November 1997 | EPI Book
Download the full text of this study 
Read the predecessor to this publication, Where’s the Money Gone?
Purchase this publication in paperback
Where’s the Money Going?
Changes in the Level and Composition of Education Spending, 1991-96
by Richard Rothstein
Executive Summary
Enrollment in American public education has grown rapidly in recent years, and schools have had to increase their real (inflation-adjusted) spending simply to maintain the same level of per-pupil resources. At the same time, however, school spending per pupil has been flat. For instance, real per pupil spending across the nation was roughly stable over the 1991-96 period, growing by only 0.7% (or 0.14% on an average annual basis). This was a significant slowdown from the growth in per pupil spending of 61% (or 2.0% on an average annual basis) from 1967 to 1991.
In the most recent period, some districts have actually had to reduce regular per pupil education spending in response to the combined pressures of enrollment growth, inflation, and shifting priorities toward spending on special populations.
This report examines shifts in the amount and composition of school spending growth over the 1991-96 period by analyzing expenditures in nine representative districts. It finds:
- The share of spending on regular education is shrinking. By the 1996 school year, regular education accounted for only 56.8% of all school spending, down from 58.5% in 1991.
- Special education spending grew to 19.0% of all school spending in 1996, up from 17.8% in 1991.
- School lunch and breakfast programs grew to 4.8% of total school spending in 1996, compared to 3.3% in 1991.
- Bilingual education programs grew to 2.5% of total school spending in 1996, up from 1.9% in 1991.
- The shift of spending away from the regular education program continues a trend observed over the 1967-91 period. However, in an era of stagnant overall school spending, such as the 1990s, this shift has translated into an actual reduction in regular education spending per pupil in several school districts.
These findings notwithstanding, it would be inaccurate to conclude that special education or other special programs have been growing at the expense of regular education: such a conclusion would have to be based not on data but rather on speculation about what would have happened to regular education in the absence of the growth of special programs. Nonetheless, the continued pattern of rapid growth in special program spending combined with slow growth in regular education spending is a phenomenon worthy of policy makers’ attention.
Introduction
Everyone from education experts to casual observers frequently speak of the “inefficiency” of U.S. elementary and secondary education. The claim rests in part on a belief that taxpayers have poured vast new resources into public education but have little in the way of results to show for this investment. “After allowing for inflation,” according to a 1994 report from the Brookings Institution, “the amount spent on each pupil in America’s schools has increased unabated for a century, with steady growth at about 3.5% a year….Despite increasingly large amounts of resources devoted to schools, student performance has shown few tangible improvements….[O]ur schools are demonstrably inefficient” (Hanushek et al. 1994).
In a previous report we showed that such conclusions rely on two serious mismeasurements (Rothstein and Miles 1995). First, we noted that the conclusion does not really “allow for inflation,” as claimed. Because elementary and secondary education is inherently a low productivity industry (using a much higher proportion of labor and less technology than the average industry), inflation adjustments based on productivity and price trends in an average industry will inevitably understate inflation in education. The Brookings report, for example, adjusts nominal school spending for inflation using the gross domestic product implicit price deflator, which includes data from all components of final demand in the economy, including the most technologically advanced manufacturing. Price reductions from technological improvements are mostly not available to schools, which continue to rely primarily on teachers’ labor for delivery of instruction. Properly adjusting for inflation in schools, therefore, requires an index based not on all components of final demand but based instead on labor-intensive low productivity growth industries. We suggested using a modified service sector index, called the net services index (NSI), to adjust school expenditures for inflation, because services, like schools, tend to use larger proportions of labor than does manufacturing, and productivity gains from new capital-based technologies are less accessible in services than in manufacturing.
Using this more accurate inflation adjustment, and examining the years 1967 to 1991, our earlier report found that about 39% of the commonly perceived doubling in per pupil spending resulted from this one mismeasurement problem alone. Allowing for inflation using the NSI, real per pupil spending increased by 61% over that period, or about 2% a year.
The second mismeasurement concerns the data used for comparison of “resources” to “student performance.” While our earlier report offered no conclusions about whether regular academic student performance has shown “tangible improvements,” we noted that the resources (per pupil spending) against which the efficiency of these improvements is normally measured are devoted to a wide range of programs, academic instruction being only one. Schools also provide special education for the disabled, vocational education, lunch programs, and special instruction for at-risk youth and for economically disadvantaged students. Comparing the combined expenditures for all these programs to the outcomes of only one of them (academic achievement of regular students) provides a misleading picture.
This mismeasurement is difficult to correct because schools do not report their expenditures by program. Instead, districts, states, and the federal government report education expenditures by “function” (administration, instruction, etc.) or by “object” (salaries, benefits, supplies, etc.). These categories cut across varied school programs, making it difficult to discern the efficiency of spending on regular, special, or vocational education or on noninstructional programs like health or nutrition. Because school finance is normally concerned with “function” and “object,” not programs, there are no conventionally accepted definitions of programmatic categories or their components.
To overcome this reporting failure, we selected nine typical districts from across the nation to mirror the experience of U.S. elementary and secondary education. We examined the expenditure reports of these districts for 1967 and 1991, and recategorized each expenditure of each of the districts into a programmatic group. In these typical districts, real (NSI-adjusted) spending on “regular education” had grown by only 28% in this 24-year period, or an average of about 1% a year. Regular education consumed 58.5% of all spending in 1991, compared to 80.1% in 1967. Of the “new” money in 1991 (i.e., total per pupil spending in 1991 in excess of the level in 1967), 37% had gone to special education, 27% to regular education, 6% to food services, 7% to programs for “at-
risk” youth, and the balance to other smaller programs. In 1967 special education consumed just 3.6% of total spending; , by 1991 its share had grown to 17.8%.
The “inefficiency” question, therefore, is whether a 1% per year annual growth in regular education expenditures was accompanied by comparable growth in regular students’ academic outcomes during the 1967-91 period. Because the state of knowledge regarding measurement of outcomes may be even more primitive than that regarding expenditure accounting, this may be an impossible question to answer with existing data. Neither the claim of inefficiency nor its opposite can be made with confidence.
This report examines whether the trends detected in the 1967-91 period continued in more recent years. We have extended the calculation of the NSI to 1996, and we returned to the nine sample districts to collect expenditure data for the 1996 school year. The following pages describe the new findings.
