Universal School Vouchers and Education Savings Accounts Are Good Policies
本文运用Levin的四维评估框架，基于现有研究分析美国普及型学校券和教育储蓄账户政策在扩大选择、提升效率、促进公平和社会凝聚力方面的效果，认为这些政策总体上是好的。
Private school choice is having a moment. After a slow start, voucher and Education Savings Account (ESA) programs experienced explosive growth in their numbers, geographic scope, and enrollments recently. Vermont and Maine established the first voucher programs in rural areas that lacked public schools in 1869 and 1873, respectively. They were the only private school choice initiatives in the United States until Wisconsin established the first modern voucher program in Milwaukee in 1990 (Witte and Wolf 2017). Ohio followed in 1995 with the Cleveland Scholarship and Tutoring Program. Arizona launched the first tax-credit-funded scholarship (voucher) program statewide in 1997 and followed with the first ESA program in 2011. By 2025, 35 states, plus DC and Puerto Rico, operated 70 tuition voucher or ESA programs serving over 1.2 million students.1 Until 2022, eligibility for private school choice programs was targeted to students disadvantaged by disability, geography, or family income in all states. That year, Arizona expanded eligibility for its pioneering ESA initiative to every K–12 student in the Grand Canyon state. In the three short years since Arizona launched the first universal private school choice program, 16 other states have followed suit, including Idaho, Indiana, New Hampshire, Tennessee, Texas, and Wyoming in 2025 alone. On July 4, 2025, President Trump signed the One Big Beautiful Bill Act, which includes a national voucher-type program that states can opt into beginning January 1, 2027. Private school choice programs initially were few, exclusively in the tuition voucher mold, and narrowly targeted. Now, choice programs are common throughout the country, often take the more flexible form of ESAs, and are either building toward or have reached universal eligibility. Is universal private school choice good policy? I think so. Levin (2001, 9) argues that private school choice programs should be evaluated based on their effects on four key dimensions: expanding consumer choice, improving the productive efficiency of K–12 education, promoting equity, and advancing social cohesion. Here, I apply Levin's evaluative framework to the existing research on school choice. Although robust evidence on the effects of U.S. universal private school choice programs is just now emerging, we can draw inferences from studies of universal choice programs in other developed countries, evaluations of targeted U.S. programs, and descriptive data from new universal initiatives in the United States. These inferences are provisional, given the inherent uncertainty regarding how such policies will evolve. Universal voucher and ESA programs expand consumer choice by allowing more parents to access the schools they prefer for their children. According to Morning Consult's recent polling, 36% of parents of school-age children would prefer a private school placement for their child, and 14% would like to homeschool them (EdChoice 2025a). Among that same representative sample of parents, only 9% of students are currently enrolled in a private school and only 5% are homeschooled. Thus, 36% of parents report the inability to access their preferred schooling option for their child. In states with large school choice programs, the enrollment decisions of parents more closely match their schooling preferences. In 2023–2024, the first year that every Florida student was eligible to receive a voucher or ESA for private or homeschooling, 643,022 students enrolled in Florida private or homeschools, representing 18.3% of the school-age population in the Sunshine State (Gibbons 2025). Sweden instituted a universal private school choice system in 1993, when its nascent private school sector enrolled less than 2% of Swedish students. In 2022, 20% of Sweden's secondary students were enrolled in private schools (World Bank Group 2024). Finally, the Netherlands instituted universal private school choice in 1917. By 2008, over 70% of Dutch elementary students and over 80% of Dutch secondary students attended publicly financed private schools (Patrinos 2011, 57). Parents who participate in choice programs tend to be more satisfied with their child's school. In a systematic review of 19 studies of the effect of choice programs in the United States on parent satisfaction with schools, Rhinesmith (2019, 114) concludes, “Whether students enrolled in their choice program through lottery or self-selected into their private school of choice… providing choice in education leads to higher levels of parent satisfaction.” An experimental study in the review drew upon data from the original congressionally mandated evaluation of the DC school voucher program. Kisida and Wolf (2015, 274) report that the overall positive impacts of the voucher program on parent satisfaction ranged from 0.34 to 0.53 standard deviation (SD), using either average school grade or a multi-item satisfaction scale. The parent satisfaction impacts were higher if the student was a lower performer at baseline, was in the middle or high school grades, or had switched into the program from a public school that was not failing. Satisfaction