Restore the Enhanced Credits

Bring back the larger Marketplace premium credits that lapsed at the end of 2025, for three more years.

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 →

Between 2021 and 2025 the federal premium credit covered a larger share of Marketplace premiums and was paid to households above four times the poverty line as well. Both features lapsed at the end of 2025, and what enrollees pay out of pocket rose by 58 percent. The House voted in January 2026 to restore the earlier schedule for three years; the Senate has not acted. Restoring it would leave the eligibility rules, the benchmark plan and the insurer market untouched and change only the size of the credit and who may claim it. This evaluation looks four years ahead, over the three restored years and the year in which the schedule would lapse again.

Balance

Balanced · 0.56 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 18 · 56 % Against 14 · 44 %

Size class: large Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 5 billion euro per year. The balance turns on one setting. A euro reaching a Marketplace household is valued here at 1.9 times a euro at median income, against 1.0 for the federal euro that pays for it. At 1.5 the two sides are level; at 2.4 the pro side leads by a wide margin. How we score →

Arguments for

Arguments against

7 arguments evaluated · Scoring v1.3 Δ absolute +4

Arguments — For

4 arguments · top 3 shown

Lower premiums for those insured either way

14of 100

Most of the money does not buy new coverage. It lowers what 15.3 million people who would be insured in any case pay each month, and those households sit largely in the bottom two fifths of American incomes. Whether that is worth the federal money is the central question of this debate, and it turns on what a euro is worth to them.

Value 5 · Household budgetsImpact 5.1Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money in household budgets, and money is money whatever it later buys. This site prices it at the middle of the scale, the level it uses for any euro that changes hands without changing what exists in the world. That the receiving households are poorer than average does not raise the value; it raises the weight the euro carries, and that is counted in the Impact. Booking it in both places — once as a good for the badly off and once as a weighted euro — would price the same fact twice. What the money does after it arrives, in health or in security, is counted in the other arguments. The value is the middle of the scale, because the stream is a transfer of money and the distribution it improves is priced in the Impact.

About 17.5 million people are expected to hold Marketplace coverage in 2026 without the restored credits ^[3]. Roughly 2.2 million of them earn above four times the poverty line and are treated separately below; the remaining 15.3 million would be insured either way, and for them the credit is not coverage but a smaller bill ^[3]^[4]. What enrollees pay rose by an average of 1,016 dollars a year when the schedule lapsed, which is 876 euro at 1.16 dollars to the euro ^[3]. Restoring it hands that back: 13.4 billion euro a year. Where those euro land is what decides the argument. About half of Marketplace enrollees live below twice the poverty line, which places them in the bottom fifth of American households, where this site counts a euro at two and a half times its worth at median income; most of the rest sit in the second fifth, at one and a half. The average used here is 1.9, in a range from 1.5 to 2.4. The Impact is the largest in this debate because it is a large sum landing where money is scarce, and it is the only argument the weighting alone can move by half.

▸ Show calculation ▾ Hide calculation

	People who keep Marketplace coverage either way [3][4]	17.5 million minus 2.2 million above the old income ceiling	15.3 million people
×	Premium relief per person what enrollees pay rose by an average of 1,016 dollars for 2026 when the schedule lapsed; converted at 1 euro = 1.16 dollars [3]	876 euro a year	13.4 billion euro
×	Weight of a euro at these incomes Setting, range 1.5 to 2.4: about half of Marketplace enrollees live below twice the poverty line, in the bottom fifth of American households where this site counts a euro at 2.5; most of the rest sit in the second fifth at 1.5	1.9	25.5 billion euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	5.1

Score 5.1 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 14 of 100

The mechanism is statutory rather than behavioural: the credit is a formula, and lowering the share of income a household must contribute lowers what it pays. There is nothing to identify and no chain of reactions to check — the counterfactual is the schedule now in force, and the schedule that would replace it is written in the bill ^[9]. What is estimated is the number of people it reaches. The budget office projects Marketplace enrolment, and its record on that particular series has been mixed over the past decade, which is why the enrolment figure carries a range rather than a point. The average credit per enrollee is drawn from what actually happened in January 2026, when the schedule lapsed and premium payments were observed to rise ^[3]. That makes it a measured quantity rather than a forecast, which is the firmer half of this argument. The looser half is the weight, and that doubt is written into the derivation as a range rather than discounted here. The Plausibility sits at the upper end of what a projection can carry: the rule itself is certain, and only the count of people it reaches is estimated.

