Check our work.
This page exists because the argument on the rest of the site is only worth as much as the arithmetic behind it. Everything here is either a table you can look up or a sum you can redo.
Three kinds of number
Every figure the calculator produces is one of these three, and the interface says which. If a number is ours rather than someone else’s, we would rather tell you than have you find out.
| Statutory | Fixed in law or in a published federal table. Exact, not an estimate. The poverty guidelines, the premium-credit schedule, the Medicare payroll rate, the COBRA 102% cap. |
|---|---|
| Survey | A real published measurement, but an average. Employer premiums come from KFF’s annual employer survey: genuine data, though your employer is not the average employer. |
| Modeled | Our estimate. Out-of-pocket spending, uninsured cash spending, and part of the universal figures. Labelled everywhere it appears. |
Poverty guidelines Statutory
2026 coverage uses the guidelines in force when open enrollment began, which are the 2025 HHS figures. Add the increment once per person beyond the first.
| Region | First person | Each additional | Household of four |
|---|---|---|---|
| 48 states & DC | $15,650 | $5,500 | $32,150 |
| Alaska | $19,550 | $6,880 | $40,190 |
| Hawaii | $17,990 | $6,330 | $36,980 |
The premium tax credit Statutory
The credit is the gap between the benchmark plan’s premium and what the law says your household is expected to contribute. That expected share comes from IRS Rev. Proc. 2025-25, and it rises linearly within each band, not in steps.
| Household income | Contribution rises from | to |
|---|---|---|
| Below 100% of the poverty line | No credit | |
| 100% to 133% | 2.10% | 2.10% |
| 133% to 150% | 3.14% | 4.19% |
| 150% to 200% | 4.19% | 6.60% |
| 200% to 250% | 6.60% | 8.44% |
| 250% to 300% | 8.44% | 9.96% |
| 300% to 400% | 9.96% | 9.96% |
| 400% and above | No credit at all | |
That last row is the cliff, and it is the single most consequential number on this site. The enhanced credits enacted in 2021 capped contributions at 8.5% of income with no upper income limit; they expired on 31 December 2025. Above 400% of the poverty line the credit now goes to zero rather than tapering, so one extra dollar of income can cost a household thousands.
The first row is the floor, and it matters most in the ten states that have not expanded Medicaid (Alabama, Florida, Georgia, Kansas, Mississippi, South Carolina, Tennessee, Texas, Wisconsin and Wyoming). An adult there below the poverty line is too poor for the credit and, unless a parent with very low income, ineligible for Medicaid as well: the coverage gap. In the states that expanded Medicaid, adults under 138% of the poverty line (215% in the District of Columbia) qualify for Medicaid, and the law gives no marketplace credit to anyone eligible for Medicaid, so the calculator shows no credit there either and says why. The 100% to 133% row used to serve lawfully present immigrants whom Medicaid excludes: they could take the credit below the poverty line and, in expansion states, below the Medicaid ceiling. The July 2025 reconciliation law (Public Law 119-21) removed that credit for plan years from 2026, and from 2027 refugees, asylees and people with temporary protected status lose it as well. The row is empty in practice now, and the floor is a floor for everyone.
The benchmark premium Statutory
The credit is measured against the second-lowest-cost silver plan available to your household, so that premium is the hinge of every marketplace figure here. It comes from the published 2026 federal plan files, computed per county, and here is exactly how.
