Return on Educational Investment: Findings
Read the full report (pdf)
This report is designed to spark a national conversation about educational productivity and to identify districts that generated more relative achievement per dollar spent. Below are our major findings:
Inefficient school systems represent a significant reform opportunity
If school systems spent their dollars more productively, many would see large gains in student achievement. Consider California, where a low-productivity school district could see as much as a 25 percent boost in achievement if it increased its efficiency from the lowest level to the highest, all else being equal. In Arizona, that jump in achievement could be more than 36 percent, according to our analysis.
Our data show that 41 states show the potential for double-digit percentage increases in achievement without necessarily spending additional funds. Such growth in student learning will not come without significant reform since the programs and policies that cause low productivity are often systemic. But at a time of sagging revenues and pending budget cuts, these results should inspire states and districts to tackle productivity head-on and consider reforms that boost achievement without incurring significant costs. (See Table A2)
Low productivity is costing the nation’s school system as much as $175 billion a year
After adjusting for variables outside a district’s control, districts with below average productivity spent over $950 more per student than did above-average districts. This estimated loss in capacity equals about 1 percent of the nation’s growth domestic product.
To be sure, inefficient districts are not necessarily “wasting” the lost capacity. Our approach cannot account for all the factors outside of a district’s control, and the extra money spent by some districts might be supporting outcomes beyond the scope of this study. But at the same time, our estimate might also be low since it does not cover the cost of poorly prepared students entering college and the workforce. Far more research needs to be done in this area in order to better understand the scope of the productivity problem.
Without clear controls on how additional school dollars are spent, more education spending will not automatically improve student outcomes
Additional dollars corresponded to higher student achievement in only 16 states. In five states, including Florida, Texas, and New Jersey, additional dollars predicted slightly lower achievement.
That does not mean that money can’t—or doesn’t—have an effect on achievement in Florida, Texas, or anywhere else. Rather, it means that money matters only if it’s spent in effective ways, at least above some threshold level, and that without a systemic approach to spending, unfocused increases in expenditures are not likely to have any impact on student outcomes.
Efficiency varies widely within states
Some districts spent thousands more per student to obtain the same broad level of academic achievement. In New York, the range of spending among the districts in the highest third of achievers was more than $7,000 per student. In California, the range was almost $8,000 per student.
The differences are starker when comparing similar districts, like Oshkosh and Eau Claire, Wisconsin. The two school systems are about the same size, have largely comparable results on state exams, and serve similar demographics. But in 2008, Eau Claire spent over $8 million more than Oshkosh, or about $300 more per student.
More than 1 million students are enrolled in highly inefficient districts that received a low rating on all three of our measures
That’s over 400 school districts, serving about 3 percent of the almost 43 million students covered by our study. These districts spent far more money than other districts, and after adjusting for characteristics outside the district’s control, these low-productivity districts spent over $2,000 more per student than the average school district.
High-spending school systems are often inefficient
Our analysis showed that after accounting for factors outside a district’s control, many high-spending districts posted limited outcomes. In Minnesota just 23 percent of the districts in the top third of spending were also among the top third in achievement. In Florida, only 17 percent of the state’s highest-spending districts were also in the highest-achieving tier.
In high-spending districts, success often comes at significant cost. Consider Howard County School District in Maryland. Many consider the affluent district to be one of the best in the country, and a magazine recently heralded the district as an international academic “powerhouse.” But after controlling for factors outside the district’s control, the school system has one of the highest rates of perstudent expenditures in the state, spending over $1,600 more per student than the state average and almost $3,000 more than the national average.
To be sure, our evaluation could not control for all the variables that go into measuring an efficient school district, and some of the money expended by high-spending districts like Howard Country may support high-quality programs and tools such as science labs and computer rooms that would not be captured in math and reading tests. But given the need to dramatically improve the nation’s reading and math outcomes, taxpayers and parents should scrutinize high-spending districts and ensure that they’re getting everything that they can for their school dollar.
Students from disadvantaged backgrounds are more likely to be enrolled in inefficient districts
Our data showed that students who participated in the subsidized-lunch program were 12 percentage points more likely to be enrolled in the nation’s least-productive districts than the most productive. This finding appears after adjusting expenditures for the higher cost of educating lower-income students and suggests that highly inefficient districts are more likely to have larger percentages of students from disadvantaged backgrounds.
