Medicare For All

Replace private health insurance with a single federal programme covering everyone for everything, with no premiums, deductibles or co-payments at the point of care.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

The United States spends about 5.2 trillion dollars a year on health care, roughly a sixth of its economy and more per head than any other country, and about 26 million people have no coverage at all. The bill before Congress replaces the whole insurance system — employer plans, individual plans, Medicaid and the existing Medicare — with one federal programme that covers every resident for hospital, medical, dental, vision, mental health and long-term care, and charges nothing at the point of use. Private insurance duplicating that coverage would be prohibited, providers would be paid on a national fee schedule, and hospitals would be paid global budgets rather than per procedure. The bill leaves how to raise the money to a separate act, which is why this evaluation counts the reorganisation and not the tax schedule. This evaluation looks ten years ahead.

Balance

Better for the future · 0.62 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 71 · 62 % Against 42 · 38 %

Size class: very large Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 20 billion euro per year. One thing is deliberately left out of this evaluation: who pays. The bill leaves the financing to a separate act, and a progressive tax and a payroll tax would produce completely different answers about winners and losers. What is counted instead is the reorganisation itself — the billing that stops, the bills patients no longer face when they are ill, the care that is used because it is free, and what happens when every provider in the country is paid on one schedule. Two numbers decide the result: how much of the 8.3 percent of American health spending that goes on administration would actually disappear, and how much of the extra care people use when it costs nothing is worth what it costs to provide. How we score →

Arguments for

Arguments against

8 arguments evaluated · Scoring v1.3 Δ absolute +29

Arguments — For

4 arguments · top 3 shown

Billing that stops

34of 100

Administration takes 8.3 percent of American health spending against 2.7 percent in Canada. Most of the difference is people on both sides of a transaction arguing about who pays for what.

Value 6 · OutputImpact 8.6Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is real resources released: the claims departments, the prior-authorisation staff, the coding contractors, the collections agencies and the hours doctors spend on paperwork rather than patients. This site places that with economic output and prosperity, because what is freed is labour and capital that go on to produce something else, rather than money moving between two parties. It is a genuine saving and not a transfer: nobody receives the money that is no longer spent arguing about a bill. Nothing is counted for the people whose jobs those are, whose displacement appears among the costs. Nothing is counted for the reduction in patient time spent on paperwork, which is real and small next to this. The value sits in the middle-upper part of the scale, because what is freed is labour that goes on to produce something else.

Administration accounted for 8.3 percent of American health care expenditure against 2.7 percent in Canada, whose system is the closest existing model to the one proposed ^[4]. On 5.2 trillion dollars of national health spending, closing that gap entirely would save about 290 billion dollars a year. Full convergence is not realistic — the American system has more providers, more sites of care and more of everything to administer — so 200 billion dollars is used, in a range from 100 to 400, which is roughly two thirds of the gap. That is 172 billion euro at the ordinary weight for resources. The published modelling of billing and insurance-related costs specifically puts the reduction at between a third and a half of those costs under a single-payer design, which is consistent with the figure used. What is not counted is the administrative apparatus the new programme itself would need, which appears among the costs. The Impact is the largest on this side and the second largest in the evaluation, and it is the number single-payer advocates and their critics disagree about least.

▸ Show calculation ▾ Hide calculation

	National health expenditure [1]	5.2 trillion dollars at 1.16 to the euro	4,483 billion euro a year
×	Gap between the American and Canadian administrative shares measured in national accounts rather than modelled [4]	8.3 percent against 2.7 percent	251 billion euro a year
×	Share of the gap that the payment system accounts for Setting, range 100 to 400 billion dollars: the American system has more providers and more sites of care to administer whatever the payer looks like [5]	about two thirds	172.41 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	8.62

