Private School Tuition Increases After Voucher Expansion: 2026 Data and Study Review

Private school tuition increases after voucher expansion do not reduce to one national number. Among the sources located and verified through August 1, 2026, a 2025 study of 11 recent expansion states concluded that recent broad or universal voucher programs probably raised private-school tuition 5% to 10% in the short run. The strongest state quasi-experiment, in Iowa, estimated 21% for kindergarten and 10% for grades 1–12 when it compared the same schools over time.

Those figures are not a pooled average. The national 5%–10% range is the authors’ synthesis across models whose individual estimates ran from approximately zero to 13%. Iowa is one state in the first implementation year. Arizona and North Carolina add useful post-expansion evidence, but their most-quoted figures are descriptive. A broader 2013–14 through 2022–23 analysis of older and more varied school-choice policies estimated elementary tuition at $925 below its modeled counterfactual after adoption.

Anyone who gives you one number without naming the study, the period, the data source, and the comparison is leaving out the part that decides what the number means.

The headline numbers

Source: The School Choice Index synthesis of Harris and Olivier, Fontana and Jennings, and Bedrick, Greene, and Burke.

The most defensible single sentence is this:

Recent broad voucher and ESA expansions have produced credible evidence of higher private-school tuition in some settings, including a 5%–10% national short-run synthesis and same-school estimates of 21% and 10% in Iowa, but the published evidence does not support one universal percentage for every state or program.

What do the data show about private school tuition increases after voucher expansion?

The strongest recent evidence points toward a positive tuition response after broad voucher or ESA expansion, but the size and confidence depend on the design. Iowa supplies the clearest state quasi-experiment. The 11-state study supplies the multi-state synthesis. Arizona, North Carolina, and Indiana document additional post-expansion movement without isolating one causal effect.

A useful way to read the evidence is to separate the question “Did tuition move?” from the question “Did the policy cause the movement?”

Evidence classes used on this page

The class is not a political score and it does not judge an organization. It tells you what sentence the evidence can carry.

Table 1. Voucher tuition evidence ledger (15 rows)

Source: The School Choice Index evidence ledger, compiled from the original producers linked in each row. “Peer reviewed” distinguishes journal or institutional review from publication by an organization; it does not make unlike designs interchangeable. Last verified August 1, 2026.

Why do the published estimates disagree?

The spread is not explained by one hidden mistake. It comes from four visible differences: what data stood in for tuition, whether the study had a credible comparison, whether the number was nominal or inflation adjusted, and which program and period were observed. A result can be correctly calculated and still answer a different question from the number beside it.

The National Center for Education Statistics calls the Private School Universe Survey the nation’s primary source of information on K–12 private schools. Its published list does not include tuition or school revenue. That leaves researchers building their own price series.

Table 2. The data-source map

Source: NCES Private School Universe Survey, Harris and Olivier, Fontana and Jennings, RAND, and Indiana Department of Education.

Table 3. The same 37 Iowa schools priced two ways

Fontana and Jennings found 37 Iowa schools with both a school-published tuition schedule and a Private School Review listing. The discrepancy is real and matters for understanding why estimates can differ.

Source: Fontana and Jennings, Appendix Table A1. Percentage differences are The School Choice Index calculations using the school-published figure as the denominator.

Table 4. Iowa–Nebraska estimates: how data source choice moves the result

Harris and Olivier used Private School Review to run a close version of the Iowa–Nebraska comparison. Reasonable-looking choices about the tuition source, sample, unit, and weighting can move the Iowa estimate from a double-digit increase to a small negative result.

Source: Harris and Olivier, Table 6 and Fontana and Jennings, Table 2.

Table 5. Carolina Forward’s selected school groups — timing pattern

A before-and-after increase answers a real question. It does not answer what would have happened without the policy. Carolina Forward’s groups show a striking pattern in the second interval, but the groups were selected rather than randomly sampled or matched.

Source: Carolina Forward. Difference rows are The School Choice Index calculations from the published group averages.

The honest original finding is narrower: In Carolina Forward’s selected North Carolina groups, the treated-group divergence was under one percentage point in the first interval and roughly 10 to 11 points in the second. Current evidence does not establish that this timing pattern generalizes to other states.

What did the 11-state national study find?

Harris and Olivier concluded that recent universal or broad voucher expansions probably increased private-school tuition 5% to 10% in the short run. That is an author judgment synthesizing several models, not one regression coefficient and not a directly observed national mean. Their direct posted-tuition model is near zero, while nonprofit-revenue models produce positive estimates that usually land around 5% to 10% and reach as high as approximately 13%.

