The State of American Schooling

How many kids ~60 million

There are roughly 60 million young people in school in the United States, between the ages of about three and eighteen.¹

Where they go to school 80% district

About 80% of them attend traditional public schools, and another 7% or so attend charter schools.² The remaining 10% attend either private schools or homeschools, with the great majority in private schools.³

From 2019-20 to 2024-25, district enrollment fell 4.4%, a loss of 1.9 million students, while charter enrollment grew 14.7% over the same five years, a gain of 492,000. Homeschool enrollment grew another 4.9% in 2024-25 alone, double the pre-pandemic rate of 2%, and 36% of states hit their highest ever homeschool enrollment.⁴

The homeschool total is genuinely contested. Estimates run from about 5.2% of school-age children on the NCES figure to 6.3% on NHERI's, and the spread is mostly an artefact of what each instrument counts: NCES counts students and excludes full-time virtual school, while the Census Household Pulse counts households with at least one homeschooled child, which reads structurally higher.⁴

Private enrollment is the hardest to pin down, though Census data shows it growing 28% against pre-pandemic numbers.⁴

Governance and funding $1T a year

Total government funding into public preK-12 education is just over $1 trillion a year, with roughly 88% coming from state and local government and the other 12% from the federal government. Given that states and localities make most of the calls on how schools operate, that split makes sense.⁵ The money moves through roughly 13,000 school districts sitting under 50 state legislatures and 50 state education agencies.⁶

For traditional public schools, funding is roughly 43% from the city, 45% from the state and 12% federal, although this varies significantly by state and district.⁷

Charter schools also use this government money in the form of per-pupil payments, but must fundraise for the rest of their expenses, including facilities and materials.⁸ Private schools are mostly funded by tuition, and homeschools are of course funded by the home.

Funding trends 26x since 1920

Funding has risen consistently since the 1900s. In 1920, per-pupil spending in today's money was just below $700. Most recently, in 2024, it was tracked at $17,619 per pupil.⁹

One current change here is education savings accounts and vouchers. Vouchers allow tax money that would go to school districts to be directed instead to other educational institutions by parents, and ESAs are state-funded education savings accounts directed by parents in the interest of their child's education.¹⁰

School expenses and spending 79% people

The average student cost is $17,619 on the 2024 Census figure, though this varies by state and city — Los Angeles and New York City run roughly $25,000 and $35,000 respectively.^{11 12}

Schools spend on average 79% of their money on salaries and benefits, with the next most significant cost being contracted services like transportation.¹³

Student academic performance PISA, TIMSS, PIRLS, PIAAC

When you look into student learning performance, no scores are standardised. Not only are there over fifty different tests in use,¹⁵ but many of the methodologies used to build them are disagreed upon. There is also no science establishing the upper limits of children's abilities, because we tend to measure what children do rather than what they can do.¹⁶ What I will be using instead are tests set relative to other countries, which at least compare different education systems against each other — though most of them do not vary that much.

These tests are either done by age, as with PISA, or by years of schooling, as with the rest.¹⁷

PIRLS tests reading in students in their fourth year of schooling. The most recent data is 2021, where the US placed 4th out of the 32 countries that participated with a score of 548. The centre of the scale is 500.¹⁸

TIMSS tests maths and science in the fourth and eighth years of schooling. For fourth-grade maths, the most recent data is 2023, where the US placed 22nd out of 63 with a score of 517, in a range running from 362 in South Africa to 615 in Singapore.¹⁹ For the science portion the US ranked 12th of 63 with a score of 532.²⁰ On the eighth-grade test the US scored 488, placing 22nd out of the 44 countries that participated.²¹

PISA is done with 15-year-olds around the world. The most recent data is 2022, where US maths scored 465, below the OECD average of 472, placing 26th out of 81 countries.²² In reading the US placed 6th of 81 with a score of 504 against an average of 476,²³ and in science 10th of 81 with a score of 499.²⁴

PIAAC is done with adults from 16 to 65. The most recent data is 2023, where the US on average placed 25th out of 31 in numeracy and close to the average in literacy. On a scale out of 500 the US scored 258 in literacy against an average of 260, 249 in numeracy against an OECD average of 263, and 247 in adaptive problem solving against an OECD average of 251.²⁵

Significantly, 34.2% of US adults are at or below Level 1 in numeracy, the fifth highest share out of the 31 countries tested.²⁶ For scale, Finland leads everything with 296 in literacy, 294 in numeracy and 276 in problem solving, while Chile sits at 218 in literacy.²⁷

As for young people, the 16-24 band performed better than older Americans in literacy with a score of 261, ranking 22nd of 31, but significantly worse in numeracy with a score of 248 and a rank of 29th of 31. No adaptive reasoning score is available for that band.²⁸

On every test the US either did worse than when it was measured before, or stagnated. On literally no international test did the US, at any age, improve significantly.²⁹

The distribution of the losses what averages hide

There is a clear pattern in the distribution of performance that I would be remiss not to mention, though it is not at all surprising to me, and that is the loss within certain groups.

