<%BANNER%>

The Effects of Unitary Status on Academic Achievement and School Finance

Permanent Link: http://ufdc.ufl.edu/UFE0042126/00001

Material Information

Title: The Effects of Unitary Status on Academic Achievement and School Finance Essays on the End of the Court-Ordered Desegregation Era
Physical Description: 1 online resource (125 p.)
Language: english
Creator: Knapp, Colin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: desegregation, district, finance, resegregation, school, status, unitary
Economics -- Dissertations, Academic -- UF
Genre: Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Over the span of three decades, many southern school districts were found to be operating racially separate school systems and were placed under court-supervised desegregation plans. While under these plans, districts were forced to remove all traces of their formerly segregated nature. Upon satisfactorily achieving the U.S. Supreme Court s goal of one unitary system for all children, a district can be released from court supervision and can subsequently be considered to have received unitary status. This study identifies how exiting a court ordered desegregation plan via the granting of unitary status affects racial imbalance and academic achievement. It finds significant evidence that the level of segregation will rise if unitary status is received. This effect is robust and appears to grow over time. Unitary status is also shown to have little effect on the achievement of the average student within each district. This result holds for both white and black students at the district level, suggesting that the end of court supervision may not lead to an increase in academic inequality. Additional consideration is given to the motives behind the decision to seek unitary status. During the most active phase of the school desegregation era, the U.S. court system took control of student assignment away from districts found guilty of operating segregated school systems. This affected both large and small districts. However, as the courts retreated from active involvement, large districts were more frequently released than small ones. One explanation may be economies of scale. If the per-capita financial benefits of receiving unitary status are small, a large number of students are required to cover any non-trivial costs associated with seeking release from the court s desegregation order. This study shows some revenue shifting may occur, saving local tax payers about $100 per student in Florida. Since there is no clear indication that any sizable expenditures are forgone, the net effects are small at best. Therefore, economies of scale may be involved in the decision to seek unitary status, explaining why large districts receive unitary status more frequently than small districts.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Colin Knapp.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Kenny, Lawrence W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042126:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042126/00001

Material Information

Title: The Effects of Unitary Status on Academic Achievement and School Finance Essays on the End of the Court-Ordered Desegregation Era
Physical Description: 1 online resource (125 p.)
Language: english
Creator: Knapp, Colin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: desegregation, district, finance, resegregation, school, status, unitary
Economics -- Dissertations, Academic -- UF
Genre: Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Over the span of three decades, many southern school districts were found to be operating racially separate school systems and were placed under court-supervised desegregation plans. While under these plans, districts were forced to remove all traces of their formerly segregated nature. Upon satisfactorily achieving the U.S. Supreme Court s goal of one unitary system for all children, a district can be released from court supervision and can subsequently be considered to have received unitary status. This study identifies how exiting a court ordered desegregation plan via the granting of unitary status affects racial imbalance and academic achievement. It finds significant evidence that the level of segregation will rise if unitary status is received. This effect is robust and appears to grow over time. Unitary status is also shown to have little effect on the achievement of the average student within each district. This result holds for both white and black students at the district level, suggesting that the end of court supervision may not lead to an increase in academic inequality. Additional consideration is given to the motives behind the decision to seek unitary status. During the most active phase of the school desegregation era, the U.S. court system took control of student assignment away from districts found guilty of operating segregated school systems. This affected both large and small districts. However, as the courts retreated from active involvement, large districts were more frequently released than small ones. One explanation may be economies of scale. If the per-capita financial benefits of receiving unitary status are small, a large number of students are required to cover any non-trivial costs associated with seeking release from the court s desegregation order. This study shows some revenue shifting may occur, saving local tax payers about $100 per student in Florida. Since there is no clear indication that any sizable expenditures are forgone, the net effects are small at best. Therefore, economies of scale may be involved in the decision to seek unitary status, explaining why large districts receive unitary status more frequently than small districts.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Colin Knapp.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Kenny, Lawrence W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042126:00001


This item has the following downloads:


Full Text





THE EFFECTS OF UNITARY STATUS ON ACADEMIC ACHIEVEMENT AND
SCHOOL FINANCE: ESSAYS ON THE END OF THE COURT-ORDERED
DESEGREGATION ERA















By

COLIN A. KNAPP


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010
































2010 Colin A. Knapp


































To my wife and daughter, beautiful in so many ways









ACKNOWLEDGMENTS

This dissertation would not have been possible without the love and support of my

entire family. From the parents who provided a world-class education, to the wife that

puts food on the table, I have received more than I deserve. Like most, I am not as

appreciative as I should be and I will try to change. I am also in debt to the most helpful

dissertation chair any graduate student could ever have. Professor Larry Kenny gently

nudged me back towards completion whenever I would stray too far off course. There

are too many other teachers, professors, classmates, family members, and friends who

have mentored me through 11 years in the military, 24 years of formal education, and

36 years of life to include them all. I am sorry for that because they have earned it.









TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ............. ..................... ............... 4

LIST OF TABLES ......... ............... ..................... ....... ............... 7

LIS T O F F IG U R E S .................................................................. 9

LIST OF ABBREVIATIONS..................... .......... .............................. 10

A B S T R A C T ......... ...... ........... ................................. ........................... 11

1 INTRODUCTION ........ ......... ......... ...... ................ .... ........... 13

2 UNITARY STATUS AND ACADEMIC ACHIEVEMENT IN FLORIDA.................... 16

A Brief History of Court-Ordered Desegregation..................................... 16
The Existing Literature on the Effects of Unitary Status...................... ......... 17
D description of the D ata ......... .............................................. .......... 20
D description of the E m pirical M odels............................................... ... .. ............... 26
R results of the E m pirica l M ode ls..................................................... ... ................. 29
Unitary Status' Effect on Segregation......................... ..... ............. 29
The Effect of Unitary Status over Time................................... ............... 30
The Use of Weighted vs. Unweighted Observations .......................... 32
The Exogeneity of Unitary Status ............................................ .............. 33
The Effect of Unitary Status on Academic Achievement.............................. 35
The Effect on the Average Student ...... .... .............. ............... 35
The Effect by Race ....................................... .......... ............... 36
Conclusions and Further Discussion .......................................... 37

3 SCHOOL DISTRICT FINANCE AND UNITARY STATUS IN FLORIDA................. 49

Introduction ....................... .... ... .... ..................................... 49
Florida's Desegregation History and How It Relates to District Size.................... 52
D description of the D ata ......... .............................................. .......... 55
D description of the Em pirical M odels............................................... .................... 58
The Effect of Unitary Status on School District Finance................... ........... 62
Unitary Status and Revenues................. ........... ......... ............... 62
Unitary Status and Expenditures ...... ..................... ............... 66
The Role of Econom ies of Scale................................ .............................. ..... 68

4 A MULTI-STATE ANALYSIS OF UNITARY STATUS AND SCHOOL FINANCE ... 79

Description of the Data ........... ................. .............. .......................... 79
D e scriptio n of the M o de ls ......... ....................................... ................................ 83
The Effect of Unitary Status on Revenues.................................. ...................... 85









The Effect of Unitary Status on Expenditures ............ ...................................... 86
Revisiting Economies of Scale and Final Thoughts......... ...... ...... ............ .. 90

5 C O N C L U S IO N ................................................................. 1 1 1

APPENDIX

A ADDITIONAL TABLES............................................ 115

LIST OF REFERENCES ....... .................. ......... ......... 123

BIOGRAPHICAL SKETCH .............. ........... ............................. 125








































6









LIST OF TABLES

Table page

2-1 D istrict-leve l variable m eans ................................................... ... .................. 39

2-2 Segregation index regressions ................................... .................................... 40

2-3 Effect of selective m igration...................................................... ................... 41

2-4 Unitary status' effect on math achievement..................................................... 42

2-5 Unitary status' effect on reading achievement .............................................. 43

2-6 Unitary status' effect on achievement by race and subject.............................. 44

3-1 District-level revenues per student ........... .............................. ............ 72

3-2 District-level expenditures per student..................................... ..................... 73

3-3 Effect of unitary status on revenues ........................................... ............ 74

3-4 Effect of unitary status on expenditures....................... ..... ............. 75

4-1 Court-ordered desegregation status by state ............ ...... .................. 92

4-2 S um m ary fiscal data by state................................................... .. ................... 93

4-3 The effect of unitary status on total revenues and total expenditures per
stud e nt ................ .................................. ........................... 94

4-4 The effect of unitary status on federal revenues per student ........................... 95

4-5 The effect of unitary status on state revenues per student.............................. 96

4-6 The effect of unitary status on local revenues per student. ............................. 97

4-7 The effect of unitary status on total current expenditures (TCE) per student...... 98

4-8 The effect of unitary status on salaries ........................... ..................... 100

4-9 The effect of unitary status on benefits.......................... ................... 101

4-10 The effect of unitary status on teachers........................ ..... ............... 102

A-1 Litigation status by district .............................................................. 115

A-2 Unitary status' effect on total local, state and federal revenue per student....... 116









A-3 Unitary status' effect on state funding from other sources.............................. 117

A-4 Unitary status' effect on general administration expenses.............................. 118

A-5 Fiscal variable means (all variables)............... ............................................ 119









LIST OF FIGURES


Figure page

2-1 Florida's elementary school enrollment by race.................................... 45

2-2 Average segregation index by unitary status, 1987-2006............................... 46

2-3 Math achievement by unitary status ...... ................... ............. 47

2-4 Reading achievement by unitary status.................................... .................. 48

3-1 Florida's school desegregation over cases over time................................... 77

3-2 Florida's unitary status receipts over tim e .................................. ................... 78

4-1 Timing of unitary status grants for eligible districts (all states)..................... 103

4-2 Timing of unitary status grants for eligible districts by size: Alabama .............. 104

4-3 Timing of unitary status grants for eligible districts by size: Florida. ................. 105

4-4 Timing of unitary status grants for eligible districts by size: Georgia. ............... 106

4-5 Timing of unitary status grants for eligible districts by size: Louisiana.......... 107

4-6 Timing of unitary status grants for eligible districts by size: Mississippi ........... 108

4-7 Timing of unitary status grants for eligible districts by size: North Carolina...... 109

4-8 Timing of unitary status grants for eligible districts by size: South Carolina. .... 110











CPI-U

CLV


DOE

FCAT

FEFP

FRPL

FSIR

FTE

LSF

NCES

SCOTUS

SPAR

TCE


LIST OF ABBREVIATIONS

Consumer Price Index for All Urban Consumers

Refers to a group of authors: Charles Clotfelter, Helen Ladd, and
Jacob Vigdor.

Department of Education

Florida Comprehensive Assessment Test

Florida Education Finance Program

Free or Reduced Price Lunch

Florida School Indicator Report

Fulltime Equivalent

Local, State, and Federal

National Center for Education Statistics

Supreme Court of the United States

School Public Accountability Report

Total Current Expenditures









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE EFFECTS OF UNITARY STATUS ON ACADEMIC ACHIEVEMENT AND
SCHOOL FINANCE: THREE ESSAYS ON THE END OF THE COURT-ORDERED
DESEGREGATION ERA

By

Colin A. Knapp

August 2010

Chair: Lawrence Kenny
Major: Economics

Over the span of three decades, many southern school districts were found to be

operating racially separate school systems and were placed under court-supervised

desegregation plans. While under these plans, districts were forced to remove all

traces of their formerly segregated nature. Upon satisfactorily achieving the U.S.

Supreme Court's goal of one "unitary" system for all children, a district can be released

from court supervision and can subsequently be considered to have received unitary

status.

This study identifies how exiting a court ordered desegregation plan via the

granting of unitary status affects racial imbalance and academic achievement. It finds

significant evidence that the level of segregation will rise if unitary status is received.

This effect is robust and appears to grow over time. Unitary status is also shown to

have little effect on the achievement of the average student within each district. This

result holds for both white and black students at the district level, suggesting that the

end of court supervision may not lead to an increase in academic inequality.









Additional consideration is given to the motives behind the decision to seek unitary

status. During the most active phase of the school desegregation era, the U.S. court

system took control of student assignment away from districts found guilty of operating

segregated school systems. This affected both large and small districts. However, as

the courts retreated from active involvement, large districts were more frequently

released than small ones. One explanation may be economies of scale. If the per-

capita financial benefits of receiving unitary status are small, a large number of students

are required to cover any non-trivial costs associated with seeking release from the

court's desegregation order. This study shows some revenue shifting may occur,

saving local tax payers about $100 per student in Florida. Since there is no clear

indication that any sizable expenditures are forgone, the net effects are small at best.

Therefore, economies of scale may be involved in the decision to seek unitary status,

explaining why large districts receive unitary status more frequently than small districts.









CHAPTER 1
INTRODUCTION

The legacy of segregated schooling is a burden that many states in the traditional

south continue to bear. Whether it is in the form of residential segregation, continued

lawsuits fighting for educational equality, or the winding down of the court-ordered

desegregation era, few parts of the South have avoided being touched in some form.

One phenomenon that many states in the area share in common is the move to end the

role the court system plays in the assignment of students. How individual districts do

this is straightforward. They must be granted unitary status, which is a declaration by

the court that the previously segregated school district has removed all vestiges of the

formerly segregated system and replaced it with one system for all students. Beyond

that simple statement, little else is so clear. Many major questions such as why districts

seek unitary status or how academic outcomes are affected by unitary status have gone

largely unasked in the economic literature even though their answers will play an

important role in the continuing policy debate. The purpose of this study is to help shed

light on what happens when districts receive unitary status.

One prominent role economists perform is to identify both the intended and

unintended consequences of different policies like the granting of unitary status. If the

court's involvement in student assignment via a court-ordered desegregation plan is a

constraint on the local district, removal of the order should produce a number of

identifiable consequences. The evaluation of unitary status as a policy outcome

requires measuring outcomes across many dimensions to ensure that negative, and

often unintended, outcomes do not outweigh the perceived benefits.









Each of the following chapters plays a part in identifying what happens when a

district is granted unitary status. Chapter 2 looks at how students are sorted between

schools before and after unitary status is received and shows that a significant rise in

segregation may occur as a result of receiving unitary status. It is not immediately clear

how this outcome should be interpreted. If the court system has successfully eliminated

the biases that created the educational inequity of the 1950s and 1960s, then such a re-

sorting should have no effect on the students within the re-sorted schools. However, if

some elements of the formerly segregated school system do remain, then students

could be harmed, especially those that gained so much from the court's initial

involvement.

Chapter 2 continues by examining academic achievement during the time that

most of the recent unitary grants were received. The results suggest that there are no

differences in the academic performance based on unitary status. This applies to the

average student, the average black student, and the average white student at the

district level. This result suggests that the re-sorting of students into more segregated

schools does not hurt students along this one dimension.

While Chapter 2 deals primarily with what happens when districts are granted

unitary status, Chapters 3 and 4 begin to study why districts seek unitary status. If

districts are presumed to act rationally, they will only seek unitary status if they believe it

will make them better off than their current status under the court's desegregation plan.

If benefits turnout to be nonexistent or relatively small when compared to the cost of

having the court order removed, then the decision to seek unitary status may be

questionable. One place districts may look to make improvements is school district









finance. The analysis in these chapters reveals that school district finance should not

play a major role in a district's decision to seek unitary status. The granting of unitary

status is shown to have very little effect on the revenues received and the expenditures

spent by districts that receive it.

Since changes in district-level financing do not provide the evidence required to

declare unitary status a beneficial public policy, are we left to assume that the policy is

largely ineffective at best and detrimental at worst? No, that would be too extreme.

Chapter 5 concludes by examining the collective body of results from the study,

identifying limitations based on the available data, and discussing avenues for future

research.









CHAPTER 2
UNITARY STATUS AND ACADEMIC ACHIEVEMENT IN FLORIDA

A Brief History of Court-Ordered Desegregation

Brown v. Board of Education did not have an immediate effect on integration in

southern school districts. It took almost another decade and passage of the Civil Rights

Act in 1964 to strengthen the Supreme Court's demands to expedite desegregation. In

many cases, even financial incentives associated with Title I funding from the U.S.

Department of Education were not enough to get some districts to desegregate. In

those instances, individual court cases were filed in various U.S. District Courts to force

school integration.

Following the court's decision, a plan was typically designed to racially balance

schools within districts by defining strict rules by which students must be assigned.

These plans were effective as the traditional South went from having the most

segregated schools to having the least segregated in about one decade.1 However,

due to the strictness of some court-ordered plans, many districts sought autonomy,

arguing that equal education could be provided without the court's supervision.

Eventually, the U.S. Supreme Court agreed and in 1968 set up a process by which

districts could be released from court supervision and subsequently allowed to assign

students to schools by means other than race. This process became known as unitary

status. 2




1 See Clotfelter (2004), 25-30.
2 For a discussion of all the steps and court cases involved, see Orfield (1996). Jansen (2001) provides a
near first-hand account of one district's attempt to satisfy (or skirt in the early stages of desegregation) the
issues involved in many of the relevant cases.









As defined in Green v. County School Board of New Kent County, a district must

eliminate all vestiges of segregation before it is deemed to be operating a unitary school

system for both black and white students. By evaluating schools on the basis of six

factors, courts determined whether districts had taken the appropriate steps to be

released from a court-ordered desegregation plan. These factors are: student

assignment, faculty assignment, staff assignment, transportation, extracurricular

activities, and physical facilities.3

Two additional cases are important in connecting the fate of unitary status and

the possibilities of resegregation in those districts where it was granted. In Board of

Education of Oklahoma v. Dowell and Freeman v. Pitts, the U.S. Supreme Court

decided that districts did not have to maintain racially integrated schools after being

declared unitary and that unitary status could not be withheld if de facto segregation

would result due to racial segregation in neighborhood residential patterns.4 Due to the

presence of residential segregation in many areas, the likelihood of resegregation would

seem to be the rule rather than an accidental confluence of factors.

The Existing Literature on the Effects of Unitary Status

The literature on unitary status necessarily begins in the area of desegregation.

From the 1970s to the 1990s, there are hundreds of studies discussing desegregation

and its effects. Schofield (2005) reviews over 200 of the earliest studies and draws

several conclusions. First, white students have been held harmless by school

desegregation. Second, there is evidence that black students benefit from



3 391 U.S. 430 (1968)
4 498 U.S. 237 (1991) and 503 U.S. 467 (1992), respectively.









desegregation more in reading than they do in math. This would seem to suggest the

presence of unambiguous gains which could be lost. Therefore, measuring the effect

unitary status may have on reducing these gains becomes a very important area for

analysis.

More recent studies on desegregation have sought to not only measure the effect

desegregation has on economic outcomes, but also the mechanisms by which

desegregation has worked. Two events happened simultaneously during school

desegregation: the racial composition of schools changed and the resources available

to black students increased. Reber (2007) finds that black students in high black

enrollment districts experienced larger improvements in educational attainment than

black students in high white enrollment districts. She attributes this to the differing

ability of school districts to redirect funding prior to desegregation and suggests that the

increases in school funding that accompanied desegregation were more important than

the increase in exposure of blacks to whites. Rivkin (2000) investigates the impact of

racial composition and school quality on twelfth-grade test scores, years of educational

attainment and monthly earnings. His study finds that raising school quality is more

effective than relocating children and trying to take advantage of possible peer effects.

Guryan (2004) finds that dropout rates decreased in districts that desegregated in the

1970s, relative to districts that did not, and that this is true for only black students.

Two recent articles make the jump from desegregation to measuring the effects

of unitary status and possible resegregation. The first, Clotfelter, Ladd, and Vigdor

(2005, CLV hereafter), analyzes data on the 100 largest districts in the traditional South

and former Confederate border states from 1994 2004. CLV argue that increased









racial isolation is not a result of granting of unitary status. Their results show that

although segregation increases over time, recent unitary status declarations are not

associated with these higher levels of segregation. They attribute increases in racial

isolation to increases in the Hispanic school population rather than changes in unitary

status. CLV conclude by stating that increases in the Hispanic population may be

making racial isolation measures irrelevant as far as policy analysis is concerned and

recommend using measures of racial imbalance since district officials cannot control the

demographic changes that cause racial isolation data to vary widely.5

The second paper in this branch of the literature is Lutz (2005). He uses a

nationally representative sample of districts to measure changes in segregation after

districts are released from their court-ordered plans. He finds immediate and gradual

increases in segregation which strengthen CLV's less definitive results. He then ties

these changes to increases in both black dropout rates and private school attendance,

but finds these effects to be statistically significant only in northern states. These

results are presented as evidence that the value to black students of attaining a public

school education falls when unitary status has been granted. However, there is no clear

explanation as to why this theory should hold in the North but not in the traditional

South. CLV and Lutz demonstrate that the literature lacks a consistent answer

concerning unitary status' effect on segregation. Additionally, there is no paper that

directly links unitary status and academic achievement. My research addresses both of

these issues.




5 The differences between racial isolation and racial imbalance are explained later in the chapter.









Description of the Data

To measure the effect of unitary status on segregation and academic

achievement, I focus on elementary school data collected at the district level in the state

of Florida from 1999 to 2007. With the exception of the litigation status data, all data is

publically available from various Florida Department of Education sources. Florida's

elementary schools make a good research subject for many reasons. Florida was one

of the first states to implement a school accountability plan that required the collection of

test scores. This makes year-to-year academic achievement results available further

back in time than in many other states. Florida also has a long history with school

integration that provides a vibrant source of variation.

Data on the desegregation status of each district was obtained from the Florida

Advisory Committee's 2007 Report to the U.S. Commission on Civil Rights Report on

Desegregation in Public School Districts in Florida. This contains detailed information

on each school district and lists whether each district has ever been to court, when it

was taken to court and when the district was granted unitary status if applicable. A

detailed list of each district's desegregation status is available in the Appendix, Table

Al.

From 1956 to 1978, 34 of Florida's 67 school districts were taken to court and

found in violation of Brown. This is significant as it means 84 percent of Florida's

schools have been affected by litigation at some point in time. Of those districts taken

to court, 18 have been granted unitary status. The first release from court order came

in 1970 and the most recent was in 2006. Eight of the 18 releases occur during the

study period. From 1999 to 2007, the percentage of schools affected by unitary status

increased from 29 percent to 66 percent of all schools in the state. This means over









one-third of all schools in Florida experienced a change in unitary status during the

period of this study. I use this variation in the timing of receipt of unitary status to

identify the effect unitary status has on segregation rates and the student test score

distribution.6

The litigation data also provide the justification for how the treatment and control

groups are determined. The receipt of unitary status is conditional on having been

taken to court. There may be differences between those districts that are taken to court

and those that avoided litigation. Therefore, when considering which districts to include

in the control group, having been taken to court is an important determinant for

inclusion. It is also important to consider that districts can only be treated once to the

policy. Since some districts were granted unitary status prior to the start of the study

period, including them in either the treatment or control group is inadvisable because

they cannot be treated again. When put together, this will leave a preferred sample of

24 districts that were eligible for treatment as of 1999. Although results are reported for

various sample configurations throughout the rest of the chapter, more emphasis should

be given to the results which focus on the preferred sample since it is the one most

likely to measure the treatment effect of receiving unitary status.

Racial composition data comes from Florida's School Public Accountability

Reports (SPAR) and is based on attendance in October of each school year. Figure 1-1

lays out the racial composition trends for the state by year. Whites are still the dominant



6 It should also be noted that although never under court order, the remaining 33 school districts are
monitored by the U.S. Department of Education's Office for Civil Rights. In some ways they can be
considered as also having unitary status. This needs to be considered when defining the relevant
treatment and control groups and will be discussed later in the chapter.









racial group, but that dominance is decreasing as the percent of students who are white

decreased from 52 percent in 1999 to 45 percent in 2006.7 The percentage of black

students has decreased slightly from 25 percent to 23 percent. These decreases in

enrollment share are not due to fewer whites and blacks attending Florida's schools.