Score 8.62 Impact × 6 Value × 6.5 Plausibility ÷ 10 = 34 of 100

The counterfactual is the American system as it stands, and the comparison is with countries that already run what is proposed. The design is a precedent transfer rather than an experiment: Canada, Taiwan and the Nordic systems all administer health care for a fraction of the American share, and the difference is measured in national accounts rather than modelled. What that comparison cannot establish is how much of the gap is the payment system and how much is everything else about the American health sector — more sites of care, more litigation, more product variety — which is the confounder and is unresolved. It is the reason the figure used is two thirds of the observed gap rather than all of it. Reverse causation does not arise. The published billing-cost models give a similar answer by a different route, which is what lifts this above a single comparison. The Plausibility is above the middle: several countries demonstrate the destination and how much of the distance is the payment system is not settled.

evidence basis: Multiple precedents · P ceiling 7 identification: Controlled · rung ceiling 7

Counterfactual: the American system as it stands, compared against countries already running single-payer administration. Design: controlled — cross-country comparison of administrative shares in national accounts, with no exogenous variation in who has which system [4][5]. Confounder: how much of the gap is the payment system rather than everything else about American health care — more sites of care, more litigation, more product variety; unresolved, and the reason two thirds of the gap is used rather than all of it. Direction: no reverse causation. Ceiling: controlled 7.0 binds below the multiple-precedents ceiling of 8.5.

No bill for being ill

18of 100

Americans pay about 505 billion dollars a year out of their own pockets for care, and it lands on whoever happens to be sick. Under this bill the same money is raised in tax and lands on everybody.

Value 5 · Household budgetsImpact 6.5Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money that households pay for care, priced at the middle of the scale like all money. What changes is not how much is paid but who pays it and when: today it lands on the household that is ill, in a lump, at the moment its earnings are most likely to have fallen; under the bill it is raised in tax across everybody. Only the difference between what a euro is worth at those two addresses is counted here, not the whole sum — the same households pay for it through tax either way, so booking both ends at full size would be counting a household's own money against itself. Nothing is counted for the premiums that also disappear, because who ends up paying for those depends on a financing act the bill leaves to be written. The value is the middle of the scale, and only the difference in what a euro is worth when it lands on the ill rather than on everybody is counted.

Out-of-pocket health spending in the United States runs to about 505 billion dollars a year, or 435 billion euro, and it is heavily concentrated: the costliest five percent of patients account for roughly half of all health spending in a year, and medical bills remain among the largest single causes of consumer bankruptcy. Borne as it is today, that money carries an effective weight of about 1.3, in a range from 1.1 to 1.8 — a mixture of ordinary spending at the ordinary weight and catastrophic bills landing on households whose income has usually fallen at the same moment. Raised in tax across the whole population it would carry the ordinary weight of 1.0. Only the difference of 0.3 is counted, giving 131 billion euro a year. This is a netted figure by declaration: the sum itself is not counted on either side, because the households paying the tax are substantially the households that were paying the bills. The Impact is the third largest here and it is a difference in weight rather than a sum of money, which is why it is smaller than the amount it concerns.

▸ Show calculation ▾ Hide calculation

	Out-of-pocket health spending [1]	505 billion dollars at 1.16 to the euro	435.34 billion euro a year
×	Difference in what a euro is worth on the ill rather than on everybody Setting, range 1.1 to 1.8: the costliest five percent of patients account for about half of all spending, and medical bills are among the largest causes of consumer bankruptcy	1.3 against 1.0	130.6 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	6.53

Score 6.53 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 18 of 100

That out-of-pocket costs go to zero is written into the bill, so the direction is not in question. The counterfactual is the present distribution of cost sharing. The counterfactual for the tax side is not written into the bill at all, which is the honest limit of this argument: the financing act would determine whether the replacement lands progressively, proportionally or regressively, and only the fact that it lands on everybody rather than on the ill is used here. The confounder that matters is the concentration figure: how much of the effective weight of 1.3 comes from genuinely catastrophic bills against ordinary co-payments is not measured for the American distribution, and the range reflects that. Reverse causation does not arise. The netting is declared rather than assumed: the same money is not counted as both a gain to households and a cost to them. The Plausibility is a little above the middle: the removal of cost sharing is statutory and the weight put on its present concentration is estimated.

evidence basis: Projection · P ceiling 6 identification: Definitional · no rung ceiling

Counterfactual: the present distribution of cost sharing; the tax side is not specified by the bill and only the fact that it falls on everybody is used. Design: definitional — cost sharing goes to zero by statute; no behavioural link carries the quantity, which is a weight difference rather than a flow. Confounder: how much of the present effective weight comes from catastrophic bills rather than ordinary co-payments is not measured for the American distribution; named and carried in the range. Direction: not applicable. Ceiling: projection 6.0 binds, because the weight is estimated rather than observed.

