After-School Enrichment: A No-Brainer?

Most American school days end around three o'clock, and most workdays end around five or six. Families fill the gap however they can, while the country's school buildings, already paid for, sit largely empty from mid-afternoon until morning.

On October 12, 2026, the labor economist Kathryn Anne Edwards made the case in Bloomberg Opinion for free universal after-school programs: "In a world of hard problems, this is an easy win" (Edwards, Bloomberg Opinion). Nearly three decades of federal funding form a working laboratory, and they show that the afternoon can be among a country's better investments in its children or produce nothing measurable, depending on what fills it. The Accord's answer is to extend Cooperative Extension, the partnership of the Department of Agriculture, the land-grant universities and county governments that has served nearly every county since 1914 and runs 4-H. It would start with programs that already work, fund them fully, and grow only as fast as they produce results worth at least twice their cost.

The gap after the bell

In the Afterschool Alliance's 2025 survey of 30,515 parents, "only 6.95 million children (13%) are enrolled in a program," while the parents of another 22.6 million would enroll them if a program were available (America After 3PM, 2025). Cost is the main barrier, named by 56 percent of parents without a child in a program. Access follows income: children in high-income families are more than twice as likely to attend as children in low-income families, and the highest-income families spend about nine times as much per child on out-of-school activities. Sports follow the same pattern: in 2024, 58 percent of American children played organized sports, against 36.3 percent of children below the poverty line (Project Play). What fills the empty hours instead, from screens to inactivity, is measurable (note 1).

The buildings already exist

The country operated 99,409 public schools in 2022–23 (NCES). A school year of 180 days at 6.64 instructional hours comes to about 1,200 hours, or 14 percent of the hours in a year (derived; note 2). In the Wallace Foundation's study of 111 programs, space and utilities made up 13 percent of a program's full cost, so programs housed in schools start with that cost largely covered (Grossman et al., 2009). Private schools already treat the afternoon as part of education, and Phillips Exeter Academy counts a term on a school team toward its physical education requirement (Exeter).

The laboratory

The Nita M. Lowey 21st Century Community Learning Centers program (21st CCLC) is the only federal funding stream dedicated to after-school programs, at $1.329 billion in fiscal 2026 (Afterschool Alliance). The program's national randomized evaluation found that "treatment-group students did not have higher levels of academic achievement," along with more negative behavior in the program group (James-Burdumy et al., 2005).

Broader reviews explain why. Across 75 evaluations, programs using structured practices (sequenced, active, focused and explicit) raised achievement test scores by 0.20 standard deviations, while "other programs do not yield significant mean effects for any outcome" (Durlak, Weissberg and Pachan, 2010). A national program should pay for the gains that structured content produces, measured independently. Evidence on crime, attendance and community schools is in note 3.

What the returns depend on

Raj Chetty, John Friedman and Jonah Rockoff linked test-score gains to adult earnings (Chetty, Friedman and Rockoff, 2014). On their estimates, a lasting gain of 0.1 standard deviations is worth about $6,100 to $8,100 per child in today's dollars. A quality school-year program costs about $6,200 to $7,100 per full-time slot, so a year pays for itself through the child's own future earnings if it produces a lasting gain of about 0.08 to 0.12 standard deviations (all derived; note 4).

Sources for each row are in note 5. These ratios count only the child's own earnings. Mothers' work adds to them: a randomized evaluation of after-school care in Chile found that "program participation increases employment by 5%" (Martínez and Perticará, 2017), which would be worth a fifth to nearly half of a slot's cost if American mothers responded the same way (derived; note 6). Sports participation has measured long-run returns for girls, a music program improved self-control, and AI tutoring shows real gains smaller than some claims (note 7).

How other countries do it

American children already spend more hours in class than the OECD average, so the American gap is in the afternoon. Sources, with Japan, England and Chile, are in note 8.

Building on Cooperative Extension

Cooperative Extension was created by the Smith-Lever Act of 1914 to carry land-grant university research to farms and families. The Department of Agriculture administers formula funds, each state's land-grant university runs the state program, and county offices deliver it, "with an office in or near almost all the Nation's 3,000 counties" (NIFA). Each state must match its federal funds dollar for dollar (7 U.S.C. 343).

Extension is best known for agriculture, which still accounts for about 40 percent of its staff, but it already does youth work at scale (USDA ERS, 2026). 4-H reaches more than 6 million young people, most of them outside rural areas: 2.6 million rural, 1.8 million urban and 1.6 million suburban (NIFA, 2024). It runs after-school programs, including in all five boroughs of New York City and on military bases (note 9).