Instead, these shifts are due to the increased number of Hispanics and other racial

minorities. Figure 1-1 clearly shows that increases in Florida's elementary school

population are mainly within these latter demographic groups.

Monitoring desegregation and measuring its effects requires variables which

express the extent to which different racial groups interact. These variables are divided

into two groups: those that measure racial isolation and those that measure racial

imbalance. Variables such as black-white exposure rates and the percent of black

students attending schools with 90 percent non-white enrollments measure the extent to

which blacks are isolated from white students. Dissimilarity and segregation indexes

measure imbalance within a district by taking into account a given district's racial

composition. They generally provide a ratio of how racial groups are distributed across

schools compared to some ideal level of integration which usually corresponds to a

subgroup's share of the district population.

A key difference between these two types of measures is that racial imbalance

variables are much better at identifying policy effects between districts than those

measuring racial isolation because measures of racial isolation pick up differences in

district-level demographics absent any relevant difference in policy. This tendency can



7 In this case, 1999 and 2006 refer to the 1999-2000 and 2006-2007 school years respectively. This
convention will be used throughout the rest of the chapter. Any deviation from this convention will be
clearly stated, especially when it is necessary to define fiscal years later in the study.









skew the analysis of district-level desegregation efforts. A comparison of two school

districts in Florida illustrates this point. Gadsden County's schools have a student

population that is nearly 79 percent black while schools in Brevard County are only 14

percent black.8 The corresponding exposure rates of blacks to other blacks are 83

percent and 24 percent respectively. A comparison across districts based on only the

two exposure rates may lead some to conclude that Gadsden County's schools are

much more segregated than Brevard County's schools. However, that is not the case.

Once the racial composition differences are accounted for, the segregation indexes

reveal a different story. Segregation in each is a relatively low 0.17, indicating that

students are distributed fairly evenly between all schools in the district given the racial

composition within the district.9 Therefore, the difference in exposure rates is an artifact

of the demographics rather than a result of differing policies since both counties were

once under court-order desegregation plans and subsequently released via the granting

of unitary status. Based on this reasoning, I focus more on racial imbalance than I do

racial isolation for the remainder of the study.

It is still necessary to calculate the exposure rate since it is a component of the

segregation index. The equation for the exposure rate of blacks to whites in any given

year can be stated as:

Z (B *w,)
Exposure Rtle =-
C-Bl (2-1)


8 With few exceptions, school districts in Florida are coterminous with county borders. Most of these
exceptions are for experimental schools assigned their own district or juvenile delinquent programs, all of
which have been excluded from this study.
9 All statistics based on calculations using SPAR data for 2004. The formulas used to calculate these
statistics are discussed later in this section.









In this case, k indexes the district, B, represents the total number of blacks in

school i, and w, is the percentage of the students in school i who are white. In Table 2-

1, the exposure rate of blacks to whites can be interpreted as meaning that from 1999-

2006 the average black student in Florida attended a school that was 55 percent white.

The exposure rate is then used to calculate a segregation index of the following

form:

SegregationIndexk (wk ExposureR,te )/wk (2-2)

Again, k indexes each district and wk represents the percent of students in district

kwho are white. This index can take on a value ranging from zero (perfect integration)

to one (complete segregation). From 1999 to 2006, the average segregation index in

Florida increased from 0.14 to 0.16. This appears to show very little change over time

given the number of schools exposed to unitary status changes and is hardly near the

U.S. Department of Education's acceptable upper-threshold of 0.60. However, a closer

look shows significant variation between district types. Column 2 of Table 2-1 shows

that the average level of segregation is higher in the 34 districts that have been taken to

court than in the state as a whole. Additionally, the districts that have been released

from court order have an average segregation rate that is more than twice that of the

districts still under court supervision.

How much of this difference is due to additional districts being granted unitary

status during the time period 1999-2006? Due to the relatively short sample period that

can be constructed using publically available data from the Florida DOE, it is not

possible to look at prior trends in the level of segregation before 1999. However, it is

possible using qualitatively similar data from the U.S. DOE's Common Core of Data









which goes back as far as 1987. Preliminary evidence is provided in Figure 2-2. From

this graph, it is evident that the average level of segregation began increasing in the

early 1990s. However, this increase was not uniform across different district types. The

only districts not to see an increase in the level of segregation are those that remained

under court supervision. Those districts released during the sample period experience

first a decrease and then an increase in segregation as more districts are released from

their court orders. Those districts never taken to court and those released prior to the

beginning of the sample period also experience an increase in the level of segregation.

These trends suggest a positive relationship between unitary status and the level of

segregation. Additionally, the divergent path taken by districts released from court order

suggests that a variable mapping the receipt of unitary status over time will identify

more than just the continuation of pre-existing trends. It appears that the court orders

did provide some constraint on assignment policies and that when removed, students

will re-sort into an increasingly segregated distribution.

All other district-level variables were obtained from the Florida School Indicator

Reports (FSIR) from each district. Due to the small sample size, only a limited set of

variables are used to control for various district characteristics. The fraction of students

using the district's free lunch program is meant to capture the socio-economic

characteristics of the school population, while the fractions of students with limited

English proficiency, disabilities and those considered gifted are meant as proxies of the

student body's ability.10 Absences, suspensions, and average teacher experience


10 There is variation across districts in what qualifies students for inclusion in the gifted program. This
may limit the variables effectiveness as a proxy for ability. However, the existence of these differences
highlights the need to control for this source of variation in districts.









proxy for school quality. Total district size and per capital real income in each district

account for potential differences in parental inputs across districts.

The remaining variables in Table 2-1 are the dependent variables used to

estimate whether the receipt of unitary status affects academic achievement. These

variables measure the amount of mass in the tails of the student testing distribution.

The FSIR lists the fraction of fourth graders and fifth graders who test into one of five

levels on the reading and math portions of the Florida Comprehensive Assessment Test

(FCAT). I combine the lowest and highest two levels in order to measure movement

into and out of the tails of the distribution caused by the policy change. From 1999 -

2006, there was a rightward shift in the testing distribution with a lower percentage of

students testing into Levels 1 or 2 and a higher percentage testing into Levels 4 or 5.

This is apparent in Figures 2-3 and 2-4. It should also be noted that in all four cases,

the trend for those districts currently under court order moves in tandem with the trend

for those districts already released from court order. This suggests that achievement

gains may have occurred for reasons other than the granting of unitary status.

Description of the Empirical Models

I use two main specifications in this chapter. The first tests to what extent

changes in the segregation index are caused by the granting of unitary status. This is a

more formal statement of the relationship illustrated in Figure 1-2. It takes the form:

SegregationIndexk, = PAk + + X 11, + //,i1 /vi/u/ + uk (2-3)

This measures the segregation index in district k during year tas a function of a

county-level fixed effect( Ak), a year fixed effect (Y,), district-level controls varying by

year (Xa), a dummy variable indicating whether district kwas granted unitary status in









year t and a random error term.11 To control any within-district correlation of the

standard errors, I cluster the standard errors at the district level. For the coefficient on

unitary \ltin\ to be properly identified the following condition must hold, unitary status

must be received randomly. Whether it is reasonable to assume that this condition

holds is discussed in the next section along with the results and specification tests.

In its most robust form, the matrix of district-level variables (Xka) contains the

percent of students in district kwho are Hispanic, the district's total elementary school

population and a county-level measure of real per-capita personal income derived from

the Bureau of Economic Analysis' Local Annual Personal Income Data Series. The first

variable is included to account for demographic variation within each county. The last

two variables are meant to capture the level of sorting which takes place within county.

Based on Tiebout's hypothesis, I expect larger counties to have higher segregation

indexes because there will be greater opportunities for sorting between a larger number

of schools. This phenomenon works through the correlation of district size, the number

of schools and the presence of residential segregation within most areas. The effect of

income should also be positive as white parents, with more income and children in

black-majority schools, can opt for private education or can more easily move to areas

of a county with schools with a racial mix they prefer.

The second model looks at the distribution of academic achievement and how it

is affected by the granting of unitary status. Its basic form is:

Yk, = ?Ak + YI + IXkt + P3unitary ,,,u + ,, (2-4)



11 I would prefer to use district-year fixed effects. However, each district-year cell contains only one
observation, making a district-year fixed effect impossible to use in this instance.









In this specification, I use the mass in the tails of the student testing distribution in

both Math and Reading as the values for Yk. Ak and Y, are district and year fixed

effects. X, controls for time varying characteristics between districts. Unitary status is

measured using a dummy variable which equals one if the district is deemed unitary in

that year and zero otherwise. Standard errors are again clustered at the district level.

Whether unitary status is received randomly is important to the identification of 83 and

the applicability of this condition will be discussed in the next section.

As in the previous specification, the matrixX, accounts for district-level variation

in size, income and demographics. In addition to the three previous variables, parental

inputs are accounted for by including the percent of students utilizing the district's Free

or Reduced Price Lunch Program. Academic achievement is expected to fall as the

percentage of poorer students in a district increases. Student inputs, and to the extent

possible student ability, are accounted for by the percent gifted, percent with limited

English proficiency and the level of absenteeism. More gifted students should increase

achievement, while higher levels of limited English students and absenteeism should

decrease it as less of the academic program can be absorbed. School inputs and the

general school environment are represented by the average years of experience of the

teachers and the percentage of students suspended during the year. I expect more

experienced teachers to be more efficient at teaching which should increase

achievement.12 More disruptive environments should reduce academic achievement.


12 It can be argued that higher test scores really do not represent an increase in student ability. Teacher
experience exemplifies this phenomenon because it is impossible to determine if more experienced
teachers are imparting more knowledge or are simply better at "teaching to the test" and making their
students more effective test takers.









Results of the Empirical Models

Unitary Status' Effect on Segregation

Up to this point, I have not addressed the issue of defining the appropriate

treatment and control groups in this quasi-experimental policy analysis. The treatment

group is easy to identify. It is the group of districts which experienced a change in

unitary status during the eight-year period starting in 1999 and continuing through 2006.

This is a group of eight districts representing over one-third of the schools in Florida.13

Defining the relevant control group is more problematic since there are several

possibilities. Three types of districts must be considered for inclusion in the control

group; those never taken to court; those taken to court and released from court order

prior to 1999; and those still under court order after 2006. To examine these

differences, I use several different samples and report the results in Table 2-2. Column

1 includes all the other districts in Florida in the control group and suggests that the

granting of unitary status will have an immediate and statistically significant effect on the

segregation level, raising the index by 2.4 percentage points. This represents an

increase of 15 percent at the sample mean of 0.15. The first alternative case to

consider involves the difference between those taken to court and those that have not

faced litigation. Most of the non-litigants are located at the lower end of the population

size distribution. If the effects of unitary status are somehow correlated with

unobserved characteristics associated with district size, the previously found positive

and significant estimate may be inconsistent.



13 This group and their respective years of receiving unitary status are: Dade (2001), Duval (2001),
Escambia (2004), Hillsborough (2001), Lee (1999) Pinellas (2000), Polk (2000) and Seminole (2006).
See Table Al in the Appendix for a detailed account of each district's unitary status history.









Therefore, I exclude non-litigants from the control group and rerun the analysis.

These results are listed in Column 2 of Table 2-2. This restriction reduces the number

of districts in the analysis to the 34 that have been taken to court. The effect of unitary

status differs by only 0.1 percentage points and the parameter estimate is still significant

at the five-percent level. This continues to suggest that unitary status will increase the

level of segregation. This is noteworthy considering that the already small sample size

was reduced by half and the statistically significant result was maintained.

Even though the result remained unchanged while restricting the definition of the

control group to only those districts that have been taken to court, it is possible that the

control group may still be improperly defined. In Column 2 of Table 2-2, the control

group contained observations from 26 counties who had either already received unitary

status prior to 1999 or were still awaiting adjudication after 2006. The noteworthy

difference here is that the first group has already received unitary status and is unable

to be treated again by the same policy. Therefore, it may be reasonable to exclude

these counties from the control group as well. The results for this specification are

listed in Column 3 and show that unitary status still has a positive and significant effect

on the segregation index. When combined, the results provide robust evidence that the

granting of unitary status may raise the level of segregation by roughly 15 percent within

the average district.

The Effect of Unitary Status over Time

Prior to this point in the chapter, I have assumed that the granting of unitary status

causes a single shock to an affected district. This is unlikely and Pinellas County's

experience is a perfect example. The U.S. District Court granted Pinellas County

unitary status in 2000. However, the agreed upon exit strategy did not alter the court-









ordered desegregation plan until 2003. In other words, even though granted unitary

status in 2000, Pinellas County continued to use the assignment plan dictated by the

court-ordered desegregation plan for two more years. The first year that significant

amounts of re-sorting could occur due to unitary status was in 2003, not 2000.

Therefore, it is misleading to assume that the treatment of receiving unitary status is

realized immediately. It is also unlikely that the parents who react to the decision to

remove the court-ordered plan do so immediately. It is likely that parents move children

between schools for different reasons and at different paces. This results in the

segregation index changing incrementally rather than instantaneously.

This scenario is at odds with the one-time policy models estimated in the

previous section. To allow the effect to vary over time, I alter the specification to include

a set of dummy variables which captures the effect of unitary status over time. I

continue this process as far as eight years post-implementation and have a specification

of the following:

8
Segegationlndexkt = 8,A, + 1,Yk + ,2Xkt + / Duration kt + ukt
= (2-5)

The previous variable of interest, unitary t\ltiln is subsumed by the duration

dummies.

The results shown in Column 4 of Table 2-2 reveal much more about how unitary

status affects segregation over time. What becomes most apparent from this

specification is the extent to which the effect grows over time. The immediate effect of

granting unitary status is still statistically significant at the five percent level and has a

magnitude of 1.7 percentage points. The coefficients on every duration variable are

not only statistically significant, but also increase monotonically. By the eighth year









following the policy change, the segregation index has increased cumulatively by 20.6

percentage points. Depending on the average district-level mean used, this represents

an increase in segregation ranging from 71 to 137 percent. Also remarkable is the

increase in the effect between years three and four. At that time, the cumulative effect

more than doubles from 3.0 percentage points to nearly 6.9. This is indicative of the

two scenarios discussed previously. First, it is unlikely that districts implement policy

changes immediately. Secondly, it is also unlikely that parents can react simultaneously

with the district's implementation of new attendance plans. Ultimately, the results in

Column 4 suggest that unitary status sorts students into an increasingly less integrated

educational environment and that this effect accelerates at the four-year point.

The Use of Weighted vs. Unweighted Observations

All of the results discussed previously were observed using weighted least

squares with the weighting based on total district size. This forces the regression to

place extra weight on the observations from larger districts. I chose to do this due to the

wide disparity in size across districts in Florida. In the specification using all 67 districts,

the difference between the largest and smallest is almost 175,000 students. With this

wide a disparity, it appears unwise to allow the smallest district to have as much weight

as the largest. However, there is disagreement over the effect this procedure may have

on the size of the standard errors and the statistical significance of the results.

To counter this possibility, I also measured the effect of unitary status on

segregation using a sample of districts with a smaller dispersion. This sample includes

the ten largest districts in the state excluding Miami-Dade.14 This shrinks the difference


14 Miami-Dade is the largest district in the state with nearly 175,000 elementary school students. The
next closest district is Broward County which has over 50,000 fewer students.









between the largest and smallest district by half. Although the size disparity is still

large, it is much more likely that these districts are similar, making weighting each

observation no longer necessary. The results of this specification are presented in

Column 5 of Table 2-2. For each duration variable, the coefficient is similar in

magnitude but measured less precisely. This may be due to the lack of weighting, but

could also be due to the extremely small sample. It should be noted that the cumulative

effect is still highly significant and still represents a near doubling of the level of

segregation, even without the advantage of weighting individual observations.

The Exogeneity of Unitary Status

In order for unitary status to have a statistically identifiable effect on the level of

segregation, the timing of release from court order must be exogenous. To that end, it

should be noted that two actions must take place before a district can receive unitary

status. The school district must decide to seek release from their court ordered

desegregation plan and the court must then grant it. Any assessment of the exogeneity

of unitary status must take into account both parts of the process. The first step

certainly appears to be affected by district size.15 From 1999 to 2006, eight districts

received unitary status. Of those grantees, seven are among the ten largest districts in

the state while the smallest ranks thirteenth out of 67 districts. This relationship

between district size and the decision to seek unitary status could confound the effect of

unitary status on the segregation index if the process were solely up to the discretion of

individual school districts. However, that is not entirely the case since the court's role

appears to add back a significant amount of randomness into the timing of release.


15 This phenomenon is the primary focus of Chapter 2.









It has been noted that district courts changed their behavior towards

desegregation cases in the early 1990s.16 Given the court's newfound desire to clear its

docket of desegregation cases, it is hard to construct a scenario where a court's

decision to release a district is dependent on district size. They were willing to release

any district, regardless of size. Nonetheless, it is possible to check if this is the case by

looking at the level of correlation between the year of release from court order and

district size. For those districts released during the period of this study, the level of

correlation is a relatively low -0.22. Therefore, it may be reasonable to assume that the

actual timing of the unitary status grants is random enough to allow identification of an

effect on the segregation index.

Furthermore, if district size were causing bias in the effect of unitary status on the

segregation index, changing the composition of the sample by using only small or large

districts should change the magnitude of the coefficients on the unitary status variable.

It is apparent from the results shown in Table 2-2 that this is not the case. As I

decrease the sample size in Columns 1, 2 and 3, it is predominantly smaller districts

that are dropped from the sample. Even after this, the coefficient on unitary status

remains almost unchanged in magnitude. The same holds in Columns 4 and 5 under a

slightly different strategy. This is especially remarkable given that column five is solely

based on a sample representing 10 of the 11 largest districts. Again, it appears that the

timing of release is quasi-random.

Another possible issue is caused by the way the segregation index is calculated.

Since it uses both the black and white student populations in its construction, the index


16 See Chemerinsky (2005) for a detailed discussion of this change.









could be susceptible to an issue called selective migration, a phenomena where one

race enters or exits a district based on the existence of, or lack of, a court-ordered

desegregation plan. Selective migration would be evident by divergent racial trends

between those districts that have been granted unitary status and those that have not.

To test whether unitary status causes selective migration, I run a set of regressions

based on the following model,

Black or White Student Populationt = Akt + unitary \//lit/ + district size +
(2-6)
unitary status White + White + uk

This specification uses either the district's white or black student population as the

dependent variable. On the right-hand side, I use total district size, a dummy variable

mapping the timing of unitary status and an interaction term between unitary status and

a dummy variable equaling one for those observations where the dependent variable is

the white student population. If the coefficient on this interaction term is significantly

different from zero, selective migration may be an issue because unitary status affects

white and black student population trends differently. The results of this regression are

shown in Table 2-3. Since the coefficient on the interaction term is not statistically

significant, it appears that selective migration is not an issue and the effect of unitary

status on the segregation index is properly identified.

The Effect of Unitary Status on Academic Achievement

The Effect on the Average Student

Since it appears that unitary status is affecting the way students sort into schools

and increasing the level of segregation, it is important to identify any academic costs

associated with these changes. To do this, I measure academic achievement as

outlined in Equation 2-5. The results for Math are presented in Table 2-4 while the









results for Reading are presented in Table 2-5. In each table, the left panel represents

specifications testing the effect unitary status has on the highest two score categories of

the FCAT.17 The right side presents the same specifications using the lower two

categories. This procedure measures the density in the tails of the achievement

distribution. If unitary status has a beneficial effect on achievement, I would expect to

see positive coefficients in the first three columns and negative coefficients in the last

three. Conversely, a shift towards the lower tail would require negative coefficients for

specifications testing the higher end of the achievement distribution and positive

coefficients on those testing the lower end.

From the Math results in Table 2-4, there appears to be no significant evidence

that the achievement distribution is affected by the receipt of unitary status. For each

specification in the table, the coefficient measuring the effect of unitary status is too

imprecisely measured to identify anything other than a zero effect. For Reading, the

only result of any statistical significance is the final specification which suggests a

possible decrease in the density at the low-end of the achievement distribution of 1.5

percent.

The Effect by Race

Based on the results of the previous section, it may seem reasonable to conclude

that unitary status has no measurable effect on student achievement in the average

affected district. Such a statement is certainly premature. The tests completed above

only account for the net effects of all students within the average district. Since the

tests above combine all racial subgroups into one observation, it is possible that gains

17 There are five categories which students can test into on the FCAT. The dependent variable is either
the sum of the percentage of students testing into the two highest categories or the two lowest categories.









by whites are offsetting losses by blacks. Considering the original purpose of school

desegregation reforms and the fears associated with the effects of resegregation,

determining if this effect on achievement varies by race is possibly the most important

policy outcome of the paper.

Unfortunately, the data used for the achievement distribution is not broken out by

race and cannot be used in that manner. However, the state does report the

percentage of students of each race that are considered proficient (at grade level) in

both Math and Reading.18 This allows the analysis to be completed by race using

equation 2-5 with kt now being the percentage of either black or white students who are

considered proficient. The results for these tests are presented in Table 2-6. Unitary

status does not appear to have a significant effect on either white or black proficiency in

either subject. This extends the findings to include not only the average student, but

also the average white and black student at the county level. It appears that neither is

being harmed by the end of the court-ordered desegregation via the granting of unitary

status.

Conclusions and Further Discussion

The purpose of this chapter was to identify the effects that unitary status has on

racial imbalance and academic outcomes. I find significant evidence that the level of

segregation will rise if unitary status is received. This effect is robust to changes in the

relevant control group and appears to grow over time. The majority of the effect

appears to occur three to four years after unitary status is received, suggesting that it

takes time not only for districts to implement new assignment plans, but for parents to

18 Unlike the achievement distribution data, the proficiency data is not collected at the elementary school
level. It is a measure of all white or black students within a district regardless of grade.









react and move their children between schools if desired. Unitary status was also

shown to have no effect on the academic achievement distribution. This finding holds

for the average student, the average white student and the average black student in an

affected district.

These results are consistent with findings in the recent literature. Reber (2007)

notes that black students in Louisiana benefitted more from the increased level of

resources given to them as a result of desegregation rather than the increased

interaction with white students. Rivkin (2000) also describes how black students

benefitted more from increases in school quality than greater interaction with white

peers. Both find that of the two changes brought by desegregation, resources appear to

matter most. If their findings are correct, then resegregation caused by the end of the

court's involvement should have a minimal effect on student outcomes. As this era

ends, unitary status involves only one part of the process seen in both Reber and

Rivkin. It allows peer groups to change while holding resources constant. I find that re-

sorting does occur, as shown by increases in the level of segregation, and that the

effects of that re-sorting on academic achievement are zero. This not only strengthens

the validity of these previous studies but suggests that there are not any identifiable

academic costs associated with the granting of unitary status.










