Real Spending Per Pupil More Than Doubled Since 1970, So Where Are the Gains

If you doubled the budget of any business in America and the product stayed the same, the board would demand an explanation. If you doubled it and the product got worse, someone would be fired. In public education we double the budget, hold a press conference about how much more is needed, and ask the taxpayer to double it again.

Here is the claim I intend to prove. Since 1970, inflation-adjusted spending per public school student has risen dramatically, while the long-term trend scores of 17-year-olds, the students nearest the finish line, have barely moved. That is the strongest single piece of evidence that how money is spent matters more than how much. I will build the argument from two public data sources, show the arithmetic, give the best case for more spending its due, and answer it.

Two sources, and why they are the right ones

The first source is the Digest of Education Statistics, published for decades by the National Center for Education Statistics in the Department of Education. It is the standard reference for school finance. Among its tables are series showing current expenditure per pupil in public elementary and secondary schools in constant dollars, which means adjusted for inflation so that a dollar in one year buys the same as a dollar in another.

The second source is the long-term trend assessment of the National Assessment of Educational Progress, which I explained in What the Nation's Report Card Measures and What Proficient Really Means. The long-term trend assessment is built to measure the same skills with the same kind of questions over decades. It tests students at ages 9, 13 and 17. I focus on age 17 because that is the outcome of the whole system. A reform that worked would show up there.

I will tell you plainly what I am not doing. I am not quoting every year's figure from memory, because the exact numbers depend on which price index the Digest uses in a given edition, and they get revised. Pull the current table yourself. The shape of the data is not in dispute, and it is the shape that matters.

The first line of the argument: spending

In round numbers, the Digest's inflation-adjusted series shows per-pupil spending in the early 1970s at a level that, in current dollars, was a bit more than half of what it is now. I will use a simple labelled illustration so the arithmetic is visible. Suppose, for illustration only, that the 1970 figure was $7,000 per pupil and the later figure was $15,000 in the same constant dollars. Those are not the table's exact numbers. They are round numbers chosen to show the method.

• Later figure divided by earlier figure: $15,000 / $7,000 = 2.14. That is a 114 percent increase.
• Increase in dollars: $15,000 minus $7,000 = $8,000 more per pupil, per year.
• For a classroom of 25 students: 25 x $8,000 = $200,000 more per classroom, per year, than in the 1970 baseline.

Whatever the exact figures in the edition you open, the pattern is that real spending per student more than doubled over the half century. That is the first premise, and no serious analyst disputes it. The people who disagree about education spending usually disagree about why it rose, not whether.

The second line: results

Now the product. The long-term trend reading results for 17-year-olds began in 1971 and the mathematics results in 1973. In reading, the average score for 17-year-olds in 1971 was 285 on the 0-to-500 scale. By 2012, it was 287. A gain of two points across more than four decades, in which spending per pupil roughly doubled, is statistically indistinguishable from nothing. In mathematics, the 17-year-old average was 304 in 1973 and about 306 in 2012, with a similar story.

After 2012, the 17-year-old scores fell. I will leave you to look up the later results, because the post-2020 declines raise questions about school closures that are covered in What the Long School Closures of 2020 and 2021 Taught Parents. For this argument, I deliberately end at 2012 so that no one can say I used the pandemic to rig the comparison. Even before that shock, the picture was flat.

Notice what the younger groups show. Nine-year-olds and thirteen-year-olds gained more over the same decades, particularly in mathematics. That is a real finding, and it is the opponent's best evidence, so I will not hide it. But it carries a problem. If early gains were real and lasting, they ought to show up at 17. They largely do not. Something in the middle and upper grades erases them, or the gains were never as deep as they looked.

Here is the second premise, then. Over the period that spending per pupil more than doubled, the scores of 17-year-olds were essentially flat.

The third line: the arithmetic of a flat result

Put the two premises together. If the scores went from 285 to 287 while real spending per pupil rose by, say, 114 percent in my illustration, then the extra spending bought about 2 points of improvement. Divide the extra $8,000 per pupil by 2 points and you get $4,000 of additional annual spending per point. I want to be careful about this: it is a rhetorical calculation, not an economic law. Spending does not buy scores in a straight line. But it shows how far the result is from anything a business would accept as a return.

Here is the conclusion the data supports, in numbered form:

1. Real per-pupil spending rose steeply from 1970 on, by more than a factor of two.
2. Seventeen-year-old reading and mathematics scores on the long-term trend were roughly flat through 2012.
3. Therefore, whatever drove the spending increases, it did not drive the outcomes that the spending was supposed to improve.
4. Therefore, "we need more money" cannot, by itself, be a sufficient answer to a question about results.

I did not say money has no effect. The claim is narrower: the aggregate relationship between more dollars and better scores, across the system, over fifty years, is weak to nonexistent.

Steelmanning the case for more spending

The best case for more money is stronger than its slogans, and it deserves a fair statement.

First, spending is not the same as resources in the classroom, and the aggregate numbers hide enormous variation. Some districts spent much more than others, and a poor district's added dollars may have gone to cover real costs, such as special education mandates, health insurance and pensions, that did not exist in 1970.

Second, the students changed. Schools today serve a larger share of children with identified disabilities and a larger share who are learning English. A flat score is not the same thing as a flat school, if the school is educating harder-to-teach students than before.

