Chrys Dougherty
University of Texas at Austin
About the Author
Dr. Chrys Dougherty obtained his Ph.D. in economics in 1992 from Harvard University, with a research focus on economic growth in advanced industrialized countries. He teaches courses in microeconomics, econometrics, and education policy at the LBJ School of Public Affairs at the University of Texas at Austin. His current research focuses on school effectiveness, public information on school performance, and school choice; and the contribution of human and physical capital investment to economic growth.
Prior to teaching at the LBJ School, Dr. Dougherty worked as an economic
consultant for Magee, Inc. In the 1970s he taught elementary school science at
the Oakland Community School in Oakland, California. He has written or
co-authored papers on educational accountability, school choice, and
international comparisons of the sources of economic growth.
A Study of Administrative
Expenditures in Texas
Public Schools
Chrys Dougherty
University of Texas at Austin
This paper reports on the results of a 1992 study of administrative expenditures in Texas public school districts in the 1990-91 school year. A research team at the LBJ School of Public Affairs at the University of Texas at Austin sought answers to the following questions:
1. What definition of "administrative expenditure" makes sense if the goal is to impose state limits on administrative expenditures by local school districts?
2. What is the relevant variable to analyze: the administrative expenditure per student, or the ratio of administrative expenditure to instructional expenditure?
3. What student or school district characteristics are associated with high administrative expenditure?
4. Is there a relationship between administrative expenditure and student learning?
5. What do districts with unusually high or low administrative expenditures do differ- ently?
6. What administrative expenditure limits make sense?
We used data from the Texas Education Agency's Public Education Information Management System (PEIMS) to analyze school district expenditures. PEIMS classifies expenditures into 18 functional categories. We classified six of these categories as administrative expenditure: General Administration, School Administration, Instructional Administration, Curriculum and Staff Development, Communication and Dissemination, and Data Processing Services. In school year 1990-91, these combined expenditures represented about 13 percent of public school expenditures in the state of Texas.
Our study was funded by the state legislature through the Educational Economic Policy Center. Overall, the study recommended specific administrative expenditure limits that would redirect an estimated $289 million per year (approximately $80 per-pupil) to the classroom by school year 1996-97.
In school year 1990-91, Texas had 1,053 school districts. Although several of these districts are large (Houston, with almost 200,000 students, is the fifth largest school district in the nation; Dallas, with 138,000 students, is the 10th largest school district), most are very small. The median Texas school district contained 775 students in 1990-91, and there were 393 districts with less than 500 students. The smallest district, Allamoore, had only two students (see Table 1).
---------------------------------------------------------------------------------- School Year ------------------------------ Characteristics 1990-91 1991-92 1992-93 Number of districts 1,053 1,050 1,048 Number of districts with: Less than 2,000 students 778 766 760 Half the students 46 46 46 More than 20,000 students 35 37 38 Enrollments Smallest district 2 2 7 Largest district 194,208 196,512 198,013 Total spending per student 95th percentile $8,136 $8,330 $8,522 75th percentile 5,203 5,605 5,893 50th percentile 4,454 4,724 5,084 25th percentile 3,978 4,247 4,547 5th percentile 3,542 3,774 4,051 State average (Texas) 4,200 4,452 4,774 U.S. average 4,890 5,103 5,334 ---------------------------------------------------------------------------------
Represents spending in the 95th percentile district, not spending in the 95th percentile student. The latter number would be substantially lower, since many of the highest-spending districts are very small
Texas and U.S. average spending per student are not adjusted for differences in the cost of living between Texas and the U.S. as a whole.
SOURCE: Texas Education Agency, Snapshot 91', Snapshot 92', and Snapshot 93'.
This proliferation of small districts has an impact on administrative expenditure per student. Even very small districts are likely to hire a superintendent or principal, or both. Of the 211 districts in Texas that had only one campus in 1990-91, 92 employed both a full-time superintendent and a full-time principal. State funding formulas provide extra money per student for districts that are very small and/or have very low population densities. As a result, the administrative spending per student and the ratio of administrative to instructional spending are higher in small districts, as shown in Figures 1 and 2. Beyond a district size of around 2,000 students, however, these apparent economies of scale vanish. Table 2 compares administrative expenditures for large and small districts in Texas.
-------------------------------------------------------------------------------------- Districts with: ---------------------------- More than Less than Spending 2,000 students 2,000 students All districts ------------------------------------------------------------------------------------------- Spending per student $4,128 $4,611 $4,200 Administrative spending per student 451 594 472 Administrative /instructional spending 22.90% 27.60% 23.60% Administrative/total spending 10.90% 12.90% 11.20% --------------------------------------------------------------------------------------
SOURCE: Texas Education Agency, Snapshot 91', Snapshot92', and Snapshot 93'.