Table 2-1. District-level variable means
(1) (2) (3) (4) (5)
Court Order Only
Released
Full From Currently Under
Number of Districts in Sample 67 34 18 16 P < 0.05
Fraction White 0.64 0.58 0.51 0.64 Yes
Fraction Black 0.19 0.24 0.29 0.29 Yes
Fraction Hispanic 0.12 0.13 0.15 0.11 Yes
Exposure Rate (Blacks to Whites) 0.55 0.48 0.37 0.57 Yes
Segregation Index 0.15 0.19 0.28 0.13 Yes
Fraction of Schools Ever Under Court Order 0.84 1.00
Fraction of Schools Granted Unitary Status 0.29 0.53
Fraction of Students with Free/Reduced Price Lunch 0.55 0.54 0.53 0.53 No
Fraction of Students Considered Gifted 0.03 0.03 0.04 0.02 Yes
Fraction of Students with Limited English Proficiency 0.05 0.05 0.07 0.04 Yes
Teachers Average Years of Experience 13.25 13.37 13.24 13.49 No
Fraction of Students Absent 21+ Days 0.08 0.07 0.07 0.07 No
Fraction of Students Receiving In School Suspensions 0.03 0.03 0.02 0.04 Yes
Fraction of Students Receiving Out of School Suspensions 0.03 0.03 0.03 0.03 No
Fraction of Students Testing in Levels 4 and 5 (Math) 0.20 0.26 0.28 0.25 Yes
Fraction of Students Testing in Levels 1 and 2 (Math) 0.56 0.49 0.48 0.51 Yes
Fraction of Students Testing into Levels 4 and 5 (Reading) 0.22 0.30 0.31 0.29 No
Fraction of Students Testing into Levels 1 and 2 (Reading) 0.50 0.38 0.37 0.39 No
District Size 18,344 30,633 46,221 17,573 Yes
Per Capita Real Income ($) 26,258 25,361 28,468 22,758 Yes
Percent of Black Students Proficient in Reading 33.28 33.97 34.64 33.21 No
Percent of Black Students Proficient in Math 34.49 34.71 35.70 33.6 Yes
Percent of White Students Proficient in Reading 61.44 63.24 65.54 60.65 Yes
Percent of White Students Proficient in Math 65.72 67.52 70.07 64.68 Yes
NOTE: Column 1 shows district-level means for all 67 school districts in Florida. Column 2 restricts the sample to the 34 districts ordered by
federal courts to desegregate. Column 3 further restricts the sample to the 18 districts that have been released from the court's
desegregation order (i.e. granted unitary status). Column 4 provides means for those districts ordered to desegregate and remaining under
court supervision through 2007. Column 5 reflects the results of a two-tailed test for equality of means at the five percent-level.









Table 2-2. Seqregation index regressions


Unitary Status Granted

Fraction Hispanic (District)


Total Student Population (District)


Per Capita Income (District)


Duration1


Duration2


Duration3


Duration4


Duration5


Duration6


Duration7


Duration8

Constant

Observations
R-squared
Weighted by District Size


(1)
0.024
(2.75)**
-0.081
(0.27)
2.13*10-6
(0.96)
7.43*10-6
(2.12)*


0.059
(0.73)
536
0.98
Yes


(2)
0.026
(2.65)*
-0.189
(0.41)
3.01*10-6
(1.29)
1.23*10-5
(1.84)


-0.061
(0.41)
272
0.98
Yes


(3)
0.028
(2.54)*
0.097
(0.18)
2.47*10-6
(0.86)
1.88*10-5
(1.96)


-0.462
(0.95)
192
0.98
Yes


-0.304
(0.62)
3.39*10-6
(2.03)
7.80*10-6
(1.08)
0.017
(2.48)*
0.024
(2.46)*
0.03
(2.27)*
0.069
(4.22)**
0.079
(3.66)**
0.096
(3.82)**
0.111
(2.82)**
0.206
(9.39)**
-0.134
(0.55)
192
0.99
Yes


(5)


-0.661
(0.72)
4.89*10-6
(1.30)
1.92*10-5
(1.64)
0.024
(1.47)
0.023
(1.11)
0.022
(0.99)
0.07
(2.42)*
0.088
(2.48)*
0.096
(2.23)
0.113
(2.15)
0.206
(4.70)**
-0.285
(1.24)
80
0.98
No


NOTE: All models include district-level and year fixed effects. Models weighted by the total district
size have errors clustered at the district-level. Absolute Value of t-statistics in parentheses, *
represents significance at the five percent level and ** represents significance at the one percent
level. Columns 1 uses the full sample of districts. Column 2 uses only those districts that were
subject to court order. Columns 3 and 4 further reduce the sample by excluding those districts
granted unitary status prior to 1999. Column 5 uses only the 10 largest districts that were under
court order and were eligible for treatment, excluding Miami-Dade.









Table 2-3. Effect of selective migration
Unitary Status Granted 1,798.77
(0.54)
Unitary Status Granted*White -6,092.33
(0.82)
White=1 4,107.74
(0.66)
Total Student Population (District) 0.258
(4.78)**
Constant -316.41
(0.10)
Observations 384
R-squared 0.52
NOTE: This model includes district-level and year fixed effects for a sample
made up of districts that were under court order and eligible for unitary status in
1999. Observations are weighted by total district size. The dependent variable
is either the white or black student population in a each district. Absolute value
of t statistics in parentheses, represent significance at the five percent level
and ** represent significance at the one percent level.










Table 2-4. Unitary status' effect on math achievement
Levels 4 and 5 (High) Levels 1 and 2 (Low)
(1) (2) (3) (4) (5) (6)


Unitary Status Granted


Fraction White, District


Fraction Absent > 21 Days


Fraction Disabled


Fraction w/ Free or Reduced Price Lunch


Fraction Gifted


Fraction Limited English Proficiency


Fraction of Staff Dedicated to Instruction


Fraction Suspended In School


Fraction Suspended Out School


Fraction of Teachers w/ Advanced Degrees


Teacher Average Years of Experience


Real Personal Consumption (Income)


District Size


Constant


Observations
Number of Districts
R-sauared


0.003
(0.47)
-0.044
(0.26)
0.064
(0.57)
-0.019
(0.08)
0.065
(1.12)
1.046
(1.82)
-0.018
(0.10)
-0.064
(0.92)
0.276
(1.20)
-0.315
(0.98)
-0.017
(0.39)
0.000
(0.24)
<.000
(1.16)
<.000
(0.38)
0.107
(0.60)
536
67
0.9


0.001
(0.15)
-0.156
(0.66)
0.049
(0.48)
0.131
(0.38)
0.148
(2.18)*
1.202
(1.75)
-0.027
(0.14)
-0.140
(1.04)
0.251
(0.69)
-0.262
(0.55)
-0.034
(0.71)
-0.001
(0.34)
<.000
(0.82)
<.000
(0.56)
0.135
(0.77)
272
34
0.91


0.002
(0.26)
0.202
(0.80)
0.194
(2.18)*
-0.016
(0.04)
0.115
(1.46)
-0.024
(0.03)
0.229
(1.17)
-0.263
(1.25)
0.098
(0.23)
0.145
(0.31)
-0.092
(2.74)*
0.003
(1.45)
<.000
(0.07)
<.000
(1.25)
0.479
(1.74)
192
24
0.91


-0.014
(1.29)
-0.034
(0.19)
-0.007
(0.05)
0.094
(0.41)
-0.07
(1.41)
-1.227
(1.92)
-0.077
(0.41)
0.058
(0.94)
-0.236
(0.92)
0.205
(0.63)
0.02
(0.43)
0.001
(0.24)
<.000
(1.47)
<.000
(0.91)
0.697
(3.75)**
536
67
0.92


-0.011
(1.12)
-0.002
(0.01)
0.014
(0.11)
0.066
(0.20)
-0.106
(1.43)
-1.524
(1.97)
-0.042
(0.24)
0.179
(1.01)
-0.324
(0.88)
0.003
(0.01)
0.013
(0.24)
0.003
(1.01)
<.000
(1.41)
<.000
(0.85)
0.76
(3.73)**
272
34
0.93


-0.008
(1.03)
-0.325
(1.05)
-0.132
(1.33)
0.11
(0.26)
-0.07
(0.85)
-0.072
(0.06)
-0.362
(1.51)
0.292
(1.14)
-0.085
(0.21)
-0.251
(0.48)
0.119
(2.96)**
-0.001
(0.30)
<.000
(0.50)
<.000
(1.64)
0.138
(0.42)
192
24
0.93


NOTE: All specifications use either the percent of students testing into Levels 4 and 5 or Levels 1 and 2 on
the Math portion of the FCAT as the dependent variable and are run with district-level and year fixed effects.
Columns vary by the sample size used in the regression. Absolute value of t statistic in parenthesis, *
represents significance at the five percent level and ** represents significance at the one percent level.










Table 2-5. Unitary status' effect on reading achievement
Levels 4 and 5 (High) Levels 1 and 2 (Low)
(1) (2) (3) (4) (5) (6)


Unitary Status Granted

Fraction White, District

Fraction Absent > 21 Days

Fraction Disabled

Fraction w/ Free or Reduced Price Lunch

Fraction Gifted

Fraction Limited English Proficiency

Fraction of Staff Dedicated to Instruction

Fraction Suspended In School

Fraction Suspended Out School

Fraction of Teachers w/Adv. Degrees

Teacher Average Years of Experience

Real Personal Consumption (Income)

District Size

Constant


Observations
Number of Districts
R-squared


-0.004
(0.52)
0.289
(2.90)**
-0.177
(2.45)*
-0.024
(0.14)
-0.008
(0.21)
0.780
(2.61)*
0.058
(0.70)
0.082
(1.50)
-0.150
(0.87)
-0.241
(1.28)
-0.014
(0.42)
0.001
(0.95)
0.000
(0.93)
0.000
(1.10)
-0.055
(0.50)
536
67
0.95


-0.005
(0.57)
0.226
(1.80)
-0.194
(2.80)**
0.037
(0.15)
0.002
(0.04)
1.071
(3.25)**
0.087
(0.83)
0.054
(0.56)
-0.283
(0.90)
-0.224
(0.72)
-0.028
(0.67)
0.000
(0.12)
0.000
(1.83)
0.000
(0.66)
-0.062
(0.41)
272
34
0.96


0.002
(0.19)
0.054
(0.31)
-0.156
(3.07)**
0.104
(0.39)
-0.075
(1.39)
0.789
(1.51)
0.185
(1.13)
-0.045
(0.31)
-0.054
(0.17)
-0.775
(1.90)
-0.058
(1.36)
0.002
(1.39)
0.000
(1.46)
0.000
(0.11)
0.116
(0.42)
192
24
0.96


-0.024
(1.16)
-0.424
(2.10)*
0.278
(2.66)**
0.251
(1.00)
-0.010
(0.22)
-1.385
(2.92)**
-0.122
(0.73)
-0.141
(1.85)
-0.005
(0.03)
0.180
(0.75)
-0.094
(1.27)
-0.003
(1.57)
0.000
(1.62)
0.000
(0.53)
0.904
(3.99)**
536
67
0.95


-0.022
(1.27)
-0.481
(1.86)
0.312
(3.10)**
0.335
(0.89)
0.046
(0.69)
-1.917
(3.79)**
-0.163
(1.03)
-0.014
(0.09)
0.020
(0.07)
0.257
(0.66)
-0.107
(1.32)
-0.001
(0.68)
0.000
(2.77)**
0.000
(0.60)
1.216
(4.64)**
272
34
0.96


-0.028
(1.88)
-0.418
(0.94)
0.261
(1.92)
0.424
(1.00)
0.114
(1.57)
-1.058
(1.17)
-0.365
(1.45)
0.240
(1.00)
-0.116
(0.34)
1.003
(1.86)
-0.041
(0.56)
-0.003
(1.02)
0.000
(1.61)
0.000
(0.43)
0.557
(1.08)
192
24
0.95


NOTE: All specifications use either the fraction of students testing into Levels 4 and 5 or Levels 1 and 2 on
the Reading portion of the FCAT as the dependent variable and are run with district-level and year fixed
effects. Columns vary by the sample size used in the regression. Absolute value of t statistic in
parentheses, represents significance at the five percent level and ** represents significance at the one
percent level.









Table 2-6. Unitary status' effect on achievement by race and subject
Black


White


Math Reading Math Reading
(1) (2) (3) (4)


Unitary Status Granted

Fraction White, District

Fraction Absent > 21 Days

Fraction Disabled

Fraction w/ Free or Reduced Price Lunch

Fraction Gifted

Fraction Limited English Proficiency

Fraction of Staff Dedicated to Instruction

Fraction Suspended In School

Fraction Suspended Out School

Fraction of Teachers w/ Advanced Degrees

Teacher Average Years of Experience

Real Personal Consumption (Income)

District Size

Constant

Observations
Number of Districts
R-squared


-0.111
(0.06)
-14.718
(0.35)
-2.308
(0.20)
33.869
(0.91)
-4.072
(0.53)
128.845
(1.18)
0.330
(0.02)
-3.658
(0.19)
8.811
(0.51)
20.748
(0.37)
5.235
(0.81)
-0.202
(0.88)
0.001
(1.90)
<0.000
(0.38)
38.232
(1.10)
120
24
0.96


-0.658
(0.35)
-13.600
(0.30)
-5.623
(0.44)
64.937
(1.59)
-11.742
(1.53)
131.893
(1.11)
2.510
(0.16)
-3.482
(0.23)
-27.832
(1.17)
42.326
(0.77)
3.398
(0.82)
-0.025
(0.12)
0.001
(1.59)
<0.000
(0.99)
35.940
(1.12)
120
24
0.95


-0.384
(0.24)
29.144
(1.25)
6.305
(0.94)
68.295
(1.79)
5.942
(1.26)
52.004
(0.76)
20.605
(1.65)
15.094
(1.41)
40.311
(2.22)*
55.001
(1.16)
4.534
(1.06)
-0.217
(1.19)
0.001
(2.46)*
<0.000
(0.60)
10.378
(0.57)
120
24
0.98


-1.221
(0.91)
10.294
(0.31)
-3.495
(0.44)
84.735
(2.05)
-2.728
(0.44)
109.703
(1.24)
6.246
(0.52)
4.183
(0.38)
10.514
(0.61)
15.164
(0.34)
8.226
(2.14)*
-0.022
(0.09)
<0.000
(0.96)
<0.000
(1.04)
31.870
(1.26)
120
24
0.97


NOTE: The dependent variable in each column is the percent of the subgroup that
is proficient in a particular subject. Absolute value of t statistic in parentheses, *
represents significance at the five percent level and ** represents significance at the
one percent level.











Total State Enrollment by Race


0 ------------
1999 2000 2001 2002 2003 2004 2005

White Black
Hispanic Other
Source: FLDOE October Membership Files


Figure 2-1. Florida's elementary school enrollment by race


2006













Average Segregation Index, 1987-2006

............







.. .. .. ...


I I I I
1990 1995 2000 2005

-- All Districts Never Under CO
-- Received Unitary Status Still Under CO in 2007
-- Granted US Before 1987
Source. USDOE, Common Core of Data


Figure 2-2. Average segregation index by unitary status, 1987-2006.











High Achieving Math Students, 1999-2006


1998


2000


2002
year


2004


2006


-*- Density, US Received-NO Density, US Received-YESI
Source: FLDOE October Membership Files and Author's Calculations


Low Achieving Math Students, 1999-2006


I I I I I
1998 2000 2002 2004 2006
year
-*-- Density US Received=NO Density, US Received=YE
Source: FLDOE October Membership Files and Author's Calculations


Figure 2-3. Math achievement by unitary status











High Achieving Reading Students, 1999-2006


1998


2000


2002
year


2004


2006


-*- Density, US Received-NO Density, US Received=YESI
Source: FLDOE October Membership Files and Author's Calculations


Low Achieving Reading Students, 1999-2006


I I I I I
1998 2000 2002 2004 2006
year
-*- Density. US Received-NO Density. US Received-YE
Source: FLDOE October Membership Files and Author's Calculations


Figure 2-4. Reading achievement by unitary status










CHAPTER 3
SCHOOL DISTRICT FINANCE AND UNITARY STATUS IN FLORIDA

Introduction

In some regards, the most active and unhindered period of court-ordered

desegregation was short lived. From its start with Brown v. School Board, only 14 years

passed before the U.S. Supreme Court (SCOTUS) started to reverse its role in Green v.

County School Board of New Kent County. While the first case struck down the

"separate but equal" standard, the second set up an exit strategy for districts placed

under federal supervision. By meeting certain criteria for the assignment of students,

faculty, and staff, transportation, extracurricular activities, and facilities, formerly

segregated districts would be deemed to operate only one unified school system for all

races rather than a separate system for blacks and another for whites. In other words,

they would be deemed "unitary".

In practice, the receipt of unitary status returns the ability to assign students to

local school authorities Rather than having to meet stringent guidelines to keep school-

level racial populations standardized throughout the district, unitary districts can choose

assignment policies regardless of the effect they may have on the distribution of

students between schools. Over time, the possibility of resegregation has been

explicitly condoned by SCOTUS and recent research suggests that it is not only a

possibility, but a reality as well.2 Knapp (2009) shows that in Florida, districts released

from court order are likely to double their level of segregation within eight years. Lutz


1 347 U.S. 483 (1954) and 391 U.S. 430 (1968), respectively.
2 De facto segregation is allowed in Board of Education of Oklahoma v. Dowell and Freeman v. Pitts, 498
U.S. 237 (1991) and 503 U.S. 467 (1992), respectively.









(2005) shows immediate and gradual increases in the level of segregation following the

receipt of unitary status in a nationally representative sample of districts. Since

SCOTUS has already stated that resegregation via the granting of unitary status is not

enough to justify sanctions on those districts that allow a return towards de fact

segregation, further evidence is required to show that students are being harmed by the

increased levels of segregation. On this, the literature is mixed. Knapp (2009) finds

that the receipt of unitary status does not change the level of achievement for the

average student, the average white student, nor the average black student at the district

level. Lutz (2005) finds evidence that both black dropout rates and black private school

attendance increase following the granting of unitary status, suggesting that the value of

a public school education decreases for black students following the court's decision.

Although there is a growing literature measuring the effects of unitary status on

various academic outcomes, there is very little, if any, research on why school districts

seek unitary status in the first place. Although SCOTUS has certainly left open the

option of seeking unitary status, there is no requirement to do so. By seeking unitary

status, school districts must be trying to benefit in some manner and there could be

many reasons why they may choose this route. They may be trying to satisfy parental

preferences, operate their schools more efficiently or integrate their schools by choice

rather than force. Regardless of the reason, the decision to seek unitary status and

obtain its benefits cannot be costless. Lawyers must be hired, depositions taken, and

hearings scheduled before the case even makes it to a judge's bench for the final









decision.3 In a classical sense, for the decision to seek unitary status to be rational the

benefits from the outcome must outweigh this myriad of costs.

To date, no one has tried to answer the basic question of why districts seek

unitary status. This paper tries to bridge this gap and add to the school resegregation

literature in several ways. First, by examining both the timing of release from court

order and the size of the student population for all the districts ever placed under court

order in Florida, it can easily be shown that larger districts receive unitary status more

frequently than smaller districts. This is especially true over the last decade. This

would seem to suggest that smaller districts do not find the tradeoffs to be in their favor

and decide not to seek unitary status while larger districts appear to do the opposite.

Secondly, it will be shown that the financial benefits, if they exist at all, are very small on

a per-capita basis. Smaller districts, without a large enough student population

available to recapitalize the upfront costs incurred while seeking unitary status, may

rationally decide to stay under the court's desegregation order in perpetuity. For larger

districts that can achieve economies of scale, the proper decision may be to seek

unitary status.

Even if districts do not try to take advantage of changes in district finance ex ante,

they may benefit ex post via changes in the flow of revenues and expenditures. One

area where this may occur is transportation. It would seem plausible that one of the

most popular tools used to integrate schools, mandatory bussing, would be more costly


3 It might seem natural to insert some discussion of the probability of success at this point. However, that
may be unnecessary for two reasons. First, according to the U.S. Civil Rights Commission, there is not a
single case in Florida that has been actively pursued and eventually denied for a significant period of
time. Secondly, some may argue that recent SCOTUS decisions have set the bar for unitary status so
low that all districts have to do to achieve it is be willing to incur the costs associated with the process.
Evidence of this can be found throughout Orfield (1996) and Chemerinsky (2009)









to carry out than bussing students to their closer neighborhood schools. If that is the

case, the receipt of unitary status may allow districts to save money on transportation

and spend it on instruction in an effort to boost student achievement. Similarly, there is

anecdotal evidence that desegregation cases impart non-trivial legal costs on the

districts involved while they are actively in court.4 Payment of these fees diverts scarce

resources away from the school districts' core mission for years at a time while they

actively seek removal of their court orders. When districts get released from their court-

ordered desegregation plans, these types of costs may go away and lead to possible

changes in district-level expenditures as funds can now be shifted from legal expenses

to other areas.5 This analysis is the first to look for these kinds of patterns.

Florida's Desegregation History and How It Relates to District Size

Court-ordered school desegregation came to Florida in 1956 when the National

Association for the Advancement of Colored People's Legal Defense Fund successfully

won a case against the Miami-Dade County School Board. Over the next 22 years,

another 33 districts would be placed under court-ordered desegregation plans.6

Eventually, this list would include districts of all sizes throughout the state. Figure 1

shows the relationship between district size and desegregation status. Over this time

period, two trends stand out. First, size did play a role in the probability of being placed

4 Examples can easily be found in newspaper coverage of individual court cases. In particular, see Dries
(2009), Hughes (2005), and Spinner (1999). Legal fees associated with court-ordered desegregation
have also been addressed by the U.S. Supreme Court. See Missouri v. Jenkins, 491 U.S. 274 (1989).
5 This type of cost is different from what might be considered compliance costs. To some extent, all
districts, regardless of litigation status, have compliance costs. Even districts that have never been taken
to court must report on their level of integration to the U.S. Department of Education's Office for Civil
Rights. Therefore, removal of a court order will not necessarily decrease this type of expense because
the reporting requirement does not go away. It simply shifts between government agencies.
6 For a complete list of all the districts in Florida and their court-ordered desegregation status see Table
A-1 in the appendix.









under court order. Fourteen of the 15 largest districts were placed under federal

supervision by 1978. At the opposite end, only three of the smallest 15 were forcefully

desegregated. This means the largest districts were nearly 4.5 times more likely to be

taken to court when compared to the smallest districts. However, it should be noted

that there was some litigation within the lower end of the size distribution. Also

significant is that the odds change dramatically when the distribution is trimmed. The

probability of receiving a court order is nearly symmetric around the median district size

of 12,196 with eight being below and nine above the median.7 When put together, there

was at least some probability of prosecution for all districts in Florida.

Even as districts were still being placed under court order, some were beginning

to be released and declared unitary. The first district to receive unitary status in 1970

did so after only seven years under court supervision. However, that is not the ordinary

path taken by districts placed under court order. Of the 34 taken to court, only 19 have

received unitary status after spending an average of almost 26 years under court order.

For the 15 still under supervision, the average time since being found guilty is just over

40 years.

For most of those still under court order, there may be no compelling reason to

seek unitary status. The reason for this appears to be that size plays an even more

important role in obtaining unitary status than it does in being placed under court order,

especially over the last decade. Figure 2 outlines the granting of unitary status for the

34 districts placed under court order. Note first that there appear to be two groups



7 This median is based on the average number of unweighted fulltime equivalents in each district from
1997-2006. Using averages from the early part of the desegregation era does not create any substantive
differences.









within this sample, early grantees and those still under court order as late as 1997, with

each being separated by a decade of time during which not a single district received

unitary status. The first group is made up of eight districts that spend a relatively

shorter period of time under court order. Together they average about 12.5 years under

court supervision. The remaining 26 districts have spent an average of nearly 39 years

under court supervision. This difference in means is significant at the 99th percentile.

There is more than time that separates these two groups, as the role of size appears to

differ for these two groups. Among the early grantees, half are larger and half are

smaller than the median district size of 34,813 for districts that have been placed under

court order.8 Within the group of later grantees only one district is below the median.