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

Counterfactual: the credit schedule in force from 2026. Design: definitional — the credit per household follows from the statutory contribution schedule, not from a behavioural response, and the average change was additionally observed in January 2026 when the schedule lapsed [3]. Confounder: none applies to a formula; the estimated element is the enrolment count, and that doubt sits in the 13 to 18 million band. Direction: not applicable, no causal claim about behaviour is made. Ceiling: a budget projection caps this at 6.0, because the enrolment count is a forecast; definitional carries no ceiling of its own.

Newly covered households stop carrying the risk

2of 100

For the 2.1 million people who are insured only because of the credit, the largest change is not the care they use but the bill they no longer face if something goes wrong. An uninsured household with a hospital admission owes the full amount. What that protection is worth is contested, and the honest answer is that it is worth less than it costs.

Value 5 · Household budgetsImpact 0.7Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money again, this time the money a household does not have to find when it is unlucky. Insurance puts a small certain payment in place of a large uncertain one, and the difference between those two positions is a real gain even where the expected amounts match. It is priced at the middle of the scale like any other money stream. The medical care the newly covered actually use is not counted here: most of it happened anyway and was written off unpaid, so paying for it moves money between the federal budget and providers without changing what exists. The distributional weight on the households receiving the protection is carried in the Impact, not here. The value is the middle of the scale, because what is gained is a money position rather than a change in health.

The 2.1 million people who hold coverage only because of the credit face an average benchmark premium of about 7,200 dollars, or 6,200 euro at 1.16 dollars to the euro — 13.0 billion euro of medical cost that an insurer now carries instead of a household ^[3]. Only a fraction of that is a gain to them. Most of it is the bill itself, which in the uninsured state would in part have gone unpaid, and moving an unpaid bill from a hospital to the federal budget changes who pays without changing anything else. The part that is a genuine gain is the protection: not having to find a large sum at the worst moment. The one direct measurement of what such coverage is worth to its recipients found that Medicaid recipients would pay between a fifth and two fifths of what their coverage costs, and that most of that figure is the transfer rather than the protection ^[8]. Fifteen percent of the premium is used here as the protection share, in a range from 5 to 30 percent: 1.95 billion euro a year, weighted at 1.9 for the incomes it reaches. The Impact is a fraction of what the coverage costs, which is the point of the argument: this stream is what the newly covered gain, not what is spent on them.

▸ Show calculation ▾ Hide calculation

	People newly insured [2]		2.1 million people
×	Medical cost now carried by an insurer instead of a household the average benchmark premium for 2026 is about 7,200 dollars; converted at 1 euro = 1.16 dollars [3]	6,200 euro a year each	13 billion euro
×	Worth of not carrying that risk Setting, range 5 to 30 percent: Medicaid recipients were measured to value their coverage at a fifth to two fifths of its cost, and most of that is the bill rather than the protection against it [8]	15 %	1.95 billion euro
×	Weight of a euro at these incomes the same weighting as the argument above	1.9	3.71 billion euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	0.74

Score 0.74 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 2 of 100

That the stream exists is not in doubt — insurance shifts risk, and that is what people buy it for. What is estimated is its size, and the source used here was not built to answer exactly this question. The Oregon study measured what Medicaid recipients would give up to keep their coverage, using the same random draw that makes the health findings credible, and separated the total into the part that is a transfer and the part that is protection ^[8]. The counterfactual is again the untreated arm of the lottery, so the direction is not in question and no reverse causation is possible. Splitting the total into a protection share and applying that share to a different population is a construction built on their result rather than their result itself. The Marketplace group is much better off than Oregon's applicants, and better-off households are both more able to absorb a shock and more able to pay for insurance in the first place, which pulls the share in opposite directions. The Plausibility sits just above the middle: the underlying measurement is randomised, and the step from it to the number used here is an inference this evaluation makes on its own.

