"Why Do some Schools Get More and Others Less? An Examination of School-Level Funding in New York City"

A.E. Schwartz presentation, Research Partnership for NYC Schools
Why Do Some Schools Get More and Others Less?
An Examination of School-Level Funding in
New York City
Amy Ellen Schwartz
New York University
Ross Rubenstein
Syracuse University
Leanna Stiefel
New York University
October 5, 2007
http://steinhardt.nyu.edu/iesp/
In 2003-2004, the typical NYC elementary
school budget was roughly $13,000 per pupil
• But:
– some schools received as little as $9,000.
– others more than $15,000.
• Why?
– Differences in the needs of students?
– Differences in costs?
– Or is it just unfair and inefficient?
2
Two Schools, Similar Poverty, Different Spending, 2003-04
P.S. A
P.S. B
Total Spending per Pupil $9,559.38 $17,928.26
Gen Ed Spending per Pupil $8,893.80 $14,243.63
83.60
83.40
% poverty
19.20
4.70
% LEP
2.96
12.80
% Special education
32.40
48.90
% more than 5 years experience
100.00
37.80
% less than 2 years in this school
60.80
47.20
% passing Math
36.60
35.80
% passing Reading
3
The Answer Matters
• If differences are driven by differences in
the needs of students then they may be both
fair and efficient.
• If they aren’t, then redistributing resources
will be critical for improving equity and can
improve student performance.
4
The Time is Ripe
•
•
•
NCLB reauthorization focuses attention on resources,
outcomes and accountability at the school level.
CFE vs. New York State brings similar focus and
additional state funds to NYC schools.
NYCDOE is reforming their budget allocation
processes in their Fair Student Funding (FSF)
Initiative.
5
Outline of Talk
•
Background on School Resource Allocation
•
Analyses of School Spending in New York City
•
Implications for Policy
6
Four Lessons from Prior Research
• Disparities in spending across schools
within a district are common.
• Large districts are more likely to have large
disparities.
• Schools with poorer children have less
experienced, less educated and less
expensive teachers.
• School “budgets” largely reflect positionbased formulas and teacher sorting.
7
Where does the money come from?
• City, State and Federal funds.
• In 2003-4 the average elementary school budget
was:
– Almost 80% City Taxes and State Operating
Aid.
– About 7% from the Federal Title I program that
targets high poverty schools
– About 13% from myriad categorical grant
programs.
8
Allocation Basics - Teachers
• Teacher positions are allocated based upon:
– Enrollment
– Class size maximums (depends on grade level)
– e.g., 100 first grade students would mean 5
teachers if the district class size max is 20.
• Salaries are determined by district schedule
and money “budgeted” to the school
depending upon the teachers hired.
9
How Does Budgeting Teacher Positions
Affect School budgets?
•
•
•
Schools with higher paid teachers get funding to
cover those positions; schools with lower-paid
teachers do not receive similar amounts.
Position-based budgeting concentrates experienced
teachers in schools where needs are lower.
Higher staff turnover concentrates new teachers in
low-performing schools, meaning they have lower
budgets.
10
Advanced Resource Allocation
• Overhead allocations for administration and
building services
• Specialized formulas target students with
special needs. E.g. Poor, English Language
Learners, Learning Disabilities
• Targeted funds based upon school
characteristics (e.g., school size) or special
programs.
• “Hold Harmless” provisions limit change.
11
Empirical Analyses
• Examine the distribution of funding across
schools
– By source (taxes and state aid, Title I)
– By grade level (elementary vs middle)
• Examine changes over time
• Goal is to shed light on the factors that drive
disparities and provide a benchmark to
gauge the impact of policy changes.