This highlights the increasing role that size has played in districts deciding to seek

unitary status.9 Due to the time span between the two groups and data issues limiting

the availability of detailed financial records earlier than 1997, the remainder of the paper

will focus on the period from 1997 to 2006.10 It will attempt to explain why size plays







8 This sample has fewer districts than the one used previously and contains districts that are larger than
on average for the entire state. This is done because it may not be appropriate to consider a district that
has never been taken to court as a control for a district that has been treated to unitary status. The more
accurate control group is going to be those eligible for treatment and that requires a district to have been
taken to court at some point in the past.

9 If the issue of what makes an appropriate control group is ignored for a moment and unitary status
grantees are evaluated against the statewide median of 12,196 unweighted FTEs, the effect of size
becomes even more striking. In this case, only two districts that receive unitary status are below while the
remaining 15 are above this less restrictive, statewide median. These two districts are both in the group
of early grantees. No district smaller than the statewide median has received unitary status in the last 23
years.
10 This strategy also fits the reasoning of Orfield (1996) and Chemerinsky (2009) who show that the rules
used by the courts to determine whether unitary status should be granted changed in the early 1990s









such an important role in why some districts choose to seek unitary status and others

do not.1

Description of the Data

Litigation data for Florida was collected from the United States Commission of Civil

Rights' September 2007 report titled, Becoming Less Separate? School Desegregation,

Justice Department Enforcement and the Pursuit of Unitary Status. It contains data on

every district in Florida including the year cases were initiated (if applicable), the case

name, the year unitary status was received and whether districts not currently seeking

unitary status plan to do so in the future. These data were used to create the key

variable of interest in this study, a dummy variable that captures the trend of districts

receiving unitary status over time. Unitary Status equals zero in the years preceding

receipt of unitary status and equals one in all others. This formulation of Unitary Status

measures the cumulative effect of unitary status over time.

Fiscal data for every district in the state is publically available from the

Florida Department of Education's website and is contained in a report for each year

titled, Financial Profiles of Florida School Districts.12 Each year's report contains over

300 elements detailing district-level revenue and expenditure data. Due to variation in

element definitions and some elements being added or dropped over time, the main

focus of this paper is on aggregate data which should net out these additions, deletions

and definition changes at finer levels of detail.



11 This refers to the 1997-1998 and 2006-2007 school years. This convention will be used throughout
the rest of the chapter.

12These files are available at http://www.fldoe.org/fefp/profile.asp (last accessed on April 13,
2010).









Revenue data are broken out into 18 different categories that vary by source and

funding objective. There are four major sources of funds for Florida's school districts;

those received by districts directly from the federal government, federal funds received

by districts which are first passed through the state budgeting process, funds provided

by the state from state-level resources, and funds received from local sources such as

property taxes. All four of these sources are managed within the Florida Education

Finance Program (FEFP) to ensure an equitable distribution of resources across

districts, although some consideration is allowed for such things as differences in the

cost of living across districts.

Within each source of funds, there are several types of revenues specific to each

source. These are listed in Table 3-1 along with the several statistics. Column 1 shows

the mean for the entire time series from 1997-2006 for all 67 districts in the state.

Column 2 and Column 3 list sample means for the preferred control and treatment

groups. When combined, these two columns make up a sample of districts that had

previously been placed under court order but not granted unitary status prior to 1997.

This excludes districts never successfully taken to court and those who received unitary

status as an early grantee. Column 4 reports the result of a two-tailed t-test for equality

of means in order to provide evidence that there is variation in the data. Each column

shows revenues per student and has been converted into real 2000 dollars using the

national consumer price index for all urban consumers (CPI-U). Table 3-1 appears to

suggest that there are some significant differences between those districts that are

granted unitary status and those that are kept under court order, absent any controls for

other covariates. Eleven of the 17 categories suggest possible avenues through which









unitary status may affect revenues. Rather strikingly, unitary status recipients appear to

have $250 less per student on net when all revenue from local, state, and federal

sources is considered, which represents a difference of about 3.3 percent.

Expenditure data are taken from the same source and listed in two ways. The

first aggregates expenditures by categories such as instruction, administration, and

transportation while ignoring what type of goods or services were purchased with the

funds. The second method places the emphasis on the objects being purchased and

tracks expenditures on salaries, benefits and capital outlays. Table 3-2 summarizes

expenditures by both category and object in real 2000 dollars per STUDENT. Each

column follows the methodology used for Table 3-1 in creating different subsamples and

testing the equality of means. Although there appears to be a significant amount of

variation between the treatment and control groups by category and object, the net

result appears to suggest that there is little significant difference in total expenditures

between the two groups.

If net revenues fall for districts receiving unitary status while expenditures are

relatively constant, this may suggest that the granting of unitary status leaves districts

worse off financially and begs the question of why would a rational district seek unitary

status? Local school boards should be trying to achieve some economically rational

goal in seeking unitary status, and leaving their districts with fewer revenues and the

same level of expenditures does not appear to meet that standard. However, it is too

early to substantiate this conjecture as a more robust model needs to be developed to

account for the time-series nature of the data and other covariates.









Description of the Empirical Models

This paper uses a time-series model which in its simplest form can be expressed

as:

y, = Unitary Status, + y, + r + (3-1)


The dependent variable, yt, will take on multiple values of per-capita revenues

(expenditures) in district i in year t. Unitary Status measures the litigation status of

district i in year t. Districts still under court order in year t and those never taken to court

receive a zero (though not all of these latter districts are included in every specification).

Those that receive unitary status in year t receive a one in that and all subsequent

years. The year fixed effect, Yt, captures all district-invariant change for year t. This

accounts for unobserved statewide policy changes that affect districts equally. An

example of this type of policy would be an administrative change to accounting rules

that cause revenues (expenditures) to be counted differently from one year to the next.

Without this variable, such a change could be mistakenly attributed to a change in

unitary status in districts where such changes occur simultaneously, causing the

coefficient on Unitary Status to be biased. The district fixed effect, r7,, captures all time-


invariant characteristics of district i.13 C't represents a random error term which allows

for district-level heterogeneity.

The reported results utilize a more robust version of this model that takes the

form:



13 Although a district-year fixed effect would help account for unobserved policy changes at the district
level on a year-by-year basis, one cannot be used in this case due to the limited sample size.









y,, = Unitary Status,, + X, + y, + s, + (3-2)

It allows for more variation between districts over time by adding a vector of time-

varying, district-level characteristics, ,t. Due to the extremely small sample size, the

contents of Xt must be limited in number. The first element of X, is the number of

schools in each district. This captures the growth in the student population which

occurred in some districts over the period of the study. It also captures the effect of

economies of scale within the revenue and expenditure data. This is especially

important for administrative costs such as school boards. The next two elements

account for economic disparities between districts that may affect how districts make

decisions concerning education. The first variable is the annual property tax base per

STUDENT in each district and the second is the percentage of students in the district

that are eligible for Free or Reduced Price Lunch.14 The final element is the percentage

of students in the district that are nonwhite. This accounts for the rapid growth in the

Hispanic population across many districts in the state and any effects this may have on

how districts receive and spend financing.

This chapter's identification strategy relies on the timing of the unitary status

grants as its prominent source of variation. These three models will allow statistical

identification of the effect of unitary status on revenues (expenditures) if unitary status

occurs exogenously. In other words, unitary status must be received randomly.

Anything that affects the randomness of that timing, could affect the identification

strategy.

14 In 1997, only the percentage of FTEs eligible for free lunches is available. The percentage eligible for
both was not available until 1998. The year fixed effect should account for this type of change since it
occurs uniformly across districts.









There are several issues to consider in ensuring the validity of this identification

strategy. First, receiving unitary status is conditional on having been taken to court.

Districts that have never been taken to court cannot receive unitary status as defined in

this paper. Secondly, the study period for this paper is only ten years long, 1997-2006.

Eight districts receive unitary status before then and cannot be treated by the policy a

second time. Since it may be questionable to include either of these types of districts in

the sample, the analysis will also be conducted with a smaller, "preferred", sample of 26

districts which includes only those districts that were eligible for treatment.

For the remaining districts, the timing of unitary status must be random in order for

it to have a statistically identifiable effect. Here, the administrative process of dealing

with the court system provides a significant amount of variation in the timing of release.

This is plausible because two actions must take place before a district can receive

unitary status. The school district must decide to seek release from their court-ordered

desegregation plan and the court must then grant it. Even if the first decision is affected

by a non-random phenomenon such as size, the second part of the process, getting the

case through the court system, returns randomness to the actual release date and the

granting of unitary status. The administrative issues involved in going to court, having

the case heard, and getting a decision made are complex and varied enough to imply

that the actual year in which unitary status is received is not directly chosen by the

school district and is somewhat random.

A simple test of this theory involves looking at the level of correlation between the

year in which unitary status is received and the number of students in the district.

Figure 2-2 already highlights the tendency of larger districts to seek unitary status. This









is indicative of the first part of the process. However, the relationship between the size

of the district and the year in which unitary status is received is not a strong one. The

correlation coefficient between the year unitary status is received and the number of

students in the nine districts that received unitary status between 1997 and 2006 is only

0.09. Those districts that are released show little sign of being released in order of their

size.15 This suggests that the judicial process makes the actual timing of release

relatively random.

The last empirical issue concerns the weighting of observations in each

regression. Districts in Florida exhibit large differences in size with the difference being

nearly 350,000 students between the largest and the smallest. This size difference

persists even with the preferred sample of 26 districts. As discussed in the next section,

weighting does affect some of the results.16 The key difference between those

regressions run with and without weights is in how any statistically significant results are

interpreted. For those regressions without weights, the results represent the effect of

unitary status on revenues (expenditures) in the average district. When weights are

used the interpretation changes, now the results are interpreted as the effect on the

average student in a district that receives unitary status. Neither specification has a

clear advantage over the other as long as the distinction between the two remains clear



15 Again, it may be useful to apply the findings of Orfield (1996) and Chemerinsky (2005). They have
noted that federal courts changed their behavior towards desegregation cases in the early 1990s. Given
the court's newfound desire to clear its docket of desegregation cases, it is hard to construct a scenario
where a court's decision to release a district is dependent on district size. They would be willing to
release any district, regardless of size. Additionally, there is no evidence that a district has ever been
denied unitary status during the later part of the unitary era. Therefore, it is not unreasonable to assume
that small districts would receive unitary status if they were willing to seek it.
16 Weighted specifications utilize the number of students in the district as the weight for each district
observation. This procedure is implemented using the weights option in Stata.









when the results are interpreted. To test the effects of weighting on the results, the

preferred sample will be reported with and without weights.

In addition to reporting results for the "preferred" sample with and without

weights, a test will be conducted on a sub-sample of districts where weighting should

not affect the results. Weighting should not be an issue when the size difference

between districts is relatively small. In that case, the effect on the average district and

the effect on the average student should be close to the same. To test this, Equation 3-

2 will be estimated both with and without weights while using a sample consisting of 10

of the 11 largest districts from the preferred sample. This excludes Florida's largest

district, Miami-Dade and cuts the span of the size difference from nearly 350,000 to

190,000.

The Effect of Unitary Status on School District Finance

Unitary Status and Revenues

The majority of the results are shown in two tables. Due to the large number of

dependent variables, only the coefficient on Unitary Status is reported for each

regression. When necessary, further discussion of the other covariates and standard

errors will occur in footnotes with additional tables provided in the appendix. Table 3-3

reports the results of using Equation 3-2 when the dependent variable, Y, takes on the

value of revenues per student from various sources. Table 3-4 provides results from

when Yl takes on the value of expenditures per student. Each table uses the following

pattern. First, the model is completed using all 67 districts in the state. Next, the

sample is limited to the 34 districts that were taken to court. Then, the sample is

reduced to the "preferred sample" that is considered eligible for treatment. Results for









this sample are reported with and without weights. Lastly, the effect of weighting is

tested using a sample of ten, relatively large districts.

Overall, the results from Table 3-3 suggest that Unitary Status has at best a

small negative effect on revenues per student. The last line reports the effect Unitary

Status has on the total revenue received from all local, state, and federal (LSF) sources.

The results are mixed. Five of the six specifications, weakly suggest that revenue per

student falls. However, at no time, regardless of the sample or weighting used, does

Unitary Status have a statistically significant impact.17

Federal funding sources are listed at the top of Table 3-3. On net, there appears

to be no clear change in funding from federal sources. Any decreases appear to be

offset within the federal funding lines by corresponding increases. Career funding and

the federal meal program show a fairly uniform sign pattern suggesting decreased

levels of funding, but there is not robust statistical significance across specifications.

Even in those specifications where there are significant results, there is little effect in

terms of magnitude. Federal funding for "other" programs appears to help offset these

potential losses by adding anywhere from $7 $25 per student. All the remaining

sources of federal funding lack a statistically significant effect for Unitary Status. This

includes the total level of federal funding received which lacks any identifiable results

and a clear message as to the sign of any possible effects. This is not surprising

considering offsetting nature of the statistically significant results that were found for

federal career, meal, and other funding.


17 Table A2 provides more detailed results from regressing total revenues on all the covariates. It should
be noted that the standard errors for Unitary Status are uniformly large. What this means may depend on
the results for other sources of revenue and will be discussed shortly in greater detail.









State-level funding shows weak signs of potential increases as a result of

receiving unitary status. State funding for "other" programs is the only line within the

state-level section to show a consistent sign pattern and have meaningful levels of

statistical significance. It suggests that districts with unitary status receive an additional

$36-$134 per student.18 This increase may be partially offset by a potential decrease in

state categorical programs where there is some weak evidence of a decrease as much

as $138 per student.19 However, the actual effect is likely smaller due to the

inconsistent sign pattern and lack of statistical significance. This offset is also apparent

in the total state funding line. Results for five of the six specifications suggest that there

is a total increase in state funding. In most of the cases, the increase in total state

funding is smaller than the increase from other programs, suggesting that decreases in

categorical programs provides some attenuation. Due to a lack in statistical

significance, it is likely that any increases in state funding are small at best.

Of the three possible sources, locally provided funding appears to be the only

one to decrease with overwhelming levels of statistical significance. Here, the effect of

Unitary Status on total local revenue is significant in four of the six specifications and

shows that the receipt of unitary status decreases the level of resources received from

all local sources by about $100 per student.20 The majority of this decrease appears to


18 Table A3 shows that even with a subsample of only ten districts, the effect of Unitary Status on other
state-level funding sources is significant up to the 82nd percentile. This shows that the effect of Unitary
Status is significant or close to significant in five of the six specifications. Considering the extremely small
sample size and the relatively small drop-off in statistical significance for the fifth specification, there
appears to be some robust evidence that an increase in this area of funding does occur as a result of
having received unitary status.
19 Categorical programs include such things as Florida's Teacher Lead Program, Transportation,
Instructional Materials, Discretionary Lottery Funds, and the mandated Class Size Reduction Program.
20 The two remaining specifications are not statistically significant but still have negatively signed
coefficients ranging from $65-$86 per FTE.









come from decreases in the amount of revenue per student sourced from local tax

sources. Some additional decreases may come from the amount of other income

received by unitary districts, but the magnitudes are smaller, and only one of the six

specifications is statistically significant at any conventional level.

When all sources are considered together, it is not immediately clear that any

significant changes in funding occur as a result of receiving unitary status. The

coefficients for the total amount of LSF funding listed at the bottom of Table 3-3 never

reach a meaningful level of statistical significance. However, the sign and magnitude of

the coefficients are in line with the results of each source when they are considered

individually. The small, negatively signed effect of unitary status on total LSF funding

appears to be largely driven by changes in state and local funding. Decreased local

funding is only partially offset by increases in state funding, resulting in a modest

decrease in overall funding. Federal levels appear to be unaffected, on net, by the

granting of unitary status.

One possibility for the large standard errors and the lack of statistical significance

for overall LSF revenues is that random noise from the federal and state categories

could be masking the effect of a drop in local revenues. The results from the federal

portion suggested that there was no overall effect from federal funding. The state

category showed weak signs of possible increases. When these noisy signals are

combined, they may drown out the statistically significant local funding results.

Therefore, it may be reasonable to interpret the negative, but statistically insignificant,

LSF results as supporting the notion that some decrease in local revenues does in fact

occur.









Unitary Status and Expenditures

It is important to keep in mind that revenues are only one portion of the school

finance picture. It is also possible that the granting of unitary status could affect the way

districts spend their finances. Therefore, it is important to also look for any possible

changes in expenditures. Table 3-4 reports the coefficient for Unitary Status after

completing regressions based on Equation 3-2 where at takes on the values of two

broad types of expenditures. The top panel lists expenditures by categories such as

instruction, food service, and transportation. The bottom panel lists expenditures by

objects such as salaries, benefits, and materials. The two panels end by showing the

same overall result. Although there appear to be sizable decreases in spending per

student, five of the six specifications return negative effects ranging $58 to $305 per

student, the standard errors are large. This suggests that any decrease in expenditures

is, at best, very imprecisely measured.

When examining the expenditure breakouts in finer detail, statistically significant

results are still hard to find. Only one category, General Administration, exhibits

significant changes when unitary status is received, but this occurs in only two of the six

specifications. The results from Table 3-4 suggest that per pupil expenditures for this

category increase by $7-$10 per student.21 This change is understandable as exit from

a court-ordered desegregation plan can entail having to implement and manage

complicated assignment policies. However, this increase does not appear to carry

through to total expenditures and it appears that the increases in General Administration

21 Table A4 presents more detailed results for the effect on General Administration expenditures. Note
that the coefficient on Unitary Status is positive in all six specifications. Additionally, the standard errors
are not so large as to rule out some positive effect at lower levels of statistical significance. Additionally,
although there is robust evidence of statistical significance, economic significance is not very prominent.









expenditures could be offset in several areas. Although they do not reach meaningful

levels of statistical significance, expenditures on Instruction, School Administration,

Transportation, Facilities (Capitalized), and Debt Services all show a trend of

decreasing following the receipt of unitary status with at least five of the six coefficients

being negative.22 Of these, only Debt Service and Facilities appear to have any sizable

magnitudes which could result in the total expenditure category appearing to be

negative.23 The dominating effect of Facilities and Debt Services is confirmed in the

bottom panel of Table 3-4 where expenditures are broken out by object. Here, only two

objects show any statistically significant changes. Unitary Status appears to increase

expenditures on Purchased Services by just over $50 per student and decrease

spending on "other" expenses by $12. However, these results occur in different

specifications, potentially weakening their impact. This net gain washes out as total

expenditures by object are predominantly negative in sign and relatively large in

magnitude. Again, this offset appears to be a result of large decreases in spending on

facilities and debt services. Overall, much like the results for the revenue data, there is

at best only weak evidence for declines in spending as a result of receiving unitary







22 Much like the revenue data, the aggregation of random noise from the sub-level expenditure data could
be why the standard errors on the total expenditure figures are so large.
23 The coefficients for these two categories should be viewed with some caution. In addition to the lack of
statistical significance, it may be difficult to interpret what these negative coefficients actually suggest. In
at least one district, construction of new schools in predominantly black neighborhoods was made a pre-
condition of release from court order. This means expenditures on capitalized facilities were higher in the
periods prior to the receipt of unitary status. Therefore, the negative coefficient could really be a signal
that expenditures are returning to normal from a previously elevated level as spending on new
construction is curtailed.









status and that the bulk of these declines are a result of changes in facility and debt

service expenditures.24

The Role of Economies of Scale

Overall, unitary status does not have a clear impact on school finances. Although

both revenues and expenditures show some statistically significant results at certain

sublevels, they do not carry over to district-level bottom lines. If they did, a definitive

statement could be made concerning the financial changes caused by unitary status.

However, in the absence of such results, it is more likely that the effect of being

released from a court-ordered desegregation plan has modest and nuanced effects on

district-level finances.

For revenues, there appears to be little effect on either federal or state funding

per student. Local funding appears to be reduced, primarily through reduction of the

local tax burden, by about $100 per student. However, this effect does not carry over to

district-level bottom lines in a statistically significant way. This could mean that unitary

status has no effect on total LSF revenues or that the effect is negative, but too

imprecisely measured given the current data.25 Expenditures also show no clear cut

results. The only statistically significant results suggest small increases for certain

subcategories, but these results also disappear when aggregated.





24 If the net effect on total revenues and total expenditures is considered, it appears that districts benefit
from unitary status because expenditures per FTE fall by more than revenues per FTE in four of the six
samples. This benefit appears to be about $100 per FTE. However, this comparison is severely
weakened by the overall lack of statistical significance.
25 It is important to remember that state and federal funding could be adding random noise to the total
LSF line. This would cause the standard errors to be large and the overall effect to be imprecisely
measured. This would also obscure the effect on local tax burdens.









Fortunately, the revenue results alone appear to be enough to explain the data.

First, consider the case where the decreases in local tax revenues are offset by gains

elsewhere to combine in a zero total effect scenario on LSF revenues. This would be

evidence of revenue shifting, total LSF funding stays the same, but the local burden is

decreased. This would be a clear benefit to a district that is able to implement such a

strategy. If unitary status provides such an avenue, the implication could be that

districts seek unitary status strategically to take advantage of this revenue shifting. If

districts know that they can decrease the local burden of school finance, but maintain

overall revenue levels, seeking unitary status would appear to be a worthwhile when

holding all else constant. The results from the previous section also allow for another

case, one which says that the true effect of unitary status on revenues is negative, but

imprecisely measured given the current data. If this case holds, the effect on academic

outcomes becomes important. If resources and outcomes decrease in tandem, unitary

status may be a poor policy. However, if resources fall and outcomes remain the same,

or even increase, this could be a sign of increased efficiency. As outlined in Chapter 1,

achievement based on standardized test scores does not change with receipt of unitary

status. If the same educational outcomes are obtained at lower revenue levels, unitary

status may allow districts to operate more efficiently. Therefore, a district's decision to

seek unitary status may be an effort to save local tax dollars and operate more

efficiently. Regardless, this case also provides a strategic reason for seeking unitary

status.

Both of the previous scenarios are also consistent with the tendency of only large

districts seeking unitary status. Keep in mind, the decision to seek unitary status cannot









be costless and it may be wise to assume that significant legal and settlement costs

may be incurred if unitary status is received.26 Assume for a moment that a district

currently under a court-ordered desegregation plan must pay a fixed cost of $1 million to

receive unitary status. Table 3-3 suggests that the local burden is reduced by about

$100 per student. As long as the district has more than 10,000 students, the district will

recoup the cost in one year. If benefit to unitary status is anything smaller than $100

per student, the district must be willing to finance the cost from reserves, forego some

current expenditures or wait for the decision to pay off over time. Therefore, it is

reasonable that the data suggest that small school districts do not find such an outcome

desirable and decide that simply living with the court's order is optimal.

What does this mean going forward? The smallest district to receive unitary

status among the later grantees had an average population of about 31,000 students

during the study period. If we assume a balanced budget, this implies that the cost of

seeking unitary status is no more than $3.1 million. Of the 15 districts remaining under

court order, only three are large enough to guarantee a beneficial payback within one

year. This may imply that the era of unitary status grants is coming to a close as fewer

and fewer districts will find it in their favor to seek unitary status. This may or may not

lead to efficient outcomes. If court-ordered desegregation plans truly constrain the

districts under such plans, then release can lead increased efficiency. However, if



26 Although good accounting data on the legal cost of seeking unitary status does not appear to be
available, there is some anecdotal evidence that the costs are large. In many cases, the defendant
districts are required to pay the plaintiffs' attorney fees. In Missouri v. Jenkins, the plaintiffs were awarded
attorney's fees of over $3 million (see 131 F.3d 71 and MISSOURI V. JENKINS, 491 U. S. 274 (1989)).
In a case against Charlotte-Mecklenburg Schools, the plaintiffs asked for but were denied nearly $1.5M
(see 274 F.3d 814 (4th Cir. 2001)). Although the potential legal costs are not broken out, the Pinellas
County, FL school board budgeted for over $4.5M in "compliance" costs when they received unitary
status.









districts decide to remain under court order because the upfront costs are large, scarce

educational resources could be lost.









