most expensive possible place to set a bone. Then the bills began arriving, and they did not stop. The facility fee. The imaging. The radiologist who read the X-ray, somehow out of network despite working inside an in-network hospital. The orthopedist. The follow-up visits. The recast when it did not heal straight. The second opinion. Every one of those charges ran first against the deductible, the multi-thousand-dollar wall the family had to pay entirely out of pocket before the insurance contributed a single dollar. The "coverage" did not begin until the family had already been hollowed out. By the time the insurer started paying its share, the savings were gone, the credit cards were full, and the rent was late.

Now multiply Timmy by a nation, and you get figures that no coverage survey will ever put in a headline, because they come from the ledger instead of the questionnaire. There is 88 billion dollars in medical bills sitting in collections⁴ in this country, touching one in five Americans. That number is not from a survey of how people feel about their bills. It is scraped from the credit bureaus, from real accounts that real collection agencies are really pursuing, with real consequences for credit scores, apartment applications, and job background checks.

And it survived a deliberate effort to make it disappear: after the three major credit bureaus voluntarily wiped most medical debt off consumer reports in 2022, an act of charity that was really an admission of how distorting the debt had become, 15 million Americans still had more than 49 billion dollars⁵ in medical collections left over, concentrated in the South and in low-income communities, which is to say concentrated in places like Timmy's. Across the working-age population, roughly 72 million adults, about 41 percent, carry medical debt or are wrestling with medical bill problems. Of the people who hold that debt, nearly half owe 2,000 dollars or more, one in five owe 5,000 dollars or more, and, in the detail that should end the argument, half of all that debt is for the treatment of an ongoing condition. Not a catastrophe. Not a freak accident. A chronic illness, billed over and over, until the having of a body that does not work correctly becomes a financial sentence served in monthly installments.

These people are not the uninsured. This is the crucial inversion. You cannot generate 88 billion dollars in collections from the 27 million people the system never let through the door, because the uninsured do not get billed into collections, they simply do not get the care. Collections are downstream of treatment. To owe a hospital, you must first have been a patient, and to be a patient with debt, you very likely had insurance that turned out to be a costume. This is the debt of the covered. It is what "insured" actually means once you stop reading the wallet and start reading the ledger. The 88 billion is the financial autopsy of the 86 million, and it lists the cause of death the coverage survey swore up and down was nothing serious.

The hospitals confirm it from the other side of the transaction, in their own audited books, which is to say in numbers they cannot fudge without committing securities fraud. American hospitals have absorbed almost 745 billion dollars in uncompensated care since 2000⁶: care delivered for which no payment was ever received, written off the books as bad debt or charity. That is the institutional shadow of every Timmy who could not pay, totaled across a quarter century. And the machinery for extracting payment has metastasized into something genuinely grotesque: hospitals now spend tens of billions a year not on care, but on the administrative war to make insurers pay for care already delivered, employing whole battalions of billing clerks and appeals specialists whose entire economic function is to fight the insurance company the patient was paying precisely so this fight would not happen. A measurable fraction of what gets triumphantly reported as American "healthcare job growth" is not healthcare in any sense a sick person would recognize. It is the labor cost of the argument over who pays the bill. That is a scandal large enough to deserve its own essay, and it will get one. For now, note only that even the jobs are partly a mirage, and keep moving.

IIIThe Cost, Laundered Two Ways

Here the method earns its keep, because the government will look you in the eye and tell you that medical costs are barely rising at all.

The number it uses for that claim is the medical care component of the Consumer Price Index⁷, and as of the spring of 2026 it runs at roughly 3 percent a year, right in line with or even below general inflation. Hold that number up. Three percent. That is the official, published, on-the-record story of what is happening to the price of American medicine. It is the comparison baseline for everything that follows. Keep it pinned to the wall, at 3 percent, and watch what the actual dollars do as they run past it.

