Academic Freedom

The typical urban school district’s personnel and budgeting systems leave principals without much say in hiring teachers or allocating resources. The decentralization movement may just change that.

The Los Angeles Unified School District is the second largest school system in the nation-and perhaps the worst. Slightly less than half of its 75,000 employees are classroom teachers, meaning that Los Angeles spends just 35 percent of its budget on teacher pay. By comparison, the school systems in Houston, Texas, and Edmonton, in the Canadian province of Alberta, spend 49 percent and 56 percent, respectively, of their budgets on teachers. Since 1980, Los Angeles Unified’s enrollment has grown by 180,000 students, but the district has added only 15 schools with a total of 20,000 seats. As a result, nearly 200,000 students must be bused to a distant campus while most attend multitrack, year-round schools that can push more students through but offer 17 fewer days of instruction.

Although elementary schoolers in Los Angeles have made real gains in literacy in recent years, among high-school students, only 23 percent in reading and 34 percent in math meet or exceed the national norm on the Stanford 9. Of the district’s teachers, 27 percent lack full credentials. The system has a chronic shortage of qualified principals.

If Los Angeles is the worst school district in America, its East Coast cousin, New York City, is a close second. And the Chicago schools, while improving, are still recovering from the day in 1988 when William J. Bennett, secretary of education in the Reagan administration, pronounced them the “worst in the nation.” Why are these three school systems in such deep disarray? Certainly not because they are the three largest. None of them has more than a fraction of IBM or Toyota’s work force, and those companies are icons of good management. Nor is it because they serve high percentages of minority children from low-income families. Houston’s schools, which are equally minority and poor, perform well relative to other urban school districts. The reason is that the school districts in New York, Los Angeles, and Chicago are too centralized, much too top-down in their management, for their size.

To be fair, I have not studied every school district in America, and I cannot prove that these three are the worst. However, I have studied these three in depth, along with several others that have managed to achieve true decentralization. As a business school professor, I have also spent 30 years studying the largest companies in the United States, and it’s clear that any organization as big as these must be truly decentralized or it will fail. A series of business-focused studies has established firmly that organizations with more than about 3,000 employees have fully developed bureaucratic structures with high overhead costs, many specialist staff positions, and extensive sets of rules. Studies by Alfred Chandler and Oliver Williamson have demonstrated that such organizations must decentralize decisionmaking, thereby granting autonomy with accountability to sub-units, in order to function effectively. Decisions made at the top of a large bureaucracy will suffer from the absence of detailed information about local conditions at each site, will be slow to cope with any unusual situations, and will tend to enforce standardized procedures on every situation, no matter how poorly those procedures may fit the situation.

Decentralization has been a popular theme in school districts for a long time. Indeed, most districts claim that they are decentralized, having latched onto the “site-based management” movement of the 1980s. Superintendents and central-office personnel point to their local school councils, staffed by parents, teachers, and school administrators, and claim that they have moved decisionmaking down to the school level. However, they have neither achieved nor even attempted true decentralization, which requires that power over the budget be given to each school-and taken away from the central office. It’s the golden rule of power: whoever has the gold makes the rules. No one has made this point more clearly than New York University scholar Diane Ravitch. Her 1974 book The Great School Wars describes how New York City has played educational three-card monte over its long public school history by moving apparent control over decisions up and down the system, between the central chancellor’s office and the local superintendents, but never yielding any fraction of control to the schools.

While Los Angeles too claims to be decentralized, the principal of one of the inner city’s most successful schools says angrily, “There is no belief system in the district now that schools can make decisions for themselves-and it’s only gotten worse.” Another principal said, “We really don’t make budgets here. The mother district really tells us how many to hire of which kind, pays them, takes care of all that.”

In Making Schools Work, Lydia Segal and I detail the results of an empirical study of nine school systems that vary dramatically in their degree of decentralization. In addition to New York City, Los Angeles, and Chicago, our study included the school systems in Houston, Edmonton, and Seattle, each of which is relatively decentralized, as well as the three largest Catholic school systems in the United States-Chicago, New York City, and Los Angeles-which are more like loose federations of schools than “systems.” The findings from our study demonstrate that true decentralization yields benefits in both efficiency and performance.