Table 3-1. District-level revenues per student


Number of Districts in Sample
Direct Federal Funding
Federal Career Funding Through State
Federal Title I & V Through State
Federal Adult Basic Education Through State
Federal Meals Funding Through State
Federal Other Funding Through State
Total Federal Funding Through State
Total Federal Funding All Sources
Florida Education Finance Program
State Categorical Programs
State Racetrack Funding
State Other Sources
Total State Funding
Local Tax Revenue
Local Investment Income
Local Other Income
Total Local Revenue
Total Local, State and Federal


67
55.80
22.16
219.99
10.02
200.50
307.97
760.64
816.43
2,505.83
994.46
37.11
355.14
3,892.53
2,543.65
92.23
312.36
2,948.25
7,657.21


(2) (3)
Treated Control
9 16
48.29 41.25
13.29 21.74
200.16 209.12
4.06 17.36
192.54 194.09
260.86 302.20
670.91 744.53
719.20 785.77
2,405.52 2,554.18
815.71 1,011.70
3.13 42.89
301.86 385.79
3,526.22 3,994.56
2,710.00 2,415.33
99.89 89.49
310.89 324.18
3,120.79 2,828.99
7,366.21 7,609.32


(4)

P<0.05
NO
YES
NO
YES
NO
YES
YES
YES
NO
YES
YES
YES
YES
YES
NO
NO
NO
YES


NOTE: Revenues are shown in real dollars (2000) deflated using the CPI-U from the
January of each school year starting with 1997-1998 and ending in 2006-2007. Column 1
shows district-level means for all 67 school districts in Florida. Column 2 restricts the
sample to the nine districts in the preferred treatment group. This includes only those
districts that received unitary status between 1997 and 2006. Column 3 restricts the sample
to the 16 that were eligible for unitary status but had not received it by 2006. These districts
represent the preferred control group. Column 4 reflects the results of a two-tailed test for
equality of means at the five percent level.










Table 3-2. District-level expenditures per student
BY CATEGORY


Number of Districts in Sample
Instruction
Pupil Personnel Services
Instructional Media Services
Instruction and Curriculum Development
Instructional Staff Training
Instructional Support Services
School Board
General Administration
School Administration
Facilities (Non-capitalized)
Fiscal Service
Food Service
Central Services
Transportation
Operating Plant
Maintenance of Plant
Total General Support Services
Total Instruction, Instructional Support and General Support
Community Services
Total Current Expenditure
Facilities (Capitalized)
Debt Services
All Expenditures


BY OBJECT
(1) (2) (3) (4)
Treated Control
Number of Districts in Sample 67 9 16 P<0.05
Salaries 3,865.99 3,764.15 3,851.00 YES
Employee Benefits 1,096.20 1,132.67 1,093.27 YES
Purchased Services 549.74 518.88 522.63 NO
Energy Services 182.81 166.17 185.21 YES
Materials and Supplies 394.63 365.52 398.38 YES
Capital Outlays (Non-Capitalized) 89.41 76.97 86.88 NO
Capital Outlays (Capitalized) 1,246.27 1,142.10 1,209.15 NO
Debt Services 292.40 407.62 321.75 YES
Other 133.73 99.50 137.45 YES
Total Expenditures 7,851.19 7,673.58 7,805.71 NO
NOTE: Column 1 shows district-level means for all 67 school districts in Florida. Column 2 restricts the sample to the
nine districts in the preferred treatment group. This includes only those districts that received unitary status between
1997 and 2006. Column 3 restricts the sample to the 16 that were eligible for unitary status but had not received it by
2006. These districts represent the preferred control group. Column 4 reflects the results of a two-tailed test for
equality of means at the five percent level.


3,

















2,
6,

6,
1,

7,


(1) (2) (3)
Treated Control
67 9 16
560.83 3,529.37 3,484.04
312.93 305.75 325.49
110.46 109.35 109.41
194.14 167.12 199.94
67.00 68.43 67.22
694.26 659.31 710.71
48.53 15.91 57.54
78.88 47.92 82.58
365.49 349.35 366.26
29.03 30.09 28.16
60.80 34.91 68.49
328.60 317.80 329.50
122.04 135.82 109.15
303.33 293.09 313.70
479.07 471.27 477.32
185.11 175.10 186.37
009.62 1,878.58 2,028.97
264.71 6,067.26 6,223.73
47.82 56.60 51.09
312.52 6,123.86 6,274.81
246.27 1,142.10 1,209.15
292.40 407.62 321.75
851.19 7,673.59 7,805.71


(4)

P<0.05
NO
YES
NO
YES
NO
YES
YES
YES
YES
NO
YES
YES
YES
YES
NO
YES
YES
YES
NO
YES
NO
YES
NO










Table 3-3. Effect of unitary status on revenues
(1) (2) (3) (4) (5) (6)
Direct Federal Funding -8.39 -11.57 -6.77 0.70 1.94 2.845
Federal Career Funding Through State -3.210* -4.442** -4.257** -0.63 -1.24 -0.28
Federal Title I & V Through State 2.33 3.60 -2.50 6.25 2.35 -2.459
Federal Adult Basic Education Through State -0.83 -1.88 -3.14 1.80 1.219* 1.531**
Federal Meals Funding Through State -4.52 -4.04 -5.45 -6.30 -7.020** -9.034**
Federal Other Funding Through State 23.55 23.50 10.12 25.36* 18.62 7.242
Total Federal Funding Through State 17.31 16.74 -5.22 26.48 13.93 -2.999
Total Federal Funding All Sources 8.93 5.17 -11.98 27.19 15.87 -0.153
Florida Education Finance Program -8.85 -22.14 -1.46 -22.57 28.65 -6.179
State Categorical Programs 44.05 -59.60 -138.4* 6.71 -34.21 1.672
State Racetrack Funding 1.69 0.38 0.50 0.12 -0.16 -0.231
State Other Sources 134.1** 85.86* 99.97* 84.18** 51.19 36.14
Total State Funding 171.0* 4.51 -39.39 68.44 45.46 31.4
Local Tax Revenue -41.64 -71.82 -58.44 -77.62* -80.12 -85.49*
Local Investment Income 2.24 1.95 5.39 13.14 7.15 8.42
Local Other Income -25.70 -16.58 -47.06* -37.43 -25.50 -25.47
Total Local Revenue -65.11 -86.45 -100.1** -101.9** -98.47** -102.5**
Total Local, State and Federal 114.80 -76.78 -151.50 -6.29 -37.14 -71.3
Number of Districts 67 34 25 25 10 10
Weighted by Number of Students NO NO NO YES NO YES
NOTE: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been
received. Other controls include year and district fixed effects, the number of schools in the district, the size of the tax roll per capital, the percent
of students eligible for free or reduced price lunches and the percent of the student population that is non-white. For brevity, these coefficients
are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**)
and one (***) percent levels Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those
districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment.
Column 4 is the same sample, but the observations have been weighted by the number of full-time equivalents in the district. Columns 5 and 6
follow the same procedure but only include the ten largest districts in the state, excluding Miami-Dade.










Table 3-4. Effect of unitary status on expenditures
BY CATEGORY (1) (2) (3) (4) (5) (6)
Instruction -33.94 -35.01 -44.05 -48.99 17.76 -7.258
Pupil Personnel Services -9.209 -14.94 -0.812 2.052 -6.881 -4.896
Instructional Media Services 2.689 1.202 1.817 3.705 2.188 1.934
Instruction and Curriculum Development -5.339 -7.296 -4.428 -1.675 14.91 5.363
Instructional Staff Training 8.893 4.817 4.111 -0.309 -8.603 -10.62
Instructional Support Services -4.583 -16.2 -0.284 1.834 0.548 -10.18
School Board 0.344 1.983 4.292 1.361 -2.217 0.206
General Administration 5.459 4.248 6.154 10.09*** 7.644** 4.443
School Administration -7.963 -15.12 -11.93 -8.035 -10.6 -10.07
Facilities (Non-capitalized) 40.19 35.95 39.27 12.69 34.74 13.79
Fiscal Service -0.943 -1.455 -1.032 -0.322 0.11 0.662
Food Service -6.647 -5.957 -4.222 3.739 -3.292 -3.545
Central Services -2.863 -0.28 2.918 7.024 3.027 9.912
Transportation -14.64 -13.42 -14.61 -10.07 -12.19 -7.787
Operating Plant -10.67 -9.504 2.614 6.597 7.071 5.153
Maintenance of Plant 21.53 34 32.97 0.585 29.85 15.19
Total General Support Services 23.01 27.85 55.91 21.59 52.69 25.59
Total Instruction, Instructional Support and General Support -15.51 -23.35 11.58 -25.57 71 8.149
Community Services 10.14 11.78 -0.672 1.403 -3.634 0.262
Total Current Expenditure -5.372 -11.57 10.91 -24.16 67.37 8.411
Facilities (Capitalized) -11.75 -176.7 -206.2 -10.22 -93.72 36.23
Debt Services -117 -117.4 -81.24 55.89 -104.2 -102.5
All Expenditures -134.1 -305.6 -276.6 21.5 -130.6 -57.87










Table 3-4. Continued
BY OBJECT (1) (2) (3) (4) (5) (6)
Salaries -59.5 -76.54 -44.86 -64.64 1.481 -29.54
Employee Benefits 10.02 -8.084 -17.32 -9.308 -7.966 -11.89
Purchased Services 3.821 35.6 37.41 32.74 52.84* 58.37
Energy Services -1.599 -1.91 -1.594 -2.517 1.968 -3.219
Materials and Supplies -8.614 -5.192 -9.607 0.461 -10.31 -7.009
Capital Outlays (Non-Capitalized) 53.99 56.57 52 19.59 35.32 11.37
Capital Outlays (Capitalized) -11.75 -176.7 -206.2 -10.22 -93.72 36.32
Debt Services -117 -117.4 -81.24 55.89 -104.2 -102.5
Other -3.485 -12.00* -5.116 -0.483 -5.964 -9.666
Total Expenditures -134.1 -305.6 -276.6 21.5 -130.6 -57.87
Number of Districts 67 34 25 25 10 10
Weighted by Number of Students NO NO NO YES NO YES
NOTE: Data represent the coefficient for Unitary Status obtained by regressing different types of expenditures on whether unitary status has been
received. Other controls include year and district fixed effects, the number of schools in the district, the size of the tax roll per capital, the percent
of students eligible for free or reduced price lunches and the percent of the student population that is non-white. For brevity, these coefficients are
not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and
one (***) percent levels Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those districts
that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4
is the same sample, but the observations have been weighted by the number of unweighted full-time equivalents in the district. Columns 5 and 6
follow the same procedure but only include the ten largest districts in the state, excluding Miami-Dade.












o
0




U)
03
C 0

0 0





x x x



d d m 12,196 (Median)

1955 1960 1965 1970 1975 1980

Year Court Order Received

x Districts Taken to Court
t- Districts Never Taken To Court


Figure 3-1. Florida's school desegregation over cases over time. Shows data for all 67
districts in Florida. Some displacement of points on the graph is caused by
adding random variation to points which might otherwise lie on top of each
other.
















0 A
A






-o









1970 1980 1990 2000 2010
C=)













Year Unitary Status Is Received
A Districts wUnitary Status







Districts Still Under Court Order

Figure 3-2. Florida's unitary status receipts over time. Shows data for the 34 districts
A















that received desegregation orders. Some displacement of points on the

lie on top of each other.
A A A A A
A A 4 34,813 (Median)
A A

1970 1980 1990 2000 2010

Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order


Figure 3-2. Florida's unitary status receipts over time. Shows data for the 34 districts
that received desegregation orders. Some displacement of points on the
graph is caused by adding random variation to points which might otherwise
lie on top of each other.









CHAPTER 4
A MULTI-STATE ANALYSIS OF UNITARY STATUS AND SCHOOL FINANCE

The results in Chapter 3 suggest that economies of scale may be a reason that

only large districts in Florida seek and receive unitary status. Although Florida is

representative of the entire country as a whole in some respects, it is unwise to extend

those results to a national level without first checking to see how school districts in other

states also react to the granting of unitary status. Therefore, the analysis in Chapter 3

will now be extended to include six other states. One key difference with Chapter 3 is

that the analysis in this chapter will use a different source of data for the fiscal data. In

the previous chapter, the fiscal data was reported by the state of Florida for all of its

districts. In this chapter, the data are reported by each district directly to the U.S.

Department of Education's, National Center for Education Statistics (NCES). The data

are qualitatively similar, but use of the NCES data will allow the inclusion of more states

into the analysis.

Description of the Data

Litigation data for the multistate analysis also comes from the U.S. Commission on

Civil Rights' (USCCR) report titled Becoming Less Separate? School Desegregation,

Justice Department Enforcement, and the Pursuit of Unitary Status. In addition to data

on Florida, it also provides unitary status data on Alabama, Georgia, Louisiana,

Mississippi, North Carolina, and South Carolina. When combined it provides

information on over 780 districts across these seven states.

Table 4-1 shows the breakout by litigation status for all the districts in each state.

Over time, more than 61 percent (479 of 780) of the districts in these states were taken

to court. Of those that faced court-ordered desegregation, 41 percent have received









unitary status (195 of 479).1 As in Florida, there were two distinct periods during which

unitary status grants were received. Figure 4-1 shows the relationship between the size

of districts and the timing of their unitary status grants for all the districts in the sample

ever taken to court. It shows clearly that few districts received unitary status in the late

1980s and early 1990s. This creates two groups of unitary status grantees and acts as

a natural break in the data. Figures 4-2 through 4-8 show the timing of unitary status

by size for each state. Two things are clear from these figures. First, the bimodal

nature of unitary status grants is uniform to all states. They each appear to experience

a pause in the granting of unitary status around the year 1990. Secondly, it appears

that Florida is an anomaly. The other six states have both large and small districts

receiving unitary status. Each has a fairly even distribution of districts above and below

the median for districts ever taken to court. Florida's distribution is skewed towards only

large districts with only one unitary grantee being smaller than the median sized

district.2 This difference across states may imply that school district finances are

affected by unitary status differently as well. Therefore, further analysis of the multistate

fiscal data is warranted. If fiscal reasons explain why districts seek unitary status, these

changes are most likely to be identifiable during the latter period of unitary status grants.

As shown in Table 4-1, more than 70 percent of all unitary status grants occur during

this latter period. Any potential fiscal effects become very important once the size of


1 This data is current as of the report's 2007 publication. Since that time, another 38 districts across all
seven states have received unitary status. Since the NCES fiscal data only covers through the 2006-2007
school year, these recent grantees change in status can be ignored. For more details on these districts,
see Holley-Walker (2010).
2 This finding draws into question the economies of scale argument established in the previous chapter. If
it were to hold uniformly, small districts in other states would not be receiving unitary status. Since they
do, we can most likely rule out economies of scale as a nationwide reason why districts seek unitary
status.









these districts is considered. Although these 145 districts represent only 18 percent of

all districts in the sample, they account for nearly one-third of all students in these seven

states.

The fiscal data for this analysis come from the NCES, Common Core of Data,

Local Education Agency (School District) Finance (F-33) Survey. This dataset provides

broad measures of district-level revenues and expenditures for all the school districts in

the country. This study uses 15 years of data going back as far as the 1990 fiscal

year.3 In that year, the survey was conducted with all districts in the country. It then

alternated in some years between population and sample surveys. This study includes

only years for which data is available from all districts in all seven states and ignores the

years in which only a sample of districts is provided.4 The raw finance data for the

seven states contains 13,304 observations across all 15 years. This total includes

extraordinary education agencies such as juvenile justice programs, vocational

programs, hospital/homebound programs, and in North Carolina, charter schools.

These types of schools account for just over 10 percent of all observations and were

excluded from the analysis. An extremely small number of observations were also

affected by consolidation of independent school districts. In all of these cases, none of

the independent city districts had been taken to court. Since they would not be eligible

for treatment to unitary status, there is no harm in excluding them from the analysis.

There are also several cases in South Carolina and Alabama where new districts were

formed from school districts already under court supervision. The USCCR considered


3 The 1990 fiscal year corresponds to the 1989-1990 school year. Except for this section when fiscal
years are explicitly stated, years listed in this study apply to the start of the school year.
4 Samples were drawn in the 1991, 1993, and 1994 fiscal years.









these districts as having never been taken to court and that notation is continued

throughout this study. Upon applying these exclusions, the final data set contains

11,649 observations across fifteen years. The natural break in unitary status receipts

helps define treatment and control groups based on schools being eligible for treatment

in 1989. This break eliminates districts that were never taken to court and those that

received unitary status prior to 1989 from the sample. This leaves 429 districts across

fifteen years and totals to 7,185 district-year observations in the panel.

The means for all of the summary fiscal variables are provided in Table 4-2.5 Data

are broken out by revenues and expenditures. Means are provided in terms of per

student funding levels in constant 2000 dollars.6 Total revenues and expenditures per

student are nearly one-quarter lower in Mississippi than in Florida. In addition, to the

differences in funding across states, there is also a tremendous size difference between

states. Although not reported on this table, the average district in Florida is nearly three

times as large as the next largest state in the sample and nearly ten times the size of

the smallest. Although some of the difference is eliminated by comparing median

district size, even that difference is striking. The median district in Florida is still twice

the size of the next closest state and five times the size of the median district in the

smallest state.7 The differences across states make it clear that any model attempting





5 These summary variables are combinations of other subcategories. For brevity, only the top-level
categories are listed in Table 4-2. For means on all potential fiscal variables, see Table A5 in the
appendix. Revenues are further broken out by source and purpose. Expenditures are broken down into
the objects on which the funds are spent.
6 The nominal data were deflated using the CPI-U for 2000 as a base year.

7 This difference may be enough to explain Florida as an anomaly.









to estimate the effect of unitary status on fiscal variables will have to account for the

large amount of variation between states or treat each state separately.

Description of the Models

The models used in this chapter are very similar to the models used in the

previous chapters. The first specification used will take the following form:

y, = Unitary Status, + X, + 7t + s (4-1)

Here, yit will take the value of various per-student fiscal measures at the district level in

each year. Unitary Status, measures the legal status of each district. If the district has

never been taken to court or has been taken to court, but has yet to receive unitary

status, the district receives a zero for that year. Once the district receives unitary

status, this variable takes on a value of one in that and all subsequent years. X ,

captures the time-variant characteristics for each district. These include the size of the

student population, the number of teachers assigned to the district, the percent of

students that are eligible to receive a free or reduced price lunch, and the percentage of

students in the district that are reported as nonwhite.8 y represents a state-year fixed

effect which captures all state-year specific variation. This strategy is worthwhile

because it will pick up any variation caused by unobserved policy changes which affect

all the districts in one state in the same manner. The most likely example of this type of

policy would be accounting changes which may affect how the data are reported from

one year to the next. As long as the change is uniform across districts in a given state,



8 The number of teachers is endogenous. Alternate specifications are also run where the number of FTE
instructors is omitted from matrix of time-varying characteristics. This change does not appreciably
change the results discussed in later sections.









this state-year fixed effect will account for the change. For this specification to be

identified the error term, s,,, must meet the following condition:

E(s,, I X,,, Ys, Unitary Status ) = 0 (4-2)

This condition is likely to hold since the noise in the judicial process makes the actual

timing of release quasi-random.9 The error term is clustered at the district level to

correct for heteroskedasticity.

Care must be taken when interpreting the results generated by Equation 4-1. This

specification treats all districts the same as to whether they have been taken to court.

Since the receipt of unitary status is conditional on having been taken to court, any

unobserved differences between districts never taken to court and those currently under

court-ordered desegregation plans may bias the results. To account for this, the

specification will be run using the entire sample and a sample containing only those

districts that are eligible for treatment during the latter period of unitary status grants.

The final specification runs separate regressions for each state in the sample. In

this instance, a state-year fixed effect cannot be implemented. District-year fixed effects

are also impossible due to a cell size of one district observation per year. Therefore,

this specification uses year and district fixed effects. This makes the specification:

y,, = Unitary Status,, + X, + y, + r, + ,, (4-3)

In Equation 4-3, y,,, Unitary Status ,, and X, are the same as in Equation 4-1. The key

difference is that the state-year fixed has been replaced by a year fixed effect, y,, and a

district fixed effect, r,. This specification is also likely to be identified due to the


9 Both Chapter 2 and Chapter 3 provide greater detail about the process involved in getting the unitary
status grant and how the court makes the end of the process sufficiently random for identification.









randomness of the judicial process. The results generated by this specification will be

important in determining whether there are any general trends in the fiscal data

associated with unitary status. Not only should the results from Equation 4-1 be

significant, but the results from Equation 4-3 should also show significant results across

many of the states if unitary status has wide ranging effects.10

The Effect of Unitary Status on Revenues

The results from using Equations 4-1 and 4-3 are broken up into several tables.

Table 4-3 highlights the overall effect of unitary status on total revenues and

expenditures. In terms of revenues, there is no clear indication of a uniform effect.

Although Column 1 shows total revenue being $166 lower in districts that receive

unitary status, the effect goes away when the control group is trimmed to include only

those districts eligible for treatment as of 1990. This suggests that the result found in

Column 1 is driven more by districts that have never been taken to court and/or those

that received unitary status as an early grantee. It should also be noted that the results

are not uniform in terms of sign and magnitude across states. This suggests that there

is no clear, bottom-line effect of unitary status on revenues, regardless of source.

The results in Table 4-3 may not provide enough detail to accurately identify

changes caused by unitary status. It may be possible that there are changes to

subcategories of funding which offset each other in the total revenue figures. Tables 4-

4 through 4-6 present revenue data in greater detail by source. These tables show that

there is little consistent effect on revenues from the state and federal levels. The

exception may be in terms of local revenues where there is very modest evidence of

10 These state-specific regressions will also utilize the smaller sample of eligible districts due to the
conditionality of unitary status on having been previously taken to court.









some weak changes in revenues. In Table 4-6, local revenues generated by District

Activity Receipts show a fairly uniform sign patter across specifications, but only reach

statistically significant levels in two of the seven specifications.11 Overall, the economic

significance of this result is questionable. In Alabama, average revenue per student is

$6,400. This $16 change represents about a 0.2 percentage point drop in revenues.

There is also a relatively consistent sign pattern in miscellaneous revenues at the local

level.12 Eight of the nine specifications have a negative sign, but only two can be

considered different from zero. These decreases are also minor in terms of economic

significance. This lack of statistically significant results appears to suggest that unitary

status does not appear to affect revenues in any general way.

The Effect of Unitary Status on Expenditures

The effect of unitary status on expenditures shows one interesting result that

appears to be consistent through much of the state-by-state data. Table 4-3 shows a

drop in Total Current Expenditures (TCE) of about $147 per student. It appears that the

bulk of this drop occurs in TCE for Elementary Education which shows a drop of about

$174 per student. Considering average total expenditure per student is about $6900 in

each district this represents a decrease of about 1.5 percent. This result is also

significant in three of the nine specifications and has a consistent sign pattern in five of

the remaining six specifications. Much like the revenue data, these categories are

summations of other subcategories which also may provide relevant information.

11 It is not clear how the NCES defines District Activity Receipts. The NCES Codebook does not include a
definition and the district survey form only says, "Gross district activity receipts for those funds under
control of the custodian of district funds should be included on line 13." These two factors make it less
likely that this is an important change in per pupil funding levels.
12 Again, it is unclear what constitutes a "miscellaneous" source of funding at the local level. In fiscal year
2007, the average district had $698,000 in miscellaneous funding or about $72 per student.