Two cautions apply. The evidence for 4-H youth outcomes comes from a study that compared members with young people who chose not to join, so it cannot separate the program's effect from the families who sign up. And Extension has shrunk: full-time staff fell an estimated 26 percent from 1979 to 2024, and youth work fell from 26 to 18 percent of staff. USDA traces the decline mainly to money, as federal funds shifted from unrestricted formula grants to restricted project grants and state and county budgets rose and fell with recessions (note 10). The lesson for this program is to fund it by formula, adequately and predictably, with a youth budget of its own.

Extending the Extension

The Accord would add a youth title to Smith-Lever: formula funding to each state's land-grant system for an afternoon program covering academics, sports, the arts, science and technology, and the humanities. Each field has a federal-state partner already in place. State arts agencies and humanities councils, which receive 40 percent of the grant funds of the National Endowments for the Arts and the Humanities, would supply teaching artists and programs in history, civics and reading. University departments and 4-H's robotics and computer science partners would supply science, local leagues and recreation departments would supply sports, and college students paid through Federal Work-Study would tutor (note 11). The Department of Education would be the second federal partner, handling school buildings and folding 21st CCLC into the new program.

The site and the circuit

Every participating school would have one site operator, chosen by the county Extension office for a multi-year term: the district, a YMCA or Boys and Girls Club, a recreation department or Extension staff. The operator runs the whole afternoon and must offer every field every week. A child never changes schools to reach a program, and a change of provider never removes a field.

A typical afternoon has three parts: homework help and tutoring, an hour with a rotating specialist, and sports, clubs or free play. A painter, a violinist, an engineer, a theater director or a soccer coach spends one afternoon a week at each of five schools, so every school sees a different specialist each day. Each million full-time slots needs about 25,000 such specialists in part-time positions, at a cost of about $630 to $790 per child per year, roughly a tenth of the per-slot cost (derived; note 12). The work puts the skills of artists, coaches, engineers, retirees and graduate students to use in the hours children are free. The land-grant universities would train and certify them, with background checks for every adult. An orchestra, a robotics team or a theater production, which needs more than one afternoon a week, meets at a regional hub with transportation provided.

Mentors and service

Professionals lead every site and teach the specialist fields, and trained mentors extend their reach in tutoring, reading and homework help: college students paid through Work-Study, community volunteers, older adults and teenagers. Volunteers can raise achievement when they follow a set method under supervision. In a randomized trial of Reading Partners, whose community volunteers tutor from a structured curriculum, more than 1,200 second- to fifth-graders gained 1.5 to 2 months of reading growth over the comparison group (MDRC, 2015). Mentoring that rests on the relationship alone has done less in school, with gains that faded within two years (note 13). Mentors would therefore work from accredited methods, with training and a professional on site.

Some 16- and 17-year-olds may prefer tutoring or coaching a younger child to playing in an orchestra. The program would offer mentoring as a committed track beside teams and ensembles, with training, pay or service credit, and no teen alone with a child. Evidence on teen tutors is thin, so mentoring models would be measured and accredited like any other, and mentors' hours count toward the community's contribution described below.

One pickup at six

Many public schools offer sports, but school teams serve the students who make them, and practice ends before most workdays do. Parents fill the gaps by driving between practice, lessons and care, and among parents whose children are not in a program, 49 percent cite the lack of a safe way to get there and back (Afterschool Alliance, 2025). Intramural play reaches more children: in one study, 35.9 percent of middle school students played school sports where intramural programs ran, against 27.3 percent where only interscholastic teams did (Kanters et al., 2013). The program would run intramural leagues at every site, count a team athlete's practice as that child's sport block, and keep every activity at the child's own school until six, so a working parent makes one pickup.

Measuring and paying for results

Outcomes are measured by independent evaluators, separate from the site operators and from the universities that design the program models, and every result is published. The National Statistics Board, the Accord's independent panel for measurement and publication integrity, then sets the program's priority. Its authority is to weigh the program's measured return against the Accord's other expenditures and the revenues that fund them, and to publish where the program ranks. It chooses no curriculum and selects no sites, and its process is evidence-based, methodology-audited and appealable.

