Immigration and Border Enforcement

Public Charge Rule

Let immigration officers weigh any use of Medicaid, food aid or housing help against a green card applicant, replacing the narrower 2022 test.

8 arguments 18 Sources
  • 13.5 million people on Medicaid or CHIP live in a household with at least one noncitizen [4]
  • 5.6 million of them are citizen children [4]
  • 260,000 fewer children were on Medicaid within nine months of the announcement of the earlier rule in 2018 [6][7]

Since 2022 an officer deciding whether a green card applicant is likely to become a public charge could look only at cash aid and long-term care in an institution. A rule of 20 July 2026 lets officers weigh Medicaid, food aid and housing help too, because the administration wants immigrants to support themselves [1]. Twenty-two states and the District of Columbia sued on 14 September 2026 to stop it, and the rule took effect while the case continues [2].

The proposal

Let immigration officers weigh any use of Medicaid, food aid or housing help against a green card applicant, replacing the narrower 2022 test.

The key question

Does the rule save public money without pushing families out of aid they are entitled to?

The debate

The administration argues that Congress wants immigrants to be self-sufficient and that the narrow test let officers ignore most of the benefits that show whether they are. It expects households with noncitizens to draw billions of dollars less in aid each year. The states answer that the rule works through fear, so that families give up food aid and health coverage they are entitled to, including for citizen children. Congress has not voted on the rule, and its fate now lies with the federal court in New York.

  • “DHS agrees that aliens applying for admission and adjustment of status who are subject to the public charge ground of inadmissibility should be self-sufficient and should not depend on the government to meet their needs.”

    argues for: Department of Homeland Security, Trump administration Final rule, Federal Register, 20 July 2026, page 45331
  • “This rule preys on that fear and counts on families forfeiting the food assistance, health care coverage, and other public benefits to which they are legally entitled.”

    argues against: Letitia James, Democrat, Attorney General of New York Press release on the states' lawsuit, 14 September 2026

The analysis

Public Charge Rule

36% of the weight is for, 64% against.

An argument's weight combines three questions: how big the effect is, how certain it is and how much the good at stake matters. The larger the slice, the more the argument weighs.

For

Against

Answer to the key question

Does the rule save public money without pushing families out of aid they are entitled to?

It saves money, but mainly by pushing families out of aid. Budgets pay less because families with an immigrant member stop claiming food, cash and medical help for fear of harming a green card application, many of them citizens. Lost coverage means more untreated illness, more deaths among adults and lower earnings for children later. Whether fewer future claimants receive green cards, the rule's stated purpose, has not been estimated.

On the current analysis, the new public charge rule is a poor proposal: nearly everything it saves is taken from households that have very little.

Very likely worse

Very likely worse: Under the assumptions used, the proposal brings more than it costs in 2 of 100 model runs.

Overall effect: large

How strongly the proposal works overall, in five steps from very small to very large, comparable across all debates. The ring is always full, even when the overall effect is small; this step says how large it is.

The states have the better of the argument, for the reason they give: the rule saves money by scaring eligible families out of aid, not by admitting fewer needy people. More than half of the weight is money changing hands, food and cash aid that public budgets keep and poor families lose. A dollar does far more good in a household near the poverty line than in a public budget, so that exchange alone costs more than it saves. The strongest argument for the rule is the real Medicaid saving, and it is only a little larger than the earnings children lose later by going uninsured. The rule's own aim, fewer green cards for people likely to need aid, carries no number, because the department never estimated it.

In detail

All arguments

Arguments for3 arguments

Food, cash and housing aid saved

8.7points

About 6.2 billion dollars a year of the expected fall in payments is food aid, cash aid and other help that families stop claiming, and it stays in federal and state budgets. The amount moves with the number of families who withdraw, which the department sets from earlier experience.

Value 5 · Public financesImpact 2.7Plausibility 6.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is public money again, this time in the food aid program, the cash aid for older and disabled people, family cash aid and help with rent. It belongs to federal taxpayers, and for some programs to state taxpayers as well. A public dollar counts at the middle of the scale. Who loses the same dollars, and how much they matter there, is the paired argument on the other side. The value is the middle of the scale, because the item is money that stays in a public budget.