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

Counterfactual: the untreated arm of the Oregon Medicaid lottery. Design: experimental — willingness-to-pay recovered from randomised assignment (Finkelstein, Hendren and Luttmer, JPE 2019 [8]). Confounder: selection of healthier or wealthier applicants into coverage, removed by the randomisation. Direction: no reverse causation, assignment was drawn by lot before any outcome. Ceiling: experimental 8.5 binds below the single-study ceiling of 9.0; a context transfer of 2.0 applies twice over, once for the population and once because the split into a protection share is this evaluation's construction rather than the study's finding. The size doubt is in the 5 to 30 percent band.

Fewer deaths among the newly covered

1.2of 100

Insurance does not only pay bills; it decides when a treatable condition is found. About 2.1 million more people would hold coverage in an average year of the restored schedule. The clearest measurement of what coverage does to survival comes from Medicaid, and carrying it across to a younger, better-off group is where the uncertainty sits.

Value 10 · LifeImpact 0.2Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the number of people alive at the end of a year who would otherwise not have been. Nothing on this site is weighted above that. The people concerned are not an abstraction: they are adults in their fifties and early sixties with a cancer, a heart condition or a diabetes found late or not at all, because the appointment that would have found it was never made. Coverage does not decide the outcome alone, but it decides whether the first visit happens. This argument prices only the deaths; illness treated without anyone dying is counted separately, and the money is counted twice over below. The grief of the people around each death is included here, at a tenth of the weight of the death itself. The value is the highest the scale allows, because the stream is human lives and nothing else is folded into it.

Without the restored schedule, average monthly Marketplace enrolment falls from 22.3 million in 2025 to about 17.5 million in 2026, after what enrollees pay out of pocket rose 58 percent ^[3]. With it, the budget office puts the additional number of insured people at 0.4 million in the first year, 3.0 million in the second, 4.0 million in the third and 1.1 million in the fourth, an average of 2.1 million ^[2]. How many of those people live who would otherwise have died is the open question. The clearest measurement comes from the Medicaid expansions: low-income adults aged 55 to 64 in expanding states saw annual mortality fall by 0.132 percentage points, a reduction of 9.4 percent against a base of 1.4 percent ^[5]. Marketplace enrollees are younger and better off, so their base mortality is roughly a third as high; applying the same relative reduction and cutting it by a further third — because the people who drop coverage when premiums rise are the healthier ones — gives one death avoided for every 3,400 covered years. That is 618 deaths a year, in a range from 210 to 1,400 depending on how much of the measured effect carries over. The grief of the bereaved adds a tenth on top. The Impact is moderate for a national programme: the number of people affected is large, but the share of them for whom coverage decides survival in a given year is small.

▸ Show calculation ▾ Hide calculation

	People insured who would be uninsured without the credits, yearly average [2]	budget office path 0.4, 3.0, 4.0 and 1.1 million	2.1 million people
÷	Deaths avoided Setting, range one per 1,500 to one per 10,000: annual mortality fell 9.4 percent for low-income adults aged 55 to 64 who gained Medicaid, carried over to the younger Marketplace age mix and cut by a third because the people who drop out when premiums rise are the healthier ones [5]	one per 3,400 covered years	618 deaths a year
×	Value of the lives the value of a statistical life used across this site	1.4 million euro each	865 million euro
+	Grief of the bereaved the surcharge this site books alongside every stream of deaths	10 % of 618 cases × 1.4 million euro	87 million euro
=	Total per year		952 million euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	0.19