Results are measured where the numbers can support them. A single school is too small to detect a gain of 0.1 standard deviations, which takes about 1,600 children in each comparison group (derived). Program models, meaning a curriculum and its training such as a tutoring or coaching method, run at hundreds of sites and can be measured by pooling (note 14). Site operators are paid for each verified attended hour and judged on what they control: attendance, retention, breadth of offerings, safety and family ratings. Models are accredited for expansion on measured gains. A model that shows gains is offered to more sites, and one that does not keeps its current sites without growing, so no field disappears from a school. Lotteries for oversubscribed programs supply comparison groups, and only measures validated against long-run outcomes count, because paying on short-run measures has been gamed before (note 15).

Rural and urban

Rural children have the least access. In a 2020 survey, 11 percent were enrolled and four more were waiting for every one enrolled, with transportation and the absence of any program as the leading barriers (Afterschool Alliance, 2021). Extension is strongest in rural counties, and there the school is often the only building that can hold the program. Rural circuits are longer, mileage and late buses are program costs, and specialties that cannot reach a small school weekly can arrive by live video with an adult on site.

Cities have many providers and no system. There the county Extension office would mainly coordinate, contracting Ys, clubs and arts organizations as operators and drawing specialists from a deep local pool on short circuits. Payments would be adjusted for local prices, and safe travel home after dark is part of every urban site plan (note 16).

What it would cost and return

The Accord would fund this program well and grow it on results, starting at the full quality cost of $6,200 to $7,100 per slot. Programs must earn their keep; people never have to earn their floor. This program sits on the first side of that line: its expansion is earned.

The rollout would be slow, as suits a candidate program. The first step would bring existing programs with measured results, such as 4-H after-school clubs, Boys and Girls Clubs, YMCAs and 21st CCLC sites, up to the full quality cost. The second would announce expansion grants for proposals built on this design: one site operator for each school, rotating specialists, trained mentors and one pickup at six. Each million full-time slots costs about $6.2 to $7.1 billion a year (derived), about five times today's 21st CCLC spending and about twice what Extension spends from all sources, so a youth program of that size would become the largest part of Extension (note 17).

The bar for expansion is a measured return of at least twice the cost. Counting only the child's own earnings, that requires a lasting gain of about 0.15 to 0.23 standard deviations a year; counting mothers' added work, about 0.12 to 0.21 (derived; note 18). The best evaluated programs sit at that line and generic programs fall far below it, so the bar admits only programs that perform like the best. Cohorts that clear it earn the next expansion. Sites and models that fall short keep running at their current size, because sites that run for years supply the long data series the evaluation needs, and they grow only when their results do. Over a decade or more, growth goes to the programs that work, and if results hold, the program approaches the 29.6 million children whose parents want one.

Who pays

Families pay no cash. A price cuts valuable and frivolous use alike, as the RAND Health Insurance Experiment found for medical care, and the harm falls hardest on the poorest (note 19). The program builds commitment through attendance and conduct instead. A child holds a seat by attending, and a seat left empty goes to the next child on a waiting list ordered by lottery. Teams, ensembles and mentoring tracks ask for a conduct agreement, and open drop-in time stays available to every child. Families are asked for time: an event each term and a check-in with the site, scheduled to fit shift work. The program encourages and counts that time and never makes it a condition of a child's seat.

The cash cost is shared two-thirds federal and one-third state, with the state share phased in. Communities contribute in kind: school buildings, including the custodial, utility and security costs of the added hours, along with transportation coordination and the hours of volunteers and mentors. The program asks for no local cash, because a cash requirement would leave it thinnest where local revenue is lowest, the inequity New Hampshire's Supreme Court found in school funding. With buildings contributed, the cash cost is about $6,200 per slot, roughly $4,100 federal and $2,100 state (derived; note 20).

Failure modes and mitigations

The evidence behind each row is in note 21.

Where it would sit in the Accord

The Accord's policy perimeter already lists after-school and summer care for ages 6 to 12 as its strongest candidate for inclusion, because "a school day is shorter than a work day and a school year is shorter than a work year." It asks for a slot gap, a cost estimate and a delivery rail before the boundary moves. This essay supplies all three, with Cooperative Extension as the rail. The age range would remain a parameter that the evidence sets, serving the ages where evaluations show a return and widening only as data and scholarship support it.

As a candidate program, it would sit in the Social Stack, the Accord's engine for supports across a lifetime, beside the childcare program for younger children. Expansion grants would be open to every district, because the gap between the school day and the workday exists in wealthy districts as well as poor ones, and a lottery among qualified applicants would set the order so that the rollout doubles as an evaluation. The measured return would show where the program adds most, since children who already have paid enrichment have less to gain (note 18). The Accord writes no curriculum and sets no wages.