Defining Decentralization

Why is budgetary control so crucial to decentralization? Consider the management steps that principals must follow if they are to take charge of their schools. They must:

• Identify the educational needs of each student in the school.

• Determine the teaching, counseling, and other staff resources that will be necessary to meet each student’s needs.

• Build a staffing plan that combines the full-time, part-time, outsourced, credentialed, and noncredentialed personnel necessary to provide those services to students.

• Design a school schedule that enables the principal to use staff creatively, in a way that frees up enough money to acquire all of the part-time and outsourced resources deemed necessary to meet the students’ needs. For example, a school might decide to have some classes with 40 students in them, meeting for two hours every third day. This might enable the school to teach reading in small groups of five students each.

No two schools have the same student populations. Thus no two schools should have identical staffing or schedules. No central-office administrator can perform this kind of detailed planning at the school level. As a result, each principal must have budgetary control in order to create the proper conditions for teaching and learning.

Our study classified the nine school systems into three organizational types, based on the taxonomy created by Oliver Williamson in his 1975 book Markets and Hierarchies. “Unitary” districts like New York, Los Angeles, and Chicago are the traditional kind of organization. In these, most important decisions are made at the central office. Schools are allocated a formulaic number of regular, special-education, physical-education, and other teachers, and they are not free to change their staffing except through an onerous process of waivers. Decisions on professional development, curriculum, and materials are made at the central office, which also controls the budgets for building maintenance, utilities, and other necessities.

In “multidivisional” districts such as Seattle, Edmonton, and Houston, the principal receives an allocation of funds and then is free to decide how best to spend that money on staff, materials, computers, school maintenance, utilities, and so on, much as the managers of corporate divisions are given wide-ranging authority and are held accountable for the bottom line. In a real sense, the principal is the CEO, having the power to decide virtually all matters except teachers’ pay, which is set by union contract. The possibility that a principal might become capricious can be held in check through the use of annual surveys of all school employees, students, and parents, who are questioned regarding their view of the principal’s leadership along with other elements of school performance.

The Catholic archdiocesan schools may be characterized as “holding companies.” In these systems, the central office has very limited power over the principals. Each school is on its own to meet its budgetary needs or to go out of business. One result is that the central-office staff in holding-company districts is very small. The central office for New York City’s archdiocesan schools, for instance, has a staff of only 22 people, including secretaries, to serve a system with 115,000 students.

These systems vary greatly in the degree of budgetary control possessed by individual school principals. We interviewed at least 5 percent of the principals in each system (but no fewer than ten principals in each system), 223 principals in all, and went through their school budgets with them. We calculated how much of the money was under the control of the principal and how much the central office. These calculations showed that principals controlled 6 percent of their budgets in New York City, a traditional “unitary” district, contrasted with 92 percent in Edmonton, one of the decentralized “multidivisional” districts. That all three of the multidivisional school districts have greater decentralization of budgets than even New York City’s archdiocesan schools is remarkable. In the traditional districts, most principals had no idea how much money was in their school’s budget. One Los Angeles principal estimated that her budget was $50 million when, in fact, it was about $20 million. The budget appears to be on a need-to-know basis: principals don’t need to know their budgets, because they don’t control how the funds are spent.

The decentralized school districts are quite different in size, location, and degree of unionization. Edmonton has less than half the enrollment of Houston’s school district, Seattle less than a quarter. Edmonton and Seattle also have very strong unions, while Houston’s unions are prohibited by state law from engaging in collective bargaining. Despite these differences, all three have successfully implemented a radical form of budgetary decentralization known as “weighted student formula.”

Innovation from the North

Weighted-student-formula budgeting was invented through trial and error by Edmonton superintendent Mike Strembitsky, beginning in 1973. The system attaches funds to individual students, rather than to schools or school districts, and then allows students to choose any public school in the district, with neighborhood students’ having preference at each school. Strembitsky implemented this system in the course of his 20-plus years as superintendent, thus giving the system time to become fully embedded in the structure, management practices, and politics of Edmonton.

