Book | Wages Incomes and Wealth

Where’s the Money Gone? Changes in the Level and Composition of Education Spending

 Download PDF

Share this page:

1995 | EPI Book

Download the full text of this study (Adobe Acrobat PDF).Adobe Acrobat / PDF
Read the follow-up to this study, Where’s the Money Going?  

Where’s the Money Gone?
Changes in the Level and Composition of Education Spending

by Richard Rothstein with Karen Hawley Miles

Executive Summary
There is a widespread perception among the general public, policy makers, and even education professionals of a “school productivity collapse.” It is based on the belief that real elementary and secondary education spending per pupil has doubled in the last quarter century, and that school outcomes have not improved to justify these new investments.

This commonplace view is flawed. First, it is based on inappropriate conversions of nominal spending growth to real, inflation-adjusted growth of school spending. With appropriate inflation adjustment, it appears that total real education spending per pupil increased by 61% from 1967 to 1991. Admittedly, this is a substantial increase, but it is much less than the “doubled” spending commonly assumed to have occurred.

Second, this view assumes that all school spending aims for a single outcome-improved academic achievement of regular students. But schools actually seek a variety of additional outcomes as well: training of the disabled, student health and nutrition, vocational education, assimilation of the non-English speaking, etc. Rather than looking only at test scores, evaluation of the effectiveness of schools should match the growth of spending “inputs” in clearly distinguished school programs to the outcomes each of these programs is attempting to improve.

A detailed examination of expenditures in nine typical U.S. school districts shows that the share of expenditures going to regular education dropped from 80% to 59% between 1967 and 1991, while the share going to special education climbed from 4% to 17%. Of the net new money spent on education in 1991, only 26% went to improve regular education, while about 38% went to special education for severely handicapped and learning-disabled children. Per pupil expenditures for regular education grew by only 28% during this quarter century-an average annual rate of about 1%.

In regular education, per pupil spending on teacher compensation grew by 23% over the 25-year period. Average salaries rose because, by 1991, teachers had greater experience (age) and credentials than in 1967 and thus had higher average placements on salary schedules. The salary schedules themselves, however, did not generally require higher pay for the same levels of experience and credentials. Per pupil spending on teacher compensation also grew as a result of more intensive staffing-in particular, the hiring of more resource- and subject-specialist teachers.

Recent research on academic achievement suggests that outcomes in regular education programs have been stable or have grown modestly, and that outcomes for minority students have improved over the last quarter century. This report does not address the validity of that conclusion, but rather shows that regular education spending also has increased modestly. We do not claim that further spending increases will generate outcome improvements, nor do we argue that school reform is not important. Our findings suggest only that reforms are not likely to be well-designed if reformers, failing to examine the varied rates of spending growth in education’s many programs, assume an unproven collapse in school productivity.

When Benno Schmidt resigned Yale’s presidency to head a private school network, he explained why he had given up on public schools: “We have roughly doubled per pupil spending (after inflation) in public schools since 1965,” but the “nation’s investment in educational improvement has produced very little return” (Schmidt 1992).

This is a conventional claim of public school supporters and critics alike. School finance expert Allan Odden notes that “real” education expenditures increased by 58% in the 1960s, 27% in the 1970s, and 30% in the 1980s, “but student performance-and thus education productivity-have not improved that much” (Odden 1992, 10-11). According to a Brookings Institution report by John Chubb and Eric Hanushek, “since the Soviets launched …Sputnik…real expenditures per student rose at an annual rate of three and three-fourths percent, nearly tripling between 1960 and 1988…Spending has nearly tripled and performance has dropped” (Chubb and Hanushek 1990). “Look at spending on public schools,” advises neo-conservative author Irving Kristol. “It goes up and up and up and the results go down and down and down” (DeParle 1993).

This declining “productivity” claim is so well established that few analysts have sought empirical verification for it. Rather, the notion is but prelude to reform prescriptions; if, after all, everyone knows that growth in public education spending has outpaced any rise in school achievement, the challenge must be to design systems that use money more effectively, and there is no need to consider proposals for additional funds. Even many of those committed to public education assume that public education funds are not being used effectively and that reform is urgently required to halt a scandalous waste.