| The plan set | Individual-market, on-exchange, non-dental silver plans, excluding child-only offerings. HealthCare.gov states come from the CMS Exchange PUFs (rate, plan attributes, service area); the 21 state-run exchanges come from the CMS State-based Exchange PUFs. Both are published federal tables; no key, no scraping of insurer sites. |
|---|---|
| Why the second lowest | The choice is statute, not ours: 26 U.S.C. §36B defines the credit as this benchmark minus your expected contribution. Congress pegged it to the second-lowest silver so the credit always leaves at least two plans within your expected share, and so a single cut-rate outlier cannot set the subsidy for a whole market. |
| Per county, not per rating area | A rating area sets the price, but each plan’s service area decides where it is sold, and the two disagree often: in 37% of HealthCare.gov counties the county’s true second-lowest silver differs from its rating area’s. So we take the plans whose service area covers your county, price them at your county’s rating area, sort, and take the second. Roughly 3,300 county figures, precomputed and shipped with the page. |
| Ages | Each person is priced at their own age on their state’s rate curve, which we derive from the state’s own rate file rather than assume. Forty-two states match the federal 3-to-1 default. Alabama, Mississippi and Oregon price children lower (0.635 against the default’s 0.765); Minnesota (0.89) and the District of Columbia (0.90) price them higher, and DC’s adult curve is its own; Utah’s curve is unlike anyone else’s; Massachusetts is a 2-to-1 state whose curve tops out at 1.999. New York and Vermont do not age-rate at all and price by family tier instead. At most the three oldest children under 21 are charged, which is the ACA’s own family rule. |
| New York and Vermont | Both ban age rating entirely and quote four family tiers instead: individual, couple, one adult with children, two adults with children. We use their published tier rates directly. A third adult in the household is priced as their own individual policy, which is how they would actually buy in. |
| ZIPs that decide the price | Alaska and Massachusetts define rating areas by the first three ZIP digits, and Los Angeles County is split between two rating areas the same way, so there your ZIP picks the rate table directly. Everywhere else your ZIP resolves to a county; a ZIP spanning two rating areas asks you which county is yours rather than guessing. |
| How we know it is right | Three independent checks, run every time the table is rebuilt: every HealthCare.gov county must match the separately published QHP landscape file’s own second-lowest silver premium to the cent; New York’s county floors must match NY State of Health’s published 2026 lowest-cost-silver table to the cent; and Los Angeles must match Covered California’s published 2026 rate booklet. A build that fails any check does not ship. |
Two judgment calls to know about. “Second lowest” counts plans, not distinct prices, so if two plans tie for cheapest the benchmark equals that tied price; the IRS defines it the same way. And the credit legally counts only the essential-health-benefits share of the benchmark premium, typically 99% to 100% of it; the plan files publish that share only as a whole percent, so we use the full premium, which can overstate a credit by a fraction of a percent (about $7 a month on a $3,081 Connecticut benchmark, checked against the state exchange’s own quote).
Employer coverage Survey
From KFF’s 2025 Employer Health Benefits Survey. The employer share is the part this whole site is about: it is compensation, spent on a product instead of paid to you.
| Coverage | Total premium | Worker pays | Employer pays |
|---|---|---|---|
| Family | $26,993 | $6,850 | $20,143 |
| Single | $9,325 | $1,440 | $7,885 |
Medicare payroll tax is 1.45% of wages, plus 0.9% above $200,000 (single) or $250,000 (joint). COBRA is capped by statute at 102% of the full premium.
What universal coverage would cost you
This is where we are most exposed, so here is the whole derivation, for all six models. Two are published household-level formulas, the only two anyone has written down. The other four are our allocation of a published national estimate: the source’s own annual spending change against current law, spread evenly across 134.79 million households (Census, 2025) and added to the Sanders formula. The range deliberately runs from the only peer-reviewed estimate to the most expensive one we know of. The argument has to survive the hostile end, or it is not an argument.
From the financing options published by Sen. Sanders’ office in 2019 (its “Option 1”), the first of the two household-level formulas in this debate that anyone actually wrote down.
4% × (income − $32,200)
+ 7.5% × income
= 4% × $62,800 = $2,512
+ 7.5% × $95,000 = $7,125
= $9,637 a year
The exemption is the standard deduction, as the 2019 list stipulates; its own worked example used $29,000 for a family of four, the figure of its day. A 2026 calculator uses the 2026 deduction: $32,200 for a joint return, $16,100 for a single filer (IRS Rev. Proc. 2025-32). The Lancet formula below keeps its paper’s literal $29,000, because that is what the paper wrote down.