When John Stanford, a retired army general, became superintendent of the Seattle schools in 1995, he traveled to Edmonton with his chief financial officer, Joseph Olchefske. (Olchefske later took over as superintendent when Stanford passed away and recently announced his intention to resign in the wake of a surprise budgetary shortfall.) They brought Strembitsky’s budgeting system to Seattle and implemented it the next year. In the Seattle system, a “basic student,” assigned a weight of 1.0, is a native English speaker with no special needs or economic disadvantage. Such students receive an annual allocation of $2,600 that can be taken to any public school in the district. The maximum allocation of $24,067, associated with a weight of 9.2, applies to a non-English-speaking student who has multiple, severe learning disabilities and comes from a low-income home. Each school also receives a “foundation grant,” which amounts to approximately $195,000 a year for an elementary school and up to $529,000 for a senior high school.

Rod Paige, then Houston’s schools superintendent, had also heard about Edmonton, and he too made the pilgrimage to meet with Strembitsky. Paige, now the Bush administration’s secretary of education, implemented weighted-student-formula budgeting in Houston during the 2001-02 school year. The district had already taken several steps under Paige to achieve decentralization, but it was the strapping of funds to individual students that hardwired the change into the Houston system.

The pairing of student-based funding with school choice seems crucial to making decentralization work. On the one hand, decentralizing power to principals will not improve student achievement unless poorly performing schools begin to lose students and their funding. On the other hand, school choice without student-based funding or some equally effective method of decentralizing school funds will also be ineffective. Giving parents a choice but not giving principals the autonomy necessary to improve their schools would elicit only frustration and pain on all sides. In a sense, by establishing public school choice, all three of the decentralized school districts have found a way to approximate the proven power of competition in a governmental setting. If a school cannot attract enough students to remain viable in these systems, it is reduced to a program rather than a school. The principal is then removed and replaced by a program director who reports to another, successful principal. In every case, it is apparent when a school performs poorly, because families will abandon it and it will stand empty.

University of Washington scholar Paul Hill rightly criticizes the current use of weighted-student-formula budgeting for ignoring the reality that staff salaries are the largest budget item in any school. In Houston, Edmonton, and Seattle, the school district pays teachers’ salaries and then charges schools based on the average salary of all teachers in the district. The result is that schools in wealthy neighborhoods, which tend to attract more experienced and therefore more expensive teachers, spend more on salaries than they actually pay for. Hill calls this a “hidden subsidy” to rich neighborhoods. The Houston district, for one, has set a timetable of ten years over which it intends to move to charging schools for actual teachers’ salaries. Though student-based funding alone is no panacea, it represents a politically acceptable form of budgetary decentralization that works.

Decentralization and Efficiency

Research on businesses and government agencies consistently finds that as an organization grows, it experiences a disproportionately large increase in central-office staff. Very large organizations must adopt a different basic structure if they are to retain enough efficiency to succeed as they grow. The advantage of the decentralized structure of the multidivisional type of organization is that it can effectively mitigate that staff growth by giving managers who know the needs of their divisions the power to make decisions, thereby reducing the necessity for central-office oversight. Our results suggest that the same is true for school systems, as Figure 1 shows. Edmonton and the Catholic systems have very small central-office staffs, and Houston and Seattle, while still carrying large staffs that built up over decades, are in the process of downsizing.

Perhaps the most remarkable finding in Figure 1 is the very small central-office staffs found in the Catholic archdiocesan school systems. These results illustrate the kind of central-staff efficiency that is possible under true decentralization. However, it should be noted that these districts have little capacity to offer custom-designed professional development courses for teachers and that their ability to search out new developments in curriculum is also limited. The limitation of the “holding company” structure is that it can provide little synergy or economies of scale to its largely independent schools. Oliver Williamson, in Markets and Hierarchies, theorized that the multidivisionary organization will outperform all other types. Time may prove him right in the case of schools, though it is possible that the loosely federated Catholic systems will continue to thrive in their present organizational form.