Table 4-7 shows that these results are traceable through other parts of the

dataset. TCE on Instruction appears to show a statistically significant drop of about

$101 per student. This result is not only significant for the entire sample, but is also

significant for the reduced sample of unitary status eligible districts and one state.

There is also a relatively consistent trend in sign and magnitude among five of the

remaining six states. This category is itself a summation with the largest component

being teacher salaries.13 Table 4-8 shows how salaries have been affected by changes

in unitary status. It shows a statistically significant drop in teacher salaries in the same

specifications and accounts for about 70 percent of the drop in TCE on Instruction per

student. 14

A drop is also notable in benefits paid to instructors. This effect is shown in Table

4-9 which shows about a $20 per student drop in the benefits paid to instructors.

However, some of the robustness is lost as it is statistically significant in only one

specification.15 When combined, the drop in salaries and benefits for instructional staff

appears to account for 90 percent of the drop in TCE on Instruction.

This is a rather significant trend which suggests that districts and/or teachers may

be reacting to unitary status in a somewhat uniform way.16 One possible explanation for

this trend in salaries and benefits is that the number of instructors in each district is

13 Salaries accounts for nearly 72% percent of all expenditures on instruction in each district.
14 These results are obtained while using the number of FTE Instructors in the district as a time-varying
control. Since the number of instructors is endogenous, the specifications were also run omitting this
variable as a control. The results for TCE on Instruction and Salaries for Instructors were similar in
statistical significance and magnitude and are available upon request.
15 The sign and magnitude pattern is maintained across the other state-level specifications.
16 Salaries and benefits paid for General Administration also show some downward trends as a result of
unitary status. However, the magnitude is smaller and there is less of a consistent pattern across
specifications.









decreasing as a result of receiving unitary status. The current dataset provides some

basic information about the number of FTEs in each district. Therefore, it is possible to

regress the number of FTEs on whether unitary status has been received and other

district-level covariates. It is also possible to construct a crude class size measure

based on the number of students in the district per FTE. Using each of these as a

dependent variable in Equations 4-1 and 4-3 tests the extent to which the number of

FTEs is affected by unitary status.17 These results are presented in Table 4-10. The

results show that there is little consistent effect on the number of FTEs or the number of

students per FTE. This suggests that the number of FTE instructors does not change in

any uniform manner when unitary status is received.

There are other possible explanations for this impact on salaries and benefits.18

First, it may be possible that the pool of instructors is becoming younger, less

experienced, or less qualified on average over time as a result of unitary status. This

might occur if older teachers are more likely to leave a district after it becomes unitary

and are replaced by younger, less experienced recruits. Since teacher salaries are

relatively formulaic, this would allow the number of FTE instructors to remain constant

while the average district-level profile is decreasing in age, experience, and/or quality.

That would result in lower salaries on average in unitary districts.19


17 In each of these specifications, the number of FTEs is removed from the set of time varying district-
level covariates contained in X, .

18 Unfortunately, given the current, limited data they are untestable.
19 It might be possible that some of the potential change in the district profile is caused by the retiring of
the baby boom generation. If there is a difference in how this phenomenon affects, say, rural versus
urban districts, Unitary Status could be biased since only large districts are seeking a change in their
status. In this case, the coefficient on unitary status could just be picking up the contemporaneous effect
of replacing the retirees with younger teachers.









Recent work by Feng, Figlio and Sass (2010) highlights another possible

contemporaneous effect. Their work suggests that teachers move in response to

increased or decreased accountability measures. Teachers facing tougher

accountability measures are more likely to leave their current school while those facing

lower accountability measures are more likely to stay. They also find some evidence

that changes in accountability can affect the distribution of teacher quality as measured

in certain value-added measures. To the extent that many unitary status grants were

received during the time of increased emphasis on school accountability measures, the

coefficient on Unitary Status, may be picking up some of the effect caused by changes

in accountability measures across districts. Considering that the exogenous shock the

authors use to identify teacher mobility occurs at the school level, it is unlikely that a

state-year or combination of year-level and district-level fixed effects would account for

this type of variation. Therefore, the increased accountability measures could be

producing bias. One thing is important to note. If the effect on per student instruction

and salary expenditures is also capturing the effect of accountability and teacher

mobility, that movement must be altering the salary and benefits profile in unitary

districts. Although there are many ways this type of change may be occurring, there is

one way that is certainly not responsible and that is within-district movement of

teachers. Given the current level of analysis, this type of movement would result in a

net effect of zero at the district level. 20




20 This assumes that teachers do not accept lower salaries when they move between schools.
Unfortunately, Feng et al. does not discuss the effect of accountability on the distribution of experience.
Therefore, any relationship is just conjecture at this point.









Lower salaries in unitary districts could also be a sign that a compensating wage

differential is paid to teachers that work in districts under a court-ordered desegregation

plan. Since districts that are still under such plans have very high levels of integration,

white teachers may need to be induced by a higher salary to teach in what are likely to

be multiracial classes.21 When the desegregation plan is lifted and schools become

more segregated as discussed in Chapter 2, white teachers may be able to teach in

classrooms more to their liking for a lower salary. Martin (2010) has found evidence of

compensating wage differentials in districts that have higher concentrations of minority

students. She also found that more highly segregated districts pay lower wages, all

else equal. This study's results, that salaries will be lower in districts treated to unitary

status, are consistent with Martin's findings as well.

Revisiting Economies of Scale and Final Thoughts

The results in this chapter make it unlikely that districts seek unitary status in order

to take advantage of any perceived fiscal benefits. There does not appear to be much

of a response in terms of revenues at the local, state, or federal levels. There are some

uniform results which suggest teacher salaries and benefits are affected by the receipt

of unitary status. However, given the current data, it is not possible to determine exactly

how and why the decrease in expenditures occurs. It is also hard to construct a story

where districts use a priori expectations about expenditures on salaries and benefits as

a justification to seek unitary status. It is more likely that any changes to teacher

salaries occur unexpectedly or as a result of a confluence of factors.



21 This would be analogous to teachers leaving teaching all together at the beginning of the
desegregation era. Some of those that stayed may have stayed only because they were paid a higher
salary.









It is also important to note that the weak findings from Chapter 3 which suggested

that the burden on local tax payers could be reduced by seeking unitary status do not

carry over to data from a different source. Table 4-6 shows no statistically significant

effect of unitary status on local property tax revenues. Since this was the potential

source of fiscal gains for unitary districts in Florida, it makes the existence of economies

of scale unlikely. This is further compounded by small districts in other states receiving

unitary status when economies of scale would seem to suggest that they shouldn't be

seeking unitary status in the first place.










Table 4-1. Court-ordered desegregation status by state
Never Taken to Still under Court Granted Unitary Status Prior to Granted Unitary Status After
Court Order 1990 1990 Total
Alabama 6 53 20 52 131
Florida 33 15 8 11 67
Georgia 71 76 3 28 178
Louisiana 9 43 5 11 68
Mississippi 53 69 4 20 146
North Carolina 77 15 8 7 107
South Carolina 52 13 2 16 83
Total 301 284 50 145 780

% of Students
Affected 28.9 26.6 12.8 31.7 100
Note: Districts counts based on districts in existence in 2007. With the exception of Florida, each state has a slight change in the number of
districts over time due to the consolidation or addition of independent city school districts. Counting the number of districts on either side
of 1990 is not problematic because no districts in the sample receive unitary status in 1990.










Table 4-2. Summary fiscal data by state.
AL FL GA LA MS NC SC ALL
(1) (2) (3) (4) (5) (6) (7) (8)
Total Federal Revenue 702 722 762 1,081 997 662 775 808
Total State Revenue 3,925 3,977 4,194 3,554 3,234 4,677 3,820 3,923
Total Local Revenue 1,784 3,000 2,515 2,378 1,600 1,895 2,768 2,193
Total Revenue 6,411 7,698 7,472 7,013 5,830 7,234 7,362 6,923


Total Current Expenditures for Elementary Education 5,703 6,186 6,463 6,157 5,230 6,316 6,316 6,020
Total Non-Elementary/Secondary Expenditures 121 162 20 30 23 44 96 62
Total Capital Outlay Expenditures 576 1,178 844 528 541 711 867 729
Payments to State Governments 0 0 0 0 0 0 9 1
Payments to Local Governments 0 0 0 0 0 0 0 0
Payments to Other School Systems 5 0 33 9 0 0 35 13
Interest on Debt 90 115 79 123 84 118 180 105
Total Current Expenditures: Services 6,495 7,642 7,439 6,847 5,878 7,189 7,503 6,928
Note: Data are in constant 2000 dollars per student and were deflated using the national CPI-U. Means were calculated with
unweighted district-level data for each state. Due to the formulaic nature of school finance, weighting does not appreciably
change any of the data.











Table 4-3. The effect of unitary status on total revenues and total expenditures per student.

(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
REVENUES ALL ELIGIBLE AL FL GA LA MS NC SC
Total Federal Revenue -2.01 56.27 36.99* -38.64 -27.99 865.20 89.82 -14.49 -69.29

Total State Revenue -58.06 -40.80 17.85 -41.90 -102.00 266.80 47.44 4.33 63.69
Total Local Revenue -105.70 -93.07 -12.59 26.47 -206.60 248.50 -1.25 99.53 65.15
Total Revenue -165.70* -77.60 42.25 -54.07 -336.50 847.00 136.00 89.37 59.56


EXPENDITURES

Total Current Expenditures for Elementary Education 173.40*** -134.60* -4.23 143.20* -180.30 114.20 -11.47 97.29 108.00
Total Non-Elementary/Secondary Expenditures -5.87 -13.38* 4.75 25.37 1.37 -5.55 -3.20 -9.69 6.43

Total Capital Outlay Expenditures 29.96 56.51 66.45 -66.70 250.30* 172.70 18.22 123.80 316.00
Payments to State Governments -0.21 -0.05 -0.07
Payments to Local Governments 0.11 -0.04
Payments to Other School Systems -0.31 1.39 -0.79 -1.90 8.13 -15.77
Interest on Debt 3.14 -2.10 3.40 3.26 6.68 -38.41 19.13 61.64 4.70
Total Current Expenditures -146.60* -92.24 69.58 -181.30 -424.50 22.61 22.68 273.10 203.30
Note: Data represent the coefficient for Unitary Status obtained by regressing different fiscal variables on whether unitary status has been received.
Controls include either a state-year fixed effect or fixed effects by year and district, the number of students in the district, the number of fulltime
equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is non-
white. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*)
percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida,
Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have
been subject to court order and had not received unitary status by 1989. Columns 3-9 show results for each state individually.











Table 4-4. The effect of unitary status on federal revenues per student.

(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
ALL ELIGIBLE AL FL GA LA MS NC SC


Fed-Thru-State: Child Nutrition Act
Fed-Thru-State: Title I
Fed-Thru-State: Child with Disabilities Idea
Fed-Thru-State: Math, Science, and Teacher
Fed-Thru-State: Safe and Drub Free Schools
Fed-Thru-State: Title V, Part A
Fed-Thru-State: Vocational and Tech Education
Fed-Thru-State: Other
Federal Revenue Nonspecified
Fed-Direct: Impact Aid
Fed-Direct: Bilingual Education
Fed-Direct: Indian Education
Fed-Direct: Other
Total Federal Revenue


-4.32
-2.57
0.16
0.96
0.87
-0.36
-0.39
9.02
-2.84
-1.25
-0.33
-0.51
2.22
-2.01


-2.34
1.91
0.87
1.05
-0.91
0.45
-0.84
46.26
-3.11
6.95
0.23
-0.80
7.11
56.27


-4.24
4.47
0.75
1.76
-2.21
-0.07
-1.29
27.40***


5.06
-0.23
-0.66
-2.98
36.99*


-4.24
-4.48
4.90
-2.48
-0.01
4.56
-6.18**
-16.61


-0.06
1.97
0.01
-18.48**
-38.64


1.12
-12.94
-4.19
-4.957**
0.06
-0.04
0.73
3.27
7.16
1.69
0.48**


9.50
14.72
-10.63
5.42
0.65
-0.46
2.15
1,086.00


-3.87
-0.37


-0.06
-13.29 -73.57
-27.99 865.20


Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been
received. Controls include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of
fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student
population that is non-white. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical
significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from
using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to
only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 3-9 show
results for each state individually.


12.17**
5.83
6.64
4.32
0.01
2.16
-3.17
98.78


-4.01
0.31
-0.06
-11.48
89.82


-5.28
-4.63
-0.75
-0.56
2.94
0.91
0.55
16.15
18.21
-1.12***
-0.89
0.38
-32.45
-14.49


6.62
-57.19**
-0.39
7.38
0.18
-0.90
3.42
-54.34
23.49
-0.94
0.05


0.29
-69.29











Table 4-5. The effect of unitary status on state revenues per student.

(1) (2) (3) (4) (5) (6) (7) (8) (9)


ALL
ALL ELIGIBLE


AL FL GA


LA MS NC SC


State Rev.: General Formula Assistance
State Rev.: Special Education Programs
State Rev.: Transportation Programs
State Rev.: Staff Improvement Programs
State Rev.: Compensatory and Basic Skills Programs
State Rev.: Vocational Education Programs
State Rev.: Capital Outlay And Debt Service Programs
State Rev.: Bilingual Education Programs
State Rev.: Gifted and Talented Programs
State Rev.: School Lunch Programs
State Rev.: Other Programs
State Rev. On Behalf: Employee Benefits
State Rev. On Behalf: Not Employee Benefits
State Rev.: Not Specified
Total State Revenue


-37.24
-0.79
9.36
3.72
-0.04
-1.54
-3.62
-4.74
3.71
-0.48**
-17.01
-1.00
0.82
-6.91
-58.06


-21.72
-5.66
0.15
0.09
1.67
4.55
0.27
-4.25
0.22
-0.46
-5.50
-0.31
0.10
-8.84
-40.80


23.12 53.59 18.33 -234.00 12.87 17.13 -5.18


0.84
-13.42
0.25
-4.84
1.58
-4.13


-41.14
-6.07
-0.57
-33.26
4.01
1.54


0.00 -2.78 -0.03 6.65
0.79
8.12
0.02 12.29 1.77
0.28 3.16 5.09
-78.71 6.17*** 17.99


-3.48
-2.33
-8.02
4.32
19.90
82.69


-0.16 -15.24


-0.19 0.23


10.91
3.09*


0.44 -2.67
-1.26 -0.48** -1.98 -0.39


-3.70 -24.55 -15.49 30.38* 24.05 -0.83
4.17 1.13 -2.23


0.00 0.24
-31.85*


2.43 -5.21


17.85 -41.90 -102.00 -266.80 47.44 4.33 63.69


Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been
received. Controls include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of
fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student
population that is non-white. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical
significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from
using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to
only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 3-9 show
results for each state individually.











Table 4-6. The effect of unitary status on local revenues per student.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
ALL ELIGIBLE AL FL GA LA MS NC SC
Local: Tuition Fees from Pupils and Parents -1.67 -1.48 0.37 0.09 8.70 -4.85 0.69 -7.21 2.28
Local: Transportation Fees from Pupils and Parents -0.20 -0.16 0.18 0.25 0.04 -0.95
Local: Textbook Sales and Rentals 0.00 -0.05 0.00 -1.35 0.00
Local: District Activity Receipts -5.60 -15.76** -17.93** 34.88 -5.24 .-3.85 .-8.92
Local: Other Sales and Services -0.79 -3.18 -0.96 10.26 -6.117*** 6.84 -8.72 -3.24
Local: Student Fees, Nonspecified -0.32 -0.20 -0.02
Local: Property Taxes -24.18 -15.54 -21.08 38.13 -36.54 176.7* -18.26 125.80
Local: General Sales Taxes -28.97 -19.70 .-4.40 34.52
Local: All Other Taxes 2.38 4.71 16.47 .0.28 12.32 .-16.79
Local: From Other School Systems 1.47 0.40 2.64 -0.23 0.98 0.03 18.24 0.59 1.33
Local: From Cities and Counties -21.54 -16.15 67.88 .-56.15 4.15 -16.16* -17.46
Local: School Lunch -8.81*** -2.87 2.36 6.19 2.92 -3.16 1.12 3.06 0.28
Local: Interest Earnings 7.61 -5.69 1.68 -3.60 -18.72 -35.85 4.56 4.65 -0.58
Local: Miscellaneous -31.80** -39.38 -60.63 -54.85 -93.04* 70.36 -7.77 -40.07 -16.26
Total Local Revenue -105.70 -93.07 -12.59 26.47 -206.60 248.50 -1.25 99.53 65.15
Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been
received. Controls include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of
fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population
that is non-white. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance
starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from
Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these
seven states that have been subject to court order and had not received unitary status by 1989. Columns 3-9 show results for each state
individually.











Table 4-7. The effect of unitary status on total current expenditures (TCE) per student.
(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
ALL ELIGIBLE AL FL GA LA MS NC SC

TCE: Instruction -101.20*** -84.39** 3.18 104.90** -120.30 -62.71 -89.16 -34.08 -55.54
TCE: Support Services -60.29** -43.08 -2.70 -37.48 -2.44 -30.48 77.95 126.30 -49.10
TCE: Other (Elementary/Secondary) -11.61* -7.07 -4.98 -0.85 -60.30 -19.38 -0.26 5.03 -3.39
TCE for Elem/Secondary Education -173.40*** -134.60* -4.23 -143.20* -180.30 -114.20 -11.47 97.29 -108.00


TCE: Support Services (Pupils) -0.56 5.54 5.35 0.97 -13.38 23.55 -5.90 23.58 -1.78
TCE: Support Services (Instructional Staff) 0.24 1.82 9.65 -17.38 22.71 -31.53 29.06 18.78 -55.20
TCE: Support Services (General Admin) -27.21*** -11.01 1.80 5.99 15.36 29.40 -11.24 -12.90 20.04
TCE: Support Services (School Admin) -8.45 -17.81** 3.74 -23.26** -7.78 -24.45 6.90 21.62 5.15
CE-Support Services-Operation and Main. of Plant -14.48 -13.16 -14.62 -0.44 2.83 100.90 77.76 24.52 24.08
CE-Support Services-Student Transportation -2.70 -5.10 -5.44 2.91 -3.61 -2.15 -6.32 17.62 -18.43
CE-Support Services-Nonspecified -0.44 -0.27 0.77
TCE: Support Services -60.29** -43.08 -2.70 -37.48 -2.44 -30.48 77.95 126.30 -49.10
Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. Controls
include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime equivalent teachers in the
district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is non-white. For brevity, the coefficients for
these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and
one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South
Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by
1989. Columns 3-9 show results for each state individually.











Table 4-7. Continued
TCE: Food Services
CE-Enterprise Operations
CE- ELSEC
TCE: Other (Elementary/Secondary)


Non-Elem./Secondary Expenditures: Comm. Services
Non-Elem./Secondary Expenditures: Adult Education
Non-Elem./Secondary Expenditures: Other
TCE: Non Elementary/Secondary


Capital Outlay: Construction
Capital Outlay: Land and Existing Structures
Capital Outlay: Instructional Equipment
Capital Outlay: Other Equipment
Capital Outlay: Nonspecified Equipment
Total Capital Outlays


-13.01**
0.79
0.90
-11.61*


0.62
-4.44
-1.48
-5.87


45.10
-5.83
-1.71
-4.27
-3.37
29.96


-4.71
-1.46
-0.54
-7.07


-1.71
-6.05
-5.46
-13.38*


55.24
-0.75
0.67
0.85
0.48
56.51


-3.10 -0.85 -3.78 -13.85 -5.24 5.03 13.61
-8.23 -0.04 -0.30 -8.75
-1.00 -3.94 -5.36 4.69 -4.33
-4.98 -0.85 -60.30 -19.38 -0.26 5.03 -3.39


4.15 11.04 -1.84 1.03 1.58 -9.40 12.06
4.74 12.24 2.81 -5.95 -0.65 -0.31 -8.94
-0.66 -1.24 -4.31
4.75 25.37 1.37 -5.55 -3.20 -9.69 6.43


74.06 -72.28 -292.30** 162.30 -3.90 139.40 407.90
-5.71 -5.27 56.33* -7.71 -11.57 -48.10
-2.90 2.15** -13.00 -8.71 -2.20 2.75 9.65
2.92 5.55 -1.51 36.98 3.76 -11.07 -28.31
-4.05 1.24 24.20* -21.13
66.45 -66.70 -250.30* 172.70 18.22 123.80 316.00











Table 4-8. The effect of unitary status on salaries


(2) (3) (4) (5) (6) (7) (8) (9)
ALL
ELIGIBLE AL FL GA LA MS NC SC


Salaries-Support Services-Pupils
Salaries-Support Services-Instructional Staff
Salaries-Support Services-General Administration
Salaries-Support Services-School Administration
Salaries-Support Services-Oper. and Main. of Plant
Salaries-Support Services-Student Transportation
Salaries-Food Services
Salaries: Instruction
Total Salaries


-1.20
1.97
-14.32***
-3.82
-5.12
-0.77
-4.99**
-70.33***


1.77 0.05


0.78
-5.53


-7.28 1.73


-7.65 14.85 18.37*


2.52 -20.01 20.94* -18.14 3.61 15.00* -9.10
1.68 1.90 14.72 5.15 -5.73 -8.22 6.43


-11.15* 6.22 -18.94**


-7.79
-2.41


-3.85 -12.27 5.94 20.28** 1.63


1.56 -28.07** 11.79* -33.33**


5.62 -0.10


-5.72** 0.72 -3.96
-72.75*** -5.56 -88.47**


-2.57 3.26


-4.37 7.60 -2.82
-12.34 5.75 -7.16


-2.77 -14.25 -6.66 -0.03 4.84


-82.40 -99.80*


-70.84 -17.90 -57.99


-104.20*** -108.60** 19.49 -161.60** -58.17 -180.20* -98.40 45.83 -88.54


Note: Data represent the coefficient for Unitary Status obtained by regressing different salary expenditures per student on whether unitary status has been
received. Controls include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime equivalent
teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is non-white. For brevity,
the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing
to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North
Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not
received unitary status by 1989. Columns 3-9 show results for each state individually.


100











Table 4-9. The effect of unitary status on benefits
(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
ALL ELIGIBLE AL FL GA LA MS NC SC
Employee Benefits-Support Services-Pupils -0.37 1.20 -1.02 2.17 1.33 2.44 -1.40 2.94 -3.83
Employee Benefits-Support Services-Instructional Staff 0.42 0.90 0.44 -3.48 7.42** -4.67 1.14 1.96 -2.75
Employee Benefits-Support Services-General Administration -3.24*** -1.16 1.33 4.19 2.44 -0.60 -1.34 -2.04 -2.45
Employee Benefits-Support Services-School Administration -1.67 -3.05** -0.06 -3.12 0.18 -4.26 1.10 3.88* 0.23
Employee Benefits-Support Services-Op. and Maint.of Pant -0.60 -1.15 -0.92 1.35 5.73** 0.97 -0.93 1.10 -0.69
Employee Benefits-Support Services-Student Transportation 1.19 0.73 -0.06 4.92 0.07 2.63 -7.33 2.35 -2.56
Employee Benefits: Food Service -0.72 -0.64 -2.70 5.58** -0.89 2.51 -2.54 3.18 0.60
Employee Benefits: Enterprise Operations 0.01 0.00 0.00 0.00 -0.04 -0.01 -0.10 0.00 -0.07
Employee Benefits: Instruction -20.05** -8.50 1.29 -16.67 -7.69 -30.71 -16.34 -3.63 -17.47
Total Employee Benefits -25.67** -11.89 -3.73 -2.76 7.74 -26.14 -27.83 14.45 -28.35
Note: Data represent the coefficient for Unitary Status obtained by regressing different types of employee benefits per student on whether unitary status has
been received. Controls include either a state-year fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime
equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is non-white.
For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level
and increasing to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana,
Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order
and had not received unitary status by 1989. Columns 3-9 show results for each state individually.