Two liberties, both ours. The 2019 list presents the income premium and the payroll tax as separate options; we pair them. It also exempts an employer’s first $2 million of payroll from the 7.5% levy; a household cannot know where its employer’s payroll stands, so we apply the levy to all wages, which if anything overstates the cost. Say both out loud rather than let someone discover them.
Galvani, Parpia, Foster, Singer & Fitzpatrick, The Lancet, February 2020: the only peer-reviewed estimate in this set, and the second published household formula: a 5% income tax above the paper’s stated $29,000 exemption, plus a 10% payroll tax replacing employer premiums (the authors’ manuscript gives 11.29% as the break-even upper bound; some summaries quote 12.29%).
5% × (income − $29,000)
+ 10% × income
= 5% × $66,000 = $3,300
+ 10% × $95,000 = $9,500
= $12,800 a year
The senior author was an unpaid adviser to Sen. Sanders’ office, disclosed for the same reason Mercatus is labeled Koch-funded. The paper leaves about $507 average out-of-pocket, which every model here excludes for comparability. The paper states one $29,000 threshold; halving it for single filers mirrors our Sanders treatment and is our convention, not the paper’s. The same team’s 2026 preprint (not peer reviewed, produced for a Rep. Khanna inquiry) re-ran the model on 2024 data: national spending falling $1,041 billion, 19.7%.
Liu & Eibner, RAND Corporation, 2019: nonpartisan, modeling the bill with long-term care at payment rates around 109% of Medicare. National health spending rises 1.8%, from $3,823 billion to $3,892 billion.
+$68.8B ÷ 134,790,000 households
= $510 per household per year
$9,637 + $510 = $10,147
RAND assumed supply constraints absorb half of new demand; without that assumption its increase is 9.8%. RAND published no household figure. The allocation is ours, same convention as Blahous.
Congressional Budget Office, December 2020: five illustrative options for 2030; CBO has never scored the bill itself. We use Option 5, the one shaped most like the bill (long-term care covered, no cost sharing), which raises national spending $290 billion, the least friendly of the five. It also assumes provider payment rates above Medicare’s, which the bill does not; CBO’s options at Medicare-like rates come out cheaper. CBO’s four options without long-term care all lower national spending, by $42 to $743 billion.
+$290B ÷ 134,790,000 households
= $2,151 per household per year
$9,637 + $2,151 = $11,788
CBO published no household figure. The allocation is ours, same convention as Blahous. Its $290 billion is a 2030 projection against CBO’s own $6,631 billion baseline; we spread the source’s dollars as published rather than false-precision adjust them to a common year.
Charles Blahous, Mercatus Center, 2018: Koch-funded, and its author opposes the policy. He modeled two scenarios. His main one has national health spending falling $2.05 trillion over a decade against current law. His alternative, where provider payments do not fall to Medicare rates, has it rising $3.25 trillion.
We use the alternative, so this site never rests on the friendly reading of an unfriendly study.
$3.252T ÷ 10 years
÷ 134,790,000 households
= $2,413 per household per year
$9,637 + $2,413 = $12,050
Blahous published no household figure, so that allocation is ours. Spreading it flat across all households is a simplification: real financing would be progressive, which would move this number down for most households and up for high earners.
Blumberg, Holahan and colleagues, Urban Institute, October 2019, in a report financed by the Commonwealth Fund (named for the same reason Mercatus’ funder is): a left-leaning institute’s “Single-Payer Enhanced”, and the most expensive estimate we know of: national spending rising $719.7 billion a year (+20.6%), covering long-term care, dental, zero cost sharing and every US resident.
We include it so the top of the range is the hardest number on offer. If the argument survives this one, it survives.
+$719.7B ÷ 134,790,000 households
= $5,339 per household per year
$9,637 + $5,339 = $14,976
Urban published no household figure. The allocation is ours, same convention as Blahous. Its “Single-Payer Lite” option, without long-term care and with income-scaled cost sharing, cuts national spending 6.0%. We use the expensive one.