Score 0.19 Impact × 10 Value × 6.5 Plausibility ÷ 10 = 1.2 of 100

Two independent designs point the same way. The Medicaid measurement compares expanding with non-expanding states before and after 2014, using death records linked to survey data, and reports the effect against a pre-period in which the two groups' mortality moved in parallel ^[5]. The obvious objection is that states which expanded differ from those which did not in ways that also move mortality; the parallel pre-period is the test for that, and it holds. Reverse causation does not arise, because the expansion decision was taken by state legislatures rather than by the people whose deaths are counted. Separately, a randomised trial in which the tax authority wrote to 3.9 million uninsured households found both higher coverage and lower mortality among adults aged 45 to 64 — one fewer death for every 1,587 households written to ^[6]. What neither study establishes is the effect for this population. Marketplace enrollees are on average two decades younger than the Medicaid group in the first study and considerably better off, and nobody has measured their survival directly. That gap is the whole reason the figure used here sits at the low end of the measured range rather than at its centre. The Plausibility is above the middle: the direction is established by a randomised trial and a state comparison that both hold up, and what is missing is a measurement on this particular group.

evidence basis: Converging studies · P ceiling 8 identification: Quasi-experimental · rung ceiling 8

Counterfactual: non-expansion states, plus the untreated arm of the outreach trial. Design: quasi-experimental — difference-in-differences with a tested parallel pre-trend on linked survey and death-record data (Miller, Johnson and Wherry, QJE 2021 [5]); direction independently supported by an experimental design (Goldin, Lurie and McCubbin, QJE 2021 [6]). Confounder: expansion states differ in underlying health trends, handled by the pre-period test and the event-study specification. Direction: no reverse causation, the expansion decision is a state legislative act; selection into treatment is at state level and tested. Ceiling: quasi-experimental 8.0 binds below the converging-studies ceiling of 9.0; the context transfer from Medicaid 55-64 to a younger Marketplace population costs a further 1.5. The size doubt is carried by the 210-1,400 band, not by P.

Illness treated instead of endured

0.4of 100

Most of what insurance buys is not survival but treatment: a depression that is diagnosed, a chronic condition that is managed, a prescription that is filled. The Oregon lottery is the one place where this was measured against a randomly chosen control group. It found a large effect on mental health and none at all on blood pressure, cholesterol or blood sugar after two years.

Value 9 · HealthImpact 0.1Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is health restored or maintained short of survival: depression that lifts, pain that is treated, a condition that is managed instead of drifting. It belongs to the same class as the deaths in the argument above, and it is set one step below the top of that class because the losses counted here are recoverable and a death is not. What is priced is the person's own condition, not what the treatment costs anyone else; the money moves in the arguments below. Where a treated depression means someone keeps working, the gain belongs to them rather than to their employer, and it is not booked twice. Nothing about the size of the group enters here — that is the Impact's job. The value sits one step below the maximum: the stream is health itself, but health that can be regained.

With 2.1 million more people insured in an average year ^[2], the question is how much health each covered year buys. The Oregon lottery assigned Medicaid at random among applicants and then measured them: two years on, the treated group showed a 30 percent lower rate of depression, higher use of preventive care and far less financial strain, and no measurable difference in blood pressure, cholesterol or blood sugar ^[7]. Taking only what was found, the depression result alone is worth roughly 0.014 healthy life years per covered year for that population. Marketplace enrollees are less poor and start from a better baseline, so a third of it is carried over: 0.005 healthy life years per covered year, in a range from 0.002 to 0.015. Across 2.1 million covered years that is about 10,500 healthy life years a year. The figure is deliberately narrow — it books the one outcome the trial actually measured and leaves out the physical markers it did not. The Impact is small next to the money in this debate, because a covered year buys a modest amount of measurable health even where it clearly buys some.