Table 4-10. The effect of unitary status on teachers


(1) (2) (3) (4) (5) (6) (7) (8) (9)
ALL
ALL ELIGIBLE AL FL GA LA MS NC SC
Number of FTEs 50.95* 61.56* 14.84 221.9 24.87 -5.301 4.908 -16.36 -5.793
Students per FTE 0.0677 0.102 0.0429 -0.239 -0.263 0.0964 -0.109 0.658 -0.149
Note: Data represent the coefficient for Unitary Status obtained by regressing different measures of the number of instructors in each
district on whether unitary status has been received. Controls include either a state-year fixed effect or fixed effects by year and district, the
number students in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population
that is non-white. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical
significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels. Column 1 shows the results from
using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the
sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989.
Columns 3-9 show results for each state individually.


102












Timing of Unitary Status Grants by Size: All States


A A A ,A


3.6 (Median)


1970


1980


1990 2000


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order
Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.


Figure 4-1. Timing of unitary status grants for eligible districts (all states).


103












Timing of Unitary Status Grants by Size: State = AL


O
h,-.0 O
U)
O
()t-
0
O CO

4-
C
0




a-

) 0
vC ^


* A.AA


A mA La aEaAIIA


1980


1990


2000


3.2 (Median)


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.



Figure 4-2. Timing of unitary status grants for eligible districts by size: Alabama.


104


S A


1970












Timing of Unitary Status Grants by Size: State = FL


0

o
oC

C CO
0
.o
= 0
CL CM
0
0-
4a-.C
C 0
U)
-0
W-


1980


1990


2000


33.0 (Median)


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status

Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.



Figure 4-3. Timing of unitary status grants for eligible districts by size: Florida.


105


A A


1970












of Unitary Status Grants by Size: State = GA


A


v)
0
00
oo


C=

3
0

-a-f

^-cc


1980


1990


2000


2.9 (Median)


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.


Figure 4-4. Timing of unitary status grants for eligible districts by size: Georgia.


106


Timing
0
CD


0
ei
i "t~A












of Unitary Status Grants by Size: State = LA


Timing
C
O -
C.
00




4-
oo
CD





CI,
A


A A


1980


1990


2000


6.2 (Median)


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.


Figure 4-5. Timing of unitary status grants for eligible districts by size: Louisiana.


107


< U'


1970














of Unitary Status Grants by Size: State = MS


Timing
0
O -

O 0

0
I--
o

0 0
o o

Ia-
(CD
_= C


1980


1990


- imap p


2000


2.6 (Median)


2010
2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.


Figure 4-6. Timing of unitary status grants for eligible districts by size: Mississippi.


108


A At.4 AtA


1970












Timing of Unitary Status Grants by Size: State = NC
o A
oa
o0
o)






o 7.6 Median
oco



I I I
-a-
4-
C)

A A- 7.6 (Median)


1970 1980 1990 2000 2010

Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.



Figure 4-7. Timing of unitary status grants for eligible districts by size: North Carolina.


109












Timing of Unitary Status Grants by Size: State = SC
C

CO
0
0

0 0
O It


Q--
m A
I)
oA


- A Z M -W


1985


1990


1995


2000


2005


3.5 (Median)


2010


Year Unitary Status Is Received

A Districts w/ Unitary Status
Districts Still Under Court Order

Note. Districts still under court order have been given a unitary year of 2010. Random variation has
been added to points that would otherwise be on top of each other. Districts never taken to
court have been excluded.



Figure 4-8. Timing of unitary status grants for eligible districts by size: South Carolina.


110









CHAPTER 5
CONCLUSION

After several decades of active involvement, the courts that largely helped

desegregate southern schools are ending their involvement in the process. The

granting of unitary status would appear, at first glance, to be a watershed moment that

could unleash a large amount of change on the districts that receive it. Aside from the

re-sorting of students within districts, this study suggests changes in these districts are

muted. As far as Florida is concerned, it is clear that court-ordered desegregation plans

place some constraint on district assignment of students. As shown in Chapter 2,

districts that are released from a court-ordered desegregation plan see their level of

racial integration fall and the level of segregation rise. This change does not happen

instantaneously. It takes time for districts and parents to react, but over time the level of

segregation can be expected to double within eight years of receiving unitary status.

Such a drastic change would be a source of concern if it were accompanied by a

decrease in academic outcomes. Fortunately, no back-sliding in academic achievement

is occurring. Unitary status appears to have no effect on academic achievement as

measured by scores of the average student, the average white student, and the

average black student. To the extent that parental preferences are being better met by

the new distribution of students across schools, these results suggest that unitary status

may be a pareto improvement over the status quo.

Parental preferences may not be the only reason why districts seek unitary status.

If court-ordered desegregation plans are placing constraints on the assignment of

students, it is likely that there are also other constraints. One possible area where

constraints could be an issue is in school district finance. If finances are further


111









constrained by court orders than they otherwise would be, seeking unitary status to

relieve that constraint would be an appropriate step for local school boards. Although

the logic is enticing, there is little evidence that this is the case for most districts

receiving unitary status. In Chapters 3 and 4, there is little evidence that school district

finances are affected by unitary status to an extent that makes changes in finances a

"smoking gun" which fully explains the decision by officials to seek unitary status. This

is contrary to the assertions of Ryan (1999) which suggests that school districts agree to

terminate mandatory desegregation plans in exchange for large payments from state-

level education agencies and Moore (2002) which suggests that court-ordered

desegregation creates costs which must be borne by local communities. No such

effects are noted in this multistate analysis. If such bribes occur, they are the exception

rather than the rule since there is no significant change in revenues caused by the

receipt of unitary status.1 Additionally, it is hard to find where costs may be avoided by

seeking unitary status. With the exception of teacher salaries and benefits, few

expenditure categories changed as a result of exiting a court-ordered desegregation

plan.2

There are several other reasons why unitary status might have no effect on

revenues and expenditures. First, the era of school district finance reform altered the


1 Ryan provides examples from five specific districts in five states. None of these states are included in
this analysis.
2 It is possible that changes in expenditures are occurring within district and within category so that no
effect is identifiable. For example, assume that a district must use bussing to meet the intent of the
court's desegregation plan. This is most likely more expensive than bussing students to their closer
neighborhood schools. When unitary status is received and the district is relieved of this bussing
requirement, expenses on student transportation should fall. The results in Table 4-7 show that there is
no statistical effect on student transportation. This can mean two things. First, the bussing plan was not
artificially raising expenditures or the district replaced the bussing plan with an alternative that was just as
costly. Both would lead to the result reported in this paper.


112









way finances are done by state-level education agencies. Most school, district and

state financing decisions are now extremely rules-based. Since few, if any, of these

rules take into account a district's desegregation status, it would be unlikely that a

change in that status would affect financing levels. Another more troubling possibility is

that the student re-sorting that is causing the increasing levels of segregation is a within

district phenomenon. This might change school-level finance measures as students

move between schools. The analysis completed in this study cannot rule out this type

of change since such changes will net out at the district level. There is a final,

straightforward reason why unitary status may have no effect on school district finances.

Simply put, court-ordered desegregation worked. If the court system did its job,

students should be able to re-sort into increasingly segregated schools with no change

occurring in the district's finances. To some extent, such an occurrence is consistent

with the results of this paper.

Although this study provides evidence that court-ordered segregation plans were

successful and districts can be released from them without disastrous results, it does

not suggest that the era of monitoring segregation and its effects should come to an

end. There are several unanswered questions within this study which provide avenues

for future research on the effects of unitary status. Additional data on student-level

movement and test scores would provide a true measure of how black students are

being affected by unitary status. Too much can be occurring at the school, classroom,

and student levels to make a district-level average of black student performance the

final word on academic achievement. Additionally, teacher-level data would help

solidify the effect unitary status has on teacher mobility and salaries. This would add to


113









a robust literature on teacher mobility at the beginning of the school desegregation era.

Answering these questions should continue to provide insight into what causes districts

to seek release from their court-ordered desegregation plans and the consequences of

unitary status once they receive it.


114









APPENDIX A
ADDITIONAL TABLES

Table A-1. Litigation status by district
Year Litigation Year Case Closed
County Started (Unitary Status Granted)
Alachua 1964 1971
Baker 1970 Pending
Bay 1966 Pending
Bradford 1970 Pending
Brevard 1966 1978
Broward 1970 1996
Columbia 1970 1987
Dade 1956 2001
Duval 1960 2001
Escambia 1960 2004
Flagler 1970 Pending
Gadsden 1970 1986
Gulf 1970 Pending
Hendry 1970 Pending
Hillsborough 1958 2001
Indian River 1965 Pending
Jackson 1970 Pending
Jefferson 1970 Pending
Lafayette 1970 Pending
Lee 1964 1999
Leon 1962 1974
Manatee 1965 Pending
Marion 1978 Pending
Orange 1970 Pending
Palm Beach 1956 1979
Pasco 1970 Pending
Pinellas 1964 2000
Polk 1963 2000
St. Johns 1970 Pending
St. Lucie 1970 1997
Sarasota 1963 1970
Seminole 1970 2006
Volusia 1960 1970
Wakulla 1970 Pending
NOTE: The following districts have never been taken to court: Calhoun, Charlotte,
Citrus, Clay, Collier, De Soto, Dixie, Franklin, Gilchrist, Glades, Hamilton, Hardee,
Hernando, Highlands, Holmes, Lake, Levy, Liberty, Madison, Martin, Monroe, Nassau,
Okaloosa, Okeechobee, Osceola, Putnam, Santa Rosa, Sumter, Suwannee, Taylor,
Union, Walton, Washington.


115









Table A-2. Unitary status' effect on total local,


Unitary Status

Number of Schools

Tax Roll per Student

Fraction FRPL Eligible

Fraction Non-White

Constant


114.80
(99.76)
-0.30
(4.07)
0.00314***
(0.0011)
1509.00
(1679)
3171.00
(3304)
4560***
(942)


(2)
-76.78
(122.30)
-5.36
(4.35)
0.0012
(0.0008)
-315.10
(1038)
8600**
(3438)
3803***
(1103)


(3)
-151.50
(141.60)
-1.40
(4.27)
0.0010
(0.0008)
533.90
(965)
8419**
(3521)
3437***
(1087)


-6.29
(127.20)
0.31
(2.15)
0.00189***
(0.0005)
1356.00
(1261)
8259***
(1822)
1486
(1394)


Weighted by FTEs NO NO NO YES NO YES
Observations 670 340 250 250 100 100
R-squared 0.655 0.721 0.669 0.894 0.929 0.925
NOTE: Data represent the results obtained by regressing the total amount of revenue received from all local, state, and
federal sources per FTE on whether unitary status has been received. Asterisks represent increasing levels of statistical
significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The numbers
in parentheses are robust standard errors. Column 1 shows the results from a sample of all Florida districts. Column 2
restricts the sample to only those districts that have been subject to court order. Column 3 represents only those districts
that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been
weighted by the number of unweighted full-time equivalents in the district. Columns 5 and 6 follow the same procedure
but only include the ten largest districts in the state, excluding Miami-Dade.


116


(5)
-37.14
(115.30)
0.96
(3.22)
0.00158**
(0.0005)
1292.00
(1755)
11494***
(2890)
1570
(1130)


(6)
-71.30
(156.50)
1.04
(3.15)
0.00142**
(0.0005)
1832.00
(1724)
13386***
(2592)
251.3
(1139)


state and federal revenue per student









Table A-3. Unitary status' effect on state funding from other sources


Unitary Status

Number of Schools

Tax Roll per Student

Fraction FRPL Eligible

Fraction Non-White

Constant


Weighted by FTEs
Observations
R-squared


(1)
134.1 **
(58.99)
-0.17
(1.63)
0.000150*
(0.0001)
971.80
(749)
-1135.00
(1366)
328.7*
(197)
NO
670
0.298


(2)
85.86*
(42.23)
-2.031**
(0.86)
0.0000
(0.0001)
442.50
(435)
191.30
(922)
230.7
(336)
NO
340
0.507


(3)
99.97*
(51.94)
-1.74
(1.04)
0.0001
(0.0001)
921.3**
(347)
170.20
(994)
40.12
(339)
NO
250
0.498


(4)
84.18**
(33.57)
-1.437***
(0.51)
0.000327**
(0.0002)
1020.00
(704)
-1053*
(570)
643.4
(425)
YES
250
0.63


(5)
51.19
(35.24)
-0.86
(0.73)
0.000524***
(0.0002)
1276.00
(993)
-1916**
(834)
487.3
(451)
NO
100
0.658


(6)
36.14
(52.78)
-1.271*
(0.65)
0.000589***
(0.0001)
1615.00
(1054)
-2450**
(975)
720.5
(551)
YES
100
0.656


NOTE: Data represent the results obtained by regressing the amount of revenue received from all "other" state-level
sources per FTE on whether unitary status has been received. Asterisks represent increasing levels of statistical
significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The
numbers in parentheses are robust standard errors. Column 1 shows the results from a sample of all Florida districts.
Column 2 restricts the sample to only those districts that have been subject to court order. Column 3 represents only
those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the
observations have been weighted by the number of unweighted full-time equivalents in the district. Columns 5 and 6
follow the same procedure but only include the ten largest districts in the state, excluding Miami-Dade.









Table A-4. Unitary status' effect on general administration expenses
(1) (2) (3) (4) (5) (6)
Unitary Status 5.46 4.25 6.15 10.09*** 7.644** 4.44
(3.86) (3.97) (5.77) (3.49) (2.97) (3.51)
Number of Schools -0.270* -0.19 -0.22 0.02 0.07 0.09
(0.16) (0.18) (0.22) (0.06) (0.10) (0.10)
Tax Roll per Student 1.77e-05* 0.0000 0.0000 0.00 0.00 0.00
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Fraction FRPL Eligible 4.58 8.16 4.16 -20.91 -43.60 -39.50
(26) (30) (38) (47) (63) (73)
Fraction Non-White 47.12 55.69 47.76 -20.96 -107.20 -230.8*
(142) (136) (155) (89) (106) (111)
Constant 61.11 49.08 61.86 56.81 83.65** 125.5**
(40) (38) (42) (46) (37) (44)
Weighted by FTEs NO NO NO YES NO YES
Observations 670 340 250 250 100 100
R-squared 0.863 0.885 0.88 0.725 0.787 0.751
NOTE: Data represent the results obtained by regressing the level of expenditure on General Administration per FTE
on whether unitary status has been received. Asterisks represent increasing levels of statistical significance starting at
the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The numbers in parentheses are
robust standard errors. Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the
sample to only those districts that have been subject to court order. Column 3 represents only those districts that were
under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted
by the number of unweighted full-time equivalents in the district. Columns 5 and 6 follow the same procedure but only
include the ten largest districts in the state, excluding Miami-Dade.


118











Table A-5. Fiscal variable means (all variables).
(1) (2) (3) (4) (5) (6) (7) (8) (9)
AL FL GA LA MS NC SC ALL N
Total Revenue 6,411 7,698 7,472 7,013 5,830 7,234 7,362 6,923 11,649
Total Federal Revenue 702 722 762 1,081 997 662 775 808 11,649
Fed-Thru-State: Child Nutrition Act 206 182 241 259 303 211 231 238 11,649
Total State Revenue 3,925 3,977 4,194 3,554 3,234 4,677 3,820 3,923 11,649
Total Local Revenue 1,784 3,000 2,515 2,378 1,600 1,895 2,768 2,193 11,649
Local: Property Taxes 591 2,395 1,636 918 1,097 0 2,091 1,295 10,686
Local: General Sales Taxes 0 0 379 1,101 0 0 0 198 10,686
Local: Public Utilities 0 0 0 0 0 0 0 0 10,686
Local: Individual and Corporate Income Taxes 0 0 0 0 0 0 0 0 10,686
Local: All Other Taxes 12 0 0 0 8 0 31 7 10,686
Local: From Other School Systems 8 2 9 3 12 2 4 7 11,649
Local: From Cities and Counties 575 0 163 30 16 0 129 152 11,649
Local: School Lunch 148 119 105 61 81 169 100 114 11,649
Local: Interest Earnings 62 102 74 124 56 22 84 69 11,649
Local: Miscellaneous 198 167 98 123 214 118 138 151 11,649
NCES Local Revenue, Census Bureau Revenue 0 0 0 0 0 0 0 0 11,649
Total Current Expenditures: Services 6,495 7,642 7,439 6,847 5,878 7,189 7,503 6,928 11,649
TCE: Instruction 3,440 3,495 4,017 3,563 3,132 3,914 3,694 3,623 11,649
TCE: Instruction 3,425 3,495 4,017 3,561 3,132 3,914 3,694 3,620 11,649
TCE: Support Services 1,821 2,360 2,033 2,143 1,701 2,009 2,209 1,989 11,649
TCE: Support Services (Pupils) 246 305 284 230 200 314 395 275 11,649
TCE: Support Services (Instructional Staff) 209 347 329 301 233 234 412 286 11,649
TCE: Support Services (General Admin) 187 128 164 168 206 179 117 170 11,649
TCE: Support Services (School Admin) 344 378 394 331 279 405 376 357 11,649
Total Current Expenditures (Other Elementary Secondary) 442 332 413 451 396 393 413 408 11,649
TCE: Food Services 438 332 395 446 366 393 368 392 11,649
Total Non-Elementary/Secondary Expenditures 121 162 20 30 23 44 96 62 11,649
Total Capital Outlay Expenditures 576 1,178 844 528 541 711 867 729 11,649
Capital Outlay: Construction 444 857 610 315 341 526 605 516 11,649
Capital Outlay: Land and Existing Structures 19 71 57 58 0 46 35 37 11,649
Total Current Expenditures for Elementary Education 5,703 6,186 6,463 6,157 5,230 6,316 6,316 6,020 11,649
Payments to State Governments 0 0 0 0 0 0 9 1 11,649


119











Table A5. Continued
Payments to Local Governments 0 0 0 0 0 0 0 0 11,649
Payments to Other School Systems 5 0 33 9 0 0 35 13 11,649
Interest on Debt 90 115 79 123 84 118 180 105 11,649
Total Salaries 3,624 3,871 4,242 3,841 3,301 4,301 4,059 3,885 11,649
Salaries: Instruction 2,481 2,381 2,929 2,533 2,279 2,933 2,707 2,629 11,649
Long Term Debt: Outstanding at Beginning of Fiscal Year 1,621 1,978 1,547 2,315 1,728 1,798 3,558 1,946 11,649
Long Term Debt: Issued During FY 309 367 348 303 253 391 826 379 11,649
Long Term Debt: Retired During FY 121 177 186 201 166 198 359 192 11,649
Long Term Debt Outstanding at End of Fiscal Year 1,812 2,168 1,707 2,421 1,814 1,994 4,028 2,133 11,649
Short Term Debt: Outstanding at Beginning of Fiscal Year 0 0 22 0 0 0 0 5 11,649
Short Term Debt: Outstanding at End of Fiscal Year 4 0 19 0 0 0 0 5 11,649
Assets: Sinking Fund 68 102 183 318 136 0 131 141 10,686
Assets: Bond Fund 522 292 862 452 406 2 554 527 10,686
Assets: Other Funds 867 1,234 994 1,470 1,218 1 873 1,009 10,686
Fed-Thru-State: Title I 234 204 212 304 333 156 211 238 10,869
Fed-Thru-State: Child with Disabilities Idea 117 130 81 118 117 95 137 109 10,869
Fed-Thru-State: Math, Science, and Teacher 24 16 18 31 22 14 23 21 10,869
Fed-Thru-State: Safe and Drub Free Schools 14 7 6 8 9 7 8 9 10,869
Fed-Thru-State: Title V, Part A 8 11 6 13 24 6 8 11 10,869
Fed-Thru-State: Vocational and Tech Education 25 21 13 17 14 16 27 18 10,869
Fed-Thru-State: Other 55 111 124 265 139 38 97 112 10,869
Federal Revenue Nonspecified 0 0 49 0 0 99 41 29 10,869
Fed-Direct: Impact Aid 8 7 10 13 5 10 4 8 10,869
Fed-Direct: Bilingual Education 2 1 0 1 0 1 1 1 10,869
Fed-Direct: Indian Education 2 0 0 0 0 1 0 1 10,869
Fed-Direct: Other 19 43 17 84 48 20 2 30 10,869
State Revenue: General Formula Assistance 3,322 1,633 3,398 3,333 2,808 3,932 1,076 2,942 10,869
State Revenue: Special Education Programs 25 577 0 37 2 170 396 123 10,869
State Revenue: Transportation Programs 210 207 0 0 0 85 63 71 10,869
State Revenue: Staff Improvement Programs 39 22 0 0 0 16 894 106 10,869
State Revenue: Compensatory and Basic Skills Programs 54 55 0 0 196 57 199 80 10,869
State Revenue: Vocational Education Programs 12 98 0 0 69 124 300 73 10,869
State Revenue: Capital Outlay And Debt Service Programs 140 331 132 0 18 60 131 108 10,869
State Revenue: Bilingual Education Programs 1 53 0 0 0 4 0 5 10,869


120











Table A5. Continued
State Revenue: Gifted and Talented Programs 0 73 0 0 0 9 35 11 10,869
State Revenue: School Lunch Programs 0 8 26 0 3 7 2 8 10,869
State Revenue: Other Programs 122 885 297 210 194 184 645 314 10,869
State Revenue On Behalf: Employee Benefits 60 0 80 3 0 0 6 29 10,869
State Revenue On Behalf: Not Employee Benefits 5 0 0 0 0 66 123 23 10,869
State Revenue: Not Specified 0 0 306 0 0 0 0 70 10,869
Local Revenue: Tuition Fees from Pupils and Parents 4 1 11 9 7 6 8 7 10,869
Local Revenue: Transportation Fees from Pupils and Parents 0 2 0 0 2 0 0 1 10,869
Local Revenue: Textbook Sales and Rentals 1 1 0 0 0 0 0 0 10,869
Local Revenue: District Activity Receipts 157 174 33 0 89 0 179 84 10,869
Local Revenue: Other Sales and Services 35 24 10 10 5 30 0 16 10,869
Local Revenue: Student Fees, Nonspecified 0 12 0 0 0 0 0 1 10,869
CE-Support Services-Operation and Maintenance of Plant 470 681 483 638 483 520 561 525 10,869
CE-Support Services-Student Transportation 248 301 262 372 228 221 208 255 10,869
CE-Support Services-Nonspecified 13 0 0 0 0 0 0 2 10,869
CE-Enterprise Operations 0 0 9 0 1 0 22 5 10,869
CE- ELSEC 4 0 8 5 29 0 13 10 10,869
Non-Elementary/Secondary Expenditures: Community
Services 39 46 12 9 4 45 45 26 10,869
Non-Elementary/Secondary Expenditures: Adult Education 16 102 8 21 5 1 51 22 10,869
Non-Elementary/Secondary Expenditures: Other 72 1 0 0 12 0 0 14 10,869
Capital Outlay: Instructional Equipment 41 4 65 78 70 68 67 59 10,869
Capital Outlay: Other Equipment 73 250 89 73 122 75 128 107 10,869
Capital Outlay: Nonspecified Equipment 0 0 23 8 14 0 36 12 10,869
Salaries-Support Services-Pupils 157 211 209 173 150 233 223 191 10,869
Salaries-Support Services-Instructional Staff 132 235 199 218 134 157 254 181 10,869
Salaries-Support Services-General Administration 104 59 93 49 119 107 50 91 10,869
Salaries-Support Services-School Administration 258 281 292 251 216 319 290 271 10,869
Salaries-Support Services-Operation and Maintenance of Plant 137 229 157 176 115 192 175 160 10,869
Salaries-Support Services-Student Transportation 131 164 151 195 111 135 100 138 10,869
Salaries-Food Services 157 113 146 179 112 147 126 140 10,869
Total Employee Benefits 1,030 1,142 1,188 1,154 879 971 1,100 1,057 10,869
Employee Benefits: Instruction 688 675 816 761 583 661 710 702 10,869
Employee Benefits-Support Services-Pupils 43 60 49 44 36 51 57 47 10,869
Employee Benefits-Support Services-Instructional Staff 35 62 60 54 33 35 65 47 10,869
Employee Benefits-Support Services-General Administration 24 22 28 23 29 25 21 25 10,869











Table A5. Continued


Employee Benefits-Support Services-School Administration
Employee Benefits-Support Services-Operation and
Maintenance of Plant
Employee Benefits-Support Services-Student Transportation
Employee Benefits: Food Service
EmDlovee Benefits: Enterprise ODerations


70 78 90 71 52 73 74 73 10,869

51 81 41 47 39 45 55 48 10,869
59 63 41 65 40 22 32 44 10,869
66 45 33 64 46 42 44 47 10,869
0 0 0 0 0 0 1 0 10.869


Total Students 5,699 34,988 7,560 11,187 3,353 10,389 7,894 9,567 11,649
NOTE: Columns show per-capita values for different categories of expenditure and revenues in 2000 dollars. Values have been deflated using the CPI-U.
Columns 1-7 show state-by-state means. Column 8 is the mean value for the entire sample. Column 9 shows the number of observations available for each
variable. There are 11,649 district-level observations across the 15 years of data in the sample. Variables with 10,869 observations lack data from 1990. The
three Asset variables are missing observations starting in 1999 which results in the lower number of observations for those three variables.