What a bad year costs without coverage Modeled
The uninsured view of section 03 prints three numbers. Two are statutory; the list-price figure is our arithmetic, so here it is in full.
| One admission, at cost | About $16,700, derived by us from HCUP’s published 2022 aggregates: $548.5 billion in hospital costs across 32.9 million stays (HCUP Statistical Brief #316). Excludes physician fees, so the real bill runs higher. |
|---|---|
| The markup to list price | Charges average 3.4× cost nationally (Bai & Anderson, Health Affairs 2015, on 2012 data; the most common ratio was 2.4). A 2020 study by National Nurses United (a union that advocates single-payer; labeled, like every source here) measures 4.2× by 2018. We use the older, smaller number. |
| One admission, as billed | ≈$57,000: $16,700 × 3.4, our multiplication, labeled modeled where it appears. Uninsured patients are billed from list prices. Protections exist but are conditional (nonprofit hospitals’ assistance policies under IRS 501(r), good-faith estimates for scheduled care under the No Surprises Act), and none of them caps the price itself. |
| The insured ceiling | $10,600 self-only / $21,200 family: the 2026 out-of-pocket maximum, statutory, from CMS’s 2025 Marketplace final rule. It applies only to people with coverage. No equivalent exists in law for the uninsured. |
| After one hospitalization | ≈$6,000 in unpaid bills for an uninsured adult four years on, against ≈$300 insured: Dobkin, Finkelstein, Kluender & Notowidigdo’s event study of about a million California admissions matched to credit records (American Economic Review, 2018). A finding both camps of the bankruptcy debate below accept. |
| Debt, and care skipped | 62% of uninsured adults carry healthcare debt, against 44% of the insured (KFF survey, 2022). Cost stopped or delayed care for three in four uninsured adults in KFF’s 2025 poll, or 38.6%, more than double the privately insured, by the narrower federal NHIS measure (2024). Two surveys asking two different questions; we print both. |
The medical-bankruptcy numbers are deliberately absent from section 03, because the famous ones disagree about what they measure. A 2019 AJPH editorial (a survey of bankruptcy filers with a 29.4% response rate, whose lead authors co-founded Physicians for a National Health Program, which advocates single-payer) found 66.5% of filers said medical bills or illness-related work loss contributed. A 2018 NEJM analysis by academic economists, linking hospitalizations to credit records, estimates admissions cause about 4% of non-elderly bankruptcies, 6% among the uninsured. Self-reported contribution and measured causation are different questions; neither number refutes the other, and we will not print one without the other.
What we know is wrong Modeled
Every model has soft spots. Ours are listed here rather than left for someone else to find.
| Ages the market does not price | Anyone 65 or over is usually on Medicare, which we do not price; we charge them the 64-year-old rate, the oldest the marketplace curve has, which overstates a Medicare household’s real premium and understates its risk. Tobacco surcharges (up to 50% in most states) are not modeled; every figure is the non-tobacco rate. |
|---|---|
| ZIP boundaries | Our ZIP table is built from the Census 2020 ZCTA-to-county file, and ZCTAs approximate postal ZIPs rather than matching them exactly. A ZIP whose counties sit in different rating areas asks you to pick; about 140 PO-box-only ZIPs carry no homes of their own and inherit the nearest deliverable ZIP’s county, which we note here rather than hide. |
| Out-of-pocket spending | Anchored to published average deductibles, not to measured spending. Real figures would come from the Medical Expenditure Panel Survey. Yours depends entirely on whether you got sick. |
| The Medicare tax asymmetry | Today’s total includes the 1.45% you already pay. The universal figures do not carry a continuing Medicare tax, which flatters them by roughly that amount. And if the employer’s premium share is really your money, so is the employer’s matching 1.45%, which we leave out, which cuts the other way. |