One schedule for every price

11of 100

Commercial insurers pay hospitals roughly twice what Medicare does for the same procedure. Paying everybody the Medicare rate moves about 400 billion dollars a year from providers to whoever funds the system.

Value 5 · Household budgetsImpact 4.3Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money that does not have to be raised because care costs less to buy, priced at the middle of the scale like all money. It is a transfer rather than a saving: the money stays in the economy, it simply stops going to hospitals, specialists and the people who own them. Only the difference in weight between where it comes from and where it goes is counted, which is why this argument is a quarter of the sum involved. The response of providers to being paid less is the argument opposite, as this site requires wherever a weighted transfer is booked. Nothing is counted for the reduction in medical debt collection, which is inside the argument about out-of-pocket costs. The value is the middle of the scale, and only the weight difference between provider incomes and the general public is counted.

Commercial insurers pay hospitals in the order of twice the Medicare rate for the same procedure, and physician payment shows a smaller but similar gap. Applying the Medicare schedule to everybody would move about 400 billion dollars a year, or 345 billion euro, from providers to whoever funds the system. Provider income is not concentrated in the way that, say, corporate profit is: it is a mix of physician incomes in the top tenth of the distribution, hospital margins that belong to companies and non-profits, and the wages of nurses and technicians around the middle. A blended weight of 0.75 is used, in a range from 0.5 to 0.95, against 1.0 for the general public that funds the system, so the difference counted is 0.25 — 86 billion euro a year. If the cut fell mainly on staff wages rather than on physician incomes and margins, this argument would be close to zero. The Impact is a quarter of the money involved, because most of a payment cut lands on people whose euro is worth much what everybody else's is.

▸ Show calculation ▾ Hide calculation

	Money moved by paying every provider the Medicare rate [3]	400 billion dollars at 1.16 to the euro	344.83 billion euro a year
×	Difference in weight between provider incomes and the general public Setting, range 0.5 to 0.95: a mix of physician incomes in the top tenth, hospital margins, and the wages of nurses and technicians around the middle	0.75 against 1.0	86.21 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	4.31

Score 4.31 Impact × 5 Value × 5 Plausibility ÷ 10 = 11 of 100

The price gap is measured: hospital price transparency data and the published comparisons of commercial against Medicare rates establish it in a way that was not possible ten years ago. The counterfactual is the present mix of payers. What is not established is who inside a provider organisation bears a cut, and that is what the whole argument turns on: a hospital that loses a third of its commercial revenue may cut executive pay, close a service line, or freeze nursing wages, and nothing in the literature predicts which. That is the confounder and it is unresolved. Reverse causation does not arise. The chain has a further behavioural link — that the cut is actually implemented rather than bargained away — which the experience of every attempt to set health prices administratively suggests should not be assumed. The Plausibility is at the middle: the price gap is measured and who absorbs its removal is not.

evidence basis: Precedent · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the present mix of commercial and public payers. Design: mechanistic — the commercial-to-Medicare price gap is measured in transparency data, but who inside a provider organisation absorbs a cut is not predicted by anything. Confounder: a cut may fall on executive pay, on service lines or on nursing wages, and the weight differs by a factor of two between them; unresolved. Direction: no reverse causation. Ceiling: set at 6.0, below the precedent ceiling, because the incidence link carries the quantity and nothing identifies it.

People who live

7.5of 100

About 26 million Americans have no coverage. When Medicaid expanded, mortality among the newly covered fell by 0.132 percentage points a year — a 9.4 percent reduction against the group's own rate.