Edwards calls universal after-school an easy win. The evidence says its return depends on what fills the hours and ranges from nothing measurable to more than twice its cost. That uncertainty is why the program remains a candidate, and it shows how the Accord considers any program: by measured benefit, with growth that follows the results.

Notes

1. Only 25 percent of high school students were physically active for at least 60 minutes every day in 2025 (CDC, Youth Risk Behavior Survey). In 2021, tweens averaged 5 hours 33 minutes of entertainment screen time a day and teens 8 hours 39 minutes (Common Sense Media). A Justice Department analysis found that "19% of all juvenile violent crimes occur in the 4 hours between 3 p.m. and 7 p.m. on school days," using data from 1991 to 1996 (Snyder and Sickmund, 1999). The income figures are 21 percent against 9 percent attendance and $6,588 against $734 in annual spending (America After 3PM, 2025); families who pay report an average of $124.10 a week.
2. Days and hours from the NCES Schools and Staffing Survey; 180 × 6.64 ÷ 8,760. Buildings see more use than instruction alone. Public schools cover about 8.1 billion gross square feet with a replacement value of $2.78 trillion in the 2021 national facilities report (21st Century School Fund).
3. A later randomized study of enhanced academic instruction in 27 centers found more math growth in the first year and no added benefit from a second (MDRC, 2009), and GAO found the Department "lacks useful data about whether the program is achieving its objectives" (GAO-17-400). For at-risk students, Lauer and colleagues found effects of 0.13 standard deviations in reading and 0.17 in math (Lauer et al., 2006). A systematic review found "a small but nonsignificant effect on delinquency" (Taheri and Welsh, 2016), and CrimeSolutions rates generic after-school programs ineffective for delinquency (CrimeSolutions). The quasi-experimental LA's BEST study estimated $2.50 in avoided juvenile-crime costs per dollar (Goldschmidt, Huang and Chinen, 2007), and RAND found that New York City's community schools improved attendance (RAND, 2020). Advocates report that 21st CCLC serves nearly 1.4 million students, about $950 of federal money each (Afterschool Alliance).
4. Chetty, Friedman and Rockoff found that a one-standard-deviation improvement in teacher quality, worth 0.163 (math) or 0.124 (English) standard deviations in elementary test scores, "raises earnings by 1.3% at age 28," and put the present value of lifetime earnings at age 12 at about $522,000 in 2010 dollars. Converting with the CPI (1.476) and assuming, as they do, a constant percentage gain across the working life gives $6,100 to $8,100 per 0.1 standard deviation. Their estimate links a one-year gain to adult earnings, so it already reflects fading test scores. The cost is Wallace's $4,320 per slot in 2005 dollars times 1.648, or about $7,100, and about $6,200 without the 13 percent space share.
5. 21st CCLC: James-Burdumy et al. (2005). Structured and other programs: Durlak, Weissberg and Pachan (2010), Table 3. At-risk students: Lauer et al. (2006). Tutoring: a meta-analysis of randomized trials found "tutoring programs conducted during school tend to have larger impacts than those conducted after school" (Nickow, Oreopoulos and Quan, NBER, 2020; published in the American Educational Research Journal, 2024, DOI), with J-PAL putting the two at 0.40 and 0.21 standard deviations. The Washington State Institute for Public Policy estimated $7.68 in benefits per dollar for out-of-school tutoring by adults, with a 94 percent chance that benefits exceed costs, an estimate archived in December 2024 (WSIPP). Across federal programs, "direct investments in low-income children's health and education have historically had the highest" returns (Hendren and Sprung-Keyser, 2020).
6. Chile's evaluation also found a 7 percent rise in labor force participation. Mexico's full-time schools raised mothers' labor force participation by 5.5 points (Padilla-Romo and Cabrera-Hernández, 2019); Germany's afternoon-care expansion "had hardly any effect" (Dehos and Paul, 2023); and the American national evaluation found a difference short of statistical significance. The calculation applies a five-point rise to women's median earnings of $1,089 a week (BLS), working half to full time.
7. Sports: "a 10-percentage point rise in state-level female sports participation generates a 1 percentage point increase in female college attendance and a 1 to 2 percentage point rise in female labor force participation" (Stevenson, 2010), for girls in the 1970s. Music: a randomized trial of Venezuela's El Sistema found modest gains in self-control and no full-sample effect on school outcomes (Alemán et al., 2017). AI: a six-week after-school GPT-4 tutoring program in Nigeria raised English scores by 0.23 standard deviations (World Bank, 2025; De Simone et al., Policy Research Working Paper 11125); a Harvard physics study found 0.73 to 1.3 standard deviations for single lessons on the researchers' own test (Kestin et al., 2025); an AI assistant for human tutors raised topic mastery by 4 points without a significant effect on end-of-year tests (Wang et al., 2025); and students with unguided GPT-4 access scored 17 percent lower once it was removed (Bastani et al., 2025). Bloom's two-standard-deviation tutoring result (Bloom, 1984) has never been replicated, and a third of a standard deviation is typical for intense, well-designed programs (von Hippel, 2024).