Students from minority backgrounds are also more likely to be enrolled in highly inefficient districts. The least-efficient districts were more likely to have larger percentages of black students (18 percent versus 5 percent) and Hispanic students (14 percent versus 7 percent) than the most efficient ones. But while some within the research community use race as an adjustment within regression models to account for the fact that students of color often perform worse than their peers, we did not do so for a number of reasons. Race variables, for one, did not add significantly to the robustness of our productivity evaluations. The Center also does not support policies that lower academic expectations for students of color or from disadvantaged backgrounds.
These findings present an important avenue for additional research. Our data cannot capture everything that goes into creating an efficient school system, and there were confounding variables that our study was not able to control for. For instance, school systems with weak tax bases may be subject to greater administrative burdens and have less control over how funds are spent, which might make it harder for them to be efficient. Such school systems also typically tend to have greater proportions of low-wealth parents, whose students often perform less well on reading and math exams. But we could not account for differences in revenue sources because the fiscal database produced by NCES does not track expenditures by source, and this may make districts with weaker tax bases appear less efficient than they really are.
Highly productive districts are focused on improving student outcomes
We surveyed a sample of the districts that performed well on each of our productivity metrics, and they shared a number of attributes, such as building support within their communities for productivity reforms and using data-mining practices to reduce inefficiencies.
States and districts fail to evaluate the productivity of schools and districts
Only two states, Florida and Texas, currently provide annual school-level productivity evaluations, which report to the public how well funds are being spent at the local level. But without consistent metrics, educators will not be able to figure out if school dollars are well spent.
The quality of education data is often poor, which impedes the study of educational productivity
In far too many cases, crucial data on school finance, operations, and outcomes are unavailable, making it difficult to accurately measure the achievement that a school district produces relative to its expenditures. For instance, we could not control for certain cost factors, such as transportation, due to a lack of robust data, and so a rural district that has high busing costs because its students are spread out over a large area might be at a disadvantage in some of our metrics relative to a more densely populated district.
When states and districts do collect key education data, they often use inconsistent definitions and weak data collection practices. For instance, policymakers in some states set the cut scores of state exams at very low levels so that many school systems report having proficiency levels at 90 percent or above. Six districts in Nebraska reported that all their students were proficient in reading and math in fourth grade, eighth grade, and high school in 2008. This makes it difficult to compare the productivity of districts, since some have essentially topped the achievement scale.
The nation’s least-productive districts spend more on administration
The most inefficient districts in the country devote an extra 3 percentage points of their budgets on average to administration, operations, and other noninstructional expenditures. This translates into large per-student spending differences, and after adjusting for students in special programs and cost of living, the least productive districts spend almost $300 more per student than the average district on student and staff support, which includes expenditures on school libraries, media centers, and guidance counselors. The least productive districts also spend over $350 more per student than the average district in administrative costs, which includes dollars spent on central services such as payroll as well as principals and other administrators.
This finding does not mean that high administrative costs cause low productivity. Inefficiencies are often buried deep within the operation of school systems. The problem might be large expenses on programs that do little to raise student achievement, or salaries paid out to staff that have little or nothing to do with the employee’s effectiveness. It’s also possible that our measures reflect the fact that districts with lower achievement are often subject to increased state regulations. In that case, it wouldn’t be high administrative spending causing low efficiency, but low efficiency causing increased administrative burdens. We also do not endorse policy proposals that require a set amount of money per student to be spent in the classroom. Such blunt formulas often do more to hinder local administrators than help them.
Some urban districts are far more productive than others
While our main results are limited to within-state comparisons, we conducted a special cross-state analysis of urban districts that recently participated in a national achievement test. And after adjusting for certain factors outside a district’s control, we found that some big-city school systems spend millions of dollars more than others—but get far lower results on math and reading tests.
Read the full report (pdf)
To speak with our experts on this topic, please contact:
Print: Katie Peters (economy, education, health care, gun-violence prevention)
202.741.6285 or firstname.lastname@example.org
Print: Anne Shoup (foreign policy and national security, energy, LGBT issues)
202.481.7146 or email@example.com
Print: Crystal Patterson (immigration)
202.478.6350 or firstname.lastname@example.org
Print: Madeline Meth (women's issues, poverty, Legal Progress)
202.741.6277 or email@example.com
Print: Tanya Arditi (Spanish language and ethnic media)
202.741.6258 or firstname.lastname@example.org
TV: Lindsay Hamilton
202.483.2675 or email@example.com
Radio: Madeline Meth
202.741.6277 or firstname.lastname@example.org
Web: Andrea Peterson
202.481.8119 or email@example.com