Value 9.5 · LifeImpact 1.3Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is people alive at the end of the period who would otherwise have died, with a tenth added on the same anchor for the families, as this site does wherever it counts avoidable death. It sits just below the top of the scale rather than at it because part of what coverage buys is illness treated rather than death avoided, and only the deaths are counted here. It is separate from the money above in the ordinary way: a household that would have paid the bill and got the treatment gains from that argument, and one that went without gains from this one. Nothing is counted for the illness that is treated without ever threatening a life, which is a large stream and is not estimated. The value sits just below the maximum: the stream is life, reached through care that is taken up rather than avoided.

About 26 million Americans have no health coverage. Miller, Johnson and Wherry, linking survey and administrative death records, found that the Medicaid expansions reduced annual mortality among the newly covered by 0.132 percentage points, a 9.4 percent reduction against that group's own rate ^[6]. Applied directly, 26 million people would give about 34,000 deaths avoided a year. Half of that is used — 17,000, in a range from 5,000 to 45,000 — because the uninsured population as a whole is younger and healthier than the low-income adults that study followed, and because the randomised Oregon experiment found no measurable mortality effect on a smaller sample over two years. Valued at the figure this site uses for a life with a tenth added for the families, that is 26.2 billion euro a year. The Impact is a sixth of the administrative saving, which is the arithmetic of comparing 17,000 lives with the cost of running the billing system of a sixth of an economy.

▸ Show calculation ▾ Hide calculation

	Americans without health coverage [2]		26 million people
×	Annual mortality reduction among the newly covered measured on linked survey and administrative death records around the Medicaid expansions [6]	0.132 percentage points	34,320 deaths a year
×	Correction for a younger and healthier uninsured population Setting, range 5,000 to 45,000 deaths: the study followed low-income adults gaining Medicaid, and the randomised Oregon experiment found no significant mortality effect on a smaller sample [7]	50 %	17,000 deaths a year
×	Value of a life, with a tenth added for the families the value of a statistical life used across this site and the surcharge this site applies to avoidable death	1.4 million euro each, plus 10 percent	26.18 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	1.31

Score 1.31 Impact × 9.5 Value × 6 Plausibility ÷ 10 = 7.5 of 100

The counterfactual is comparable adults in states that did not expand Medicaid, which the study constructs by comparing expansion and non-expansion states around 2014 in linked survey and administrative data — a difference in differences with individual death records rather than aggregate rates, which is what makes it stronger than the earlier work on the same question. The confounder that matters is that expansion states differ from non-expansion states in more than Medicaid, and the paper addresses it by restricting to people who were eligible by income and by showing parallel pre-trends. Reverse causation does not arise from a state's expansion decision at the individual level. What holds the plausibility below what the design would allow is the transfer, which is large: the population studied was low-income adults gaining Medicaid, and the uninsured as a group are younger and healthier, which is why the figure used is half of what the estimate would give. The Oregon randomised experiment, which found no significant mortality effect, is the counter-evidence and it is smaller and shorter. The Plausibility is at the middle to upper part: the effect is measured on individual death records, and the group it was measured on is not the group this would cover.

evidence basis: Study · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: income-eligible adults in states that did not expand Medicaid. Design: quasi-experimental — difference in differences around the 2014 expansions in linked survey and administrative death records, with reported parallel pre-trends (Miller, Johnson and Wherry) [6]. Confounder: expansion states differ in more than Medicaid; addressed by restricting to the income-eligible and by the pre-trend check. Direction: reverse causation does not arise at the individual level from a state decision. Ceiling: quasi-experimental 8.0 less one point for the transfer from low-income Medicaid gainers to a younger, healthier uninsured population; the Oregon randomised experiment found no significant mortality effect on a smaller sample.

Arguments — Against

4 arguments · top 3 shown

Care that is used because it is free

24of 100

When care costs nothing at the point of use, more of it is used. The randomised experiment that measured this found the extra care produced no detectable improvement in health for the average adult.

