The story everyone tells is that hospitals write off unpaid bills, deduct them, and hand the cost to the taxpayer. Most of that is not what happens. A doctor in private practice deducts nothing at all. And the number the write-off column hides is much larger than the one it shows.
Start by fixing the mechanism, because the popular version is wrong in a way that matters.
An unpaid medical bill does not travel down one road with a tax deduction waiting at the end. It travels down four. The amount of federal money involved is different at the end of each one. And at the end of the first road it is nothing at all.
A cash-basis physician practiceThey never counted the money as income, so there is nothing to deduct. The unpaid bill is a loss to the practice and costs the Treasury exactly nothing. This is the single biggest misunderstanding in the whole subject.Federal cost: none
A for-profit hospital on accrualSince a 2014 accounting change, most unpayable bills get recorded as a discount rather than as a loss. Either way the hospital reports less income. But that is the government declining to tax money nobody ever received, which is not the same thing as writing a check.Federal cost: forgone tax on income never collected
A nonprofit hospitalPays no income tax anyway, and counts the charity it provides toward the community benefit that justifies being tax-free. This is the one road where real tax money is genuinely forgone, and it is the one nobody calls a write-off.Federal cost: $13.2 billion in 2022, about $28 billion including state and local
Medicare bad debtNot a tax matter at all. Medicare simply pays hospitals back for unpaid deductibles and copays, minus 35 percent. It hands over 65 cents on the dollar, in cash.Federal cost: a direct outlay, not a deduction
So the honest summary is that the write-off road is real, small, and badly described. The exemption is where actual tax money goes, and how much is contested. The hospital industry’s own commissioned study puts federal revenue forgone at $13.2 billion in 2022, against $149 billion of community benefit, a ratio of eleven to one. The Lown Institute counts more than 1,900 nonprofit hospitals whose tax breaks exceed what they give back, a shortfall of $25.7 billion in 2021.
Both numbers cannot be right. This piece does not settle it, because the argument they are having is not where the money is.
Twenty years of counting the wrong thing
The fight over medical bankruptcy is one of the most quoted arguments in health economics, and one of the least useful, because both sides are measuring the same narrow thing.
Himmelstein and colleagues surveyed people who had already filed for bankruptcy and put the medical share at about 62 percent. Dobkin, Finkelstein, Kluender and Notowidigdo took California hospital records, matched them to credit reports, and looked at what actually changed after someone was admitted rather than what they said afterward. They put the share caused by hospitalization at about 4 percent.
That is a fifteenfold disagreement. It has run for the better part of two decades. And the criticism is largely fair: asking bankrupt people why they went bankrupt does not establish what caused it.
None of which matters much. Because the same 2018 paper that shrank the bankruptcy number contains a finding an order of magnitude more important, and almost nobody quotes it.
The bankruptcy share is contested by a factor of fifteen and is the smaller quantity either way. For adults aged 50 to 59, a hospital admission was followed by a 20 percent decline in earnings and an 11 percent decline in employment. Roughly 17 million adults aged 18 to 64 are hospitalized each year. Sources: Himmelstein et al. 2009; Dobkin, Finkelstein, Kluender and Notowidigdo, American Economic Review 2018 and NEJM 2018; HCUP.
Earnings down 20 percent. Employment down 11 percent. For people in their fifties, admitted once, with insurance.
Bankruptcy is the visible tail of that distribution. Which is exactly why everyone counts it: it produces a court filing. The earnings collapse produces no document anywhere. And that is where nearly all of the damage actually lives.
A bankruptcy leaves a record. A person who quietly earns a fifth less for the rest of their working life leaves nothing but a smaller W-2, and there is no office anywhere in the federal government whose job is to notice.
This is the sixth time this framework has hit the same wall, and at this point the wall is the finding. The gauge could not see subprime. The national numbers could not reconcile corporate profits with small business failure. The survey of who goes without is being switched off. Unemployment cannot register a person working two jobs to cover one rent. Tariff tables report the smallest burden for the families priced out entirely. And here, an entire academic literature argues about the one medical hardship that generates paperwork.
What that is worth to the Treasury
The arithmetic below is this piece’s own, not a sourced figure. It is offered as an order of magnitude, with every input shown, so anyone who dislikes an assumption can move it.
17.0 million hospital stays a year, adults 18 to 64
× 60 percent who were working beforehand = 10.2 million
× the share who take a lasting hit (10 to 30 percent)
× how big the hit is (5 to 20 percent of $62,000)
× 25 percent federal income and payroll tax
→ $0.8 billion to $9.5 billion a year, from one year’s patients
Two things about that range. It looks smaller than the $13.2 billion exemption, and for a single year it is. But it counts only one year’s patients, and the earnings damage lasts for years afterward. Every year adds a new group on top of the ones still impaired. If the average effect runs five years, the standing annual loss is roughly $4 billion to $47 billion, which brackets the entire exemption and may be several times larger.
And it is cautious in three ways that all point the same direction. It counts only federal income and payroll tax, not state tax, not sales tax on the shopping that no longer happens. It excludes everyone over 64, where hospital stays are most common. And it excludes the family members who quit work to provide care, whose own earnings fall without any of them ever being admitted to anything.