impacts were higher for the parents of children whose achievement impacts were higher, and vice versa. The extent to which school voucher and ESA programs improve the productive efficiency of K–12 education is fiercely contested. The question is multi-faceted, involving three primary considerations: (1) participant effects of choice programs on student achievement and attainment, (2) competitive effects of choice on the performance of affected public schools, and (3) fiscal effects. The effects of private school choice programs on the subsequent test scores of participants are mixed (Wolf 2020, 16–22). They are more positive in earlier evaluations of urban pilot programs, in reading, when neutral tests are used, in the final evaluation year, and when the study was experimental. They are more negative in recent evaluations of statewide programs, in math, when curriculum-aligned state tests are used, in the first evaluation year, and when the study was nonexperimental. In a statistical meta-analysis of achievement impacts of voucher programs limited to experimental studies, Shakeel et al. 2021 (527) report, “Voucher impacts display a timing effect, growing more positive over time.” In the final year of the experimental evaluations, the average achievement impact of U.S. voucher programs is a gain of .05 SD in reading and slightly less than 0.05 SD in math, with the reading impact statistically significant and the math impact not significant. Recent nonexperimental evaluations of expansive statewide choice programs have reported mixed achievement effects involving large positive effects in North Carolina (Egalite et al. 2020), null findings in DC (Webber et al. 2019), mixed findings in Indiana (Waddington and Berends 2018), and large negative effects in Ohio (Figlio and Karbownik 2016). The effects of private school choice programs on educational attainment outcomes—including high school graduation, college enrollment, and college completion—are more reliably positive. Four of five findings on the effects of choice programs on high school graduation rates are positive (Cowen et al. 2013; Erickson and Scafidi 2020; Warren 2011; Wolf et al. 2013), while the remaining finding is null (Austin and Pardo 2021). Seven of eight findings on the effects of choice programs on college enrollment are positive overall (Austin and Pardo 2021; Chingos et al. 2019, 2025; Erickson and Scafidi 2020; Wolf et al. 2019) or at least for one or more subgroups (Chingos and Peterson 2015; Erickson et al. 2021), while the remaining finding is null (Chingos 2018). All five findings on the effects of choice programs on college completion are positive overall (Chingos et al. 2019, 2025) or at least for disadvantaged subgroups (Chingos and Peterson 2015, 2021; Wolf et al. 2019). Students tend to remain engaged in the educational project longer and go further if they participate in a private school choice program. The competitive effects on the test scores of the much larger set of students who remain in public schools, pressured by the launch or expansion of choice programs, are overwhelmingly positive or neutral. The most recent systematic review of 30 empirical studies on this question (EdChoice 2025b, 24) reports that 27 of them include at least one finding of a positive and statistically significant competitive effect, while only two report any significant finding of negative effects, and one reports exclusively null findings. In a statistical meta-analysis, Jabbar et al. 2022 (264) determine that the average competitive effect of U.S. voucher programs on school-level achievement in affected public schools is a statistically significant gain of 0.22 SD. A recent pioneering study finds that the statewide Ohio EdChoice Program pressured public schools to improve student college-going and college completion rates in addition to test scores (Chingos et al. 2025). The fiscal effects of universal voucher and ESA programs are largely unknown. Existing evidence suggests that targeted choice programs can generate taxpayer savings because they are limited to students who would otherwise attend public schools and because the voucher amount is smaller than the state's per-pupil expenditure in those schools (Lueken 2024). However, these fiscal benefits may not carry over to universal programs, which are more likely to subsidize students already enrolled in private schools, resulting in new public costs rather than savings. The general formula for determining the fiscal impact of a school choice program is the savings minus the subsidy. Savings come from students switching from higher-cost public schools to private schools via a lower-cost voucher or ESA. Subsidies come from governments funding private and homeschool students who previously self-financed their nonpublic education. The fiscal impact of a choice program is positive when the fiscal savings exceed the required subsidy. That tends to be true when the public school “switcher rate”—the proportion of participants diverted from public school enrollment—is above 36%–57% (Lueken 2024). The fiscal impact of universal voucher and ESA programs will vary annually as the switcher rate varies. Typically, universal programs start incrementally, in Year 1, by absorbing students in the state's