8. Germany (federal family ministry; Bundestag), with €3.5 billion in federal capital and operating aid rising to €1.3 billion a year by 2030. Sweden (Skolverket). Denmark (BUPL). South Korea (Financial News). Japan's municipal after-school clubs served about 1.6 million children in 2026, with 14,713 on waiting lists (Cabinet Office). England is building wraparound care from 8 a.m. to 6 p.m. at primary schools with £289 million in start-up funding, paid for by parents (Department for Education). The OECD counts 75 percent or more of children aged 6 to 8 in out-of-school care in Denmark and Sweden, against a European average of 31 percent for ages 6 to 11 (OECD Family Database). American children spend about 8,917 hours in compulsory classes through lower-secondary school, against an OECD average of 7,604 (Statista, citing OECD). Chile lengthened its school day by about 30 percent between 1997 and 2010 (Ley 19.532).
9. 4-H works through 3,500 Extension professionals and 500,000 trained volunteers (NIFA, 2024), and young people take part "through clubs, in school-based and after-school programs" (NIFA, 4-H). Cornell Cooperative Extension reaches families "in all five boroughs of New York City" (CCE NYC). On the Defense Department partnership, see NIFA, military youth programs. The system includes 58 universities designated in 1862, 19 in 1890 and 36 tribal colleges designated in 1994 (APLU).
10. The Tufts study followed more than 7,000 adolescents from 2002 to 2010 and found 4-H members "nearly 4 times more likely to make contributions to their communities" (University of Maine summary). Extension's measured returns come from agriculture, with estimated internal rates of return of 16 to 110 percent from statistical studies (Wang, Choices). The ERS report finds that federal funding shifted "from unrestricted capacity grants" to restricted grants, that several states consolidated county offices in the 1990s and 2000s, and that because salaries dominate budgets, "funding levels have direct implications for personnel number and programs." USDA's Extension funding fell from $1.3 billion in 1973 to $498 million in 2024, the federal share fell from 42 to 17.6 percent, and other sources made up about half the federal decline.
11. The NEA sends 40 percent of its grant funds to 56 state and territorial arts agencies and six regional organizations (NEA), and "almost every state has a grant program that supports paying teaching artists to teach in a learning space" (NASAA). The NEH sends 40 percent of its program funds to 56 humanities councils (NEH). Federal Work-Study requires each college to spend at least 7 percent of its allocation on community service jobs, including reading tutors, and allows a 100 percent federal share for tutors of elementary school children (Federal Student Aid Handbook).
12. Assumptions: a specialist leads two one-hour sessions of 20 children each afternoon, five afternoons a week, reaching 200 child-sessions a week, and is paid a loaded $40 to $50 an hour for three and a half hours a day, including travel and preparation. The cost is part of the per-slot cost.
13. In a randomized study of Big Brothers Big Sisters' school-based program, first-year gains in school performance faded, and "mentored students' performance declined, and became statistically indistinguishable from the non-mentored group" in the second year (Herrera et al., 2011). In a randomized trial of Experience Corps, in which older adults tutor early readers, 883 students showed an effect size of about 0.08, larger among those who received at least 35 sessions (Lee et al., 2012). The tutoring meta-analysis in note 5 found larger effects from teachers and paraprofessionals than from volunteers and parents. Reading Partners cost $3,610 per student in its trial, of which schools paid about $710. The Foster Grandparent Program already pays a stipend to adults 55 and older with limited incomes who serve children with special or exceptional needs (AmeriCorps Seniors). The Valued Youth Program pays secondary students at risk of dropping out to tutor younger children, and the What Works Clearinghouse found no study of it that met its standards (WWC).