Third, and most seriously, there is real research on spending that shows an effect. In 2016, economists Kirabo Jackson, Rucker Johnson and Claudia Persico published a study in the Quarterly Journal of Economics using court-ordered school finance reforms from the 1970s and 1980s. They found that for children from low-income families, a sustained increase in per-pupil spending during the school years was associated with more completed schooling, higher adult wages and a lower chance of adult poverty. That is a serious paper by serious scholars, and conservatives who dismiss it do their cause no favors.

The result for poor districts is the claim that most deserves an answer. Here it is.

The answer: how much versus how it is spent

I accept the findings I just described, and I do not think they contradict the argument. They sharpen it. Here are four points.

One: the effect is conditional on what the money bought. The Jackson study found gains when spending went to things that plausibly help students: lower student-teacher ratios, longer school years and higher teacher salaries. The reforms that worked were not a blank check. The question is not whether any dollar helps. It is where the dollar goes. A dollar spent on a capable teacher in a poor school is not the same as a dollar spent on a new administrative layer.

Two: the best-funded experiment in American education found no miracle. In the 1980s and 1990s, a federal court in Kansas City, Missouri, ordered a desegregation remedy that included a vast capital and operating program, with new facilities that included a planetarium, a television studio and a swimming pool built to Olympic size. Spending in the Kansas City district rose to among the highest of any large urban district in the country. The Supreme Court heard the dispute in Missouri v. Jenkins (1995), and by the time the court's supervision ended, the achievement gains the remedy was meant to produce had not materialized. I do not offer this as proof that money never matters. It is a warning that spending without a theory of improvement fails even when the spending is lavish.

Three: costs that rise on their own do not buy more instruction. A large share of the increase went to things that did not put more teaching in front of children. We take that up in The School Staffing Surge, Hiring Outpaced Enrollment for Decades, where the ratios of non-teaching staff to students tell much of the story. Part of the answer is mandates, which we should be honest about: federal special education law, beginning with the Education for All Handicapped Children Act of 1975, put real obligations on schools, and those obligations cost real money. Other costs came from health and retirement benefits, which grow faster than anything a student sees on Monday.

Four: the incentive structure does not reward results. In most districts, a teacher's pay is set by years of service and degrees held, not by how much students learn. We looked at what that looks like in the contract in What a Teachers Union Contract Says and Why Every Parent Should Read One. When the system pays for inputs, not outputs, more money buys more inputs, and we should not be surprised when outputs stay put.

What the evidence says about where to spend

I do not want to leave this at a complaint. If how the money is spent matters, then what should we spend it on?

The evidence points to a few places worth a legislator's attention.

• Teachers. Researchers who study teacher effectiveness consistently find that the variation among teachers is large, and that a good teacher moves a class farther than any other single school factor. A system that cannot pay its best teachers more, or retain them in hard-to-staff schools, is wasting its most valuable resource. The salary schedule rewards seniority, not skill.
• Early reading. The research summarized in The Reading Wars, How Whole Language Lost a Generation and Phonics Won Back shows that a change in method, not a change in budget, moved scores in Mississippi, and did so in one of the poorest states in the country.
• Tutoring and time on task. High-dosage tutoring, in which a small group of students works with a trained tutor several times a week, has produced some of the larger measured gains in education research. It is a use of money with a direct line to a child.
• Choice. When a family can take a share of the funding to a school that serves their child better, the school has a reason to compete. We look at that argument in Tax-Credit Scholarships, the Choice Program Built on Private Donations and in other posts in this series.

The conservative principle underneath

Why does this matter beyond education? Because it is a test case for a principle conservatives have defended in every part of government: spending is a measure of effort, not of success. When a program's defenders cite how much it costs as proof that it cares, they have confused the input for the output.

The Founders understood this. Hamilton's argument in the Federalist for a government with the means to meet its ends was always joined to an insistence on accountability, in the people's power to remove those who fail. A taxpayer is entitled to ask what he bought. When the answer to "what did we get for doubling the budget" is a shrug and a request for a bigger budget, the question has not been answered.

Look at the arithmetic of a public budget on the national debt clock and on the numbers page, and remember that every dollar spent on a program that does not work is a dollar taken from a family that could have used it.

An objection about fairness, and a reply

One more objection deserves an answer, because it is made in good faith. Some say that holding schools to results punishes the schools that serve the poorest children, who need the most help.

The reply is that it works in exactly the opposite direction. The families who can least afford a failing school are the ones who cannot buy their way out of it. A system that is shielded from the question "what did you accomplish with this money" fails them first. Accountability for results is the poor child's best friend, and a refusal to measure is the friend of the institution.

The case in brief

1. Inflation-adjusted spending per pupil more than doubled after 1970, according to the National Center for Education Statistics.
2. Long-term trend NAEP scores for 17-year-olds were essentially flat through 2012, with reading at 285 in 1971 and 287 in 2012.
3. The best evidence for more spending, the Jackson, Johnson and Persico study of school finance reforms, finds gains when money goes to classroom inputs that help children, and that is a point about how money is spent.
4. The Kansas City desegregation spending of the 1980s and 1990s, and the growth of non-teaching staff, show that dollars without a theory of improvement do not raise scores.
5. The right question for any budget is not how much, but what it bought, and the answer should be measured against a yardstick the schools cannot move.

What you can do: open your school district's published budget, find the share spent on classroom instruction, and ask the board at its next meeting what that share bought.