In the public districts, Seattle is an outlier, with a relatively large central staff. This may be the result of the dramatic decline in enrollments in Seattle after the onset of forced busing, during which enrollment dropped from nearly 100,000 in 1970 to only 39,000 in 1990. The district has yet to downsize its central staff to match that decline in enrollments. As a result, Seattle spent $9,700 per pupil in 2000-01 (excluding debt service and capital expenses). Meanwhile, Edmonton and Houston spent just $5,800 (Canadian) and $5,600, respectively, per pupil. By comparison, New York City spent $11,800 per pupil; Los Angeles, $9,600; and Chicago, $8,200.

Overall, it is clear that the decentralized systems are far more efficient in the number of central-office staff. Officials in New York City, Los Angeles, and Chicago might complain that the comparison is unfair because most of the people who appear on the central-office payroll actually work in schools. That is true, but keep in mind that the schools have no choice about which kinds of staff they receive from the central office. The principals we interviewed said that if they were given the money instead of the person assigned to them by the central office, they would choose to deploy those funds in other, more efficient ways.

Decentralization and Achievement

Decentralization also appears to improve student performance. Only a few of the districts could be directly compared because they use the same standardized achievement test. Houston and Los Angeles, the sixth and second largest school districts in the United States, respectively, each use the Stanford 9 test. Students in both systems are about 90 percent minority (Houston has more African-Americans, Los Angeles more Latinos), and about 80 percent come from low-income families. Both have witnessed improvements in test scores over the past few years. However, Houston’s scores are significantly higher than those of Los Angeles in both reading and math, and the city has experienced greater gains in recent years. In 1997-98, Los Angeles students in grades 2-8 scored in the 24th percentile in reading on the SAT 9, while Houston scored in the 32nd percentile, a gap of 8 national percentile ranking points. By 2000-01, the gap had grown to 13 points, with Houston scoring in the 46th percentile and Los Angeles in the 33rd percentile.

Our further analysis revealed that the “race gap” has narrowed each year in Houston since decentralization began there, even before the introduction of student-based funding, while the race gap in Los Angeles widens each year. A similar 2002 analysis by the Council of the Great City Schools confirmed both the improvement in Houston and the narrowing of that city’s racial achievement gap. Note, though, that comparing test results across cities is difficult even when both cities use the same test. In this case, for example, the results for Los Angeles are likely to be overstated, because in 2000-01 Los Angeles counted SAT 9 scores only for those students who had also taken the test the year before.

New York City public schools do not use a standardized test that can be compared with any of the other public school districts. However, a 2001 study by Raymond Domanico of the Metro New York Industrial Areas Foundation (prepared for the New York University Program on Education and Civil Society) compared the scores of New York City Catholic and public schools with similar student populations. He found that the archdiocesan students outscored their public school counterparts on standardized tests in grades 4 and 8 in both 1999 and 2000. University of Chicago scholar Anthony Bryk and his colleagues, in Catholic Schools and the Common Good, reported similar results from comparisons of public and Catholic schools. Of course, here too it is difficult to make comparisons because the act of choosing a private school may reflect other factors, such as a family’s commitment to education, that influence student achievement.

Today’s urban school districts have more than enough money in their budgets to do their jobs well. Now the challenge is to organize them so as to maximize their efficiency and performance. There is hope that change on a larger scale is coming. Joel Klein, chancellor of the New York City schools, announced in August of this year that he is implementing the “My Galaxy” computer-driven system, which will enable principals to assume control over their school budgets. This is the first step toward a full-blown implementation of weighted-student-formula budgeting in that city. Cincinnati adopted its own version of this budgeting system during the 1999-2000 school year, and other districts are considering following suit.

School districts are like other large institutions. They work well only if they are organized and managed properly. And as in other large organizations, faraway bureaucrats in central offices are in a poor position to know what should be done in each of a hundred or a thousand schools. Thus local autonomy along with accountability is essential to success in education. Decentralizing power to school principals and funding students rather than schools would be a good start.

-William G. Ouchi is a professor at UCLA’s Anderson School of Management. More information on his study of school district management is available at

Last Updated


Notify Me When Education Next

Posts a Big Story

Business + Editorial Office

Program on Education Policy and Governance
Harvard Kennedy School
79 JFK Street, Cambridge, MA 02138
Phone (617) 496-5488
Fax (617) 496-4428

For subscription service to the printed journal
Phone (617) 496-5488

Copyright © 2024 President & Fellows of Harvard College