Value 5 · OutputImpact 8.1Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is real resources consumed: appointments, scans, prescriptions and hospital days that are provided because nothing stands between the patient and the service. It is priced at the middle of the scale, and it is a genuine resource cost rather than a transfer, because the staff hours and the equipment are used up. Only the part of the additional care that is worth less than it costs is counted here; the part that does people good is not counted as a cost at all, and its benefit sits partly in the argument about lives above. Nothing is counted for the crowding-out of urgent care by routine care, which is a separate consequence and appears in the argument after this one. The value is the middle of the scale, and only the additional care that is worth less than it costs to provide is counted.

Removing cost sharing raises use. The randomised health insurance experiment that measured this found that free care raised spending by roughly a third against a plan with substantial cost sharing, and — the finding that matters here — that the extra care produced no detectable improvement in health for the average adult, though it helped the poorest and sickest. A six percent rise in national health spending is used, in a range from two to fifteen percent, which is well below the experimental figure because most Americans already have insurance with moderate cost sharing rather than none. On 4,483 billion euro that is 269 billion euro of additional care. Sixty percent of it is treated as worth less than it costs, in a range from 30 to 85 percent, giving 161 billion euro of resources consumed for no return. The remaining forty percent is care that does people good and is not counted against the measure. The Impact is the largest cost here and it is roughly the size of the administrative saving on the other side, which is the central trade in this debate.

▸ Show calculation ▾ Hide calculation

	National health expenditure [1]	5.2 trillion dollars at 1.16 to the euro	4,483 billion euro a year
×	Rise in use once care is free at the point of service Setting, range 2 to 15 percent: well below the third measured under randomisation, because most Americans already have insurance with moderate cost sharing rather than none [8]	6 %	268.97 billion euro a year
×	Share worth less than it costs to provide Setting, range 30 to 85 percent: the randomised experiment found no detectable health improvement for the average adult from the extra care, though it helped the poorest and sickest	60 %	161.38 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	8.07

Score 8.07 Impact × 5 Value × 6 Plausibility ÷ 10 = 24 of 100

The counterfactual is the same population under present cost sharing. The design behind the response is experimental: households were randomly assigned to plans with different cost sharing and their use was measured over three to five years, which is the strongest evidence anywhere in this evaluation. Its weakness is age and transfer — it was run between 1974 and 1982, on a working-age population, in a health system that no longer exists — and that is a real limit rather than a formality. The confounder that matters is what has changed since: care is now more supply-driven, more of it is prescribed rather than sought, and the price a patient faces may matter less than it did. That is named and unresolved. The second and larger uncertainty is not the response but the valuation: the experiment's finding that the extra care did not improve average health is what justifies treating most of it as waste, and it is the single most contested result in American health economics. Reverse causation does not arise from random assignment. The Plausibility is above the middle: the response is measured under randomisation, in a health system half a century old.

evidence basis: Study · P ceiling 6.5 identification: Experimental · rung ceiling 8.5

Counterfactual: the same households randomly assigned to plans with cost sharing. Design: experimental — random assignment of households to insurance plans, use measured over three to five years [8]. Confounder: the system in which it was run no longer exists, and care is now more supply-driven, so the price a patient faces may matter less; named and unresolved. Direction: reverse causation does not arise from random assignment. Ceiling: experimental 8.5 less two points for the transfer from a 1974 to 1982 trial on a working-age population to today's system.

Waiting for the care

9.5of 100

Paying every provider the Medicare rate cuts hospital revenue by roughly a third in commercial-heavy markets while demand rises. Something has to give, and in every system that has tried this it is time.

Value 9.5 · HealthImpact 2Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is health lost to delay: a hip replaced a year later, a diagnosis made at a later stage, a specialist appointment that takes four months. This site places that in the class it uses for life and health, one step below the top because most of what is lost is function and comfort rather than years. It is the required counterpart to the argument about paying every provider the Medicare rate, which is a weighted transfer and therefore carries its behavioural response on this side. Nothing is counted for the people who receive care sooner because somebody else is no longer jumping the queue with better insurance, which is a real effect in the other direction and is not estimated. The value sits just below the maximum, and this argument exists because a payment cut booked as a gain has to carry the response it provokes.