The study compares 11 treated states with 34 comparison states using Private School Review data (approximately 19% of U.S. private schools), National Center for Charitable Statistics data (approximately 11%), and staggered difference-in-differences models. That is why both of these statements are true:

1. The paper’s direct posted-tuition result is statistically close to zero.
2. The paper’s overall conclusion is that tuition probably increased 5% to 10%. The second is the paper’s synthesis across imperfect measures.

Table 6. State coefficients from the 11-state study

The dollar column below is an original rough-scale conversion: coefficient × the study’s reported pre-treatment mean. Because the dependent variable uses the inverse hyperbolic sine transformation, the coefficients approximate proportional changes at these magnitudes. The dollar column is nota family bill and is not the authors’ preferred national specification.

Source: Harris and Olivier, Table 3. Dollar scale calculated by The School Choice Index and rounded to the nearest dollar.

• These are model-specific state estimates, not 11 stand-alone state studies.
• The national posted-tuition model has failed pretrend tests.
• The paper’s preferred conclusion comes from triangulating several outcomes, not averaging these 11 coefficients.

What happened to private school tuition in Iowa?

Iowa provides the clearest recent state evidence because eligibility differed by grade in the same first program year. Kindergarten was universally eligible, grades 1–12 were partly eligible, and pre-K was ineligible. The same-school estimates were 21% in kindergarten, 10% in grades 1–12, and no statistically detectable increase in pre-K.

Iowa’s first-year ESA amount was $7,635. That award sat above the average tuition reported in the study, which gave lower-price schools room to raise posted tuition without pushing the full increase onto an award recipient in the first year.

Table 7. Iowa’s modeled estimates

Source: Fontana and Jennings, Table 2 and abstract.

Table 8. Iowa and Nebraska raw tuition changes

Source: Fontana and Jennings, Table 1. The final column is transparent subtraction from the published means; it is not the regression estimate.

The study’s final analytic sample covered 51% of Iowa private schools educating 62% of Iowa private-school students, and 44% of Nebraska private schools educating 51% of Nebraska private-school students.

What happened to private school tuition in North Carolina?

North Carolina removed the Opportunity Scholarship income cap beginning in 2024–25. The maximum 2024–25 award was $7,468. Public Schools First NC found expansion-year average increases of 13.9% to 15.4% in three small matched cohorts. Neither design proves how much of the change the policy caused.

Table 9. Matched North Carolina cohorts

Source: Public Schools First NC, 2025 tuition fact sheet.

Table 10. North Carolina tuition bands, same 299 schools

Source: Public Schools First NC. Percent changes are The School Choice Index calculations from the published counts.

Of the 48 schools in the $7,000–$7,999 band: 12 charged between $7,400 and $7,500; two charged exactly $7,468, the maximum 2024–25 award. That clustering is worth documenting. It is not proof that every school set its price to capture the award.

Pungo Christian Academy’s posted tuition rose from $5,293 to $7,400 — a 39.8% recorded increase.In an on-the-record response to WUNC, the school’s head said books, technology, and insurance charges had been moved into tuition and that the total cost of attendance rose about 18%, not 40%. Both facts belong together. A posted-tuition series can correctly record a 39.8% change while a school can correctly say the all-in price moved less.

What happened to private school tuition in Arizona?

Arizona has four widely cited tuition figures, and none is a duplicate of another. They use different periods, statistics, price bases, and comparison methods. RAND’s report did not attribute the increases definitively to universal ESAs. Goldwater’s 14.7% and RAND’s 12% look similar on the page and are not comparable measurements.

Table 11. Four Arizona tuition measures

Source: Goldwater Institute, Hechinger Report, RAND, and Harris and Olivier.

RAND also documents market growth alongside price growth: estimated private-school count grew from 451 in 2022 to 515 in 2025 (+14%), and estimated enrollment grew from 66,032 to 77,783 (+18%). A voucher expansion can affect price, enrollment, and the number of providers at the same time.

Why does the Heritage analysis find a lower elementary tuition path?

Heritage’s 2023 report estimates that adopting a school-choice policy was associated with elementary tuition $925 below the modeled no-adoption path. The result is real. It studies a broader mix of vouchers, tax-credit scholarships, and ESAs over 2013–14 through 2022–23, not the recent group of broad or universal expansions studied by Harris and Olivier.

Source: Bedrick, Greene, and Burke, “Does School Choice Affect Private School Tuition?”