| What the universal figure is | A contribution, not a cost of care. The other two bars show what covering this household costs in full; this one shows the share of it the household would pay. Its coverage would still cost roughly what it costs now (for the worked household above, $32,171 a year through an employer), and on the six models the household would pay $9,637 to $14,976 of that, 30% to 47%, the rest coming from progressive financing and the federal health spending that already exists. Read as a like-for-like total, it flatters universal coverage. Putting a number on that remainder would mean allocating national spending per household, a second modeled figure stacked on the one we already publish, so for now we say this instead. |
| Mixed vintages in the allocations | The four allocated models spread each source’s own dollars (Blahous’s 2020s decade, Urban’s 2020, RAND’s 2019, CBO’s 2030 projection) over one 2025 household count. Inflating them all to one year would add false precision to numbers whose real uncertainty is the model, not the deflator; leaving them as published understates the older sources slightly. We chose the disclosure over the adjustment. |
| One income number | The statutes want three different ones: the subsidy runs on MAGI, the Medicare payroll tax on wages, and the Sanders premium on income above an exemption. We run all three from the single figure you enter. Close for a household whose income is mostly wages; retirement deferrals drive them apart, since Medicare wages include a 401(k) deferral and MAGI does not. We also treat any household larger than one as filing jointly. |
| Uninsured spending | Modeled, and low for a reason worth stating plainly: uninsured households spend less because they skip care, not because care is cheaper for them. It usually is not. Our figure is the cost of the care an uninsured household uses in a year. Cash out of pocket runs lower still (roughly $500 to $750 in the last published MEPS breakdowns), because providers collect little of what they bill. Either way it describes a quiet year; the bad-year arithmetic has its own section above. |
| Households the calculator does not price | Three kinds of coverage are not on the route control, so their households should not read the employer figure as theirs. TRICARE: the Military Health System covers 9.4 million service members, families and retirees on a FY2026 request of $64 billion, about $6,800 a person, a figure that includes military hospitals and medical readiness. Active-duty families pay no enrollment fee and a catastrophic cap of $1,000 (Group A) or $1,324 (Group B); retirees pay $375 to $1,191 a year in fees with caps of $3,000 to $4,635, by plan and group. Medicare: 69.7 million people, financed partly by the payroll tax this site counts and partly by premiums it does not model. VA care: 9.1 million enrolled veterans, mostly free at the point of use. For all three the employer-share argument does not apply, and under the financing models here most such households would pay more than they do today. The site says so rather than price them. |
| Repaying the credit | The credit shown is the advance credit for a household whose income comes in as entered. Since tax year 2026 there is no cap on repaying excess advance credit (Public Law 119-21): a household that earns more than it projected repays the whole difference at filing, and one that crosses the 400% line repays every dollar of it. The page prices the year as entered and does not model repayment; section 03 shows what crossing the line costs. Two later changes narrow who can claim the credit rather than what it pays: from 2027 refugees, asylees and people with temporary protected status are excluded, and from 2028 every enrollee must re-verify eligibility before each year’s coverage instead of being re-enrolled automatically. |
| Wage incidence | The claim that the employer’s premium share is really your compensation is well supported among economists but it is not a law of nature, and it plays out over years rather than on the day your premium changes. |
Sources
Each link goes to the primary document, not a summary of it. Every one was retrieved and checked on 4 September 2026; if a link has since moved, the citation still stands on its own.