Two things push in the same direction at once: demand rises because care is free, and supply tightens because hospitals in commercial-heavy markets lose something like a third of their revenue while rural hospitals that already lose money on public patients lose their cross-subsidy entirely. The result in every system that pays administratively set prices for universal coverage is a queue. One million quality-adjusted life years a year are used, in a range from 200,000 to three million — about a tenth of a day per American per year on average, which sounds trivial and is not, because it lands almost entirely on the minority who need something done. At the value this site uses for a healthy year that is 40 billion euro. The figure is the most uncertain in this evaluation and it is the one that a phased implementation, a transition fund for rural hospitals, or a less aggressive fee schedule would most change. The Impact is a quarter of the largest cost here and it is the argument least anchored in any American measurement.

▸ Show calculation ▾ Hide calculation

	Quality-adjusted life years lost to delay each year Setting, range 200,000 to 3 million: about a tenth of a day per American on average, landing almost entirely on the minority who need something done [4]	one million	1,000,000 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site	40,000 euro each	40 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	2

Score 2 Impact × 9.5 Value × 5 Plausibility ÷ 10 = 9.5 of 100

The counterfactual is the American system as it stands, with its shorter waits and its 26 million uninsured. The evidence is precedent rather than experiment: Canada, the United Kingdom and Australia all combine universal coverage with administratively set prices and all report waiting lists that the United States does not have. What that comparison cannot do is separate the payment system from the level of funding — those countries spend half what the United States does per head, and a single payer spending American amounts might not queue at all, which is the strongest counter-argument and is unresolved. The confounder is therefore funding level rather than payment design. Reverse causation does not arise. Nothing has measured what a payment cut of this size does to American capacity, because nothing of this size has been attempted, and the quantity here is a construction rather than a finding. The Plausibility is at the middle: every comparable system queues, and every comparable system also spends far less, so what is being observed may be the budget rather than the payer.

evidence basis: Multiple precedents · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the American system as it stands. Design: mechanistic — several countries combine universal coverage with administered prices and report waiting lists, but nothing separates the payment design from the funding level. Confounder: those systems spend about half what the United States does per head, so what is observed may be the budget rather than the payer; named and unresolved, and it is the strongest counter-argument. Direction: no reverse causation. Ceiling: set at 6.0, below the multiple-precedents ceiling, because the quantity is a construction rather than a finding.

Moving 160 million people at once

7.1of 100

Roughly 160 million Americans would leave employer coverage, about 1.8 million people work in health insurance, and a federal agency would have to be built to pay every medical bill in the country.

Value 5 · Enforcement costImpact 2.6Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is what a change of this size consumes on the way: the systems built, the staff hired and retrained, the claims processed twice while two systems run in parallel, and the working lives interrupted for people whose jobs disappear. It is priced at the middle of the scale, the level this site uses for the cost of running a rule, and it is a real cost rather than a transfer, because the hours are consumed. It is counted as an annual figure across the ten-year horizon of this evaluation rather than as a one-off, because that is how a transition of this size is actually paid for. Nothing is counted for the risk that the transition fails outright, which is not a cost but a different scenario. The value is the middle of the scale, and the transition is counted as an annual figure across the horizon rather than as a one-off.

About 160 million Americans hold employer-sponsored coverage and would move; roughly 1.8 million people work in health insurance and most of those jobs would not exist. A federal agency capable of paying every medical bill in the country would have to be built, which is the largest administrative construction the government has attempted since the Second World War, and the existing programme it would be built on top of currently covers a fifth of the population. Sixty billion dollars a year across the ten-year horizon is used, in a range from 20 to 150 — about 52 billion euro. That figure covers systems, parallel running, retraining and the earnings displaced workers lose before they find other work. It excludes the administrative cost of the finished programme, which is inside the administrative saving on the other side rather than here. The Impact is a third of the largest cost here, and unlike the others it stops once the change is complete.