This study re-examines the apparent consensus that real school spending has roughly doubled in the last quarter century. It finds that, while spending has risen substantially, the increase is both smaller and more complex than most assume:

Real school spending increased by 61% from 1967 to 1991, 40% less than the real growth conventionally assumed.

Only about one-fourth of this increase was directed at “regular education,” the traditional school activities whose outcomes can be measured in test scores, graduation rates, etc.

This report concerns “inputs” only-expenditures for education-and makes no contribution to literature on school outcomes or productivity. However, the prevailing view among education researchers seems to be that academic outcomes have been mostly stable or have grown modestly, particularly because minority student outcomes have improved, closing some of the gaps with whites (see, for example, Koretz 1986 and 1987; Grissmer et al. 1994). If claims of modest improvement are accurate, then it may be that both regular education inputs and outcomes have grown modestly.

Productivity measures are based on two components-what inputs are utilized, and what outcomes result. Inputs are often measured as dollars spent per pupil. Measuring outputs is more complex, since schools perform a variety of functions. Despite public fascination with test scores, education researchers and policy professionals lack agreement about appropriate definitions of outcomes and how well test scores or other indices measure them. How should we measure children’s health to ask whether school breakfast and lunch funds are well spent? What outcomes do we seek for severely mentally retarded or for trainable mentally handicapped children to evaluate the effectiveness of spending growth in special education? Do low-income student’s graduation rates respond sufficiently to investments in compensatory education or dropout prevention programs? What academic progress do we expect, and after how long, for immigrant children whose families arrive with varied levels of literacy? How do we judge the goals of vocational education?

To understand school spending and to determine whether school program productivity is truly declining (i.e., whether additional dollars have been well used), expenditures must be linked to the program and to the specific outcomes the spending w
as designed to enhance. If, for example, schools have used much of their new money to improve training of mentally handicapped youngsters, it would make no sense to judge the effectiveness of this spending by whether SAT scores improved for the college bound.

This report differs from conventional analyses of changes in school spending in two ways. First, it uses a more appropriate method to adjust for inflation. Second, because public education consists of many programs with a variety of goals, this report examines changes in school expenditure by program (e.g., regular education, special education, transportation, school lunches, etc.), not by function (instruction, support services, etc.) or by “object” (salaries, supplies, etc.), as is commonly done. A list of programs examined in this report is shown in Figure 1.

Expenditures are examined for nine school districts in 1967 and 1991. The districts were chosen to be illustrative of U.S. districts in size, expenditure growth over the period, geographic location, urbanicity, minority enrollment, and relative affluence (see Appendix 2). Because neither school districts, states, nor the federal government tracks expenditures by program, this study required a research team to examine expenditure records for each of the nine districts. Every expenditure was categorized by the program it was designed to support.

Over any period, expenditures may rise solely because of inflation, or they may rise because more goods and services are purchased (or because of a combination of both factors). When, for example, Benno Schmidt claimed that “real” per pupil spending doubled from 1967 to 1991, he meant that schools now utilize twice the inputs they utilized a quarter-century earlier. To say this, however, requires a calculation of what it would have cost, in 1991, to purchase exactly those inputs (teachers, textbooks, administrators, etc.) that schools purchased in 1967. Only if this cost of 1967 resources (in “1991 dollars”) was less than half of actual 1991 spending can we say that per pupil expenditures did indeed double.

This report analyzes the portion of new spending from 1967 to 1991 that was due to inflation, the portion that was used to purchase additional resources, and the amount of additional resources received by each school program.

It is commonplace to use the consumer price index (CPI) to adjust expenditures for inflation. Yet as shown in the first section of this report, the CPI for all items understates inflation of school costs because prices grow faster in sectors like education where cost efficiencies are harder to achieve. To adjust expenditures more accurately for the inflation experienced by schools, we develop an adjustment measure based on inflation in service industries like education. We call this new index the “net services” index (NSI). Real school spending growth is the amount of new money spent in excess of inflationary changes measured by the NSI.