targeted choice programs, which are saving the state money, and adding rising kindergarteners and new categories of eligible students, some of whom are already in nonpublic placements and represent new subsidies. The newly subsidized students likely will outnumber the previous voucher users and public-school switchers who save the state money, so choice programs building to universality likely will have a net fiscal cost in Year 1. Year 2 will be like Year 1 but on a larger scale, costing the state an amount less than the program appropriation due to some savings from new or past school switchers (Daniels et al. 2025, 25-26). Year 3, typically the year in which programs achieve universal eligibility, will be costlier than the first two years, because all fee-paying private school students and homeschoolers become eligible for the state subsidy, dwarfing the number of participants who save the state money. Year 4, the first year after universal eligibility, should differ fiscally from Year 3. The only possible sources of new program participants will be public-school switchers, students relocating from other states, and rising kindergarteners. All the students in that set will save the state money by participating in the choice program instead of enrolling in public schools, except the relocators and rising kindergarteners who would have paid for nonpublic schooling absent the program. From the first year after universal eligibility on, new cohorts of choice participants will save state education dollars. If those new cohorts become large, the programs could change from costing the state money to saving the state money. The only program old enough to have experienced its first post-universal eligibility year is the Arizona ESA Program. Lueken (2025, 10) estimates that Arizona experienced a net cost of $37 million for the program that year, representing 0.2% of state and local spending on public education. Program critics claim that 80% of the participants were private school students and therefore are being newly subsidized by the state (Center for Evaluation and Education Policy 2023), but over 10,000 students in the universal program had switched from public schools into Arizona's previous choice programs. Yes, those students were enrolled in private schools in the prior year, but school choice had previously drawn them out of the public schools, saving the state money. The enrollment data are more complete regarding the ESA program in Arkansas, which is approaching Year 3 of program implementation and thus its first year of universal eligibility (Daniels et al. 2025; Johnson et al. 2025). In Year 1 (2023–2024), 1437 participants saved the state money because they were absorbed into the ESA program from the state's voucher program for students with disabilities, or switched from a public school, or were rising kindergarteners diverted from enrolling in public school.2 The “savers” composed 25.9% of program participants in Year 1, with the remaining 74.1% of participants representing newly subsidized private school students or “costers.” (Johnson et al. 2025) The state spent $34.9 million on the ESA program in Year 1, but the net cost to the state was estimated to be $31–$33 million due to savings from the modest number of “savers” participating. The net cost of the choice program in Year 1 represented 0.5% of total public education funding in Arkansas that year. In Year 2, a $97.5 million appropriation funded 14,388 students in the ESA program. Since 6489 (45.1%) of the participants were “savers,” the net cost of the program to the state was $75–$85 million, representing 1.7% of total public education spending that year (Johnson et al. 2025). In the current Year 3, the Arkansas program has universal eligibility. The state has appropriated $277 million to support the program, which could provide $7000 ESAs to 39,500 students in private or homeschools. If the entire appropriation in support of the program is spent in 2025–2026, it will represent approximately 5.9% of state public education spending. Under most conditions, universal private school choice programs are a net cost to states as they build toward and achieve universal eligibility. Still, they represent a small fraction of total K–12 spending even when they become large. Will universal voucher and ESA programs promote productive efficiency in K–12 education? The answer, at this point, is unknown. Choice programs have demonstrated mixed achievement effects along with mostly positive attainment effects, which generate economic and social gains for choice students and society (Patrinos and Rivera-Olvera 2025). Competitive pressure from choice programs tends to improve the productivity of affected public schools, to the benefit of public-school students and society. Choice programs require extra educational resources as they grow to and achieve universal eligibility, though they realize some cost savings along the way and could be net cost savers in the long run. Do universal voucher and ESA programs promote equity? Compared to what? Before there were any modern, private school choice programs, K–12 education was highly inequitable. Access to a high-quality public or private school depended heavily on family wealth (Wolf 2005). Families of school-age