▸ Show calculation ▾ Hide calculation

	People insured who would be uninsured without the credits [2]		2.1 million people
×	Healthy life years gained Setting, range 0.002 to 0.015: the Oregon lottery measured a 30 percent fall in depression and no change in blood pressure, cholesterol or blood sugar; the depression finding alone is worth about 0.014 healthy life years a year for that much poorer group, and a third is carried over here [7]	0.005 per covered year	10,500 healthy life years
×	Value of the healthy life years the value of a healthy life year used across this site	40,000 euro each	420 million euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	0.08

Score 0.08 Impact × 9 Value × 6 Plausibility ÷ 10 = 0.4 of 100

The finding rests on a genuine lottery. Oregon had more applicants than places for its Medicaid expansion in 2008 and drew names at random, so the compared groups differ only in whether they were drawn ^[7]. The counterfactual is therefore the untreated arm of that draw rather than a group selected by anything about the people in it, and neither reverse causation nor selection into treatment can arise. The confounder that would otherwise dominate — that healthier people seek out coverage — is removed by the randomisation itself. Two things limit what the trial carries. It ran in one state on a much poorer population than the Marketplace serves, and its physical-health results were null, so anyone citing it must either restrict the claim to what was found or leave the ground it stands on. This argument restricts it. What remains open is whether the mental-health gain transfers at all to households with three times the income. The Plausibility is a little above the middle: the finding is as clean as health research gets, and the distance between the group it was measured on and the group it is applied to is large.

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

Counterfactual: the untreated arm of the Oregon Medicaid lottery. Design: experimental — randomised assignment among applicants (Baicker and others, NEJM 2013 [7]). Confounder: healthier people selecting into coverage, removed by the randomisation. Direction: no reverse causation possible, assignment preceded outcomes and was drawn by lot. Ceiling: experimental 8.5 binds below the single-study ceiling of 9.0; the context transfer from an Oregon Medicaid population to national Marketplace enrollees costs 2.0, and the null result on physical markers is respected by booking only the depression stream rather than by a further deduction.

Arguments — Against

3 arguments

What the credits cost the federal budget

13of 100

The restored schedule costs about 30 billion euro a year, of which 22.3 billion pays for the streams counted on the pro side. The money is borrowed: the federal deficit already runs at 5.8 percent of output. Nothing in the design makes the cost fall over time.

Value 5 · Public financesImpact 4.5Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is federal money, priced at the middle of the scale like any other euro. This site does not treat public money as cheaper or dearer than private money; the difference in what a euro is worth to the people at either end of a transfer is counted in the Impact, not here. Nor is a deficit treated as a harm on top of the spending: borrowing shifts who pays and when, and that shift is already inside the euro figure. What the money buys is counted in the four arguments above, so this argument is the payment and nothing else. The value is the middle of the scale, the level this site uses for public money whatever it is spent on.

The budget office scores a permanent restoration at 350 billion dollars over ten years, an annual rate of about 35 billion dollars, or 30.2 billion euro at 1.16 dollars to the euro ^[1]. A three-year restoration costs roughly the same per year while it runs, with a partial year at either end. Of that total, 7.9 billion euro goes to households above four times the poverty line and is booked in the argument below, leaving 22.3 billion here ^[4]. That sum splits two ways: about 13.4 billion lowers the bills of people who would be insured anyway, and about 8.9 billion buys coverage for people who otherwise would not have it. The federal euro carries the standard weight of one for public money, so the figure passes through unchanged. The range around it is set by enrolment — 18 to 27 billion euro depending on how many people return to the Marketplace — and it is the same range that sets the benefit, so the two move together. The Impact is the second largest in this debate and is the mirror of the largest: the same money, seen from the side that pays it.