SOURCE: Texas Education Agency, Public Education Information Management System (PEIMS) data.
SOURCE: Texas Education Agency, Public Education Information Management System (PEIMS) data.
Since the policy issue we were concerned with was whether to penalize districts with excessive administrative expenditures, we used a broad definition of these expenditures to discourage creative accounting. We thought it particularly important to include school administration (category 23) in our definition, since measured administrative expenditure could be reduced by paper reassignments of central office personnel to specific campuses. Likewise, omission of instructional administration (category 21) from the state's definition of administrative expenditure might result in a proliferation of curriculum coordinators in school district offices. Table 3 shows the types of expenditures that were classified as administrative and non-administrative expenditures.
------------------------------------------------------------------------------------ Expenditure Categories Defined as Administrative Cost: Category Number Definition ------------------------------------------------------------------------------------ 21 Instructional Administration 23 School Administration 25 Curriculum and Staff Development 26 Communication and Dissemination 41 General Administration 75 Data Processing Services Other PEIMS Expenditure Categories: Category Number Definition ------------------------------------------------------------------------------------ 11 Instruction 22 Instructional Resources and Media Services 31 Guidance and Counseling Services 32 Social Work Services 33 Health Services 34 Student Transportation 36 Co-curricular/Extracurricular 37 Food Services 42 Debt Services 51 Plant Maintenance and Operations 52 Facilities Operation and Construction 81 Community Services ------------------------------------------------------------------------------------
SOURCE: Texas Education Agency
We considered several types of administrative expenditure ratios for use in our analysis. In particular, we might have focused on:
Defining instructional expenditure as category 11 in the PEIMS data, we based most of our recommendations on the ratio of administrative to instructional expenditure, for three reasons:
1. Districts could improve their ratio in four ways, all of which are desirable:
a. reduce administrative expenditure per student;
b. shift resources from administration to instruction;
c. shift resources from other non-instructional areas to instruction; and
d. use increased tax revenues to increase overall spending for instruction.
Use of a per-student administrative spending measure sacrifices the incentive for options (c) and (d), while the administrative/total operating expenditure measure does not provide an incentive for (c).
2. There would be no need to change the "allowable" ratio every year to adjust for inflation and increases in school district expenditures and revenues, as would be the case with a per-pupil measure.
3. Fewer variables would need to be taken into account in adjusting allowable district administrative expenditure for factors that are beyond the district's control. A per- pupil measure would require consideration of seven such variables, while the ratio measure requires adjustment only for size and the district's percentage of Limited English Proficient (LEP) students.
School spending in Texas consists of expenditure from the general fund (Fund 10), and a large number of special revenue funds dedicated to categorical programs, such as Chapter 1 and the Job Training Partnership Act. The administrative expenditure ratio for all of these programs combined is only slightly higher than for the general fund; however, this ratio varies widely across programs. For example, a grant to write a new curriculum might be counted almost entirely as administration.
In order not to penalize districts for receiving those grants, we separated the general fund (Fund 10) from categorical funds, and recommended excluding the categorical programs when calculating the ratio of administrative to instructional expenditure.
We used ordinary least squares regression to determine which variables are systematically associated with administrative expenditure, measured on a per-pupil basis or as a ratio of administrative to instructional expenditure. Our working assumption was that causality runs one way from each of the variables in Table 4 to administrative expenditure per student or the administration/instruction expenditure ratio. Our initial hypotheses are shown in the right-hand column of Table 4.
------------------------------------------------------------------------------------------- Variable Initial hypothesis ------------------------------------------------------------------------------------------- District size Larger districts should have lower administrative costs per student relative to instructional costs. However, beyond around 2,000 students, there are no additional cost savings from additional size. District wealth Wealthier districts should spend more per student, but it is not obvious whether they would spend more relative to instruction. Average campus size For a given district size, a larger campus size implies fewer campuses, saving on both measures of the administrative cost. Student-teacher ratio A higher student-teacher ratio implies fewer teachers per student to supervise, lowering the administrative cost per student; it is not obvious what happens to the administrative/instructional cost ratio. Percent of LEP students More bilingual students implies more expense in curriculum development, raising the administrative cost per student; it is not obvious what happens to the administrative/instructional cost ratio. Percent of students in Same as for LEP students. special education Percent of low income Same as for LEP students. students Percent mobil students Higher student mobility increases the cost of keeping track of students, raising the administrative cost per student and the administrative/instructional cost ratio. Administative salary index Higher administrative salaries in neighboring districts increase the administrative cost per student; if teacher salaries are also higher, it is not obvious what happens to the administrative/instructional cost ratio. Five-year percent change Districts may adjust their administrative spending with in enrollment a time lag when enrollments increase, causing a negative relationship between this variable and both measures of administrative cost. -----------------------------------------------------------------------------------------------
SOURCE:Chrys Dougherty's hypotheses.