Throughout this study we focus on three measures of spending:

  • Shares of total expenditures. This is the percentage of all spending devoted to a particular program.
  • Shares of net new spending in 1991. This is calculated by subtracting each program’s spending in 1967 from its spending in 1991, and then determining what share of the change in total spending from 1967 to 1991 is attributable to that program’s growth.
  • Real growth. This is the change in per pupil spending over the period.

Each of these statistical types has uses and limitations. For example, while it makes sense to speak of regular education’s 28% real per pupil spending growth, it makes no sense to say that real per pupil spending on bilingual education grew by nearly 1,000%, since real spending on this program in 1967 was so insignificant. It does make sense, however, to say that bilingual education received 4% of net new funds spent by districts in 1991 compared to 1967. Each statistic in this report is clearly labeled so that readers can understand which type is being used.

All expenditure figures are calculated per pupils enrolled in all district programs. For example, per pupil spending for special education is calculated by dividing special education expenditures by total enrollment in the district in both special and regular education programs. Thus, when it is reported here that a district spent about $1,000 per pupil on special education in 1991, it does not mean that services for the average special education student cost only $1,000; in reality, spending on the average special education student was much greater. And when this report shows that the district spent about $3,800 per pupil on regular education, it does not mean that each regular education student cost $3,800; in reality, spending on the average regular student was less.

This convention enables a clear view of how district priorities have changed, exclusive of changes caused by rising or falling total enrollment. But the convention does not permit an understanding of how much is spent on each student in any particular program. For example, a rising per pupil cost of school lunches is calculated as total school lunch expenditures divided by all students, whether they receive subsidized lunches or not. The increase may reflect more expensive meals or more students in need of subsidized food, or a combination of both. Rising per pupil cost of special education also may reflect more resource-intensive instruction or the classification of more students in the special education program. Per pupil spending for any program can go up either because the client base of that program has expanded or because more real resources were devoted to a constant number of clients.

More sophisticated analyses of school program productivity would have to distinguish more intensive use of resources in each program from more students receiving these resources. This report, however, is but a first step and (except for some aspects of its analysis of regular education instruction) does not make those distinctions.

An examination of expenditure changes in the nine districts shows that:

  • As noted above, real per pupil spending, appropriately adjusted for inflation, grew by 61% between 1967 and 1991, a growth rate 40% less than conventionally reported;
  • The share of all spending received by regular education (what most people think of as a schools normal academic function) declined from 80% in 1967 to 59% in 1991; nonetheless, per pupil expenditures grew 28% over the period; regular education received 26% of the net new money spent in 1991;
  • Special educations share of all expenditures rose from 4% in 1967 to 17% in 1991; special education received 38% of the net new money spent in 1991;
  • About 8% of net new money went to expansion of the school lunch and breakfast programs. Another 7% went to attendance, dropout prevention, alternative instruction, and counseling;
  • In both 1967 and 1991, about two-thirds of regular education funds were spent on teachers’ compensation;
  • In regular education, higher average teacher salaries were mainly due to teachers greater experience (age) and credentials (e.g., masters degrees) in 1991 compared to 1967. Real salaries for teachers of similar experience and training did not significantly increase during this period and declined in many cases.

Growth in regular education staffing intensity was more marked at the elementary than at the secondary level. Elementary class sizes declined, but about half the reduction in pupil-teacher ratios was caused by more subject specialists and resource teachers, supporting more planning time for regular classroom teachers.

The growth of regular education spending was more marked in suburban than in urban districts. In the urban districts studied, per pupil regular education spending grew hardly at all, and might have decline
d in real terms if the districts had not cut back on operations, maintenance, and general administration spending. But in some suburban districts, regular education resources grew substantially.

The rest of this report is organized as follows. The first section describes the net services inflation index, how it was created, and why it is preferred over the CPI. The empirical methodology for both selection of sample districts and the categorization of expenditures is outlined in the second and third sections. The findings-how real spending for districts various programs has changed during the last quarter century-are found in the final sections.

See related work on Wages Incomes and Wealth

See more work by Richard Rothstein