Impact

Of the 13.05 billion dollars a year the department expects, about 6.86 billion is Medicaid and CHIP [1][2][5]. The remaining 6.19 billion is everything else: food aid, which alone accounts for about 1.02 billion dollars of federal payments [2], cash aid for older and disabled people, family cash aid and help with rent. Food aid averaged 188 dollars per person a month in 2025 [16], and the cash aid for older and disabled people averaged 739 dollars a month in August 2026 [17]. In its proposal the department expected about 447,000 people to leave food aid and about 64,000 to leave the cash aid for older and disabled people [6]. The money stays in the public budgets, where a dollar counts at the standard weight of one. The range follows the number of families who withdraw, from half to 1.7 times the department's central rate, and how the total splits between Medicaid and the rest. That gives a Normalised Impact of about 5.3, in a range from 2.58 to 9.29, the same sum the families lose in the paired argument. The Impact is large because it is the full amount of aid not paid, before asking who loses it.

▸ Show calculation ▾ Hide calculation
13.05 billion dollars a year
Expected fall in benefit payments
− 4.05 billion federal ÷ 59 % federal share counted in the argument on Medicaid spending; Setting for the federal share, range 50 to 70 percent
6.19 billion dollars a year
Medicaid and CHIP part
÷ 1.16 dollars to the euro the rate used throughout this evaluation
5.34 billion euro-equivalent a year
In euro-equivalent
× 1.0 the standard weight for public money on this site, against 2.5 on the paying side
5.34 billion euro-equivalent a year
Weight of a public dollar
÷ 2 billion euro a point scale of this evaluation
2.67
Normalised Impact
Score 2.67 Impact × 5 Value × 6.5 Plausibility ÷ 10 = 8.7 of 100

Plausibility

The claim is that families who fear for a green card stop claiming food, cash and housing aid they are entitled to. The best evidence comes from the first attempt at this rule. After the 2018 announcement of the earlier public charge rule, child enrollment fell more in counties with many noncitizens than in counties with few, measured against each county's own trend since 2015 [7]. For Medicaid the study scaled that to about 260,000 fewer children nationally within nine months [6][7]. What could distort the comparison is that places with many immigrants were changing for other reasons; the study controls for local unemployment and state trends, and enrollment had moved in parallel before the announcement. It cannot run the other way, because the announcement came from Washington, not from the counties. For food aid the evidence is weaker: the county study found a fall in Texas that was not statistically certain [7]. Federal participation data show eligible citizen children in households with noncitizen adults falling from 80 to 64 percent between 2016 and early 2020 [6]. The 2025 budget law already removes some immigrants from food aid, which may shrink the group the department counts [1]. Read back: in about six or seven of ten comparable cases the aid not claimed comes to roughly this much. The Plausibility is above the middle and the same as on the families' side, because both describe one sum changing hands.

Evidence: Precedent · plausibility at most 7 Evidence of effect: Quasi-experimental · at most 8

Less Medicaid and CHIP spending

7.7points

Federal and state governments pay about 6.9 billion dollars a year less for Medicaid and CHIP when families with an immigrant member leave or never enrol. Once unpaid hospital care is taken off, about 5.5 billion remains. The amount moves one for one with how many families withdraw, which the department sets from what happened after the 2018 announcement.

Value 5 · Public financesImpact 2.4Plausibility 6.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is public money: the federal and state budgets that pay for Medicaid and CHIP. It belongs to taxpayers in every state, and a dollar not spent here can fund something else or reduce borrowing. Public money counts at the middle of the scale, the same weight as a dollar at median household income. The coverage itself and what it does for health are counted on the other side, in the arguments on deaths, illness, protection against bills and children's later earnings. The value is the middle of the scale, because what is saved here is money and nothing more.

Impact

The department expects benefit payments to households that include noncitizens to fall by about 13.05 billion dollars a year, federal and state combined [1]. The federal Medicaid and CHIP part of that is about 4.05 billion dollars a year [2]. The department's method charges the states a uniform 41 percent of Medicaid and CHIP costs [5], so the two programs together account for about 6.86 billion dollars a year. The rest is food, cash and other aid, counted in the next argument. Not all of it is saved, because people without coverage still fall ill and end up in emergency departments. When Tennessee cut adults from Medicaid in 2005, each newly uninsured person added close to 900 dollars a year of unpaid hospital care [13]. Hospitals expect about 1,400 dollars a person for the 2025 Medicaid cuts [18]. A fifth of the saving is taken off for this, in a range from 5 to 35 percent, leaving about 5.49 billion dollars a year. How many families withdraw is the largest doubt: the department's own scenarios run from a third to 1.7 times its central rate [5], and the range used here runs from half to 1.7 times. That gives a Normalised Impact of about 4.7, in a range from 2.25 to 8.46, a little below the food and cash aid in the next argument. The Impact is large because a small share of a very large group of enrollees stops claiming coverage.

▸ Show calculation ▾ Hide calculation
4.05 billion dollars a year
Federal Medicaid and CHIP payments not made
÷ 59 % the uniform federal match the department's method applies; Setting, range 50 to 70 percent
6.86 billion dollars a year
Federal share of these costs
[5]
× 80 % Setting, range 65 to 95 percent: about 900 dollars a year of unpaid hospital care per newly uninsured person in Tennessee, about 1,400 dollars expected for the 2025 cuts
5.49 billion dollars a year
Share not returning as unpaid hospital care
÷ 1.16 dollars to the euro the rate used throughout this evaluation
4.73 billion euro-equivalent a year
In euro-equivalent
× 1.0 the standard weight for public money on this site
4.73 billion euro-equivalent a year
Weight of a public dollar
÷ 2 billion euro a point scale of this evaluation
2.37
Normalised Impact
Score 2.37 Impact × 5 Value × 6.5 Plausibility ÷ 10 = 7.7 of 100

Plausibility

The claim is that fear of the rule makes eligible families give up Medicaid and CHIP, so the money is not spent. The best evidence comes from the first attempt at this rule. After the 2018 announcement of the earlier public charge rule, child enrollment fell more in counties with many noncitizens than in counties with few, measured against each county's own trend since 2015 [7]. For Medicaid the study scaled that to about 260,000 fewer children nationally within nine months [6][7]. What could distort the comparison is that places with many immigrants were changing for other reasons; the study controls for local unemployment and state trends, and enrollment had moved in parallel before the announcement. It cannot run the other way, because the announcement came from Washington, not from the counties. The current rule differs in detail from the 2019 one, and the 2025 budget law already narrows which immigrants qualify for Medicaid, which may shrink the group the department counts [1]. That is why its figure is used with a wide range rather than taken as given. Read back: in about six or seven of ten comparable cases, spending falls by roughly this much. The Plausibility is above the middle, because the withdrawal has been measured once already under a near-identical rule.

Evidence: Precedent · plausibility at most 7 Evidence of effect: Quasi-experimental · at most 8

Fewer green cards for likely claimants

0points

The rule's stated purpose is to admit fewer immigrants who would later need public aid. The department does not estimate how many applications it will deny or change, so this argument is recorded but not yet assessed.

Value 5 · Public financesImpact 0Plausibility 0
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is public money over the longer run: aid that would otherwise go to immigrants admitted today who later fall on hard times. It belongs to federal and state taxpayers. It counts at the middle of the scale, like every public dollar in this debate. Withdrawals by families already here are counted in the two savings arguments, not here. The value is the middle of the scale, because the item is money.

Impact

The department states the aim of the rule as making sure that immigrants can support themselves, but its analysis gives no number of applications it expects to deny or to see withdrawn [1]. Under the new rule an officer weighs age, health, family, income, education and skills case by case, without a fixed list of benefits that count [3]. Without an estimate of denials, there is no quantity of later aid avoided to count. Any saving from families already here who stop claiming is counted in the first two arguments. The Impact is therefore left at zero and the argument is recorded as open. The Impact is not set, because the step it depends on has no number.

Plausibility

Whether officers will deny more green cards under the new test, and how many, is not stated in the rule [1]. A denial also does not mean that a person leaves the country or never claims aid. Applicants exempt by law, such as refugees and asylees, are not affected at all [3]. Denial counts under the 2019 rule would be informative, but none are given in the sources used here. Until such a number exists, any plausibility for the size would be a guess. The Plausibility is not set, because there is nothing yet to judge.

Recorded, not yet scored: The department gives no estimate of how many green card applications the rule will deny or change, so the later aid it avoids cannot be sized; the argument is recorded and left unscored.

Evidence: Plausibility · plausibility at most 5

Arguments against5 arguments

Families lose food and cash aid

18points

Families who stop claiming lose about 6.2 billion dollars a year of food, cash and housing aid, almost all of them at or near the poverty line. A dollar in such a household counts for far more than the same dollar in a public budget, and that weighting decides this argument; its size moves with the number of families who withdraw.

Value 5 · Household budgetsImpact 5.5Plausibility 6.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is the money families live on: food aid, cash aid and help with rent. Food aid goes to households with incomes up to 130 percent of the poverty line, and the cash aid for older and disabled people to those with almost no income of their own [16][17]. In these households the money is the week's groceries, a month's rent or the bus fare to work. In the bottom fifth of incomes the American income bands count a dollar at 2.3 times its weight at the median, and that weight sits in the Impact, not in the value. The value is the middle of the scale; what makes the argument heavy is who loses the money.

Impact

The amount is the same 6.19 billion dollars a year as in the paired argument, the aid the families no longer claim [1][2]. Part of it is not cash: help with rent is worth less to a family than it costs, and some families turn to food banks instead. The loss is therefore set at 90 percent of the amount, in a range from 70 to 100 percent. Everyone who qualifies for these programs lives at or near the poverty line, in the bottom fifth of American households, where the weight on the American income bands is 2.3, in a range from 1.85 to 2.7. For a family that loses a quarter of its income the weight would rise further, so 2.3 is the cautious choice. The result is a Normalised Impact of about 12.0, in a range from 5.4 to 21.7, more than twice the saving it mirrors. The Impact is the largest in this debate because the same dollars weigh more than twice as much where they are lost as where they are saved.

▸ Show calculation ▾ Hide calculation
6.19 billion dollars a year
Food, cash and housing aid not claimed
÷ 1.16 dollars to the euro the rate used throughout this evaluation
5.34 billion euro-equivalent a year
In euro-equivalent
× 90 % Setting, range 70 to 100 percent: rent help is worth less than its cost, and some families find food banks
4.8 billion euro-equivalent a year
Share that is lost as money
× 2.3 the American band of the bottom fifth of incomes, 1.85 to 2.7, against 1.0 in a public budget
11.05 billion euro-equivalent a year
Weight of a dollar at these incomes
÷ 2 billion euro a point scale of this evaluation
5.52
Normalised Impact
Score 5.52 Impact × 5 Value × 6.5 Plausibility ÷ 10 = 18 of 100

Plausibility

The claim is that families who fear for a green card stop claiming food, cash and housing aid they are entitled to. The best evidence comes from the first attempt at this rule. After the 2018 announcement of the earlier public charge rule, child enrollment fell more in counties with many noncitizens than in counties with few, measured against each county's own trend since 2015 [7]. For Medicaid the study scaled that to about 260,000 fewer children nationally within nine months [6][7]. What could distort the comparison is that places with many immigrants were changing for other reasons; the study controls for local unemployment and state trends, and enrollment had moved in parallel before the announcement. It cannot run the other way, because the announcement came from Washington, not from the counties. For food aid the evidence is weaker: the county study found a fall in Texas that was not statistically certain [7]. Federal participation data show eligible citizen children in households with noncitizen adults falling from 80 to 64 percent between 2016 and early 2020 [6]. The 2025 budget law already removes some immigrants from food aid, which may shrink the group the department counts [1]. Read back: in about six or seven of ten comparable cases the aid not claimed comes to roughly this much. The Plausibility is above the middle and the same as on the budgets' side, because both describe one sum changing hands.

Evidence: Precedent · plausibility at most 7 Evidence of effect: Quasi-experimental · at most 8

Lower earnings for children later

5.1points

About 530,000 children a year spend a year without coverage. The Congressional Budget Office assumes each year of Medicaid in childhood raises adult earnings by about 0.4 to 0.5 percent, which makes this the largest harm after the lost aid, and the least directly measured.

Value 8 · Life chancesImpact 1.6Plausibility 4
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is a child's chances in life: health in the years that shape later schooling and work. It is lost by children, most of them citizens, whose parents stop their Medicaid or CHIP out of fear for a green card in the family. What they lose shows up decades later as lower earnings, which this site counts as lost life chances, the value it gives to education. The weight sits between education and health, because the budget office traces the effect through both. The value is high, because the loss lands on children and lasts a working life.

Impact

At the department's central rate about 1.39 million people leave Medicaid or CHIP, and about 45 percent of them are children, since 5.6 million of the 13.5 million enrollees concerned are citizen children [4][5]. With 85 percent going uninsured, about 530,000 child-years of coverage are lost each year. The budget office assumes that each year of Medicaid in childhood raises adult earnings by typically about 0.5 percent, and by 0.4 percent on average for a policy that keeps children covered [14]. Its figure is half of what the literature implies, including studies that follow children born just before and after eligibility cut-offs into adulthood [14][15]. The value used here is 0.4 percent of the earnings of a working life, in a range from 0.2 to 0.8 percent. Across 530,000 child-years that is the equivalent of about 2,100 whole working lives of earnings each year. The Normalised Impact is about 3.2, in a range from 1.17 to 7.74, nearly five times the deaths in weight. Most of it falls after the five years counted here, but it is caused by coverage lost within them. The Impact is large because a small loss in each year of childhood adds up over forty years of work.

▸ Show calculation ▾ Hide calculation
1.39 million people a year
Enrollees who withdraw
× 45 % Setting, range 40 to 50 percent: 5.6 of the 13.5 million are citizen children
0.63 million children a year
Children among them
[4]
× 85 % Setting, range 70 to 95 percent
0.53 million child-years a year
Share left uninsured
× 0.4 % of a working life Setting, range 0.2 to 0.8 percent: the budget office assumes about 0.5 percent per year of childhood Medicaid, 0.4 on average for a policy that keeps children covered
2,127 working lives of earnings
Adult earnings lost per child-year
× 1.5 million euro-equivalent the German lifetime earnings anchor, used here as a declared setting because no American figure has been adopted
3.19 billion euro-equivalent a year
Earnings of a working life
÷ 2 billion euro a point scale of this evaluation
1.6
Normalised Impact
Score 1.6 Impact × 8 Value × 4 Plausibility ÷ 10 = 5.1 of 100

Plausibility

The claim is that children who go without coverage earn less as adults. The direct evidence compares children born just before and just after cut-off dates that gave younger cohorts more years of Medicaid eligibility, and follows them into adulthood in tax and hospital records [15]. Those with more eligibility paid more in federal tax by age 28 and used hospitals less [15]. What could distort the comparison is that neighbouring birth cohorts differ in other ways; the cut-offs are narrow enough that this is unlikely. It cannot run the other way, because eligibility was set by law and by birth date. The budget office takes half of what these studies imply and projects it over a lifetime [14]. The largest step is the transfer: the studies measured children gaining eligibility in the 1980s and 1990s, while this argument concerns children losing coverage for a few years now. Read back: in about five of ten comparable cases the earnings fall by roughly this much. The value of a working life is the German anchor of 1.5 million euro, set here because no American figure has been adopted, which takes off one more step. The Plausibility is just below the middle: the effect is measured, but its size here rests on a projection and on a borrowed anchor.

Evidence: Projection · plausibility at most 6 Evidence of effect: Quasi-experimental · at most 8

Less protection against medical bills

2.8points

Families who leave Medicaid or CHIP lose the protection a policy gives against a large bill, worth about 1 billion dollars a year at what recipients would pay for it. The share comes from a model built on the Oregon lottery and is the least certain number in this argument.

Value 5 · Household budgetsImpact 1.0Plausibility 5.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is the household budget, and in particular its protection against a bill it cannot pay. A family without coverage that meets an appendicitis, a premature birth or a broken arm faces a bill of thousands of dollars, which for a household near the poverty line means debt or unpaid rent. Money counts at the middle of the scale, and the low incomes of these households sit in the Impact as a weight of 2.3. The care itself and the health it buys are counted in other arguments. The value is the middle of the scale, because what is priced here is financial security.

Impact

The coverage these families give up costs about 6.86 billion dollars a year [1][2][5]. What it is worth to the people who lose it is less than what it costs. The one systematic estimate was built on the Oregon lottery with a model of household spending [12]. It puts the value of Medicaid to recipients at a fifth to two fifths of its cost, most of which is the care itself rather than the protection against the bill [12]. Fifteen percent is used here as the protection share, in a range from 5 to 30 percent, as in this site's evaluation of the Medicaid work requirement, giving about 1.03 billion dollars a year. These households sit in the bottom fifth of incomes, where a dollar counts at 2.3 on the American income bands. That gives a Normalised Impact of about 2.2, in a range from 0.59 to 5.49. The care and the health it buys are counted in the arguments on deaths, illness and children's earnings, not here. The Impact is moderate: the protection is a small share of the cost, but it falls on households with little to fall back on.

▸ Show calculation ▾ Hide calculation
6.86 billion dollars a year
Medicaid and CHIP coverage given up
× 15 % Setting, range 5 to 30 percent: recipients value Medicaid at a fifth to two fifths of its cost in the model on Oregon lottery data, most of it the care itself
1.03 billion dollars a year
Share that is protection, not the bill
÷ 1.16 dollars to the euro the rate used throughout this evaluation
0.89 billion euro-equivalent a year
In euro-equivalent
× 2.3 the American band of the bottom fifth of incomes, range 1.85 to 2.7
2.04 billion euro-equivalent a year
Weight of a dollar at these incomes
÷ 2 billion euro a point scale of this evaluation
1.02
Normalised Impact
Score 1.02 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 2.8 of 100

Plausibility

The claim is that leaving coverage exposes a household to bills it cannot pay. The Oregon lottery compared people drawn for Medicaid with people not drawn; those covered had far less financial strain and fewer catastrophic medical costs [11]. The money value of that protection comes from a model on those data, not from a price anyone paid [12]. What could distort it is the model's assumption about how households cope with a bill, which decides the figure; the study reports a range rather than a point. It cannot run the other way, because the lottery decided who was covered. Whether families withdraw at all is the doubt the two savings arguments carry too. Read back: in about five or six of ten comparable cases the loss is roughly this large. The Plausibility is just above the middle, because the protection is measured but its money value comes from a model.

Evidence: Study · plausibility at most 6 Evidence of effect: Mechanistic · at most 6

Deaths among adults without coverage

2points

About 650,000 adults a year go without coverage once they leave Medicaid, and at the rate measured when Medicaid expanded that means about 430 more deaths a year. The rate is the contested number; its range runs from a third of that to three times as much.

Value 10 · LifeImpact 0.3Plausibility 6
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is life itself. It is lost by adults in immigrant and mixed-status families who stop seeing a doctor once their coverage ends, and whose cancer or heart disease is then found late. A life lost counts at the top of the scale, against which every money amount in this debate is measured. The grief of those left behind is added as a tenth on top. The value is the highest on the scale, because the loss cannot be undone.

Impact

The department's central rate implies that about 10.3 percent of the 13.5 million Medicaid and CHIP enrollees who live with a noncitizen withdraw, about 1.39 million people [4][5]. Roughly 55 percent of them are adults, since 5.6 million of the 13.5 million are citizen children [4]. Most adults who leave have no other insurance to move to; 85 percent are taken to go uninsured, in a range from 70 to 95 percent, about 650,000 people a year. Annual mortality among low-income adults aged 55 to 64 fell 9.4 percent when Medicaid expanded [8]. Applied to the whole adult age range, as in this site's evaluation of the Medicaid work requirement, that is one death for every 1,500 years without coverage [8]. The range runs from one per 460 to one per 4,000 [10]. That gives about 430 deaths a year, and the grief of the bereaved adds a tenth on top. The Normalised Impact is about 0.67, in a range from 0.20 to 2.46, smaller than either saving but on the gravest item in the debate. Deaths among children are not counted, because their mortality is too low for a change to be measured. The Impact is moderate because few of those who withdraw die in any one year, but each case is a life.

▸ Show calculation ▾ Hide calculation
1.39 million people a year
Enrollees who withdraw
× 55 % Setting, range 50 to 60 percent: 5.6 of the 13.5 million are citizen children
0.77 million people a year
Adults among them
[4]
× 85 % Setting, range 70 to 95 percent
0.65 million people a year
Share left uninsured
÷ one per 1,500 years without coverage Setting, range one per 460 to one per 4,000: mortality fell 9.4 percent for low-income adults aged 55 to 64 who gained Medicaid, applied to all adults as in the Medicaid work requirement evaluation
433 deaths a year
Deaths that follow
× 1.4 million euro-equivalent each, plus a tenth the value of a statistical life used across this site, and the surcharge booked with every stream of deaths
0.67 billion euro-equivalent a year
Value of the lives, with the grief of the bereaved
÷ 2 billion euro a point scale of this evaluation
0.33
Normalised Impact
Score 0.33 Impact × 10 Value × 6 Plausibility ÷ 10 = 2 of 100

Plausibility

Two things have to hold: that families leave coverage, and that going without coverage costs lives. The first was measured after the 2018 announcement, against counties with fewer noncitizens [7]. The second rests on the Medicaid expansions, which compared death records in expanding and non-expanding states before and after 2014; mortality had moved in parallel beforehand [8]. The obvious objection, that expanding states differ in ways that also move mortality, is what that parallel period tests. A randomised trial, in which the tax authority wrote to uninsured households about coverage, points the same way [9]. It cannot run the other way, because expansion was a state legislative act. Immigrants tend to be healthier than people born in the country, so the measured rate may overstate the loss, and the range carries that. Read back: in about six of ten comparable cases the deaths come to roughly this many. The Plausibility is above the middle: that coverage saves lives is established under two designs, and only how many is open.

Evidence: Converging studies · plausibility at most 6.5 Evidence of effect: Quasi-experimental · at most 8

Illness that goes untreated

1.5points

About 1.2 million people a year, adults and children, go without coverage they would otherwise have kept. At the effect measured in the Oregon lottery, mostly on depression, that costs about 14,000 healthy life years a year.

Value 9 · HealthImpact 0.3Plausibility 6
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The good is health: the depression that is treated, the prescription that is filled, the check-up that finds a problem early. It is lost by the adults and children in these families who go without a doctor. The Oregon lottery found the clearest gain from coverage in mental health, and that is what is counted here [11]. Health sits one step below life on the scale, because much of what is lost can be recovered once care resumes. The value is close to the top, because the harm is real but often reversible.

Impact

About 1.39 million people a year leave Medicaid or CHIP at the department's central rate, and 85 percent of them, about 1.18 million, are taken to go uninsured [4][5]. Oregon drew names by lot for its Medicaid expansion in 2008 [11]. Two years on, the people drawn had 30 percent less depression, more preventive care and far less financial strain, with no measurable change in blood pressure, cholesterol or blood sugar [11]. The depression finding alone is worth about 0.014 healthy life years for each year of coverage. As in this site's evaluation of the Medicaid work requirement, 0.012 is used, in a range from 0.005 to 0.020, and the same rate is applied to children for lack of a trial on them. That gives about 14,200 healthy life years a year. Only what the trial found is booked; the physical markers it did not find are left out. The Normalised Impact is about 0.57, in a range from 0.18 to 1.19, close to the deaths in weight. The Impact is modest per person and sizeable in total, because a year of coverage buys a little measurable health across many people.

▸ Show calculation ▾ Hide calculation
1.39 million people a year
Enrollees who withdraw
× 85 % Setting, range 70 to 95 percent
1.18 million people a year
Share left uninsured
× 0.012 per year without coverage Setting, range 0.005 to 0.020: the Oregon lottery measured 30 percent less depression and no change in physical markers
14,183 healthy life years a year
Healthy life years lost
× 40,000 euro-equivalent each the value of a healthy life year used across this site
0.57 billion euro-equivalent a year
Value of the healthy life years
÷ 2 billion euro a point scale of this evaluation
0.28
Normalised Impact
Score 0.28 Impact × 9 Value × 6 Plausibility ÷ 10 = 1.5 of 100

Plausibility

The comparison is between the Oregon applicants whose names were drawn and those whose names were not. Assignment by lot means the groups differ only in whether their name came up, so neither selection into coverage nor reverse causation can arise [11]. Two limits apply. The trial ran in one state on adults, while here the same rate is applied to families that include children and recent immigrants. It also measured people gaining coverage, while this argument concerns people losing it. Whether families withdraw at all is the doubt the savings carry too, measured after the 2018 announcement [7]. Read back: in about six of ten comparable cases the loss of health is roughly this large. The Plausibility is above the middle: the effect is randomised, and what is stretched is the population it is applied to.

Evidence: Study · plausibility at most 6.5 Evidence of effect: Experimental · at most 8.5

How robust the result is

Under the assumptions used, the proposal brings more than it costs in 2 of 100 model runs.

gains more than it costscosts more than it gains

Each dot stands for 200 of the 20,000 model runs.

This is mainly a distributional debate: aid that public budgets keep and poor families lose carries more than half of the circle. How a dollar is weighted carries this result. Aid lost by families at or near the poverty line counts 2.3 times a dollar in a federal or state budget, the American weight for the bottom fifth of incomes, drawn between 1.85 and 2.7. So weighted, the rule comes out ahead in none of the runs. If every dollar counted the same, the food and cash aid would only change hands, leaving the Medicaid saving against the deaths, illness and lost earnings; the rule would then come out ahead in about four of ten runs and behind on average. Coverage lost within the five years counted here is charged with the earnings it costs children over their whole working lives, most of which fall after 2031. Those earnings are valued with the German anchor for a working life, 1.5 million euro, as a setting of this evaluation until an American figure is adopted; the argument carries one step less plausibility for it. The verdict would turn only if a dollar in these households counted less than about 0.86 of a budget dollar.

The verdict hardly depends on how incomes are weighted: if a euro at a low income counted somewhat less or somewhat more than assumed, it would stay on the same side, between “Likely worse” and “Very likely worse”. How incomes are weighted

Show the distribution of the model runs

20,000 simulated futures

Very likely worse · 2 %

In 2 of 100 simulated futures the proposal brings more than it costs.

No effect is known exactly; every argument has a range. The calculation draws a possible size for each argument from its range 20,000 times and counts how often the benefits exceed the costs.

Net effect in points per future. Left of zero the drawbacks outweigh the benefits, right of zero the benefits win. Eight in ten futures fall between −29 and −1.8 points.
Scale and assessment

Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 2 billion euro per year.

Size class:large

Effect large (−26) · 27 % of the effect is positive

Areas show individual argument weights. The positive share of the effect nets out money transfers between the two sides; the probability of a positive net effect is determined separately.

How we score →

Review and sources

AI evaluation · not yet approved by a human

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

Last reviewed by Claude Opus 5.5 · November 13, 2026 · 6 reviews since 11/2026 · 5 corrections · not yet approved by a human
  1. November 13, 2026AI reviewClaude Opus 5.5revised

    wertschoepfung-pflichtpruefung/1.0: note added to the derivation (effect booked in another argument); no number changed.

  2. November 10, 2026AI reviewClaude Opus 5.5re-scored

    Wave 2 of the rebooking (decisions of 31.10. and 06.11.2026): bottom-fifth weight 2.5 to the American 2.3 with its own band on the paying leg con-4, the net cards pro-1 (care not consumed) and con-3 (protection) booked as real and value-creating streams, German lifetime-earnings anchor declared as a setting with P 5 to 4 on con-5; P(D > 0) 0.000 to 0.019, still very likely worse.

  3. November 10, 2026AI reviewClaude Opus 5.5revised

    Kernfrage von Julian Ohm im Chat freigegeben; Zahlen nicht geprüft.

  4. November 9, 2026AI reviewClaude Opus 5.5revised

    Key question and its answer added to the page texts (draft); numbers unchanged.

  5. November 1, 2026AI reviewClaude Opus 5.5revised

    Page texts (layout G) added before first publication.

  6. November 1, 2026AI reviewClaude Opus 5.5First evaluation

    KI-Entwurf aus der Themen-Automatisierung übernommen; Anzeige-Maßstab gesetzt, Rechenweg und P-Deckel an validate angepasst.

Every evaluation is reviewed again and again: an AI model looks for errors, gaps and new developments and corrects them. Rounds approved by a member of the editorial team carry their name. Each round is listed here with its date, and earlier versions stay readable.How we review →

Show all 18 sourcesHide sources
  1. Federal Register: Public Charge Ground of Inadmissibility, final rule, 91 FR 45324 (20 July 2026). federalregister.gov
  2. Office of the Minnesota Attorney General: Attorney General Ellison sues to block illegal Trump attempt to punish immigrants (14 September 2026). ag.state.mn.us
  3. Boundless: DHS Rescinds 2022 Public Charge Rule: What It Means for Green Card Applicants (2026). boundless.com
  4. KFF: New Public Charge Policies Will Likely Lead to Decreased Enrollment in Medicaid and CHIP (20 July 2026). kff.org
  5. Georgetown University Center for Children and Families: Comments on the 2025 public charge proposed rule (December 2025). ccf.georgetown.edu
  6. Northwest Iowa Community Action (comment letter): Comment on the 2025 proposed public charge rule (1 December 2025). media.nwica.org
  7. Barofsky, Vargas, Rodriguez and Barrows, Health Affairs 39(10): Spreading Fear: The Announcement Of The Public Charge Rule Reduced Enrollment In Child Safety-Net Programs (2020). ideas42.org
  8. Miller, Johnson and Wherry, Quarterly Journal of Economics 136(3): Medicaid and Mortality: New Evidence From Linked Survey and Administrative Data (2021). academic.oup.com
  9. Goldin, Lurie and McCubbin, Quarterly Journal of Economics 136(1): Health Insurance and Mortality: Experimental Evidence from Taxpayer Outreach (2021). academic.oup.com
  10. Gaffney, Himmelstein and Woolhandler, Annals of Internal Medicine: Projected Effects of Proposed Cuts in Federal Medicaid Expenditures on Medicaid Enrollment, Uninsurance, Health Care, and Health (2025). acpjournals.org
  11. Baicker and others, New England Journal of Medicine 368: The Oregon Experiment: Effects of Medicaid on Clinical Outcomes (2013). nejm.org
  12. Finkelstein, Hendren and Luttmer, Journal of Political Economy 127(6): The Value of Medicaid: Interpreting Results from the Oregon Health Insurance Experiment (2019). journals.uchicago.edu
  13. Garthwaite, Gross and Notowidigdo, American Economic Journal: Applied Economics 10(1): Hospitals as Insurers of Last Resort (2018). aeaweb.org
  14. Congressional Budget Office: Exploring the Effects of Medicaid During Childhood on the Economy and the Budget, Working Paper 2023-07 (November 2023). cbo.gov
  15. National Bureau of Economic Research: Childhood Medicaid Coverage Improves Adult Earning and Health, NBER Digest (May 2015). nber.org
  16. Pew Research Center: What the data says about food stamps in the U.S. (14 November 2025). pewresearch.org
  17. Social Security Administration: SSI Monthly Statistics, August 2026, Table 2. ssa.gov
  18. Commonwealth Fund: The Impact of Proposed Federal Medicaid Work Requirements on Hospital Revenues and Financial Margins (September 2025). commonwealthfund.org

More debates in this area

All debates in Immigration and Border Enforcement →

Ask the bot

The bot knows the evaluation of this debate and explains it and the method. It answers only from Argupedia's content and does no research of its own.

Before you ask

  • The answers are written by an AI model (Claude Haiku by Anthropic). They may be wrong or incomplete. The content of the pages is authoritative.
  • The bot gives no legal, financial, medical or voting advice.
  • What you type is sent to Anthropic in the USA. Do not enter names, contact details or other personal information.

The terms of use (section 9a) and the privacy notice for the chat apply.