Our actual analysis, as shown in Tables 5 and 6, revealed the following results:
--------------------------------------------------------------------------------------------------- Dependent variable ------------------------------------------------------------------- Administrative/instructional ratio Admin exp/student Small Large Small Large Variables districts districts districts districts Natural logarithm of district size - (-) Natural logarithm of distric wealth + + + + Average campus size (in hundreds) (-) - + Student-teacher ratio + - - - Percent LEP students + + + + Percent special education students (-) - Percent low income students (+) Student mobility rate + Administrative salary index 5-year percent enrollment change (-) (-) ---------------------------------------------------------------------------------------------------
Parentheses imply significance at the 10% level; no parentheses imply significance at the 5% level or better.
SOURCE: Chrys Dougherty's statistical analysis.
------------------------------------------------------------------------------------- Dependent variable ----------------------------------------------------------------- Administrative/instructional ratio Adminexp/student ---------------------------------- ----------------------------- Districts with Districts Districts Districts with greater than less than greater than less than Variables 2000 students 2000 students 2000 students 2000 students ------------------------------------------------------------------------------------- Constant 0.249 0.137 217.1 -187.8 (3.76) (1.68) (1.02) (-1.14) Natural logarithm of -0.060 -0.005 -213.8 -2.26 size (-11.5) (-1.68) (-12.7) (-0.38) Natural logarithm of 0.03 0.013 217.7 83.2 wealth (7.75) (2.76) (17.0) (8.52) campsize (in hundreds) -0.005 -0.005 25.6 -3.69 (-1.77) (-2.94) (2.38) (-1.01) stu/tch 0.006 0.000 -52.8 -20.0 (2.99) (0.02) (-8.76) (-4.88) Percent LEP 0.002 0.001 6.81 2.47 (3.41) (2.62) (4.77) (4.18) Percent speced -0.001 -0.001 -11.1 -3.39 (-1.82) (-1.39) (-5.36) (-1.60) Percent comped 0.000 0.000 -0.318 0.662 0.69 0.89 (-0.58) (1.81) mobility 0.000 0.001 -0.583 0.895 (0.04) (2.42) (-0.66) (1.55) salindex (in thousands) 0.002 0.001 -3.13 2.53 (1.35) (0.91) (-0.80) (1.37) Percent sizchg 0.000 -0.000 -0.412 -0.566 (1.29) (-1.68) (-0.79) (-1.96) R 0.462 0.137 0.762 0.440 Adjusted R 0.455 0.104 0.759 0.418 ----------------------------------------------------------------------------------
Note: T-statistics are in parentheses
SOURCE: Chys Dougherty's regressions using PEIMS and TASS data from Texas Education Agency.
To examine the relationship between administrative expenditure and student learning, we regressed test score data on a set of demographic and expenditure variables. The dependent variable we used was the average of the third- and fifth-grade reading, writing, and mathematics scaled scores on the Texas Assessment of Academic Skills (TAAS) test, which was administered in Texas elementary schools in October 1990. These scaled scores had a mean of 1,600 and standard deviation of 73.3. We had data on these scores in 1,323 schools.
Gain scores-- the average difference between individual students' test scores in consecutive years--would be a more appropriate dependent variable to use in this analysis. However, the TAAS at the time was not designed to be gain-scored, and was not administered to the same students in successive years. Thus, we lacked the data to implement this more desirable alternative.
The independent variables we used were:
The results, as shown in Table 7, indicate that there is little association between administrative spending per teacher and student learning. However, there is also no evidence that administrative spending has a negative effect on instruction. We were unable to explain the opposite-sign relationships between campus and instructional administrative spending and student learning.
----------------------------------------------------------------------------- TASS TASS Variables Scaled Score Scaled Score ----------------------------------------------------------------------------- Constant 1,658.70 1,649.24 Percent low-income students -1.83 -1.84 (-16.8) (-17.00) Percent black students -0.592 -0.571 (-5.64) (-5.45) Percent Hispanic students -0.236 -0.216 (-2.25) (-2.06) National logorithm of district size 4.778 5.798 (4.48) (3.98) campus size -0.012 -0.009 (-1.64) (-1.12) Student-teacher ratio -2.37 -2.244 (-3.28) (-3.03) Teachers average years 1.223 1.358 of experience (2.07) (2.30) 1.36 -- Expenditure per teacher (1.34) on total administration (in thousands) Expenditure per teacher -- 1.311 on central administration (0.80) (in thousands) Expenditure per teacher -- 3.877 on school administration (2.3) (in thousands) Expenditure per teacher -- -5.57 on instructional administration (-2.10) (in thousands) Expenditure per teacher 0.573 0.449 on classroom instruction (1.38) (1.08) (in thousands) R 0.636 0.638 Adjusted R 0.632 0.634 -----------------------------------------------------------------------------
Note: T-statatistics are in parentheses. Variables which were not statistically significant are not shown: percent LEP students, percent special education students, percent mobile students, and expenditure per teacher on counseling, health, and social work services.