▸ Show calculation ▾ Hide calculation

	Cost of the enhanced credits converted at 1 euro = 1.16 dollars [1]	350 billion dollars over ten years, so about 35 billion a year	30.2 billion euro
−	Money reaching households above the old income ceiling booked separately in the contra argument on who receives the credit [4]	7.9 billion euro	22.3 billion euro
×	Weight of a euro in the federal budget the standard weight for public money on this site, against 1.9 on the receiving side	1.0	22.3 billion euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	4.46

Score 4.46 Impact × 5 Value × 6 Plausibility ÷ 10 = 13 of 100

The claim is that a restored credit schedule costs money, which is arithmetic on a formula rather than a prediction ^[1]^[9]. The counterfactual is current law with the enhanced schedule lapsed, and nobody disputes that restoring it raises outlays. What is estimated is the amount, and that depends entirely on how many people enrol. The ten-year score and the year-by-year coverage path come from the same model, so cost and coverage move together rather than independently: if enrolment disappoints, this figure and the benefit above shrink at once ^[1]^[2]. The one asymmetry is that a credit paid for someone who is not in fact enrolled costs money without producing coverage; where that happens it belongs in the enrolment figure above, which is where it is subtracted, rather than being added here as a second cost. The Plausibility is at the top of what a budget projection can carry: the rule is certain, and only the number of people it reaches is estimated.

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

Counterfactual: current law with the enhanced schedule lapsed. Design: definitional — an outlay that follows from the statutory credit formula, with no behavioural link carrying the amount other than the enrolment count. Confounder: none for the formula; the enrolment count is projected and banded at 18 to 27 billion euro. Direction: not applicable. Ceiling: a budget projection caps this at 6.0, because the amount rests on the budget office's enrolment forecast; definitional carries no ceiling of its own. The coupling to pro-3 and pro-4 is deliberate — one enrolment number drives both sides.

A tenth of the money lands near the median

0.7of 100

The enhanced schedule pays a credit to households above four times the poverty line, who received nothing before. For an older couple just above that line it can be worth more than 10,000 dollars a year. Those households are not poor, and a euro reaching them is worth slightly less than the euro that left the federal budget.

Value 5 · Household budgetsImpact 0.2Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money once more, at the middle of the scale. The objection is not that these households deserve nothing; it is that the euro ends up where it is worth slightly less than where it started, and the difference is a loss. Only that difference is counted here — the payment itself already sits in the argument above, on the paying side. Treating the recipients' income as a separate question of fairness would price the same fact twice, once as fairness and once as weight. The value is the middle of the scale, and the distributional point is carried entirely by the Impact.

Around one in ten Marketplace enrollees, some 2.2 million people, have incomes above four times the poverty line — 62,600 dollars for a single person signing up for 2026 ^[3]^[4]. Under the pre-2021 rules they received nothing at all; the enhanced schedule caps their premium at 8.5 percent of income instead, which is worth most to older enrollees, for whom the benchmark premium can run to a fifth of income ^[4]. An average of 3,600 euro a year is used here, in a range from 2,000 to 6,000, giving 7.9 billion euro. The weighting is where the loss appears. A single person at four to five times the poverty line earns 63,000 to 78,000 dollars and sits close to the American median; a family of four at the same multiple earns 128,000 to 160,000 and sits in the fourth fifth, where this site counts a euro at 0.7. An average of 0.85 against the federal euro's 1.0 leaves 1.19 billion euro of value lost in the passage. The Impact is small: the sum is large, but the gap between what a euro is worth at either end of it is narrow, and only the gap is counted.

▸ Show calculation ▾ Hide calculation

	Enrollees above four times the poverty line [3]	about one in ten of 22.3 million	2.2 million people
×	Credit they receive only because the ceiling was removed Setting, range 2,000 to 6,000 euro: the credit is largest just above the old ceiling and for older enrollees, where the benchmark premium can exceed a fifth of income [4]	3,600 euro a year each	7.9 billion euro
×	Difference in what a euro is worth a single person at four to five times the poverty line sits near the American median at 1.0; a family of four at the same multiple sits in the fourth fifth at 0.7 — the euro is worth slightly less there than in the budget it came from	1.0 minus 0.85	1.19 billion euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	0.24

Score 0.24 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 0.7 of 100

Who receives the credit and how much follows from the income schedule in the statute, so the direction is not open to dispute ^[9]. The counterfactual is the pre-2021 rule, under which the credit stopped entirely at four times the poverty line, and the difference between the two schedules is arithmetic rather than behaviour. Two things are estimated. The first is the number of enrollees above the line, drawn from enrolment data and reasonably firm at about one in ten ^[3]. The second is the average credit they receive, which varies enormously with age and is the reason for the wide range around it. The weight applied to those households is a judgment rather than a measurement, and it matters less than it might because the sum it multiplies is small. The Plausibility is somewhat above the middle: the rule is certain, and both the average payment and the weight applied to it are estimates.