Based on the data, children in the top 10 percentiles, although still performing worse, lost less performance than those in the bottom percentiles. The gap widens here as well.³⁰

Time in school 20% of a childhood

Although there is some variation between states in how much time children spend in school, it is generally around 178 days and 1,227 total hours in school, of which typically about 1,000 hours is actual instruction.³¹ The lowest day count is Colorado at 160 and the highest is Kansas at 186.³² Districts typically determine how much of that time is instructional, which can vary depending on recess, assemblies and other draws on time.

The split of instructional time is heavily favoured towards reading, writing, maths and science, at close to 50% of the time spent on those subjects alone.³³ To put that in perspective, it varies a lot by country, with places like Japan and Israel opting for more instructional days but fewer hours per day — around 200 days at roughly four hours.³⁴

Nonetheless, a child by the age of 18 in the US will have had about 105,120 hours of awake time, assuming eight hours of sleep. That child will have spent about 20,579 hours on school work. Which means that when they graduate at 18 they will have spent 20% of their awake time on school.³⁵

Additionally, we know breaks are important to learning, so I did some research on average recess times in the US. The average amount of time spent at recess, based on surveys, is likely only about 25 minutes. Some states mandate minimums of only 20 minutes, and many have no mandate at all.³⁶ Twenty minutes out of an average six hours of schooling.

Chronic absenteeism 22%

Another piece in the current state of education is absenteeism. Are children attending school consistently?

The answer is that almost a quarter, based on the last available data, are not. We know from PISA and other studies that being absent correlates with lower grades, and it is also easy to see the connection between a student's interest and engagement in school and how much they attend.³⁷

Data from AEI and RAND shows that in the 2018-19 school year chronic absenteeism, defined as missing 10% or more of school days, was 15%. Over the following two years during COVID it climbed to 28% and then 26%, before gradually decreasing to 22% in the 2024-25 school year.³⁸

Teachers 1 in 8 uncertified

Another thing I want to look into is teachers. Their salaries are a huge percentage of expenditure in schools, and their teaching methods have huge potential to affect students. They deserve some look into how they are doing and where they are coming from.

After the research, none of this surprised me. Three in four teachers are women, and the average age is 42 with a bimodal distribution — many later in their career and many early.³⁹

Average starting salary is $48,000 and typically works its way up to around $70,000, plus solid benefits and pension. The caveat is that qualified teachers, meaning those with a degree, certifications and training, still make about 17% worse money including benefits than their peers with the same qualifications.^{40 41}

The number of unqualified teachers is now one in eight around the country.⁴²

A note on qualifications. There is little evidence to support the idea that teachers with qualifications achieve better performance in students than uncertified teachers. What is typically shown to move the needle for traditional teachers is subject-matter knowledge and years teaching, rather than licences or degrees.⁴³

Turnover is around 16%, with 8% every year leaving the teaching field entirely.⁴⁴ Vacancy is around 1-2%, which is a healthy rate, and it costs between $12,000 and $25,000 to replace a teacher.⁴⁵

Phonics 2013 onward

Another trend is phonics. From 2013, when Mississippi first implemented policies to use phonics in schools, their children's literacy rank improved dramatically in NAEP, showing significant gains from the beginning of its use up until 2022, where the gains in performance stagnated somewhat.⁴⁶

Since Mississippi's dramatic rise in the rankings, 43 states have implemented laws to increase the use of phonics in schools within their jurisdiction.⁴⁷

Grade inflation GPA up, ACT down

Then there is the separation of GPA from actual performance. One proof of this is ACT's analysis of four million students across 4,800 schools around the country, which found that while students' maths GPAs rose from 3.02 to 3.32, their ACT performance fell from 20.8 on average to 20.3.⁴⁸

A separate and independent proof is NAEP, which shows student performance on its tests falling while GPA rose. To say it plainly, A students in the past performed better than A students of the present.⁴⁹

AI and tech adoption 2,982 apps

I would be remiss to miss mentioning AI and technology adoption in schools today, and its effects.