The one necessity you cannot refuse
The Toilet Paper Theory says a family at its limit absorbs a squeeze by going without, and that the going without is invisible because a purchase that never happens leaves no trace. Healthcare is the one necessity where that is only half true, and the exception is the interesting part.
Two completely different things hide under one word. Check-ups and elective care behave exactly as the theory predicts. The family skips the physical, cancels the specialist, splits the pills in half, and the amount consumed simply falls.
But emergency care is delivered by law whether or not you can pay. The ambulance comes. The operation happens. And the family consumes something it could not possibly finance.
Ordinary necessity: cannot pay → does not buy → quietly buys less
Emergency care: cannot pay → gets it anyway → a debt is created
The shortfall does not vanish into things not bought.
It gets turned into a debt, and the debt has a name on it.
That conversion is the whole subject here. In every other necessity, being unable to pay destroys demand. In this one, it manufactures a liability. About 41 percent of American adults carry medical or dental debt, something like 100 million people, against a total near $220 billion outstanding.
That pile is not a measure of care consumed. It is a measure of the gap between care delivered and money available, frozen into a document.
And because the check-up you skipped and the emergency you cannot refuse are the same system, the first reliably produces the second. A family that skips the physical for four years does not avoid the cost. It moves the cost to a later, larger, more frightening moment, and one nobody can insure against.
The circle closes
Here is the part that makes this urgent rather than sad, and it closes into a circle worth walking through in order.
Money gets tight for a family. Nothing real has run out: the hospital is there, the doctors are there, the drugs are there, and an earlier piece found no physical shortage standing in the way of covering everyone.
The family consumes the care anyway, because the law says it must be given. The shortfall becomes a debt. The debt, and the illness that created it, cut the family’s earnings by something like a fifth. That cuts income and payroll tax receipts, permanently. Which reduces what the government can afford.
And what the government can afford is the reason given, in the next budget, for why covering everyone is impossible.
The fake shortage of money destroys the tax base that would have paid to end the fake shortage. That is not a metaphor about vicious circles. It is an accounting identity with a delay built in.
And the pattern is the one this framework keeps running into. The care goes to families at the bottom. The debt gets attached to those families. The tax exemption benefits institutions whose surpluses flow upward. And the lost revenue is carried by everyone through a smaller tax base.
Nobody in that chain behaves badly. The hospital is required to treat and cannot collect. The doctor is out of pocket with no deduction. The patient consumed something they genuinely needed. And the result is a transfer that no participant chose and no statistic records.
What would prove this wrong
20% and 11%
The drop in earnings and in employment after a hospital stay, ages 50 to 59, from the study famous for shrinking the bankruptcy number.
62% vs 4%
The disputed medical share of personal bankruptcies. A fifteenfold argument about the smaller of the two numbers.
$13.2B
Federal tax not collected because nonprofit hospitals are exempt, in 2022. About $28 billion counting state and local.
$220B
Medical debt Americans are carrying, held by roughly 41 percent of adults. The gap between care given and money available.
Claim
Measured by
Falsified if
Horizon
The write-offs cost less than the wrecked earnings do
Tax not collected because of how hospitals are treated, against tax not collected because patients earn less afterward.
A full count finds the hospital side bigger than the household side once both are added up properly.
by 2031
Emergency care turns inability to pay into debt instead of into going without
Emergency admissions by income group, against check-ups and elective care in the same groups.
Emergency visits fall with income just as much as check-ups do, meaning people skip both equally.
testable now
Skipping the check-up raises the later bill rather than avoiding it
Emergency admissions and their cost, among people with known gaps in routine care.
People who skip routine care show no higher emergency costs once their starting health is accounted for.
by 2031
The circle is real: medical trouble shrinks the tax base
Tax paid over time by people who had a major medical event, against similar people who did not.
Those people catch back up in what they pay within a few years.
generational
The research measures the one outcome that generates paperwork
How much research exists on medical bankruptcy, against research on what happens to earnings afterward.
The earnings research turns out to be as large and as prominent as the bankruptcy research.
standing
Three warnings, and the first is about this piece’s own starting point. The question that prompted it assumed hospital write-offs were the main way tax money leaks out. The research does not support that. A cash-basis practice deducts nothing. An accrual hospital mostly records a discount rather than a deduction. The real tax expenditure is the exemption, which is a separate argument with its own contested numbers. Saying so seemed better than writing the piece that was expected.
Second, the Treasury arithmetic in the middle is a construction, not a measurement, and it rests on a persistence assumption nobody has estimated cleanly. Treat the range as an argument about which column is bigger, not as a number to quote. Third, the exemption figures come from people with positions. The eleven-to-one ratio was commissioned by the hospital industry. The shortfall comes from an institute that campaigns on the issue. Both are cited here precisely because they disagree.
What survives all three is the ordering. This framework asks which shortage is real, and in American healthcare the answer has been consistent. The buildings are standing. The staff are trained. The medicine exists. What binds is money.
What this adds is that the money does not stay put. It descends into the household, turns into a debt because the law will not let the care be refused, takes a fifth of that household’s earnings with it, and comes back to the public accounts as a permanently smaller tax base. Money was always the shortage. What is new is that the shortage has started manufacturing more of itself.