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

Counterfactual: the pre-2021 credit schedule, which stopped at four times the poverty line. Design: definitional — eligibility and amount follow from the statutory schedule. Confounder: none for the rule; the estimated elements are the count above the line and the average credit, banded at 2,000 to 6,000 euro. Direction: not applicable. Ceiling: a budget projection caps this at 6.0, because the average credit is a projection; definitional carries no ceiling of its own.

The same cliff is built again three years later

0.1of 100

A three-year restoration does not settle the question; it moves it to 2029. Everyone holding Marketplace coverage goes through the same re-shopping in and out that they went through in January 2026. The cost is time, and it falls on 21.5 million people twice.

Value 9 · Life timeImpact 0.0Plausibility 4.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is time people have to spend and get nothing for: comparing plans again, re-entering income details, finding a new doctor when a plan disappears. This site treats time of that kind as part of a person's life rather than as a matter of convenience, which places it near the top of the scale rather than near the bottom. It is not the same as time somebody chooses to spend; nobody re-shops an insurance plan for pleasure. What an hour is worth and how the hours are weighted are separate questions, and both are set out in the derivation. This argument counts only the hours, not any coverage lost in the process — that loss is already the difference between the two futures. The value is high because the stream is hours of life spent under compulsion, not a comfort that is lost.

About 21.5 million people would hold Marketplace coverage under the restored schedule, and each meets the transition twice: once when the credit returns and again when it lapses in 2029 ^[3]. Three hours per person across both transitions is used here, in a range from one to six — comparing plans, re-entering income data, and for those whose plan disappears, finding a new provider. That is 64.5 million hours, or 16.1 million hours in each of the four years this evaluation covers. The hours are valued at the rate this site uses for time spent under compulsion with nothing in return. What is not counted here is the coverage people lose in each transition; that loss is the difference between the two futures and is already inside the enrolment figures above. Nor is the insurers' own repricing counted, which is real but small against a market of this size. The Impact is the smallest in this debate by a wide margin: the burden is real and it is spread very thinly over a very large number of people.

▸ Show calculation ▾ Hide calculation

	People re-shopping coverage at each edge of the three-year window [3]		21.5 million people
×	Forced hours per person across both transitions Setting, range 1 to 6 hours: comparing plans, re-entering income data, and for those who lose a plan, finding a new provider	3 hours	64.5 million hours
÷	Spread over the four years of this evaluation	4	16.1 million hours a year
×	Value of forced time the rate this site uses for time a person must spend with nothing in return	6.85 euro an hour	110 million euro
÷	Normalised Impact scale of this evaluation	5 billion euro a point	0.02

Score 0.02 Impact × 9 Value × 4.5 Plausibility ÷ 10 = 0.1 of 100

The claim has a precedent that was observed rather than predicted. In January 2026 the enhanced schedule lapsed, sign-ups fell by about three million, and enrollees across the country went through exactly the re-shopping described here ^[3]. The counterfactual for that episode is the preceding year, in which no schedule change occurred, and the difference in enrolment behaviour between the two is documented. What the precedent does not supply is the quantity. Nobody has measured how many hours a Marketplace transition costs a household, and the three-hour figure is a construction rather than a finding. The chain from a schedule change to hours spent is short and every link in it is visible, and the counter-argument — that people would re-shop anyway — is answered by the fact that most enrollees are auto-renewed when nothing changes. What is missing is only the measurement. The Plausibility is at the top of the range for a claim whose steps are all visible and whose size nobody has counted.