What does the journal-reviewed research say?

Hungerman and Rinz is the journal-peer-reviewed article in this evidence ledger. Published in the Journal of Public Economics in 2016, it studies earlier subsidy programs using nonprofit tax-return data and finds that program design changes where the public subsidy appears:

• When eligibility was restricted to defined student groups, participating schools expanded enrollment and prices changed little.
• When subsidies were broadly available within the studied programs, per-student revenue and prices rose while enrollment changed little.

That earlier result is directly relevant to current design debates, but modern universal ESAs are not identical to the older programs in the paper. The article supplies a mechanism and historical comparison, not a current national effect size. The two most-cited studies of the post-2021 expansions — Fontana and Jennings and Harris and Olivier — remain working papers as of August 1, 2026. “Peer-reviewed research shows” should point to Hungerman and Rinz unless a sentence also identifies the different publication status of the newer work.

Source: Hungerman and Rinz, NBER working-paper version and the published article.

What do Indiana’s administrative records show?

Indiana publishes a direct participant-based administrative series. From 2023–24 to 2024–25, reported tuition and fees rose 7.99%, while the average award rose 4.35%. The arithmetic difference between those two averages widened 23.34%.

Table 12. Indiana Choice Scholarship participant averages

Source: Indiana Department of Education 2023–24 report and 2024–25 report. Calculations by The School Choice Index from the unrounded published values.

The phrase “arithmetic difference”is deliberate. It should not be called every family’s unfunded balance. Some participants receive an award equal to their full reported tuition and fees; others receive the 90% amount and may owe more. The population also changed from 70,095 participants in 2023–24 to 76,067 in 2024–25, and the mix of schools, grades, regions, and newly eligible families can move the average.

For a broader award-to-tuition comparison, see School Voucher Tuition Gap by State: 2026–27 Data.

How much of Iowa’s first-year award matched the measured tuition response?

Dividing Iowa’s reported tuition effects by the official $7,635 first-year ESA amount produces an implied tuition-response share of approximately 10% to 19% of the award, depending on the grade band and whether the modeled or raw comparison is used.

Table 13. Iowa tuition response as a share of the 2023–24 ESA amount

Source: Tuition responses from Fontana and Jennings; award amount from the Iowa Department of Education. Division by The School Choice Index.

This table does not mean every family “kept” the rest. It measures one narrow ratio: the estimated or raw posted-tuition response divided by the published first-year award. It is also a one-year Iowa calculation.

Are the tuition figures adjusted for inflation?

Some are and most are not. Iowa, Harris and Olivier’s posted-tuition work, North Carolina, Hechinger, and Indiana report nominal values. RAND reports Arizona tuition in 2025 dollars. Goldwater reports nominal medians and places CPI beside them. Heritage’s before-and-after table is inflation adjusted. A nominal difference-in-differences coefficient and an inflation-adjusted before-and-after change are not made comparable by subtracting CPI from one of them.

Table 14. Price basis by source

Source: Method sections and table notes in the original reports.

Which private schools showed the largest tuition response?

Harris and Olivier report larger estimated effects among nonreligious schools, smaller schools, and schools with lower baseline tuition. North Carolina’s distribution also shifted away from lower price bands and toward the $6,000–$7,999 range. These are patterns in specific samples, not rules that every school in those categories followed.

The low-baseline result has a straightforward program mechanism: when an award is above a school’s existing price, the school can raise posted tuition while an award recipient still has public funds available. Iowa’s first-year $7,635 ESA was above the study’s average tuition. North Carolina’s maximum award sat inside the band that gained schools.

That mechanism does not prove motive at any individual school. Schools can also raise prices because of compensation, facilities, insurance, enrollment mix, new services, or a change in which fees are included in tuition. Modern ESAs add another constraint: funds can often be used for more than tuition, so a tuition increase can leave a family with less ESA money for tutoring, curriculum, transportation, or therapies.

Does a higher sticker price mean families paid more?

Not necessarily, and the current evidence cannot answer that question for every family. Four amounts can differ:

1. Posted tuition
2. Required tuition and fees
3. Net price after school aid, denominational subsidy, or sibling discounts
4. Family payment after the voucher or ESA

The studies do not observe all four for every family. Harris and Olivier’s Private School Review sample estimates a roughly 0.9-percentage-point increase in the share of students receiving financial aid and an approximately 2.5%increase in average aid after expansion — but the same table has failed pretrend tests. The Pungo example shows the fee problem concretely: posted tuition rose 39.8%, while the school said its total cost of attendance rose about 18% after separately billed charges were moved into tuition.