First, set aside the sheer oddity, which everyone has agreed not to notice, that the Bureau of Labor Statistics, an agency created to count workers and wages and strikes and housed inside the Department of Labor, is the body that decides for 335 million people what their healthcare costs. The department whose institutional purpose is the management of labor produces the number that sets the Social Security raise, indexes the tax brackets, and serves as the deflator that determines whether the country got richer or poorer last year. There is no clean reason for that arrangement. There is only the fact of it, and the fact that it has never once been inconvenient for the people who benefit from a low number.

But take the index entirely on its own terms, grant it every good intention, and it still cannot tell you the truth about what care costs, because of how it is built. It quality-adjusts: if this year's treatment is judged better than last year's, the statisticians subtract the value of the improvement from the price, so that a bill which went up in dollars can be recorded as flat or even falling in the index. It weights to out-of-pocket spending, which mutes the enormous share of medical cost that flows through premiums. And hospital and physician services together, the most expensive and most consequential encounters in an American life, the encounters that bankrupt people, make up less than half of the index, about 47 percent. The result is a number that can sit placidly at 3 percent while the thing it claims to measure does something violent. The index is not fabricated. Nobody is making up data in a back room. It is constructed, lawfully and transparently, and the construction smooths and mutes and adjusts the real growth down to less than half its true magnitude.

This is the deeper thing, and it has to be named precisely, because it is not the same as "the numbers are wrong." The official numbers are not merely imperfect or noisy. They are engineered to mislead, and the proof of engineering rather than error is the direction. Random error scatters. These errors all point the same way, the flattering way, every single time, across every metric, decade after decade. The uninsured count is built small. The cost index is built low. An honest instrument makes mistakes in both directions. This one only ever errs toward comfort. And there is a trap waiting for anyone who notices the rot, which is to lunge for some "true" number peddled by a skeptic, a figure that almost always turns out to be the government's own number with a panicked multiplier bolted onto it, the identical lie wearing the opposite coat. The honest move is to trust neither the soothing official figure nor the hysterical correction. The honest move is to refuse the index altogether and go to the dollars, because the dollars were reported under legal compulsion and have no opinion about your peace of mind.

So go to the dollars. The Centers for Medicare and Medicaid Services keeps a second set of books on American healthcare, through its Office of the Actuary, and that set is not an index. It is an accounting. It counts the actual dollars spent on healthcare⁸ in the United States, assembled from claims and provider records, the same hard administrative species of data that a tax return belongs to. The count: 5.3 trillion dollars in 2024. Fifteen thousand four hundred and seventy-four dollars for every living person in the country, infant and centenarian alike. Up 7.2 percent in a single year, the second consecutive year above 7 percent, now devouring 18.0 percent of the entire American economy, and projected by CMS itself to keep outrunning economic growth every year through 2033.

Put the two numbers in the same sentence and let them fight. The index says the price of medical care rose about 3 percent. The accounting says the nation's actual medical spending rose 7.2 percent. Same year, same country, same federal government issuing both. One number is a smoothed, quality-adjusted, out-of-pocket-weighted basket designed to sit still. The other is the sum of the dollars that actually moved. And the dollars grew at more than twice the rate of the index built to track them. A defender will say, correctly, that spending growth includes volume, that Americans consumed more care as the population aged and chronic disease spread, not only higher prices. That is true, and it is also the point. The index strips the volume out and calls the remainder "inflation," which lets it report 3 percent while the actual burden on the actual economy grows at 7. The thing that lands on the country is the spending, not the index. The index is what they show you. The spending is what happens to you.

And the single most honest number in the entire domain is the one with a household's signature on it: the premium, the bill a family actually pays to be allowed to keep the card. The average family premium for employer-sponsored coverage reached 26,993 dollars in 2025⁹, with the worker personally paying 6,850 of that out of their own paycheck and the employer covering the rest, which is simply deferred wages the worker never sees. That premium rose 52 percent over the decade¹⁰ from 2014 to 2024. The deductible underneath it, the wall Timmy's family slammed into before their coverage paid anything, rose 47 percent over the same decade for single coverage. A family health plan now costs about as much as a new car, paid for in full, every year, forever, with nothing to show for it at the end of the year but the right to do it again.