▸ Show calculation ▾ Hide calculation

	Transition cost a year across the horizon Setting, range 20 to 150 billion dollars: systems, parallel running, retraining, and the earnings of about 1.8 million insurance workers before they find other work [2]	60 billion dollars	60 billion dollars a year
÷	In euro, at the standard weight for money exchange rate used throughout this evaluation	1.16 dollars to the euro, weight 1.0	51.72 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	2.59

Score 2.59 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 7.1 of 100

The counterfactual is no change. The components are estimable and none of them is measured for a project of this size, because nothing of this size has been attempted. The chain is complete and its links are ordinary — systems have to be built, staff have to be retrained, two systems run in parallel for a period, displaced workers spend time out of work — and the size of each is judgement. The confounder that would raise the figure substantially is that large government technology projects overrun, and the one that would lower it is that the existing Medicare programme already pays a fifth of the country's medical bills and could be scaled rather than replaced; both are named and neither is resolved. Reverse causation does not arise. The insurance-sector job losses are the firmest part, since the employment figures are published and the bill's own transition provisions acknowledge them. The Plausibility is a little above the middle: every component is ordinary and the size of a transition on this scale has no precedent to be measured against.

evidence basis: Projection · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: no change. Design: mechanistic — the components are ordinary and each is a judgement, because nothing of this size has been attempted. Confounder: large government technology projects overrun, which would raise the figure, while the existing Medicare programme already pays a fifth of the country's bills and could be scaled rather than replaced, which would lower it; both named, neither resolved. Direction: no reverse causation. Ceiling: projection 6.0 binds and mechanistic gives the same.

Medicines that are not developed

1.5of 100

The American market pays roughly twice what other rich countries do for branded medicines and funds a corresponding share of the world's pharmaceutical research. A single buyer negotiating one price would change that.

Value 9.5 · HealthImpact 0.5Plausibility 3.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is health that never arrives: treatments that are not developed, or arrive later, because the revenue that would have funded them was negotiated away. It sits just below the top of the scale, in the class this site uses for life and health. The effect is global rather than American, and it is counted anyway, because a life year is priced the same here wherever it is lived. Nothing is counted for the money patients and governments save on medicines, which is inside the argument about paying every provider a single rate. Nothing is counted for research that is redirected rather than abandoned. The value sits just below the maximum, and the effect is counted although it falls mostly outside the United States, because this site prices a life year the same wherever it is lived.

Global pharmaceutical research runs at roughly 250 billion dollars a year and the American market supplies a disproportionate share of the revenue that funds it. A single buyer negotiating one national price would cut industry revenue by something like a fifth. The published elasticity of new drug approvals to market size is about 0.23, so a fifth less revenue implies about 4.6 percent fewer new medicines. New medicines are estimated to account for a substantial part of the annual gain in life expectancy across rich countries; five million quality-adjusted years a year globally is used as that contribution, in a range from one to fifteen million. Applied together that gives 230,000 quality-adjusted years a year, or 9.2 billion euro. Every number in that chain is contested and the range on the result spans a factor of ten. The Impact is the smallest here, a fortieth of the largest cost, and it is the argument whose chain has the most links and the least measurement.

▸ Show calculation ▾ Hide calculation

	Global health gain attributable to new medicines each year Setting, range one to fifteen million: new medicines account for a substantial but disputed part of the annual gain in life expectancy across rich countries [9]	five million quality-adjusted years	5,000,000 quality-adjusted years
×	Fewer new medicines once industry revenue falls by a fifth at a measured elasticity of new approvals to market size of about 0.23 [9]	4.6 %	230,000 quality-adjusted years
×	Value of the years the value of a healthy life year used across this site, applied wherever the year is lived	40,000 euro each	9.2 billion euro a year
÷	Normalised Impact scale of this evaluation	20 billion euro a point	0.46

Score 0.46 Impact × 9.5 Value × 3.5 Plausibility ÷ 10 = 1.5 of 100

The counterfactual is the present American pricing of branded medicines. The chain is named — lower prices, lower revenue, less research, fewer medicines, less health — and only the middle link has a measured parameter: the elasticity of new approvals to market size, estimated on the expansion of Medicare drug coverage and on demographic changes in market size. The confounder that matters is that pharmaceutical revenue and research spending are only loosely coupled, because much research is funded from capital markets on expectations rather than from current revenue, and because a large share of basic research is publicly funded already; that is named and unresolved. Reverse causation is a live concern for the elasticity itself, since firms lobby for the markets they expect to serve. The last link — from fewer medicines to less health — rests on an attribution that health economists dispute sharply. Because the chain is long and one of its links has a finding against it, this sits below the middle. The Plausibility is well below the middle: the mechanism is coherent, one link of five is measured, and the coupling between revenue and research is weaker than the argument needs.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain open · P 3–3.5