children paid thousands of dollars a month in premiums on their mortgages to live in residential areas zoned to high-performing public schools (Barrow 2002). As former Milwaukee Public School Superintendent Howard Fuller often says, “School choice is widespread in America, unless you are poor.” Will universal vouchers and ESAs create a more even playing field than what occurs in the absence of choice? Universal choice programs promote equity as they march toward universality. The Arizona ESA program was targeted to low-income students, Native Americans, students with disabilities, students in D- or F-rated public schools, students with severe disabilities, children of active-duty military, and children of first responders before it became universal in 2022 (EdChoice 2022). The 12,127 disadvantaged students in the program during the 2021–2022 school year had the advantage of “first movers.” They received access to their preferred schools of choice before advantaged students did. Most universal ESA programs have adopted similar eligibility limits as they scale, with over 48% of Year 1 and over 36% of Year 2 participants in the Arkansas ESA program having a disability (Daniels et al. 2025, 9). Universal ESA programs are attractive to students with disabilities because they provide them access to an education customized to fit their unique needs without the notorious proceduralism that dominates special education in the public-school sector (Wolf and Hassel 2001). An increased number of disadvantaged students will have access to school choice through universal programs, and they get to go first as those programs expand. Universal voucher and ESA programs advance social cohesion by reducing the compulsion attached to the residential assignment to public schools and replacing it with voluntaristic behavior. Horace Mann (n.d.), the father of the public school system in the United States, stated clearly, “We who are engaged in the sacred cause of education are entitled to look upon all parents as having given hostages to our cause.” That is hardly a vision for social comity! Forced segregation has typified public schooling in the Unites States throughout its history (Glenn 1988). From 1890 to the 1960s, the public school system in the South enforced strict racial segregation on students (Foreman 2005), while northern cities achieved de facto racial segregation in their public schools by redlining African American families into specific neighborhoods, concentrating their children in racially uniform schools (DeRoche 2020). Achieving cohesion within public schools also has been a challenge, as political battles have raged over public-school curricula, library books, and lessons on sexuality, race, religion, patriotism, and other contentious topics (Cato Institute, n.d.). In contrast, many European countries have achieved a high level of social cohesion by instituting universal private school choice. In Belgium, Denmark, Sweden, and the Netherlands, parents can choose the public or private school their child attends. Both public and private schools receive equal per-pupil funding from the state. Berner (2016) characterizes the European system of private school choice, which also operates in most parts of Canada and Australia, as “educational pluralism.” The Netherlands enshrined its universal school choice policy in its constitution in 1917, expressly to promote social cohesion (Vermeulen 2005). Citizens engage in more social cooperation if they have the ability to send their child to the school of their choice. Thus, it is unsurprising that a meta-analysis of the effects of private schooling on the civic outcomes of political tolerance, political behavior, political knowledge and skills, and community engagement concludes that private schools of choice tend to outperform government-run public schools in instilling these civic values in young people in the United States and globally (Shakeel et al. 2024). Universal voucher and ESA programs are still a new development in the United States. Much remains to be learned about their long-term effects. However, the available evidence—from both targeted U.S. programs and universal choice systems abroad—offers a promising foundation for evaluating these reforms. Applying Levin's (2001) framework, the evidence thus far suggests that universal choice programs meaningfully expand consumer choice, enhance productive efficiency under the right conditions, promote equity through phased implementation that prioritizes disadvantaged students, and have the potential to foster social cohesion by allowing families to select schools that their values within a educational these programs require public during their to long-term fiscal may improve as the of public-school switchers on student achievement effects on participants remains in the short but outcomes like high school graduation and college completion tend to be for students in private school choice programs to those in public competitive pressure from choice programs outcomes in public these the for universal private school choice is These programs families more in educational that fit their while the for a more and education should to expand and these their is and their are The evidence suggests that universal choice, when and is good