Three percent, says the index. Seven percent, says the spending. Fifty-two percent over a decade, says the bill with the signature on it. Only one of the three is assembled from money, and it is the one that put Timmy's family in the car.

IVThe Door Marked Charity, and the State That Bolts It Shut

There is supposed to be a backstop under all of this, and there is, sort of, which makes the way it fails even more revealing. For the uninsured and the underwater, the federally funded community health center, the FQHC, is the door that is not the emergency room. It is where a person without coverage is supposed to be able to get primary care on a sliding scale tied to income, so that a manageable problem gets managed early and cheaply instead of being neglected until it explodes into a crisis that forces an ER visit at ten or twenty times the cost. These centers served 32.4 million patients in 2024¹¹, and about 18 percent of them, roughly 5.8 million people, were uninsured. The clinics catch perhaps one in five of the officially uninsured and route them toward cheap preventive care instead of catastrophic emergency care. It is, measured honestly, almost the only thing in this entire story that actually saves anyone money, the system's one genuinely rational organ.

But the backstop is not hung evenly across the country, and whether it is there to catch you depends overwhelmingly on a political decision made by your governor and your state legislature, people Timmy's parents may never have voted for. The single largest determinant of whether an American has coverage at all is whether their state expanded Medicaid under the law that made expansion optional. In the states that refused expansion, the uninsured rate runs near 14.5 percent; in the states that accepted it, about 8 percent¹². Nearly twice as likely to be bare, for no reason other than the accident of which side of a state line you happened to break your arm on. About four in ten of all uninsured nonelderly Americans live in the ten holdout states, a concentration of exposure produced not by poverty or demography but by deliberate policy.

Timmy lives in Texas, so the story is as bad as the country permits it to be. Texas has the highest uninsured rate in the nation: 16.7 percent of the whole population by one federal measure, more than 19 percent of the nonelderly by another, and above 21 percent, better than one in five, among working-age adults. Massachusetts, which expanded Medicaid and built its own coverage architecture, sits near 3 percent. That is a six-fold gap between two states flying the same flag, a difference so large it makes the phrase "American healthcare" almost meaningless as a single category. In the community health centers of Texas and Georgia, the share of patients who are uninsured reaches 46 percent, against 19 percent in New York, because a state's refusal to expand Medicaid does not make its poor people disappear, it simply dumps them onto the safety net's doorstep with no coverage behind them, doubling the load on the one rational organ while starving it of the Medicaid dollars meant to keep it funded. The door marked charity is the busiest door in Texas, and the state has spent a decade making sure it stays that way, session after session, refusing the expansion that roughly 1.4 million low-income Texans would qualify for, on principle, the principle being that someone else's suffering is a price they are willing to have other people pay.

VThe Bubble That Pops in the Dark

Stand all the way back now and the shape of the whole machine resolves. The country spends 5.3 trillion dollars a year on medicine and buys, in exchange, a life expectancy of 79 years¹³: two full years below the developed-world average, third worst among all the wealthy nations of the OECD, ahead of only Mexico and Türkiye. More than twice the per-person spending of comparable countries, in exchange for a shorter life than their citizens enjoy. In any functioning market, paying double for a worse outcome is a signal that triggers correction, customers flee, the bad product dies. This market cannot send that signal, because nobody comparison-shops a heart attack from the floor of the emergency room, and no price is posted anywhere before the procedure is done and the bill arrives weeks later with numbers on it that bear no relationship to anything. So the prices climb and the premiums climb and the debt climbs, and the one line on the chart that refuses to climb is how long an American gets to live.

It is fair to call that a bubble. But it is essential to understand that this bubble will never give you the clean, cathartic satisfaction of a crash, and the people who say "it's unsustainable, it has to pop" are waiting for a detonation that by its nature cannot come. A financial bubble pops once, loudly, all at once, everywhere, and the price round-trips down to where it started. This thing has no single price to round-trip and no single day to circle on the calendar, because it does not pop globally. It pops continuously and locally, at the funding seams, one quiet failure at a time, while the aggregate sails on serene.