| Figure | Source |
|---|---|
| Poverty guidelines | HHS, 2025 Poverty Guidelines |
| Premium credit schedule | IRS Rev. Proc. 2025-25, applicable percentage table for 2026 |
| Standard deduction, 2026 | IRS Rev. Proc. 2025-32 (the Sanders exemption) |
| Employer premiums | KFF Employer Health Benefits Survey, 2025 |
| 2026 marketplace changes | KFF, 2026 Marketplace enrollment, premiums and deductibles |
| Single-payer financing | Office of Sen. Bernie Sanders, financing options, 2019 |
| Single-payer, peer-reviewed | Galvani et al., Improving the prognosis of health care in the USA, The Lancet 395:524-533, 2020; the authors’ open manuscript on PubMed Central |
| Single-payer national cost | Blahous, The Costs of a National Single-Payer Healthcare System, Mercatus Center, 2018 |
| Single-payer, nonpartisan | Liu & Eibner, National Health Spending Estimates Under Medicare for All, RAND RR-3106, 2019 |
| Single-payer, five options | CBO, How CBO Analyzes the Costs of Proposals for Single-Payer Health Care Systems, Working Paper 2020-08, December 2020 |
| Single-payer, most expensive | Blumberg, Holahan et al., From Incremental to Comprehensive Health Insurance Reform, Urban Institute, October 2019 (financed by the Commonwealth Fund) |
| Lives saved, 114,174 | Pandey, Wells, Ye, Fitzpatrick & Galvani, Projected economic gains and lives saved under universal healthcare in the United States, Yale School of Public Health preprint on medRxiv, July 2026 (not peer reviewed; the two components on the page, 62,863 and 51,311, are the paper’s own and sum exactly) |
| Marketplace enrollment decline | KFF analysis of 2026 Marketplace enrollment; CMS effectuated enrollment reports, February 2026 against February 2025, for the “nearly three million” |
| Household count | US Census Bureau, Total Households, 2025 |
| Hospital stay costs | HCUP Statistical Brief #316, AHRQ, February 2026 (2022 data) |
| Hospital markups | Bai & Anderson, Extreme Markup, Health Affairs, 2015; National Nurses United, Fleecing Patients, 2020 (union-funded) |
| 2026 out-of-pocket maximum | CMS, Marketplace Integrity and Affordability final rule, June 2025 |
| Hospitalization aftermath | Dobkin, Finkelstein, Kluender & Notowidigdo, American Economic Review, 2018 |
| Health-care debt | KFF Health Care Debt Survey, 2022 |
| Care skipped over cost | KFF Health Tracking Poll, May 2025; National Health Interview Survey, 2024 |
| Medical bankruptcy, both sides | Himmelstein et al., AJPH, 2019 (PNHP-affiliated authors); Dobkin et al., NEJM, 2018 |
| Benchmark premiums, HealthCare.gov states | CMS Exchange PUFs (rate, plan attributes, service area) and the QHP Landscape file, plan year 2026 |
| Benchmark premiums, state exchanges | CMS State-based Exchange QHP PUFs, plan year 2026 (data as of June 2026) |
| Benchmark cross-checks | NY State of Health, 2026 Lowest Cost Silver Plan by County; Covered California, 2026 QHP Rates by County |
| Rating area definitions | CMS, State Specific Geographic Rating Areas |
| ZIP to county | Census 2020 ZCTA-county relationship file; GeoNames (CC BY 4.0) for PO-box-only ZIPs |
| Medicaid expansion & CHIP limits | Medicaid.gov, Medicaid and CHIP eligibility levels |
| Military Health System | CRS, FY2026 Budget Request for the Military Health System, In Focus IF13108; TRICARE, 2026 health plan costs |
| Medicare and VA enrollment | CMS, Medicare Monthly Enrollment, November 2025; Veterans Health Administration, About VHA |
| Early retirement | EBRI, 2026 Retirement Confidence Survey (press release, 21 April 2026): nearly half of retirees retired earlier than planned; the median retirement age is 62. Retrieved 18 September 2026. |
| Members and cosponsors | unitedstates/congress-legislators (every current member, keyed by bioguide id); govinfo BILLSTATUS for H.R. 3069 and S. 1506 (cosponsors); the OpenSourceActivismTech ZIP-to-district crosswalk. Rebuilt by hand when Congress moves; the retrieval date shows on the card. |
| Member portraits | The @unitedstates images collection (official congressional photos, public domain), loaded per member from unitedstates.github.io |
Found an error? That is the most useful thing you could send us: contact@thewholebill.org. The calculation layer is a single file with no dependencies, and every constant on this page appears in it by name.