evidence basis: Precedent · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the 2025 enrolment year, in which no schedule change occurred. Design: mechanistic — a short behavioural chain (schedule change → re-shopping → hours), with the January 2026 lapse as an observed precedent for the first two links [3]. Confounder: enrollees who would have re-shopped anyway, answered by the auto-renewal default. Direction: no reverse causation, the schedule change is a legislative act. Ceiling: mechanistic 6.0 binds below the precedent ceiling of 8.5, because no source carries the hours figure. Finding: the chain is closed but unmeasured — every link is named, and what is missing is only the count of hours.

Nothing speaks against the claim; what is missing is a measurement of the hours. The steps are all named and the one counter-mechanism — enrollees who re-shop anyway — is answered by the auto-renewal default. Read back: the transition described here happens in fewer than half of the cases assumed, or costs less time, about as often as it happens as described.

Open: A survey of Marketplace enrollees on time spent during the January 2026 transition would replace the three-hour setting with a measurement and could carry P to 6.0, or shrink the argument to nothing if the true figure is under an hour.

Summary

The question this measure poses is narrow: is 30 billion euro a year of federal money well spent on lowering what Marketplace enrollees pay? Most of it — about 13 billion — buys nobody any coverage at all, but lowers the bills of 15.3 million people who would be insured either way, and the whole case for that half rests on the fact that those households are poor. The coverage the credits genuinely do buy is worth less than it costs, which is the ordinary finding for health insurance rather than an argument against it; the deaths avoided are real and measured, but modest at this scale. Against them sit the cost itself and a tenth of the money that lands on households near the American median, where it is worth slightly less than where it came from. The two sides finish close enough that the income weighting decides the outcome: at 1.5 they are level, at 2.4 the case is clear.

Outlook — effect over time

Balanced · 0.56 previous scale

Sources

Congressional Budget Office: The Estimated Effects of Enacting Selected Health Coverage Policies on the Federal Budget and on the Number of People With Health Insurance. cbo.gov
Congressional Budget Office: Estimated Budgetary Effects of S. 3385, the Lower Health Care Costs Act. cbo.gov
KFF: What We Know So Far About 2026 ACA Marketplace Enrollment, Premiums, and Deductibles. kff.org
KFF: A Steep Subsidy Cliff Looms for Older Middle-Income Enrollees if ACA Enhanced Tax Credits Expire. kff.org
Miller, Johnson and Wherry, Quarterly Journal of Economics 136(3): Medicaid and Mortality: New Evidence From Linked Survey and Administrative Data. academic.oup.com
Goldin, Lurie and McCubbin, Quarterly Journal of Economics 136(1): Health Insurance and Mortality: Experimental Evidence from Taxpayer Outreach. academic.oup.com
Baicker and others, New England Journal of Medicine 368: The Oregon Experiment: Effects of Medicaid on Clinical Outcomes. nejm.org
Finkelstein, Hendren and Luttmer, Journal of Political Economy 127(6): The Value of Medicaid: Interpreting Results from the Oregon Health Insurance Experiment. journals.uchicago.edu
Congressional Research Service: Enhanced Premium Tax Credit and 2026 Exchange Premiums: Frequently Asked Questions. congress.gov

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 an allen 7, normalisierung erstmals an allen 7 (nur globale Anker), Eintrittsgruppe deckungswirkung-uebertragbar, massstab_hinweis und summary_text ohne r. Das Paar con-1 ⇄ pro-3 bleibt unverdrahtet (P 6,0 gegen 5,5) — numerisch folgenlos, beide definitorisch. Kategorie steigt von Ausgeglichen (r 0,56) auf Deutlich besser (P(D>0) 0,93).
September 6, 2026AI reviewClaude Opus 5First evaluation
Created for the English side: seven arguments scored against the budget office coverage path and the randomised and quasi-experimental coverage literature.

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