It pops at the fiscal seam, where the Medicare trust fund grinds toward a depletion date that every actuary knows and no politician will say into a microphone, the moment when the program cannot cover its scheduled obligations and someone must choose, by law, to cut benefits or cut provider payments or raise taxes, any of which is the burst, just denominated in legislation instead of share price.

It pops at the provider seam, where the rural hospital that could no longer survive on the reimbursement rates simply closes its doors, and a county becomes a place where you can be born and grow old but you cannot be born safely or die comfortably, because the nearest open bed is now ninety minutes away. That burst is happening right now, in specific named towns, while the national spending line stays perfectly smooth.

And it pops at the household seam, which is where Timmy is, where a family with a card and a job and a healed arm loses the apartment and moves into the car, because the national spending line stays smooth precisely by absorbing its collapse one broke household at a time. The system socializes the inflation, spreading it across everyone's premiums and everyone's taxes, and privatizes the ruin, dropping the whole of it onto whichever family happened to get sick this quarter. There is no crash to point at and assign blame for. There is only Timmy, in the Honda, a single data point the aggregate was specifically built to swallow without a ripple.

This is the final reason the smooth official headline is the misdirection and the hard local number is the warning that matters. The closed-hospital count, the medical-bankruptcy filing, the collections balance on a credit report, the trust-fund actuary's table: those are the leading indicators, and they are visible in the administrative records years before they ever disturb the serene national index. The index is the lagging number, and it hides the burst the same way the coverage survey hides the debt and the medical CPI hides the cost. Smoothness, in a system like this one, is not evidence of health. Smoothness is the symptom. It is the sound the machine makes while it eats the failures one at a time so that none of them ever get large enough to force a reckoning.

VIThe Method, and the Family in the Car

Every official healthcare number in this essay is built from a survey or an index, and every one carries a definition engineered to minimize the problem it claims to measure. The uninsured count uses a 365-day trick that erases anyone covered for even a day. The cost index quality-adjusts and reweights the real price growth down out of sight. The coverage rate measures whether you hold a card and pointedly never asks whether you can afford to use it. Each of these is defensible on its own narrow technical terms, which is exactly what makes the system so durable: every individual lie is a respectable methodological choice, and only the pattern gives it away. And the pattern is unmistakable, because they all err in the same direction, the comfortable one, the press-release one, every time.

And every single time you set the survey down and pick the money up, the dollars somebody was legally compelled to report, the number roughly doubles and the story turns inside out. Eight percent uninsured becomes 86 million inadequately covered. Three percent medical inflation becomes 7.2 percent spending growth and a family premium that doubled in a decade. A system that "fundamentally works" becomes 88 billion dollars in collections, 745 billion dollars in care the hospitals were never paid for, and a life expectancy below that of nations our politicians condescend to from podiums.

That is the method, and its whole value is that it outlives this one subject. It will work on the employment numbers and the inflation numbers and the poverty numbers and every other headline a government issues about the daily lives of the people it governs. Do not argue with the official figure on its own ground, because its ground was graded and leveled in advance to make you lose. Find the number the government was forced to keep in dollars, money that moved under legal compulsion, set it down beside the number the government chose to publish in a survey, and measure the gap between them with your own hands. The official figure tells you what they need to be true. The money tells you what is true. The distance between the two is not a discrepancy or a rounding error or a difference in methodology to be split. The distance is the entire story, and learning to read it is the closest thing to a superpower an ordinary citizen can acquire in a country that has industrialized the production of flattering numbers.

In American healthcare, that distance is measured in trillions of dollars and in millions of human bodies the survey was deliberately built not to count. One of those bodies belongs to a boy who fell off the monkey bars on an ordinary Tuesday. His father had health insurance the entire time, before the fall and after it, all the way into the parking lot. The insurance was never the thing that was going to save them, and the number that said they were fine was never measuring whether they were fine. It was measuring whether you would believe they were.