SOURCE: Chrys Dougherty's regressions using PEIMS and TAAS data from the Texas Education Agency.
We selected seven Texas school districts with unusually low or high ratios of administrative to instructional expenditure. For site visits our judgment about which districts' administrative costs are unusually low or high was based on residuals from a regression equation similar to that used in Table 5.
When administrative expenditures are especially high, how do school districts spend the money? In some cases, the district uses the school district administrative budget as an employment program. One high-expenditure district with 18,000 students had a staff of 2,500 of whom 995 were teachers. This district had the state's 13th-highest ratio of non-teachers to total staff. Judging from interviewees' comments in a number of districts, "kicking the bad principals upstairs" into the central office is a fairly common practice.
Other districts have special circumstances. One small, suburban high-wealth district hired extra staff to process the thousands of job applications received each year from teachers anxious to work in that district. Another district was paying three superintendents, two of whom had been dismissed from multi-year contracts in the previous two years. One of these former superintendents had sued the district for wrongful termination, creating high legal costs as well.
Districts with below-average administrative expenditures employ several expenditure-saving measures. First, they limit expenditures on instructional administration, relying on their teachers or the state's Regional Educational Service Centers for curriculum development services. Second, they expect senior administrative staff to share clerical and support staff. Third, they pay their administrators less. This option may not be available to districts that hope to attract top-flight principals and superintendents, however.
An underutilized option is the formation of multi-district cooperatives to share expenditures in areas such as curriculum development and data processing. To examine the use of cooperatives, our study contacted 48 school districts, 25 with a high administrative/instructional expenditure ratio and 23 with a low administrative/instructional expenditure ratio. While 46 of these districts participated in cooperatives to pool resources for special education and several do the same for vocational education, only two districts were members of cooperatives designed to achieve economies in general administrative expenditures. One district was part of a seven-district cooperative designed to share data processing expenses. The second cooperative served 13 districts, providing services in data processing, staff development, and technology support.
As a result of our study, we made the following recommendations to the Texas legislature in 1993:
We projected that this approach would redirect $269 million into the classroom by the 1996-97 school year, or about $70 per student based on an enrollment of 3.8 million students in Texas. Compelling small districts' administrative/instructional expenditure ratio to conform to the state average would redirect an additional $20 million.
Distracted by school finance issues and the threat of a court-ordered shutdown of public schools, the 1993 Texas Legislature paid relatively little attention to administrative expenditure. The administrative expenditure control measure that passed was considerably different from the one recommended in the Administrative Expenditure Study. The Legislature divided school districts into five size categories, and specified that the Commissioner of Education would set allowable ratios of administrative to instructional expenditure for each category. "Administrative expenditure" as defined by the legislature excludes campus administration, but includes state and local categorical programs.
This legislation was first implemented in the 1993-94 school year. As actual expenditure data from subsequent years become available, it should be possible to assess the impact of this legislation.
Hughes, M. F. 1991. "A comparison of social, economic, and educational variables in high and low expenditure school districts of Virginia, 1988-89." Journal of Education Finance 17.
Monk, D. H., and J. Underwood. eds. 1988. Microlevel school financing: Issues and implications for policy. Yearbook of the American Education Finance Association, No. 9. Cambridge: Ballinger Publishing Co.
Monk, D. H., and B. D. Walker. 1991. "The Texas cost of education index: A broadened approach." Journal of Education Finance 17.
Performance Evaluation and Expenditure Review Committee. 1993. School districts' FY 1992 spending for central office administrators' and principals' salaries and potential administrative savings available for redirection to classroom instruction. Jackson, MI: Mississippi State Legislature. ERIC Document Reproduction Service No. ED 360 745.
Texas Education Agency. 1990. Standards for school district data delivery, 1990-91. Austin, TX.
Texas Education Agency. 1992. Financial accounting manual. Bulletin 679, Change 26. Austin, TX.
Texas Education Agency. 1992. Snapshot '91: 1990-91 school district profiles. Austin, TX.
Texas Education Agency. 1992. Texas school directory, 1992-93. Austin, TX
Texas Research League. 1991. Bench marks: School district budgets, 1990-91. Austin, TX.