Counterfactual: the present American pricing of branded medicines. Design: mechanistic — only the elasticity of new approvals to market size is measured; the links on either side of it are not. Confounder: research is funded from capital markets on expectations rather than from current revenue, and much basic research is publicly funded, so revenue and research are loosely coupled; named and unresolved. Direction: reverse causation is live for the elasticity itself, since firms lobby for the markets they expect to serve. Ceiling: mechanistic 6.0 binds. Band: chain open — the coupling between revenue and research carries the quantity and has a finding against it.

The chain is named but the link that carries the quantity is weak: research spending tracks expected future markets and capital availability more closely than current revenue, and a large share of basic research is publicly funded. Read back: rather less often than not, a price cut of this size produces the loss of medicines assumed here.

Open: Other rich countries have negotiated branded prices down for decades, and approval counts by therapeutic area are published. Comparing research intensity against changes in the share of global revenue those countries supply would test the coupling directly and could carry this above 5.

Summary

This is the largest single reorganisation evaluated on this site and it comes out ahead, on a ledger whose two biggest entries almost cancel. Ending the American billing apparatus is worth about 172 billion euro a year of freed staff and systems; the extra care people use when it costs nothing consumes about 161 billion euro of resources that the one randomised experiment on the question suggests mostly does not improve health. What tips the balance is everything around those two: 17,000 people a year who do not die because they are covered, medical bills that stop landing on whoever happens to be ill, and provider prices roughly halved for the commercially insured. Against them stand a queue that every comparable system has, a transition that moves 160 million people at once, and a global research effort that the American market currently funds. The financing is deliberately not scored — the bill leaves it to a separate act, and a payroll tax and a progressive income tax would produce entirely different answers about who wins.

Outlook — effect over time

Better for the future · 0.62 previous scale

Sources

Centers for Medicare and Medicaid Services: National health expenditure data. cms.gov
Congress.gov: H.R. 3069, Medicare for All Act. congress.gov
Congressional Budget Office: Economic effects of five illustrative single-payer health care systems. cbo.gov
Center for American Progress: Excess administrative costs burden the U.S. health care system. americanprogress.org
Scheinker and colleagues, Health Services Research: Reducing administrative costs in US health care: assessing single payer and its alternatives. onlinelibrary.wiley.com
Miller, Johnson and Wherry, Quarterly Journal of Economics: Medicaid and Mortality: New Evidence from Linked Survey and Administrative Data. academic.oup.com
Baicker and colleagues, New England Journal of Medicine: The Oregon Experiment: effects of Medicaid on clinical outcomes. nejm.org
RAND Corporation: The Health Insurance Experiment: a classic study speaks to the current health care reform debate. rand.org
Dubois and colleagues, RAND Journal of Economics: Market size and pharmaceutical innovation. onlinelibrary.wiley.com

Last reviewed by Claude Opus 5 · September 6, 2026 · 2× AI, not yet reviewed by a human

September 6, 2026AI reviewClaude Opus 5record updated
i_spanne und normalisierung an allen 8 Argumenten, Eintrittsgruppe fee-schedule-bites, massstab_hinweis ohne r. Kategorie bleibt Besser (P(D>0) 0,88). Befund des Laufs: das Spannen-Werkzeug rechnet an Gewichts-Schritten zu eng, weil die Setzungsspanne dort die Differenz zu 1,0 skaliert und nicht das Gewicht — betrifft jeden genetteten Transfer im Bestand.
September 6, 2026AI reviewClaude Opus 5First evaluation
First evaluation: the reorganisation scored and the financing deliberately left out, because the bill leaves it to a separate act.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →