💬 Discussion
Greek discussion partners compare a tax ledger, a scholarship fund, a school model and a student's books.

School Scholarship Credit

The 2025 budget law created the first federal tax credit for school scholarships, and it starts in January 2027 in every state whose governor opts in. Republicans in Congress wrote it as school choice without new state spending; Democrats in both chambers have filed bills to repeal it before it begins. When the Treasury published its rules in October 2026, 30 states had opted in [2].

  • 4 billion dollars a year is the expected cost of the credit to the Treasury once it has ramped up [5][30]
  • 3 in 4 users of Arizona's state program did not come from a public school [11]
  • 49.6 million children attend public schools in the United States [19]

The proposal

Let every taxpayer send up to 1,700 dollars of federal income tax to scholarship organizations for school children instead of the Treasury, from January 2027.

The debate

Supporters in Congress say the credit brings private money to scholarships, gives families more options and takes nothing from public school budgets. Opponents answer that it moves public money to private schools that choose their students and answer to no one, and that most of it reaches families already paying private tuition. The credit passed as part of the budget law signed on 4 July 2025, without Democratic votes. Repeal bills followed in the Senate in April 2026 and in the House on 11 June 2026; neither has come to a vote.

  • “This innovative credit will unlock billions of private dollars to fund scholarships for K-12 students across the country, including in your state.”

    argues for: Tim Walberg, Republican, Michigan Letter to the governors, 17 October 2025

  • “Research consistently shows that the majority of families participating in voucher programs already send their children to private schools.”

    argues against: Suzanne Bonamici, Democrat, Oregon Press release, 11 June 2026

The analysis

School Scholarship Credit

53% of the weight is for, 47% 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

Whether the school scholarship credit is a good measure, the analysis leaves open.

Unclear

Unclear: Under the assumptions used, the proposal brings more than it costs in 48 of 100 model runs.

 Overall effect: medium

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 analysis leaves the question open, and the size that would settle it plays almost no part in Congress. The two largest items are money changing hands: tuition help for families already in private schools, paid with income tax the Treasury does not collect. The forgone revenue weighs a little more, because a dollar does more good in the federal budget than with these better-off families. For the credit stand savings in state budgets and the worth of a chosen school; against it, lower math scores of switchers and the cost of running scholarship groups. What decides is how much public schools improve when families can leave: at the top of the Florida range, the credit would probably come out ahead.

In detail

All arguments

Arguments for6 arguments

Tuition help for families

19points

From 2027 a taxpayer can send up to 1,700 dollars of federal income tax to a scholarship organization instead of the Treasury, and the organization must spend at least 90 percent on scholarships. Most of that money is expected to reach families whose children already attend a private school, and for them it lowers what they pay out of pocket. At about 4 billion dollars a year in credits, that is about 2.7 billion dollars a year, of which part ends up with the schools through higher tuition.

Value 5 · Household budgetsImpact 3.9Plausibility 9.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money that stays in household budgets because a scholarship pays part of tuition or home-schooling costs the family would have paid anyway [4]. It does not buy a different school: these children were already outside public schools, so what changes is who pays. A dollar in a household is counted at the standard value of money; how much it matters to a family of a given income is carried by the weight in the amount, not here. The other end of the same money, the income tax the Treasury does not collect, stands against this measure in the argument on federal revenue, and only the difference between the two weights moves the balance. The value is the middle of the scale, because the stream is money for households.

Impact

After a few years of ramp-up, the Urban Institute expects the Treasury to forgo 2.7 to 6.1 billion dollars a year [5]; figures of the Joint Committee on Taxation put the enacted credit at about 3 to 4 billion dollars a year [30], and about 26 billion dollars over ten years were reported during the Senate debate [7]. The central value here is 4 billion dollars a year over 2027 to 2036. At least 90 percent must go to scholarships [4], which gives 3.6 billion dollars. In comparable state programs most users were already outside public schools: about three quarters in Arizona [11], about two thirds in Indiana [10] and almost all in the first year in Arkansas [10]. Three quarters gives 2.7 billion dollars. Not all of it stays with families: when Iowa opened its program to every kindergarten child, private schools raised tuition by 21 to 25 percent, and by 10 to 16 percent in grades where only some children were eligible [28], and earlier programs open to all show the same pattern [29]. A fifth is therefore counted as reaching the schools at the standard weight of 1, and four fifths as reaching families at a weight of 0.8, because private school families are better off than average [12] and about 96 percent of students are eligible [33]. The combined weight is 0.84, which gives a Normalised Impact of 3.91 on the scale of this evaluation, in a range from 1.63 to 7.16. The Impact is large because the whole tuition sum moves, and smaller than the sum because most of it reaches families above the middle of incomes.

▸ Show calculation ▾ Hide calculation
4 billion dollars a year
Credits claimed each year, 2027 to 2036
[5][7][30]
× × 0.9 at least 90 percent of an organization's income
3.6 billion dollars a year
Spent on scholarships
[4]
× × 0.75 Setting, range 0.5 to 0.9, from Arizona, Indiana and Arkansas
2.7 billion dollars a year
For children already outside public schools
[10][11]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 0.97 to 4.26 billion euro-equivalent (about 1.1 to 4.9 billion dollars)
2.33 billion euro-equivalent a year
In euro-equivalent
× × (0.8 × 0.8 + 0.2 × 1.0) = × 0.84 Setting: four fifths stay with families at a weight of 0.8, one fifth passes to schools through higher tuition at a weight of 1 (Iowa: tuition up 10 to 25 percent); range 0.81 to 3.58 billion euro-equivalent at this weight
1.96 billion euro-equivalent a year
Weight of a dollar where it ends up
[12][28][29][33]
÷ 500 million euro-equivalent a point scale of this evaluation
3.91
Normalised Impact
Score 3.91 Impact × 5 Value × 9.5 Plausibility ÷ 10 = 19 of 100

Plausibility

The counterfactual is no credit, which is what the repeal bill S. 4297 would restore [8][21]. That the money reaches families and schools follows from the law: scholarships are paid directly to schools or reimbursed against receipts, and organizations must spend 90 percent of their income on them [4]. No behavioural link has to hold for the payment to occur, so it counts as close to certain. How large the sum is depends on how many states take part, 30 as of October 2026 [2], and on how many taxpayers donate; that doubt sits in the range of the Impact, not here. The confounder that matters is that donors would have given to the same schools anyway, which would move part of the money to donors and not to families; that is a question of who receives, not of whether the money flows. Reverse causation does not arise, because the amount is set by donors and the law. This argument carries the same Plausibility as the paying side, because both are the same sum seen from two ends. The Plausibility is very high: the payment is a legal consequence of the credit, and only its size is estimated.

Evidence: Mechanism · plausibility at most 9.5 Evidence of effect: Definitional

Public schools that compete

19points

When families can take a scholarship elsewhere, public schools have a reason to hold on to them. In Florida, as the state's tax-credit scholarship grew to about four in a hundred students, test scores rose more in public schools with many private schools nearby than in schools facing less competition. The federal credit is expected to move about 180,000 children, fewer than one in a hundred of the 25 million public school students in the participating states.

Value 8 · Educational opportunityImpact 4.7Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is what children in public schools learn, measured in test scores that carry into later earnings and choices. That is an educational opportunity, valued one step below health and life on this site. Every student counts the same here; that the Florida gains were largest for lower-income students is reported, not weighted again [13]. The effect is counted after any money public schools lost to the program, because the Florida estimate already includes it. Costs that districts cannot cut when a few children leave are part of that net figure and are not counted again [32]. The value is high, because it is lasting learning for children in public schools.

Impact

Public schools enroll 49.6 million students [19]; the 30 participating states [2] are assumed to hold half of them, so 24.8 million, and each year one cohort of about 1.9 million leaves school. In Florida, the program grew from about 15,000 to more than 100,000 students, about 4 percent of enrollment, and after 15 years scores in schools facing more competition stood about 0.15 standard deviations above those in schools facing less [13][25]. That figure is a difference between two groups of schools, not a gain for the average student; if schools with little competition gained nothing, the average is at most half of it. The federal credit is expected to move about 180,000 children [9][10][11], 0.7 percent of students in the participating states against Florida's 4 percent, so less than a fifth of the pressure. Scaled this way, the gain for an average student lies between 0.3 and 1.5 percent of a standard deviation; it is set at 0.5 percent, near the lower end, because the wider research finds smaller effects than Florida [22][23], and the range runs from 0.1 to 1.5 percent. Two reductions follow. The program takes years to grow and the first cohorts leave school having seen little of it, so 0.7 of the full gain is counted over 2027 to 2036; and most of these students live in states that already run large programs of their own, where the same number of added scholarships adds less pressure, counted at 0.75. A lasting standard deviation is valued at about a quarter of a lifetime income, set at about 2.3 million dollars for the United States. That gives about 5,000 standard deviations a year, a Normalised Impact of 4.71 on the scale of this evaluation, in a range from 0.94 to 14.12. The Impact is the largest on this side because a very small gain is spread over millions of children.

▸ Show calculation ▾ Hide calculation
24.8 million students
Public school students in participating states
[2][19]
÷ ÷ 13 grades
1.91 million students a year
One school-leaving cohort a year
× × 0.005 standard deviation Florida gap of about 0.15 standard deviations, at most half as an average, scaled from 4 percent of students to the 0.7 percent expected to switch here: 0.003 to 0.015; Setting 0.005, range 0.001 to 0.015
9,538 standard deviations a year
Gain in achievement once the program is fully grown
[13][22][23][25]
× × 0.7 Setting, range 0.6 to 0.8: the program grows over several years and early cohorts leave school before it is felt
6,677 standard deviations a year
Share reached over 2027 to 2036
× × 0.75 Setting, range 0.5 to 1: where families already have a state scholarship, the same added number adds less pressure
5,008 standard deviations a year
Allowance for states with existing programs
× × 470,000 euro-equivalent value of a lasting standard deviation, 23.5 percent of a United States lifetime income set at 2.0 million euro-equivalent (about 2.3 million dollars); range 0.47 to 7.06 billion euro-equivalent, from the range of the gain alone
2.35 billion euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
4.71
Normalised Impact
Score 4.71 Impact × 8 Value × 5 Plausibility ÷ 10 = 19 of 100

Plausibility

The counterfactual is Florida public schools that faced fewer private schools nearby before the program existed. The carrying source is Figlio, Hart and Karbownik (2021), a quasi-experimental difference-in-differences design that compares schools by their competition measured before the program, with student fixed effects and records linked to birth data [13]. The confounder that matters is that areas with many private schools were on different trends anyway; measuring competition before the program and following the same students over time addresses that. Direction: the competition measure predates the program, so the outcomes cannot have caused it. What the design identifies is the gap between schools, and reading it as a gain for all students assumes that schools with little competition did not lose, which the study does not test. Other programs point the same way with smaller effects: neutral to small gains in Louisiana [23], modest gains in Ohio [24], and a review of the research that finds small positive effects whose school-level estimate cannot be told apart from zero [22]. The transfer to the federal credit costs two steps, because Florida's program was means-tested and required most children to come from public schools. Read back: in about half of comparable cases, a program of this size raises public school results by this much. The Plausibility is middling: the direction is supported by several programs, the size by one study of a differently built program.

Evidence: Study · plausibility at most 5 Evidence of effect: Quasi-experimental · at most 8

States spend less on leavers

5.3points

Every child who moves from a public to a private school with a scholarship is a child the state and district no longer have to teach. Public schools spend about 15,500 dollars per student, but when a few children leave each school, only part of that can be cut. With about 180,000 children switching, state and local budgets save about 1.1 billion dollars a year.

Value 5 · Public financesImpact 1.9Plausibility 5.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money that state and local budgets no longer spend on teaching children who have left. It is public money, counted at the standard weight. Whether states use it to lower taxes or to spend more on the students who remain is open; either way it is money, and any effect on remaining students is part of the competition argument. The savings fall to state and local budgets, which in most states share the cost of each student through aid formulas [9]. They are counted once, as money, and not again as better schooling for anyone. The value is the middle of the scale, because the stream is public money.

Impact

A quarter of 3.6 billion dollars in scholarships, 0.9 billion, is assumed to pay for children who switch [10][11]. At the average tax-credit scholarship of 4,997 dollars [9], that is 180,000 children, 0.7 percent of the students in the participating states. Public schools spend 15,542 dollars per student [9]. Not all of it falls when a child leaves: buildings, administration and debt stay, while instruction and student support follow students [9]. For North Carolina, Ladd and Singleton put the cost a district cannot shed at 3,500 of 8,000 dollars for each student who leaves for a charter school, which leaves a little over half as saved [32]. Because the switchers here are spread thinly, a few children per school, four tenths are counted as falling, with a range of 0.2 to 0.6. That saves 1.12 billion dollars a year, with a range of 0.33 to 2.8 billion dollars. The Impact is sizeable because each child who leaves frees more than the scholarship that moved them.

▸ Show calculation ▾ Hide calculation
180,000 children
Children who switch, at any time
[9][10][11]
× × 15,542 dollars
2.8 billion dollars a year
Public spending per student
[9]
× × 0.4 Setting, range 0.2 to 0.6: instruction and support follow students, buildings and administration do not; lower than the North Carolina estimate because few children leave each school
1.12 billion dollars a year
Share that falls when a child leaves
[9][32]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 0.28 to 2.4 billion euro-equivalent (about 0.33 to 2.8 billion dollars)
0.96 billion euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
1.93
Normalised Impact
Score 1.93 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 5.3 of 100

Plausibility

The counterfactual is the same children in public schools. The chain is short: a scholarship moves a child, enrollment falls, and state aid and district costs fall with it. Whether costs actually fall depends on how districts respond when they lose a few students per grade. A fiscal analysis by a school-choice organization counts savings of 3,300 to 7,800 dollars per participant across existing programs [9], while the North Carolina estimate shows that close to half of the spending stays behind in the district [32]. The confounder is that children who leave are spread thinly over many classes, so costs stay fixed for years; that is why the share counted as falling is four tenths and can go down to two. Reverse causation does not arise. The number of children who switch is taken from other states' programs and sits in the range. The Plausibility is moderate: the savings follow from fewer students, but how fast districts adjust their costs is not measured for this program.

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

Choosing the school

2points

A scholarship lets a family pick a school it could not afford before. Families who switch still pay part of the tuition themselves, which shows that the new school is worth more to them than the free place they gave up. In Washington, D.C., families who won a scholarship saw their schools as safer.

Value 5 · Choice of schoolImpact 0.8Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is the ability of a family to decide where its child goes to school. That is more than convenience, because a school shapes a child's days for years, and less than a basic right, because a free public school place remains open to every child. It is therefore placed one step above everyday choices on this site and at the level of money. The families concerned are those who switch, not those whose children were in private school already; for the latter the scholarship is money, counted in the argument on tuition help. What children learn in the new school is counted in the arguments on test scores and college, not here. The value is in the middle of the scale: a real gain in self-determination, with the public school still available.

Impact

About 180,000 children are expected to switch with a scholarship [9][10][11]. A place in a private elementary or middle school cost 9,210 dollars on average in 2020-21, and 9,720 dollars in Catholic schools of all levels [31]; at about 9,000 dollars, a scholarship of about 5,000 dollars [9] leaves 4,000 dollars uncovered. Families are assumed to pay 2,500 dollars of that themselves and schools or parishes the rest, which is a setting with a range of 1,000 to 4,000 dollars. A family that pays this sum for a school when a free one is available values the choice at least that highly. Together these families pay 450 million dollars a year, and the choice is counted at that amount, with a range of 110 million to 1.1 billion dollars once the uncertain number of switchers is included. That is about half of what the federal budget spends on the same children's scholarships. The Impact is small because few children switch, and it is counted only at what families show they are willing to add.

▸ Show calculation ▾ Hide calculation
180,000 children
Children who switch, at any time
[9][10][11]
× × 2,500 dollars Setting, range 1,000 to 4,000 dollars: tuition of about 9,000 dollars less a scholarship of about 5,000 dollars leaves 4,000 dollars, shared between families and schools
450 million dollars a year
What families add to the scholarship
[9][31]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 98 to 990 million euro-equivalent (about 110 million to 1.1 billion dollars)
387.9 million euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
0.78
Normalised Impact
Score 0.78 Impact × 5 Value × 5 Plausibility ÷ 10 = 2 of 100

Plausibility

The counterfactual is the same families with only the public school place. The chain is short: a scholarship lowers the price, a family that then switches and pays a share has shown that it prefers the new school. No study measures this value for the federal credit. The two lottery evaluations in Washington, D.C., point the same way without agreeing on details: in the first, parents were more satisfied and saw schools as safer [35]; in the second, students were more satisfied and schools were seen as safer, while parents were not more satisfied [27]. The confounder is that satisfaction with a school one has chosen partly reflects the choosing itself. Direction: the lottery assigns the scholarship before satisfaction is asked. The size rests on what families pay on top, which is assumed and sits in the range. The Plausibility is middling: the preference follows from the families' own decision, and its size is not measured.

Evidence: Plausibility · plausibility at most 5 Evidence of effect: Mechanistic · at most 6

More switchers go to college

0.74points

About a quarter of the scholarships is expected to move children from public into private schools, roughly 180,000 children at any time. In Florida, low-income students who used the state's tax-credit scholarship were more likely to enroll in college and slightly more likely to earn a bachelor's degree than similar students who did not. A bachelor's degree raises lifetime earnings by about three quarters.

Value 8 · Educational opportunityImpact 0.3Plausibility 3
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is a degree that changes what a young person can earn and do over a working life. That is an educational opportunity in the sense of a threshold to an independent living, which this site values one step below health and life. The children concerned come mostly from households below the income cap, and many from low-income families [2]; their situation is carried by this value, not by an extra weight on the money. The earnings are counted as what the degree adds, not as a payment from anyone, so no other argument carries their other end. The value is high, because a degree shapes a whole working life.

Impact

Of 3.6 billion dollars in scholarships a year, a quarter, 0.9 billion, is assumed to go to children who switch from public schools [10][11]. At the average tax-credit scholarship of about 5,000 dollars [9], that is 180,000 children at any time, or about 13,800 in each year's school-leaving cohort. Florida's scholarship users were 1 to 2 percentage points more likely to earn a bachelor's degree [14]; at 1.5 points, in a range of 0.5 to 2, that is about 210 additional degrees a year. A bachelor's degree raises lifetime earnings by 75 percent over a high school diploma, 2.8 against 1.6 million dollars [20]. Half of that gap is counted as caused by the degree and not by who chooses to study, which is a setting. Lifetime income is set at about 2.3 million dollars, the value this site uses for the United States until a measured one is available. That gives 78 lifetime incomes' worth of gain, about 180 million dollars a year, with a range of 37 to 490 million dollars. The Impact is small because few of the scholarships move children, and only a small share of those earn a degree they would not otherwise have earned.

▸ Show calculation ▾ Hide calculation
180,000 children
Children who switch, at any time
[9][10][11]
÷ ÷ 13 grades
13,846 children
In each year's school-leaving cohort
× × 1.5 points Florida, range 0.5 to 2 points
208 degrees a year
Additional bachelor's degrees
[14]
× × 0.75 × 0.5 a degree adds 75 percent; half counted as caused by it (Setting, range 0.3 to 0.8)
77.9 lifetime incomes a year
Lifetime earnings gained per degree
[20]
× × 2.0 million euro-equivalent (about 2.3 million dollars) lifetime income set for the United States; range 32 to 420 million euro-equivalent (about 37 to 490 million dollars)
155.8 million euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
0.31
Normalised Impact
Score 0.31 Impact × 8 Value × 3 Plausibility ÷ 10 = 0.7 of 100

Plausibility

The counterfactual is the same low-income students without a scholarship. The carrying source compares Florida scholarship users with students of similar characteristics and prior test scores who never used the scholarship (Chingos and co-authors 2019) [14]; the design is controlled, without random assignment. The confounder is that families who apply may be more ambitious for their children, which matching on prior scores reduces but does not remove. Direction: the scholarship comes before college entry, but selection into the scholarship can run the other way. Against it stands Louisiana's lottery-based evaluation, where the effect on college entrance was positive but not distinguishable from zero [15]. Supported is only part of the claim: more enrollment in Florida, a small and uncertain gain in degrees, and no confirmation from the one lottery. Read back: in about three of ten comparable programs, switchers earn this many more degrees. The Plausibility is low: one matched study supports the gain, and the stronger lottery evidence does not confirm it.

Evidence: Study · plausibility at most 5 Evidence of effect: Controlled · at most 7 Placement: Partial aspect supported · plausibility 2.5–3

Open: A lottery-based follow-up of a large program to degree completion showing a gain of one point or more would lift this to about 5.

Tutoring in public schools

0points

Scholarships are not limited to private school tuition. The rules also allow tutoring, after-school programs and learning materials for children who stay in public schools, which is why some Democratic governors have opted in. How much of the money will go this way is unknown. The argument is included but not yet scored.

Value 8 · Educational opportunityImpact 0Plausibility 3
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream would be additional learning for children in public schools, paid for by a scholarship. That is an educational opportunity, valued one step below health and life on this site, the same value as in the other arguments on learning. The children concerned would be mostly from families that cannot pay for tutoring themselves. Low-income children already in public benefit programs or school-based tutoring are eligible without a further test [3]. The money itself is counted in the arguments on federal revenue; here only what it would achieve is at issue. The value would be high, because tutoring changes what a child learns.

Impact

The rules list tutoring, after-school programs and other supplementary services for public school students among the expenses a scholarship may cover [3]. Scholarship organizations decide which students they serve and what they pay for, and states may not steer them [33]. Vermont has enacted rules that would restrict scholarships to public school and low-income students, in conflict with the federal rules [3]. No estimate exists of the share of scholarship money that will go to public school students. The state programs used here to estimate who receives scholarships pay almost only for private school tuition, so they say nothing about this use. The Impact is therefore not set. The Impact is not scored: the share of scholarships that will pay for tutoring is not known.

Plausibility

The chain would be: organizations offer tutoring scholarships, public school students take them up, and their results improve. The counterfactual is the same students without paid tutoring. The design is mechanistic; that the rules allow this use is documented [3][33], but whether services bought with such scholarships raise achievement is not shown. The confounder is that families who apply for tutoring scholarships are more engaged than others. Direction: the scholarship comes before the outcome. With the link that carries the quantity untested, the Plausibility stands at three in ten. The Plausibility is low as long as nobody knows how much of the money reaches public school students.

Recorded, not yet scored: Included but not scored: there is no estimate of how much scholarship money will pay for tutoring and after-school programs of children who stay in public schools. The first annual reports of the scholarship organizations for 2027 will show the share, and the argument is to be scored then.

Evidence: Plausibility · plausibility at most 5 Evidence of effect: Mechanistic · at most 6 Placement: Chain open · plausibility 3–3.5

Open: First-year reports of scholarship organizations for 2027 showing the share of scholarship money paid for public school students would allow a quantity and could lift P to about 5.

Arguments against7 arguments

Revenue the Treasury forgoes

22points

Every dollar credited is a dollar of income tax the Treasury does not collect, and the federal budget runs a deficit, so the gap is borrowed. For the part of the scholarships that pays tuition families would have paid anyway, about 2.7 billion dollars a year, the money moves from all taxpayers to those families. Senate Democrats want to repeal the credit for this reason.

Value 5 · Public financesImpact 4.7Plausibility 9.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money that leaves the federal budget as a tax credit. It is public money, counted at the standard weight, and it stands for the taxpayers who will service and repay the borrowing that replaces it. The receiving end of the same money, families who pay less tuition, is counted for this measure in the argument on tuition help. Only the difference in what a dollar is worth to the two sides moves the balance. The value is the middle of the scale, because the stream is public money.

Impact

The credit is expected to cost about 4 billion dollars a year: the Urban Institute estimates 2.7 to 6.1 billion after ramp-up [5], and figures of the Joint Committee on Taxation put the enacted credit at about 3 to 4 billion [30]. This argument counts the part that pays for children already outside public schools: 90 percent goes to scholarships [4], and three quarters of those are assumed to go to such children [10][11]. That is 2.7 billion dollars a year. A public dollar carries the standard weight of 1, and the range runs from 1.1 to 4.9 billion dollars. The scholarships for children who switch and the share kept for administration are counted in their own arguments. The Impact is large because it is the whole sum that moves to families who would have paid anyway.

▸ Show calculation ▾ Hide calculation
4 billion dollars a year
Credits claimed each year, 2027 to 2036
[5][30]
× × 0.9 at least 90 percent
3.6 billion dollars a year
Spent on scholarships
[4]
× × 0.75 same share as the receiving side; range 0.5 to 0.9
2.7 billion dollars a year
For children already outside public schools
[10][11]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 0.97 to 4.26 billion euro-equivalent (about 1.1 to 4.9 billion dollars)
2.33 billion euro-equivalent a year
In euro-equivalent
× × 1.0 standard weight of public money
2.33 billion euro-equivalent a year
Weight of a public dollar
÷ 500 million euro-equivalent a point scale of this evaluation
4.66
Normalised Impact
Score 4.66 Impact × 5 Value × 9.5 Plausibility ÷ 10 = 22 of 100

Plausibility

The counterfactual is no credit, the state the repeal bill S. 4297 would restore [8][21]. That revenue is lost follows from the law: the credit is dollar for dollar against income tax, up to 1,700 dollars per taxpayer, with unused amounts carried forward [2][4]. No behavioural link has to hold beyond the donation itself, so the loss counts as close to certain. How large it is depends on how many states take part and how many taxpayers give, and that doubt sits in the range of the Impact. The confounder is that some donors would have given to scholarship funds anyway and now get their gift refunded, which does not change the revenue lost. Reverse causation does not arise. This argument carries the same Plausibility as the receiving side, because both are the same payment. The Plausibility is very high: the revenue loss is a legal consequence of the credit, and only its size is estimated.

Evidence: Mechanism · plausibility at most 9.5 Evidence of effect: Definitional

Federal money for switchers

7.4points

The quarter of scholarships that moves children from public to private schools is also paid for with income tax the Treasury does not collect, about 0.9 billion dollars a year. What it buys is counted in other arguments: private school places, effects on these children's learning, and savings in public schools. Here stands what it costs.

Value 5 · Public financesImpact 1.6Plausibility 9.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is federal money spent on private school places for children who leave public schools. It is counted at the standard value of public money. What the money achieves is counted where it happens: in the learning of the children who switch and in what states no longer spend on them. The money comes from the same forgone income tax as in the argument on federal revenue, but it buys something different: a change of school rather than a lower bill for a family. Counting it as a cost and its results as their own arguments keeps the same dollar from being counted twice. The value is the middle of the scale, because the stream is public money.

Impact

Of 3.6 billion dollars in scholarships a year [4][5], a quarter is assumed to go to children who switch from public schools. In Arizona, the share of new users coming straight from public schools rose from 21 to 57 percent as the program matured [11], while in Indiana about a third of users came from public schools [10]. That gives 0.9 billion dollars, with a range of 0.22 to 2.7 billion dollars. Each switcher is assumed to receive the average tax-credit scholarship of about 5,000 dollars [9], so this sum corresponds to about 180,000 children at any time. The same share and the same children are used in the arguments on savings in public schools, college, test scores and the cost of the private school place. The Impact is moderate because only a quarter of the scholarships is expected to move children.

▸ Show calculation ▾ Hide calculation
3.6 billion dollars a year
Scholarships each year
[4][5]
× × 0.25 Setting, range 0.1 to 0.5, from Arizona and Indiana
0.9 billion dollars a year
For children who switch from public schools
[10][11]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 0.19 to 2.37 billion euro-equivalent (about 0.22 to 2.7 billion dollars)
0.78 billion euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
1.55
Normalised Impact
Score 1.55 Impact × 5 Value × 9.5 Plausibility ÷ 10 = 7.4 of 100

Plausibility

The counterfactual is no credit [8]. That the money is spent follows from the law, which ties the credit to scholarships [4], so the outlay counts as close to certain. The split between switchers and other users is an allocation of the same sum and sits in the range of the Impact. The confounder is how the program matures: in Arizona, the share coming from public schools grew over time [11], which would raise this cost and the savings in public schools together. Reverse causation does not arise. The Plausibility is very high: the money is spent by law, and only its split is estimated.

Evidence: Mechanism · plausibility at most 9.5 Evidence of effect: Definitional

Lower scores after switching

5.6points

Children who used vouchers to move to private schools in Louisiana, Indiana and Ohio scored lower in math than comparable children who stayed, and in Louisiana and Indiana the loss did not go away. In Washington, D.C., scores were no different after three years. About 13,800 switchers leave school each year.

Value 8 · Educational opportunityImpact 1.2Plausibility 6
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is what children who switch schools learn, measured in test scores that carry into later earnings and choices. That is an educational opportunity, valued one step below health and life on this site, the same value as the gains counted for the credit. Every child counts the same here; that many switchers come from lower-income families is carried by this value, not by an extra weight [2]. The loss is counted as lasting only in the averaged amount, which already allows for the recovery seen in some programs. Gains in college going for the same children are a separate argument, because degrees and test scores have moved apart in this research. The value is high, because it is lasting learning for the children the program is meant to help.

Impact

About 180,000 children are expected to switch at any time, a quarter of the scholarships at the average amount [9][10][11], or about 13,800 in each year's school-leaving cohort. In Louisiana, attending a voucher school lowered math scores by 0.4 standard deviations in the first year [16], and scores were still lower after four years [18]. In Indiana, switchers lost 0.15 standard deviations in math, a loss that persisted however long they stayed, with no change in reading [17][26]. In Ohio, voucher users scored lower than closely matched peers who stayed in public schools [24]. In Washington, D.C., reading and math scores were no different three years after the lottery [27]. Averaged over math and reading and over the four programs, the lasting loss is set at 0.09 standard deviations, with a range of 0.03 to 0.15. Valued at about a quarter of a lifetime income, set at about 2.3 million dollars for the United States, that is 1,246 standard deviations a year, a Normalised Impact of 1.17 on the scale of this evaluation, in a range from 0.26 to 3. The Impact is moderate because the loss per child is sizeable, but few children switch.

▸ Show calculation ▾ Hide calculation
180,000 children
Children who switch, at any time
[9][10][11]
÷ ÷ 13 grades
13,846 children
In each year's school-leaving cohort
× × 0.09 standard deviation Louisiana, Indiana, Ohio and Washington, D.C., averaged over subjects; Setting, range 0.03 to 0.15
1,246 standard deviations a year
Lasting loss in achievement
[16][17][18][24][26][27]
× × 470,000 euro-equivalent value of a lasting standard deviation at a United States lifetime income set at 2.0 million euro-equivalent (about 2.3 million dollars); range 130 million to 1.5 billion euro-equivalent
585.7 million euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
1.17
Normalised Impact
Score 1.17 Impact × 8 Value × 6 Plausibility ÷ 10 = 5.6 of 100

Plausibility

The counterfactual is lottery losers in Louisiana and Washington, D.C., and matched non-users in Indiana and Ohio. The carrying source is Abdulkadiroglu, Pathak and Walters, an experimental design using the admission lotteries of Louisiana's program [16]. The published Indiana study, with matched peers, finds a math loss that persists [26], and the Ohio evaluation finds lower scores for voucher users as well [24]. The confounder in Indiana and Ohio is that families who switch differ in ways the matching does not see; the Louisiana lottery removes that. Direction: the lottery assigns the voucher before any test, so the outcome cannot have caused it. Against the loss stands the second lottery study, in Washington, D.C., which found no difference after three years [27], and the Louisiana losses were concentrated in low-tuition schools [16], so a different mix of schools could avoid them. Read back: in about six of ten comparable programs, switchers lose this much. The Plausibility is above the middle: three of four recent programs show the loss, one with a lottery, and the other lottery shows none.

Evidence: Converging studies · plausibility at most 6 Evidence of effect: Experimental · at most 8.5

Running scholarship groups

3.3points

Scholarship organizations may keep up to a tenth of what they receive to raise money, check eligibility, pay schools and pass audits. On about 4 billion dollars a year, that is up to 400 million dollars that does not reach a child. Treasury rules also require annual audits and an IRS registration for every organization.

Value 5 · Enforcement costImpact 0.7Plausibility 9.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is work that has to be done to run the program: fundraising, eligibility checks, payments to schools and audits [2][4]. It uses people and money that could do something else, and is counted at the standard value of money. Public schools have administrative costs too, but those stay in the comparison because they mostly do not shrink when a few children leave. Only the extra layer that the credit creates is counted here. Treasury and IRS work on the portal and on the lists of organizations is left out as small. The value is the middle of the scale, because the stream is the cost of running the program.

Impact

Organizations must spend at least 90 percent of their income on scholarships [4][6], so up to a tenth can go to running them. On about 4 billion dollars a year in credits [5], a tenth is 0.4 billion dollars. The full tenth is counted, in line with the 90 percent used for scholarships in the other arguments. Large organizations may need less and the credit volume may differ, so the range runs from 5 to 10 percent of 2.5 to 6.1 billion dollars, or 0.13 to 0.61 billion dollars. The cost to the IRS and to state agencies of keeping lists and a portal is small beside this and not counted. The Impact is moderate because it is a tenth of the whole sum at most.

▸ Show calculation ▾ Hide calculation
4 billion dollars a year
Credits claimed each year
[5]
× × 0.1 the allowed tenth, as in the other arguments; range 5 to 10 percent
0.4 billion dollars a year
Kept for running the organizations
[4][6]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 0.11 to 0.53 billion euro-equivalent (about 0.13 to 0.61 billion dollars)
0.34 billion euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
0.69
Normalised Impact
Score 0.69 Impact × 5 Value × 9.5 Plausibility ÷ 10 = 3.3 of 100

Plausibility

The counterfactual is no credit and no scholarship organizations to run [8]. That running them costs something follows from the law and the rules, which require separate accounts, IRS registration, annual audits and reports [2][4]. The cost arises whenever the credit exists, so it counts as close to certain. How much is estimated from the spending rule, which sets the upper end, and that doubt sits in the range of the Impact. What could make it smaller is that existing scholarship organizations already have staff and systems. Reverse causation does not arise. The Plausibility is very high for a cost that follows from the rules, and its size is in the range.

Evidence: Mechanism · plausibility at most 9.5 Evidence of effect: Definitional

Cost beyond the scholarship

3.1points

A scholarship of about 5,000 dollars does not pay for a whole school place. Private elementary schools charge about 9,000 dollars on average, and the difference is covered by the family, the school or a parish. For 180,000 children who switch, that is about 0.7 billion dollars a year of teaching that has to be paid for in addition.

Value 5 · Household budgetsImpact 1.2Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is the part of a private school place that the scholarship does not cover and that families, schools and parishes pay. It is money spent on teachers, rooms and materials, counted at the standard value of money. The children were taught at public expense before, and that saving is counted for the credit in the argument on state spending; this is the other side of the same move. Families who switch are not assumed to be better off than average, so their dollar carries the standard weight of 1. What families get for the money, a school of their choice, is its own argument for the credit. The value is the middle of the scale, because the stream is money for schooling.

Impact

About 180,000 children are expected to switch at any time, a quarter of the scholarships at the average amount of about 5,000 dollars [9][10][11]. Private schools charged 12,790 dollars on average in 2020-21, 9,210 dollars in elementary and middle schools and 9,720 dollars in Catholic schools [31]. Schools that take scholarship students tend to be the cheaper ones, so a place is set at 9,000 dollars, with a range of 7,000 to 11,000 dollars. That leaves 4,000 dollars per child that the scholarship does not pay. Over 180,000 children it comes to 720 million dollars a year, with a range of 210 million to 1.7 billion dollars. The scholarship itself is counted in the argument on federal money for switchers and is not included here. The Impact is moderate: a few thousand dollars per child, for the small share of children who switch.

▸ Show calculation ▾ Hide calculation
180,000 children
Children who switch, at any time
[9][10][11]
× × (9,000 − 5,000) dollars Setting for the place, range 7,000 to 11,000 dollars: average tuition 9,210 dollars in private elementary and middle schools
720 million dollars a year
Cost of the place not covered by the scholarship
[9][31]
÷ ÷ 1.16 exchange rate used throughout this evaluation, 1.16 dollars to the euro; range 180 million to 1.5 billion euro-equivalent (about 210 million to 1.7 billion dollars)
620.7 million euro-equivalent a year
In euro-equivalent
÷ 500 million euro-equivalent a point scale of this evaluation
1.24
Normalised Impact
Score 1.24 Impact × 5 Value × 5 Plausibility ÷ 10 = 3.1 of 100

Plausibility

The counterfactual is the same children in public schools, where no tuition is due. The chain is short: a child switches, the school place costs more than the scholarship, and someone pays the difference. That private places cost more than the average scholarship follows from tuition statistics [31]; what is not measured is which schools scholarship students will attend and what those schools charge them. Tuition may also rise where scholarships are widely available, as in Iowa [28], which would raise this cost. The confounder is that schools with empty seats can take additional children at little extra cost, so tuition overstates the resources used. Reverse causation does not arise. The number of children who switch is taken from other states' programs and sits in the range. The Plausibility is middling: the gap follows from prices, but the schools and their true extra cost are not known.

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

Donations lost elsewhere

0points

The credit refunds a gift to a scholarship organization in full, while a gift to any other charity saves the donor at most the tax on it. Donors who would otherwise have given to a food bank, a hospital or a church collection have a reason to redirect up to 1,700 dollars. How much is redirected is unknown. The argument is included but not yet scored.

Value 5 · Charitable givingImpact 0Plausibility 3
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream would be money that other charities no longer receive because donors redirect gifts to scholarship organizations. It is money, counted at the standard value. Much of what charities receive reaches people with low incomes, so a dollar lost there would weigh more than a dollar in an average household, and that weight would stand in the amount. The scholarship organizations' gain is not an offset here, because their funding is already counted in full in the arguments on tuition help and on switchers. Donors themselves lose nothing, since the credit refunds the gift [4]. The value would be the middle of the scale, because the stream is money for those whom other charities serve.

Impact

A gift to a scholarship organization is refunded dollar for dollar up to 1,700 dollars, and each spouse qualifies separately [4][33]. A gift to any other charity lowers the donor's tax only by the deduction, so it costs the donor most of its amount. If a quarter to a third of about 4 billion dollars a year in credited gifts [5] came out of gifts that would otherwise have gone elsewhere, other charities would lose 1 to 1.3 billion dollars a year. That share is an illustration, not an estimate. No study of the federal credit or of the state credits used in this evaluation measures how much giving is redirected. The Impact is therefore not set. The Impact is not scored: the share of gifts taken from other causes is not measured.

Plausibility

The chain would be: a full refund makes one kind of gift free, donors with a fixed budget for giving redirect part of it, and other charities receive less. The counterfactual is the same donors without the credit. The design is mechanistic; the price difference between the two kinds of gift follows from the law [4], but the response of donors is untested. The confounder is that a free gift may come on top of existing giving, because it costs the donor nothing. Direction: the credit precedes the giving decision. With the link that carries the quantity untested, the Plausibility stands at three in ten. The Plausibility is low as long as the shift in giving is not measured.

Recorded, not yet scored: Included but not scored: nobody has measured how much of the credited gifts replaces gifts to other charities. Tax return data for 2027 will show whether filers who claim the credit deduct less for other gifts, and the argument is to be scored then.

Evidence: Plausibility · plausibility at most 5 Evidence of effect: Mechanistic · at most 6 Placement: Chain open · plausibility 3–3.5

Open: Tax return data for 2027 comparing itemized charitable deductions of filers who claim the credit with earlier years would show the shift and could lift P to about 5.

No protection against exclusion

0points

Private schools that take scholarship students may select whom they admit, and the federal rules attach no ban on discrimination and no test of results. States may not add stricter conditions or exclude organizations. In Wisconsin, nearly half of the voucher schools reviewed had rules that appear to exclude LGBTQ students or children with disabilities. How many children would be turned away is unknown. The argument is included but not yet scored.

Value 9 · Equal treatmentImpact 0Plausibility 3
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream would be children who are refused or expelled by a school paid for with tax-funded scholarships because of a disability, their religion or their sexual orientation. Equal treatment by institutions that public money supports is a basic right, valued near the top of the scale on this site. The children concerned cannot change what they are excluded for. A public school place remains open to them, which limits the harm and is why the value is not at the very top. The value does not depend on how many children are affected; that would stand in the amount. The value would be very high, because exclusion on such grounds touches a basic right.

Impact

The federal rules require annual audits of scholarship organizations but set no conditions for the schools on admission or on academic results [3]. States must list every organization that meets the federal requirements and may not add stricter ones [33]. Wisconsin Watch reviewed the public materials of about a third of that state's 373 voucher schools and found policies that appear to discriminate against LGBTQ students or students with disabilities in nearly half of them [34]. That shows such rules are common among schools that take vouchers, but not how many children are refused or leave because of them. Nor is there an agreed quantity for a refused admission when a public school place remains open. The Impact is therefore not set. The Impact is not scored: the number of children excluded is not known.

Plausibility

The chain would be: scholarships flow to schools with exclusionary rules, children apply or attend, and some are refused or made to leave. The counterfactual is no credit, under which these schools receive no federally supported scholarships. The design is mechanistic; that the rules allow it follows from the regulations [3][33], and that such school policies exist is documented for one state [34], but the number of children affected is untested. The confounder is that families rarely apply to schools whose rules exclude their child, so written policies overstate actual refusals. Direction: the rules precede the admissions. With the link that carries the quantity untested, the Plausibility stands at three in ten. The Plausibility is low for any given number, although the missing protection itself is a fact of the rules.

Recorded, not yet scored: Included but not scored: the rules set no ban on discrimination and no test of results for schools that receive scholarships, but nobody has counted how many children are refused or leave for that reason. A count from one participating state would allow a score.

Evidence: Plausibility · plausibility at most 5 Evidence of effect: Mechanistic · at most 6 Placement: Chain open · plausibility 3–3.5

Open: A count of refused admissions and expulsions in schools that receive scholarships, by reason, from one participating state would allow a quantity and could lift P to about 5.

How robust the result is

Under the assumptions used, the proposal brings more than it costs in 48 of 100 model runs.

gains more than it costscosts more than it gains

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

The tuition money and the income tax behind it are the same sum; only their weights differ, 0.84 for a dollar where the scholarships end up and 1 in the federal budget. At equal weights the two would cancel, which moves the result only a little toward the credit. A lasting standard deviation in a child's achievement is valued at about a quarter of a lifetime's earnings, and gains and losses are assumed not to fade. The gain in public schools is set at half a percent of a standard deviation, near the lower end of a range that reaches one and a half percent. The credit comes out ahead in nearly five of ten runs and a little ahead on average. At one and a half percent it would come out ahead in nearly eight of ten runs, the result would be likely better, and the effect would rise from medium to large.

Show the distribution of the model runs

20,000 simulated futures

Unclear · 48 %

In 48 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 −19 and +30 points.
Scale and assessment

Scale of this evaluation: Normalised Impact, unitless and calibrated to this topic. For comparison, one point here corresponds to about 500 million euro-equivalent a year.

Size class:medium

Effect medium (+2) · 54 % 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 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 →

Last reviewed by Claude Opus 5.5 · October 29, 2026 · 6× AI, not yet reviewed by a human
  1. October 29, 2026AI reviewClaude Opus 5.5record updated

    Page texts for the new page layout (draft G) with quotes from primary sources; numbers unchanged.

  2. October 21, 2026AI reviewClaude Fable 5.1record updated

    Money amounts in argument texts stated in dollars per style-en.md; scores, chains and balance unchanged

  3. October 20, 2026AI reviewClaude Fable 5.1record updated

    Summary and scale note rewritten to the style of 20 Oct 2026; numbers unchanged

  4. October 16, 2026AI reviewClaude Opus 5.5record updated

    Ranges of pro-2, pro-4, pro-5, con-3 and con-5 rederived by combining the factor ranges in quadrature instead of multiplying their extremes; the result stays unclear.

  5. October 16, 2026AI reviewClaude Opus 5.5record updated

    Texts updated to Bilanz 2.2

  6. October 12, 2026AI reviewClaude Opus 5.5First evaluation

    KI-Entwurf der Themen-Automatisierung, am 12.10.2026 unabhängig nachgeprüft und korrigiert übernommen

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

Show all 35 sourcesHide sources
  1. Internal Revenue Service (8 June 2026): More than half the U.S. states signed up to participate in the federal scholarship tax credit program, IR-2026-76 — 27 participating states, credit of up to 1,700 dollars. irs.gov
  2. Ballotpedia News (7 October 2026): Internal Revenue Service issues regulations for Education Freedom Tax Credit — 30 participating states, dates, income limit, rules for scholarship organizations. news.ballotpedia.org
  3. Education Week (October 2026): Federal education tax credit rules are finally here: what we know so far — 31 states in or planning to join, opt-in by 1 January 2027, pending bills. edweek.org
  4. Current Federal Tax Developments (1 October 2026): Federal scholarship tax credit regulations: technical analysis of temporary and proposed regulations under section 25F — 90 percent spending rule, priorities, qualified expenses, start date. currentfederaltaxdevelopments.com
  5. Urban Institute, Kristin Blagg (2 September 2025): What comes next for federal tax credit scholarships? Estimates of take-up — 2.7 to 3.6 million filers, 2.7 to 6.1 billion dollars a year in federal revenue. urban.org
  6. Education Week (September 2026): Federal tax credits for K-12 scholarships: what we know and don't — donation scenarios, eligibility, share for administration. edweek.org
  7. Deseret News (27 June 2025): Senate cuts education tax credits: Joint Committee on Taxation estimate of about 26 billion dollars over ten years, reported while the Senate version carried a cap of 4 billion dollars a year. deseret.com
  8. Ballotpedia News (24 April 2026): Senate Democrats introduce bill to repeal federal education tax credit scholarship — S. 4297, Keep Public Funds in Public Schools Act. news.ballotpedia.org
  9. EdChoice (October 2024): Fiscal effects of private K-12 education choice programs — average tax-credit scholarship of 4,997 dollars, public school spending of 15,542 dollars per pupil, cost categories that follow students. edchoice.org
  10. EdChoice (29 August 2024): What states tell us about who is using school choice — share of participants who came from public schools in Indiana and Arkansas. edchoice.org
  11. The 74 (4 December 2025): In Arizona, the typical ESA recipient already attends private school, study finds — about three quarters outside public schools before joining. the74million.org
  12. Murnane, Reardon, Mbekeani and Lamb, Education Next (2018): America's private elementary schools increasingly serve the affluent — gap in private enrollment between high- and middle-income families. educationnext.org
  13. Figlio, Hart and Karbownik, CESifo Working Paper 9056 (2021): Effects of scaling up private school choice programs on public school students — Florida, test scores, absences and suspensions. ifo.de
  14. Chingos, Kuehn and co-authors, APPAM (2019): The effects of the Florida Tax Credit Scholarship Program on college enrollment and graduation: an update. appam.confex.com
  15. Erickson, Mills and Wolf, APPAM (2019): The effects of the Louisiana Scholarship Program on college entrance — lottery-based evaluation. appam.confex.com
  16. Abdulkadiroglu, Pathak and Walters, NBER Working Paper 21839: Free to choose: can school choice reduce student achievement? — Louisiana Scholarship Program, lottery-based estimates. nber.org
  17. Austin, Waddington and Berends, RSF Journal of the Social Sciences 5(3) (2019): Voucher pathways and student achievement in Indiana's Choice Scholarship Program. pmc.ncbi.nlm.nih.gov
  18. K-12 Dive (25 April 2019): Study: after four years, scores drop among students in voucher program — Louisiana Scholarship Program, Mills and Wolf. k12dive.com
  19. National Center for Education Statistics: Fast facts: back-to-school statistics — 49.6 million public and 5.5 million private school students. nces.ed.gov
  20. Georgetown University Center on Education and the Workforce (2021): The college payoff — lifetime earnings of 2.8 million dollars with a bachelor's degree and 1.6 million dollars with a high school diploma. cew.georgetown.edu
  21. GovInfo: S. 4297 (introduced), Keep Public Funds in Public Schools Act — bill text. govinfo.gov
  22. Jabbar, Fong, Germain, Li, Sanchez, Sun and DeVall, Educational Policy 36(2) (2022): The competitive effects of school choice on student achievement: a systematic review, meta-analysis with small positive effects; school-level estimate 0.06 with an interval that includes zero. pmc.ncbi.nlm.nih.gov
  23. Egalite, School Choice Demonstration Project, University of Arkansas (2016): The competitive effects of the Louisiana Scholarship Program on public school performance: neutral to positive effects, small in size. scholarworks.uark.edu
  24. Figlio and Karbownik, Thomas B. Fordham Institute (2016): Evaluation of Ohio's EdChoice Scholarship Program: selection, competition, and performance effects; modest gains for eligible students who stayed in public schools, lower scores for voucher users. fordhaminstitute.org
  25. Graff, Education Next (2024): School choice or school spending? Florida's 15-year experiment: summary of Figlio, Hart and Karbownik; gap of 0.166 standard deviations after 15 years between schools with more and less competition; author works for a school-choice organization. educationnext.org
  26. Waddington and Berends, Journal of Policy Analysis and Management 37(4) (2018): Impact of the Indiana Choice Scholarship Program: achievement effects for students in upper elementary and middle school; math loss of 0.15 standard deviations that persists. ideas.repec.org
  27. Institute of Education Sciences, NCEE 2019-4006 (May 2019): Evaluation of the DC Opportunity Scholarship Program: impacts three years after students applied; no effect on reading or math, students more satisfied and schools seen as safer. ies.ed.gov
  28. Fontana and Jennings, EdWorkingPaper 24-949, Annenberg Institute at Brown University (April 2024): The effect of taxpayer-funded education savings accounts on private school tuition: evidence from Iowa; tuition up 21 to 25 percent where every child was eligible, 10 to 16 percent where some were. edworkingpapers.com
  29. Hungerman and Rinz, NBER Working Paper 21687 (2015): Where does voucher funding go? How large-scale subsidy programs affect private-school revenue, enrollment, and prices. nber.org
  30. Institute on Taxation and Economic Policy, Carl Davis (2 July 2025): Trump megabill will give $30 billion a year to private school vouchers? Joint Committee on Taxation figures of about 3 to 4 billion dollars a year for the enacted credit. itep.org
  31. National Center for Education Statistics, Digest of Education Statistics 2023, table 205.50: Private school enrollment and average tuition, 2020-21: 12,790 dollars on average, 9,210 dollars in elementary and middle schools, 9,720 dollars in Catholic schools. nces.ed.gov
  32. Lake and Hill, Brookings Institution (17 May 2018): School funds should follow students, not protect institutions: reports Ladd and Singleton's estimate of 3,500 dollars of uncompensated cost per departing student in a district spending 8,000 dollars. brookings.edu
  33. EdChoice (October 2026): The federal tax credit rules are here: here's what we know; Treasury estimate that about 96 percent of students in participating states are eligible, states may not add stricter requirements or exclude organizations, spouses each qualify for 1,700 dollars. edchoice.org
  34. Wisconsin Watch (5 May 2023): False choice: Wisconsin taxpayers support schools that can discriminate; nearly half of the voucher schools reviewed had policies that appear to exclude LGBTQ students or students with disabilities. wisconsinwatch.org
  35. Institute of Education Sciences, NCEE 2009-4050 (2009): Evaluation of the DC Opportunity Scholarship Program: impacts after three years; parents more satisfied and schools seen as safer. ies.ed.gov

More debates in this area

Very likely better · 100 %
Size: very large
Lifting The Payroll Cap

Charge the Social Security payroll tax on earnings above the taxable maximum of 184,500 dollars, which today stop being taxed altogether.
Very likely worse · 0 %
Size: large
Keep the Tips Deduction

Make the deductions for tipped and overtime income permanent instead of letting them lapse after 2028.
Very likely better · 100 %
Size: medium
California Billionaire Tax

Charge everyone who lived in California on 1 January 2026 with a net worth above 1 billion dollars a one-time tax of 5 percent, and spend nine tenths of it on health care.
Very likely better · 100 %
Size: large
A Fully Refundable Credit

Pay the full child tax credit to every eligible child, including those whose parents earn too little to owe income tax.
All debates in Taxes, Tariffs and the Federal Budget →