14. The sample size comes from the standard power calculation at 80 percent power. Models with strong evidence already exist: Saga Education's tutoring raised math scores in a Chicago trial, though during the school day (Guryan et al., 2023); Houston schools assigned to partnerships with community arts organizations saw fewer disciplinary infractions, better writing and more empathy (Bowen and Kisida, 2023); and adventure programs of the Outward Bound type showed an average effect size of 0.34 (Hattie et al., 1997).
15. In the federal job-training program, "the short run measures used to monitor performance are weakly, and sometimes perversely, related to long run impacts" (Heckman, Heinrich and Smith, 2002), and gaming of the same incentives had "a negative impact on the true goal of the organization" (Courty and Marschke, 2004).
16. In the 2020 survey, 50 percent of rural parents cited transportation and 45 percent said no program was available. About 27 percent of public schools and 19 percent of public school students are rural (NCES; NCES). Boys and Girls Clubs report "nearly 5,500 Club locations" (BGCA, 2026) and the YMCA more than 2,600 locations serving 6.6 million young people (YMCA, 2025). Urban Extension practitioners have met as the National Urban Extension Leaders since 2013 (NUEL).
17. Smith-Lever's core extension funding is $325 million in fiscal 2026 (University of Maryland government relations), and Extension's total funding from all sources is about $2.7 to $2.9 billion a year in 2020 dollars (ERS), or about $3.4 to $3.6 billion in 2025 dollars. A million slots at $6.2 to $7.1 billion is 1.7 to 2.1 times that total and 4.7 to 5.4 times 21st CCLC's $1.329 billion.
18. The 2× bar applies the earnings link in note 4 to twice the per-slot cost. The mothers' work component uses the range in note 6. Part of the program's cost displaces spending that already occurs: families who pay an average of $124.10 a week spend about $4,500 over a 36-week school year. The return on the marginal dollar falls as coverage extends from children with the least access to families who already buy enrichment.
19. RAND's summary: "Cost sharing reduced the use of both highly effective and less effective services in roughly equal proportions," and the health gains from free care "were concentrated among the sickest and poorest patients" (RAND, 2006). In Kenya, demand for insecticide-treated bednets "dropped by 60 percentage points when the price was increased from zero to 40 Kenyan Shillings," and families who received nets free were as likely to use them as families who paid (Cohen and Dupas, 2010). A small charge for deworming drugs "reduced take-up by 80%" (Kremer and Miguel; published in the Quarterly Journal of Economics, 2007, DOI). The Accord's childcare plan for children under five keeps a family share, which the Universal Child Allowance covers. This program differs because its case rests on the child's measured gains, and a price would withhold them from the children with the least access.
20. The split is the Accord's choice. The federal government pays the larger share because the larger part of the fiscal return arrives as federal income and payroll tax on the children's adult earnings, wherever they later live, and because a larger state share would leave the program thinnest in the states with the least revenue. The state third gives each legislature a stake in quality, and phasing it in lets states budget for it. Extension operates in every state with a dollar-for-dollar state match. Head Start's federal share "shall not exceed 80 percent," and its local share "may be in cash or in kind, fairly evaluated, including plant, equipment, or services," with waivers where a community lacks the resources (42 U.S.C. 9835). On July 1, 2025, New Hampshire's Supreme Court held in the ConVal case that base aid of $4,182 per pupil fell short of the more than $7,356 an adequate education costs (Boston Globe). Arithmetic: $6,195 × 2/3 = $4,130 and $6,195 × 1/3 = $2,065. The buildings replace the 13 percent space and utilities share in Wallace's cost (note 4); the totals elsewhere in this post include that share.
21. Extension staff fell 26 percent and USDA's Extension funding fell by about 60 percent (note 10). Congress rescinded $1.1 billion for public broadcasting in 2025 and its corporation announced it would close (Michigan Public), and humanities council grants were terminated in April 2025 and later restored (Ohio Humanities). Ten states still decline Medicaid expansion more than twelve years after it began (KFF). The 21st CCLC evaluation and the gaming evidence are in notes 3 and 15.

Figures marked derived are calculated in this post from the cited sources and are not estimates made by those sources; the arithmetic and assumptions are in the notes. Effect sizes are average returns of evaluated programs, and a national program reaching children who already have access would earn less on its marginal dollar. The rural enrollment figures come from a 2020 survey, the 21st CCLC enrollment figure comes from advocates, and network sizes are the organizations' own figures. Cost totals are gross and unscored.