The average school district uses 2,982 different apps in their learning.⁵⁰ Specific to AI, the usage numbers show anywhere from 54% to 80% of teachers were using AI in 2025, and around the same among students.⁵¹

In terms of hardware, 38 states and many districts have banned phones in school, but still use Chromebooks and other laptops.⁵² Adoption of 1:1 device programmes with laptop-type hardware is close to 88% in high school and around 80% in earlier years.⁵³

Birthrates and the future of school 2028 - 2035

Lastly, I want to talk about the future of school, and the fundamentals of any market or space: supply and demand. At the most basic level there are fewer children now than there were a few years ago, and that trend is continuing downward not only in the US but around the world, with places like South Korea at 64% below the replacement rate.⁵⁴

While the US population continues to hold at healthy numbers, that is only because of immigration of adults rather than more children being born.⁵⁵ In the ten-year span from 2010 to 2020, the number of children under five went from 20.2 million to 18.4 million.⁵⁶

Class sizes through hiring have taken a clear trend, from a 27:1 student-to-adult ratio in 1950 to roughly 15:1 now in traditional schools — but as enrollment falls, hiring must fall as well.⁵⁷ Studies show some effects from ratio differences, especially at the lower end, but the ratio is not near enough to cause significant positive gains.⁵⁸

It is worth putting this next to the switching story from earlier, because they are the same story. District schools lost roughly 1.9 million students. Charters picked up about 490,000 of them, and private schools some further share. What is left over after that is not switching at all — it is children who were never born. That distinction decides which problem districts actually have: if the losses were pure switching, they have a competition problem, and a better product wins the students back. If a large share is demographic, they have a structural problem that no reform fixes.

The falling birthrate is also affecting enrollment in schools. In 2019 enrollment peaked at 50.8 million children in public school; as the years continued, enrollment dropped off to 49.4 million in 2024 at the most recent count.⁵⁹ It is forecast to drop 8% by 2030,⁶⁰ and because there are fewer people having fewer babies, this will compound.

The problem here is that children are revenue for schools. If schools are losing full percentages of revenue, they must also cut full percentages of expense, which is considerably harder with facilities and other static costs. Additionally, there continue to be more children with disabilities and more children who need to learn English, increasing the cost per student for every district where that is true.⁶¹

The cliff is also less hypothetical than the projection makes it sound, because it has already landed in the lower grades. Since fall 2019, PK-8 enrollment has fallen 4.5% to 33.9 million, while grades 9-12 actually rose 2%, from 15.2 million to 15.6 million.⁷⁵ The high schools are still full of children born before the decline. The elementary schools are not, and that gap works its way upward one year at a time.

These problems are a few years out for the US, so why mention them? Because of what is happening in Asia right now.

In Japan, 1.2 schools a day are closing, and the country has been below a replacement birthrate since 1974.⁶² This problem gets worse, and we know what happens there. Korea is the same. There are significantly fewer students, and the number of schools enrolling zero first graders is up 81%.⁶³

Summary and insights the takeaways

As a reflection, I see much of this as negative. As an optimist and an entrepreneur, I see it as a huge opportunity.

• Time is a tax, and it is paid disproportionately. Taxes are not the only thing paid disproportionately by the middle and lower classes, due to their lower status and know-how with tax offsetting. Time, through school, is also paid as a tax, at close to the same total effective rate — 20-30% of a child's life, or 20-30% of an adult's income.⁶⁴
• The report card no longer describes the product. It is clear to see that while GPAs are improving, the actual academic performance of US students is decaying, which puts a lot of weight on reporting systems that are not carrying it.⁶⁵
• Learning science has traction, not just potential. Implementation of learning and motivation science into the school system has huge potential, because of the way the use of phonics in Mississippi inspired action across the nation.⁶⁶
• Teachers are not one of the worst paid professions. That myth is disproven by this data. While they are making worse money on average than their peers from college, they are making about the average wage in many places.⁶⁷
• The system has no quality feedback loop. It cannot yet improve, because it is not even measured effectively. There is no test applied nationally by our country, and the current systems like GPA are inflated. Critics may say NAEP is that test, but its standards are both controversial and hold no backing. Although it is federally run to balance the states, it carries no consequences and is only ever a sample of students rather than everyone.⁶⁸
• We are roughly a generation behind Korea and Japan. Around 2028 to 2035 is when enrollment decay in US schools will accelerate to more obvious effect, as it is in Asia. That makes the most volatile period in the history of America's education system one that is coming soon — and makes now one of the greatest times to start building, as the market softens to better solutions.⁶⁹
• People are already leaving. Absenteeism and private enrollment are increasing similarly, which signals disengagement from the current system.⁷⁰
• Satisfaction is falling and merit is being noticed. 35% of people in the US are satisfied with the public education system, and it is decreasing: 8% fewer people are satisfied with public school this year than in 2024, per Gallup.⁷¹ More and more people are realising that competing for merit matters.⁷² And there are huge advances in our understanding and ability to teach children well, given that learning science really began roughly only 60 years ago.⁷³
• More money has not bought better results. Even as we put more money into the education system, our scores continue to fall.⁷⁴

That last one is the one I keep coming back to. It means the system is broken. Just like the machine of a car that many of us drive, at some point it is best to call it a total loss.

Last but not least, we have a chance to change this course. While birthrates are falling and our students are less capable, there will be greater and greater malleability in the education system.

In changing our education system, we have the opportunity to uproot the practices that got us here. What has gotten us here cannot get our society out.