For current award-versus-listed-tuition comparisons, see School Voucher Tuition Gap by State. For program participation and prior-school-status definitions, see Voucher Students Already in Private School.

Why this evidence matters now: the 2027 federal scholarship tax credit

Internal Revenue Code §25F allows an individual taxpayer to claim a nonrefundable federal credit of up to $1,700 for qualifying cash contributions to listed scholarship-granting organizations beginning January 1, 2027. As of July 24, 2026, the IRS listed 30 participating states for 2027.

Source: IRS Federal Scholarship Tax Credit page and IRS implementation announcement.

The evidence on this page points to a simple reporting standard. A state that wants to monitor prices should publish, by school and year: posted tuition by grade, required fees, net tuition charged to program participants, award amount, school-provided aid and discounts, number of participating students, prior-school status, whether tuition and fees were reclassified, and a stable school identifier that permits year-to-year matching. Without those fields, every national study will keep rebuilding an incomplete price series from commercial profiles, tax returns, archived school pages, or participant averages.

What this evidence ledger adds

The original contribution is not a new regression. It is a reproducible map of unlike evidence that records what each number measures, what sentence it can support, and what cannot be combined. The assembled evidence includes:

1. A design-and-scope classification separating recent quasi-experiments, multi-state models, comparative descriptions, before-and-after records, and broader historical evidence.
2. A normalized ledger that keeps price basis, statistic, observational unit, comparison group, publication status, and causal status beside every result.
3. The 37-school Iowa price-source discrepancy, calculated as −37.1% at the median and −22.4% at the mean.
4. The close Iowa–Nebraska cross-paper comparison, showing a +10% estimate beside −1.3% and −1.8% estimates while making the sample and unit differences visible.
5. Rough dollar-scale conversions for all 11 Harris–Olivier state coefficients, with the inverse-hyperbolic-sine and model warnings attached.
6. Raw Iowa treated-comparison arithmetic in dollars and percentage points, labeled as descriptive rather than causal.
7. North Carolina distribution and selected-group calculations, including the 48.2%, 32.9%, 54.8%, 11.31-point, and 9.91-point results.
8. Iowa award-share calculations that show exactly what is being divided.
9. Indiana’s exact two-year administrative arithmetic using the unrounded state figures.
10. A non-pooling rule that prevents the most tempting and least meaningful “national average.”

How we built this evidence ledger

What this evidence does not show

The ledger can tell you what the published evidence measures and how strong each design is. It cannot manufacture a national tuition census, a long-run effect, a school’s motive, or a family’s final bill from data that do not contain those fields.

• No federal tuition census. NCES maintains the best federal frame for private schools but does not collect tuition or revenue. Every national price estimate therefore depends on a partial commercial, nonprofit, school-published, or administrative source.
• Short post-expansion periods. The recent studies mostly observe one to three post-expansion years. That is enough to measure short-run movement and not enough to settle a long-run market response.
• Working-paper status. Fontana and Jennings and Harris and Olivier are working papers, not journal articles. Their estimates and methods can still change before journal publication.
• National pretrend problems. Harris and Olivier’s posted-tuition models fail parallel-trends tests around the pandemic. The 5%–10% result must remain attached to the synthesis process.
• Selective and incomplete samples. Private School Review supplies usable tuition for approximately 19% of U.S. private schools. NCCS data cover approximately 11%. Iowa’s most-complete cohorts contain 51% of schools. Public Schools First NC’s most-quoted complete cohorts contain 10, 30, and 49 schools. None of those is the entire market.
• Price definitions change. Tuition can exclude registration, books, technology, insurance, transportation, activities, or other mandatory charges. A school can move a charge into tuition without raising total cost by the same percentage.
• Posted price is not net price. School aid, congregational subsidy, sibling discounts, voucher amount, other scholarships, and eligible ESA expenses all sit between the posted schedule and the family’s final payment.
• Administrative averages change with the population. Indiana’s reported average can rise because the same schools raised price, because participants moved into higher-price schools or grades, or both.
• Program labels hide major differences. A targeted voucher, tax-credit scholarship, universal ESA, and federal scholarship tax credit do not create the same demand shock.
• No pooled average. Averaging the −$925 Heritage coefficient, Iowa’s 21%, RAND’s 12% real average, North Carolina’s unweighted cohort means, and Indiana’s participant average would produce a number with no coherent unit, population, period, or counterfactual.

Frequently asked questions

Downloads

How to cite this page

Primary sources