12
Data
• NYC Department of Education, Annual School Reports and
School-Based Expenditure Reports, 2000-01 to 2003-04
• School-level spending variables by source of funds
• Key school-level variables on student/school needs:
•
•
•
•
•
•
Eligible for free lunch
Limited English proficient
Enrollment (school size)
Resource room and special education
Recent immigrant
Student performance – percent level 1 fourth and eighth grade
• Sample: N = 911 in 2003-04
13
Five Measures of Resources
1) Total spending per pupil
2) General Education Funding per pupil
(includes PTSE funds)
3) Tax levy and state operating aid - Gen Ed
4) Title I – Gen Ed
5) Other – Gen Ed
14
Total Spending
0
5000
10000
15000
20000
25000
Elementary Schools, 2003-04
0
20
40
60
Percent Free Lunch Eligible
80
100
15
R2 = 0.08 Slope = 27.3 n=682
General Ed Spending
0
5000
10000
15000
20000
25000
Elementary Schools, 2003-04
0
20
40
60
Percent Free Lunch Eligible
80
100
16
R2 = 0.08 Slope = 19.9 n=682
Tax Levy and State Operating Aid
0
5000
10000
15000
20000
25000
Elementary Schools, 2003-04
0
20
40
60
Percent Free Lunch Eligible
80
100
17
R2 = 0.01 Slope= -5.1 n=682
Title I Revenues
0
1000
2000
3000
4000
Elementary Schools, 2003-04
0
20
40
60
Percent Free Lunch Eligible
80
100
18
R2 = 0.61 Slope= 15.6 n=682
Revenues From Other Sources
0
1000
2000
3000
4000
Elementary Schools, 2003-04
0
20
40
60
Percent Free Lunch Eligible
80
100
19
R2 = 0.29 Slope = 9.4 n=682
12000 15000
Elementary Schools, 2003-04
Total Spending
9000
Gen Ed Spending
3000
6000
Tax Levy and
State Operating
Aid
Other Revenues
0
Title I Revenues
0
20
40
60
Percent Free Lunch Eligible
80
100
20
Regression Analyses Describe the
Determinants of Spending
• First using only school and student
characteristics for 2003-04 (Cross-sectional)
– Including Pct Poor, LEP, PTSE and school size
• Next examining change over time
– Between 2002-03 and 2003-04 (Short)
– Between 2000-01 and 2003-04 (Long)
21
Poverty Coefficients
27.3
0.08
Short
Adjusted Changes
Total Spending
12.0
0.2
0.61
0.75
Pct Poor
R2
19.9
0.08
General Ed
15.4
-1.6
0.42
0.71
Pct Poor
R2
Tax Levy and Operating Aid
-5.1
-6.4
10.0
0.01
0.33
0.7
Pct Poor
R2
Pct Poor
R2
Pct Poor
R2
Long
Changes
6.0
0.96
4.1
0.94
-1.1
0.93
15.6
0.61
Title I Only
15.3
1.0
0.65
0.71
4.2
0.89
9.4
0.29
Other Sources
6.5
0.2
0.52
0.75
1.1
0.87
22
Key Results
• Funding reflects student need
– i.e., a 1% point increase in the pct. PTSE is
associated with $69 more in funding per pupil
• Higher poverty means more Title I and
Other funds; less Tax Levy/Operating Aid.
• Change matters but the response is sluggish.
• Last year’s funding is an excellent predictor
of this year’s funding
• Much variation is unexplained.
23
Weighted Student Funding (WSF)?
• Resources “follow the student.”
• Allocates dollars and not positions.
• Amount depends upon student needs
(characteristics) with higher “weights”
given to students with greater needs.
• Transparent targeting of resources
• Effectively reduce unexplained variation.
24
Cautions
• No adjustments for concentration
– Economies/diseconomies of scale ignored
• Transparent incentives may have
unintended consequences.
– Getting weights right is critical
– Evidence base is thin.
• Rapid change can be difficult
• Implications for teachers.
25
Agenda for Further Research
• We need to know the cost of student needs.
• Should there be a weight for immigrants?
• For student mobility?
• We need to understand (dis)economies of scale.
• We should learn from experience:
• Does Title I money get where it’s intended and
does it improve student performance?
• What was the impact of the recent changes in
seniority transfer rules?
• Evaluation of FSF has to start now – to study
implementation and school responses.
26