122









LIST OF REFERENCES


Ampadu, Osei, and Lei Zhou. 2009. Documentation for the NCES Common Core of
Data School District Finance Survey (F-33), School Year 2006-07 (Fiscal Year
2007) (NCES 2009-339). Washington D.C.: U.S. Department of Education.
Retrieved June 30, 2010 from
http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2009339.

Chemerinsky, Erwin. 2009. "The Segregation and Resegregation of American Public
Education" in School Resegregation, Must the South Turn Back. eds. John C.
Boger and Gary Orfield. Chapel Hill: The University of North Carolina Press.

Clotfelter, Charles T. 2004. After Brown: The Rise and Retreat of School
Desegregation. Princeton: Princeton University Press.

Clotfelter, Charles T., Helen F. Ladd and, Jacob L. Vigdor. 2005. "Federal Oversight,
Local Control and the Specter of Resegregation in Southern Schools." NBER
Working Paper Series, W11086.

Dries, Bill. "County Schools Move on After Desegregation Case." The Memphis Daily
News. May 26, 2009.

Feng, Li, David N. Figlio and Tim Sass. 2010. "School Accountability and Teacher
Mobility." NBER Working Paper Series, W16070.

Florida State Advisory Committee to the United States Commission on Civil Rights.
2007. Desegregation of Public School Districts in Florida: 18 Public School
Districts Have Unitary Status 16 Districts Remain Under Court Jurisdiction. A
Report to the United States Commission on Civil Rights. Washington D.C.

Guryan, Jonathan. 2004. "Desegregation and Black Dropout Rates." American
Economic Review. 94(4): 919-943.

Holley-Walker, Danielle. 2010. "After Unitary Status: Examining Voluntary Integration
Strategies for Southern School Districts." North Carolina Law Review, 88(3): 877-
910.

Hughes, Bayne. "The $814,187 Question." The Decatur Daily News. July 18, 2005.

Janssen, Julie. 2001. "An Analysis of the Legal and Historical Context of the Pinellas
County School District's Separation from Court Ordered Desegregation
Established in Bradley v. Board of Public Instruction." Unpublished Ph.D.
Dissertation, University of South Florida, Tampa, FL.

Lutz, Byron F. 2005. "Post Brown vs. the Board of Education: The Effects of the End of
Court-Ordered Desegregation." Finance and Economics Discussion Series,
Federal Reserve Board.


123









Martin, Stephanie M. 2010. "Are Public School Teacher Salaries Paid Compensating
Wage Differentials for Student Racial and Ethnic Characteristics?" Education
Economics. Forthcoming.

Moore, Monika L. 2002. "Unclear Standards Create an Unclear Future: Developing a
Better Definition of Unitary Status." The Yale Law Journal, 112(2): 311-351.

Orfield, Gary. 1996. "Turning Back to Segregation." In Dismantling Desegregation: The
Quiet Reversal of Brown v. Board of Education, eds Gary Orfield and Susan E.
Eaton. New York: The New Press.

Pinellas County School Board, Annual Budget Summary for 2000-2001. Last accessed
at http://www.pcsb.org/budget/0001.html on May 6, 2010.

Reber, Sarah J. 2007. "School Desegregation and Educational Attainment for Blacks"
NBER Working Paper Series, W13193.

Rivkin, Steven G. 2000. "School Desegregation, Academic Attainment, and Earnings."
The Journal of Human Resources, 35(2): 333-346.

Ryan, James E. 1999. "Schools, Race and Money." The Yale Law Journal, 109(2): 249-
316.

Schofield, Janet W. 2005 "Review of Research on School Desegregation's Impact on
Elementary and Secondary School Students." In Handbook of Research on
Multicultural Education, eds. James A. Banks and Cherry A. McGee, San
Francisco: John Wiley and Sons.

Spinner, Jackie. "Prince George's County to Pay Legal Costs for NAACP as Part of
Settlement." The Washington Post. March 9, 1999.

United States Commission on Civil Rights. 2007. Becoming Less Separate? School
Desegregation, Justice Department Enforcement and the Pursuit of Unitary Status.
Washington D.C.


124











BIOGRAPHICAL SKETCH

Colin Knapp was born and raised in St. Petersburg, Florida. After receiving

degrees in economics and finance from Ohio University, he served honorably for 11

years in the United States Air Force in various academic and financial management

positions. Upon leaving the military, he returned to the University of Florida to complete

his doctorate. He is married, has one daughter, and considers himself very lucky on

both accounts.


125





PAGE 1

1 THE EFFECTS OF UNITARY STATUS ON ACADEMIC ACHIEVEMENT AND SCHOOL FINANCE: ESSAYS ON THE END OF THE COURT ORDERED DESEGREGATION ERA By COLIN A. KNAPP A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Colin A. Knapp

PAGE 3

3 To my wife and daughter, beautiful in so many ways

PAGE 4

4 ACKNOWLEDGMENTS This dissertation would not have been possible without the love and support of my entire family. From the parents who provided a worldclass education, to the wife that puts food on the table, I have received more than I deserve. Like most, I am not as appreciative as I should be and I will try to change. I am also in debt to the most helpful dissertation chair any graduate student could ever have. Professor Larry Kenny gently nudged me back towards completion whenever I would stray too far off course. There are too many other teachers, professors, classmates, family members, and friends who have mentored me through 11 years in the military, 24 years of formal education, and 36 years of life to include them all I am sorry for that because they have earned it.

PAGE 5

5 TABLE OF CONTENTS ACKNOWLEDGMENTS .................................................................................................. 4 page LIST O F TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATIONS ........................................................................................... 10 ABSTRACT ................................................................................................................... 11 1 INTRODUCTION .................................................................................................... 13 2 UNITARY STATUS AND ACADEMIC ACHIEVMENT IN FLORIDA ....................... 16 A Brief History of Court Ordered Desegregation ..................................................... 16 The Existing Literature on the Effects of Unitary Status .......................................... 17 Description of the Data ........................................................................................... 20 Description of the Empirical Models ........................................................................ 26 Results of the Empirical Models .............................................................................. 29 Unitary Status Effect on Segregation ............................................................... 2 9 The Effect of Unitary Status over Time ............................................................. 30 The Use of Weighted vs. Unweighted Observations ........................................ 32 The Exogeneity of Unitary Status ..................................................................... 33 The Effect of Unitary Status on Academic Achievement ......................................... 35 The Effect on the Average Student .................................................................. 35 The Effect by Race ........................................................................................... 36 Conclusions and Further Discussion ...................................................................... 37 3 SCHOOL DISTRICT FINANCE AND UNITARY STATUS IN FLORIDA ................. 49 Introduction ............................................................................................................. 49 Floridas Desegregation History and How It Relates to District Size ....................... 52 Descriptio n of the Data ........................................................................................... 55 Description of the Empirical Models ........................................................................ 58 The Effect of Unitary Status on School District Finance .......................................... 62 Unitary Status and Revenues ........................................................................... 62 Unitary Status and Expenditures ...................................................................... 66 The Role of Economies of Scale ............................................................................. 68 4 A MULTI STATE ANALYSIS OF UNITARY STATUS AND SCHOOL FINANCE ... 79 Description of the Data ........................................................................................... 79 Description of the Models ....................................................................................... 83 The Effect of Unitar y Status on Revenues .............................................................. 85

PAGE 6

6 The Effect of Unitary Status on Expenditures ......................................................... 86 Revisiting Economies of Scale and Final Thoughts ................................................ 90 5 CONCLUSION ...................................................................................................... 111 APPENDIX A ADDITIONAL TABLES .......................................................................................... 115 LIST OF REFERENCES ............................................................................................. 123 BIOGRAPHICAL SKETCH .......................................................................................... 125

PAGE 7

7 LIST OF TABLES Table page 2 1 District level variable means ............................................................................... 39 2 2 Segregation index regres sions ........................................................................... 40 2 3 Effect of selective migration ................................................................................ 41 2 4 Unitary status effect on math achievement ........................................................ 42 2 5 Unitary status effect on reading achievement .................................................... 43 2 6 Unitary status effect on achievement by race and subject ................................. 44 3 1 District level revenues per student ..................................................................... 72 3 2 District l evel expenditures per student ................................................................ 73 3 3 Effect of unitary status on revenues ................................................................... 74 3 4 Effect of unitary status on expenditures .............................................................. 75 4 1 Court ordered desegregation status by state ..................................................... 92 4 2 Summary fiscal data by state. ............................................................................. 93 4 3 The effect of unitary status on total revenues and total expenditures per student. ............................................................................................................... 94 4 4 The effect of unitary status on federal revenues per student. ............................. 95 4 5 The effect of unitary status on state revenues per student. ................................ 96 4 6 The effect of unitary stat us on local revenues per student. ................................ 97 4 7 The effect of unitary status on total current expenditures (TCE) per student. ..... 98 4 8 The effect of unitary status on salaries ............................................................. 100 4 9 The effect of unitary status on benefits ............................................................. 101 4 10 The effect of unitary status on teachers ............................................................ 102 A 1 Litigation status by district ................................................................................ 115 A 2 Unitary status effect on total local, state and federal revenue per student ....... 116

PAGE 8

8 A 3 Unitary status effect on state funding from other sources ................................ 117 A 4 Unitary status effect on general administration expenses ................................ 118 A 5 Fiscal variable means (all variables). ................................................................ 119

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Floridas elementary school enrollment by race .................................................. 45 2 2 Average segregation index by unitary status, 19872006. .................................. 46 2 3 Math achievement by unitary status ................................................................... 47 2 4 Reading achievement by unitary status .............................................................. 48 3 1 Floridas school desegregation over cases over time ......................................... 77 3 2 Floridas unitary status receipts over time .......................................................... 78 4 1 Timing of unitary status grants for eligible districts (all states). ......................... 103 4 2 Timing of unitary status grants fo r eligible districts by size: Alabama. .............. 104 4 3 Timing of unitary status grants for eligible districts by size: Florida. ................. 105 4 4 Timing of unitary status grants for eligible districts by size: Georgia. ............... 106 4 5 Timing of unitary status grants for eligible districts by size: Louisiana. ............. 107 4 6 Timing of unitary status grants for eligible districts by size: Mississippi. ........... 108 4 7 Timing of unitary status grants for eligible districts by size: North Carolina. ..... 109 4 8 Timing of unitary status grants for eligible districts by size: South Carolina. .... 110

PAGE 10

10 LIST OF ABBREVIATIONS CPI U Consumer Price Index for All Urban Consumers CLV Refers to a group of authors : Charles Clotfelter, Helen Ladd, and Jacob Vigdor. DOE Department of Education FCAT Florida Comprehensive Assessment Test FEFP Florida Education Finance Program FRPL Free or Reduced Price Lunch FSIR Florida School Indicator Report FTE Fulltime Equivalent LSF Local, State, and Federal NCES National Center for Education Statistics SCOTUS Supreme Court of the United States SPAR School Public Accountability Report TCE Total Current Expenditures

PAGE 11

11 Abstract of Dissertation Presented to the Graduate School of the University of Flor ida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECTS OF UNITARY STATUS ON ACADEMIC ACHIEVEMENT AND SCHOOL FINANCE: THREE ESSAYS ON THE END OF THE COURT ORDERED DESEGREGATION ERA By Colin A. Knapp August 2010 Chair: Lawrence Kenny Major: Economics Over the span of three decades, many southern school districts were found to be operating racially separate school systems and were placed under court supervised desegregation plans. While under these plans, districts were forced to remove all traces of their formerly segregated nature. Upon satisfactorily achieving the U.S. Supreme Courts goal of one unitary system for all children, a district can be released from court supervision and can su bsequently be considered to have received unitary status. This study identifies how exiting a court ordered desegregation plan via the granting of unitary status affects racial imbalance and academic achievement. It finds significant evidence that the level of segregation will rise if unitary status is received. This effect is robust and appears to grow over time. Unitary status is also shown to have little effect on the achievement of the average student within each district. This result holds for bot h white and black students at the district level, suggesting that the end of court supervision may not lead to an increase in academic inequality.

PAGE 12

12 Additional consideration is given to the motives behind the decision to seek unitary status During the most active phase of the school desegregation era, the U.S. court system took control of student assignment away from districts found guilty of operating segregated school systems. This affected both large and small districts. However, as the courts retreated from active involvement, large districts were more frequently released than small ones. One explanation may be economies of scale. If the per capita financial benefits of receiving unitary status are small, a large number of students are required to cov er any nontrivial costs associated with seeking release from the courts desegregation order. This study shows some revenue shifting may occur, saving local tax payers about $100 per student in Florida Since there is no clear indication that any sizable expenditures are forgone, the net effects are small at best. Therefore, economies of scale may be involved in the decision to seek unitary status, explaining why large districts receive unitary status more frequently than small districts.

PAGE 13

13 CHAPTER 1 INTRODUCTION The legacy of segregated schooling is a burden that many states in the traditional south continue to bear. Whether it is in the form of residential segregation, continued lawsuits fighting for educational equality, or the winding down of the court ordered desegregation era, few parts of the S outh have avoided being touched in some form. One phenomenon that many states in the area share in common is the move to end the role the court system play s in the assignment of students. How individual districts do this is straightforward. They must be granted unitary status, which is a declaration by the court that the previously segregated school district has removed all v estiges of the formerly segregated system and replaced it with one system for all students Beyond that simple statement, lit tle else is so clear. Many m ajor questions such as why districts seek unitary status or how academic outcomes are affected by unitary status have gone largely unasked in the economic literature even though their answer s will play an important role in the continuing policy debate. The purpose of this study is to help shed light on what happens when districts receive unitary status. One prominent role economists perform is t o identif y both the intended and unintended consequences of different polic ies like the granting of unitary status If the courts involvement in student assignment via a court ordered desegregation plan is a constraint on the local district, r emoval of the order should produce a number of identifiable c onsequences T he evaluation of unitary status as a policy outcome requires measuring outcomes across many dimensions to ensure that negative, and often unintended, outcomes do not outweigh the perceived benefits.

PAGE 14

14 Each of the following chapters plays a part in identifying what happens when a district is granted unitary status. Chapter 2 looks at how students are sorted between schools before and after unitary status is received and shows that a significant rise in segregation may occur as a result of receiving unitary status It is not immediately clear how this outcome should be interpreted. If the court system has successfully eliminated the biases that created the educational inequity of th e 1950s and 1960s then such a re sort ing should have no effect on the students within the re sort ed schools. However, if some elements of the formerly segregated school system do remain, then students could be harmed, especially those that gained so much from the courts initial involvement. Chapter 2 continues by examining academic achievement dur ing the time that most of the recent unitary grants were received. The results suggest that there are no difference s in the academic performance based on unitary status. This applies to the average student, the average black student, and the average whit e student at the district level. This result suggests that the re sort ing of students into more segregated schools does not hurt students along this one dimension. While Chapter 2 deals primarily with what happens when districts are granted unitary status Chapter s 3 and 4 begin to study why districts seek unitary status. If districts are presumed to act rationally, they will only seek unitary status if they believe it will make them better off than their current status under the courts desegregation pla n. If benefits turnout to be nonexistent or relatively small when compared to the cost of having the court order removed, then the decision to seek unitary status may be questionable. One place districts may look to make improvements is school district

PAGE 15

15 finance. The analysis in these chapters reveals that school district finance should not play a major role in a districts decision to seek unitary status. The granting of unitary status is shown to have very little effect on the revenues received and the expenditures spent by districts that receive it. Since changes in district level financing do not provide the evidence required to declare unitary status a beneficial public policy, are we left to assume that the policy is largely ineffective at best and detrimental at worst? No, that would be too extreme. Chapter 5 concludes by examining the collective body of results from the study, identifying limitations based on the available data, and discussing avenues for future research.

PAGE 16

16 CHAPTER 2 U NITARY STATUS AND AC ADEMIC ACHIEVMENT IN FLORIDA A Brief History of Court Ordered Desegregation Brown v. Board of Education did not have an immediate effect on integration in southern school districts. It took almost another decade and passage of the Civi l Rights Act in 1964 to strengthen the Supreme Courts demands to expedite desegregation. In many cases, even financial incentives associated with Title I funding from the U.S. Department of Education were not enough to get some districts to desegregate. In those instances, individual court cases were filed in various U.S. District Courts to force school integration. Following the courts decision, a plan was typically designed to racially balance schools within districts by defining strict rules by whic h students must be assigned. These plans were effective as the traditional South went from having the most segregated schools to having the least segregated in about one decade.1 However, due to the strictness of some court ordered plans, many districts sought autonomy, arguing that equal education could be provided without the courts supervision. Eventually, the U.S. Supreme Court agreed and in 1968 set up a process by which districts could be released from court supervision and subsequently allowed to assign students to schools by means other than race. This process became known as unitary status.2 1 See Clotfelter (2004), 25 30. 2 For a discussion of all the steps and court cases involved, see Orfield (1996). Jansen (2001) provides a near first hand account of one districts attempt to satisfy (or skirt in the early stages of desegregation) the issues involved in many of the relevant cases.

PAGE 17

17 As defined in Green v. County School Board of New Kent County, a district must eliminate all vestiges of segregation before it is deemed to be operating a unitary school system for both black and white students. By evaluating schools on the basis of six factors, courts determined whether districts had taken the appropriate steps to be released from a court ordered desegregation plan. These fact ors are: student assignment, faculty assignment, staff assignment, transportation, extracurricular activities, and physical facilities.3 Two additional cases are important in connecting the fate of unitary status and the possibilities of resegregation in those districts where it was granted. In Board of Education of Oklahoma v. Dowell and Freeman v. Pitts, the U.S. Supreme Court decided that districts did not have to maintain racially integrated schools after being declared unitary and that unitary status could not be withheld if de facto segregation would result due to racial segregation in neighborhood residential patterns. 4The Existing Literature on the Effects of Unitary Status Due to the presence of residential segregation in many areas, the likelihood of resegregation would seem to be the rule rather than an accidental confluence of factors The literature on unitary status necessarily begins in the area of desegregation. From the 1970s to the 1990s, there are hundreds of studies discussing deseg regation and its effects. Schofield (2005) reviews over 200 of the earliest studies and draws several conclusions. First, white students have been held harmless by school desegregation. Second, there is evidence that black students benefit from 3 391 U.S. 430 (1968) 4 498 U.S. 237 (1991) and 503 U.S. 467 (1992), respectively.

PAGE 18

18 desegregation more in reading than they do in math. This would seem to suggest the presence of unambiguous gains which could be lost. Therefore, measuring the effect unitary status may have on reducing these gains becomes a very important area for analysis. Mor e recent studies on desegregation have sought to not only measure the effect desegregation has on economic outcomes, but also the mechanisms by which desegregation has worked. Two events happened simultaneously during school desegregation: the racial composition of schools changed and the resources available to black students increased. Reber (2007) finds that black students in high black enrollment districts experienced larger improvements in educational attainment than black students in high white enrol lment districts. She attributes this to the differing ability of school districts to redirect funding prior to desegregation and suggests that the increases in school funding that accompanied desegregation were more important than the increase in exposure of blacks to whites. Rivkin (2000) investigates the impact of racial composition and school quality on twelfthgrade test scores, years of educational attainment and monthly earnings. His study finds that raising school quality is more effective than relocating children and trying to take advantage of possible peer effects. Guryan (2004) finds that dropout rates decreased in districts that desegregated in the 1970s, relative to districts that did not, and that this is true for only black students. Two recent articles make the jump from desegregation to measuring the effects of unitary status and possible resegregation. The first, Clotfelter, Ladd, and Vigdor (2005, CLV hereafter), analyzes data on the 100 largest districts in the traditional South and former Confederate border states from 1994 2004. CLV argue that increased

PAGE 19

19 racial isolation is not a result of granting of unitary status. Their results show that although segregation increases over time, recent unitary status declarations are not assoc iated with these higher levels of segregation. They attribute increases in racial isolation to increases in the Hispanic school population rather than changes in unitary status. CLV conclude by stating that increases in the Hispanic population may be mak ing racial isolation measures irrelevant as far as policy analysis is concerned and recommend using measures of racial imbalance since district officials cannot control the demographic changes that cause racial isolation data to vary widely.5 The second paper in this branch of the literature is Lutz (2005). He uses a nationally representative sample of districts to measure changes in segregation after districts are released from their court ordered plans. He finds immediate and gradual increases in segregation which strengthen CLVs less definitive results. He then ties these changes to increases in both black dropout rates and private school attendance, but finds these effects to be statistically significant only in northern states. These results are presented as evidence that the value to black students of attaining a public school education falls when unitary status has been granted. However, there is no clear explanation as to why this theory should hold in the North but not in the traditional South. CLV and Lutz demonstrate that the literature lacks a consistent answer concerning unitary status effect on segregation. Additionally, there is no paper that directly links unitary status and academic achievement. My research addresses both of these i ssues. 5 The differences between racial isolation and racial imbalance are explained later in the chapter.

PAGE 20

20 Description of the Data To measure the effect of unitary status on segregation and academic achievement, I focus on elementary school data collected at the dist rict level in the state of Florida from 1999 to 2007. With the exception of the litigati on status data, all data is publically available from various Florida Department of Education sources. Floridas elementary schools make a good research subject for many reasons. Florida was one of the first states to implement a school accountability pl an that required the collection of test scores. This makes year to year academic achievement results available further back in time than in many other states. Florida also has a long history with school integration that provides a vibrant source of variation. Data on the desegregation status of each district was obtained from the Florida Advisory Committees 2007 Report to the U.S. Commission on Civil Rights Report on Desegregation in Public School Districts in Florida. This contains detailed information on each school district and lists whether each district has ever been to court, when it was taken to court and when the district was granted unitary status if applicable. A detailed list of each districts desegregation status is available in the Append ix, Table A1. From 1956 to 1978, 34 of Floridas 67 school districts were taken to court and found in violation of Brown This is significant as it means 84 percent of Floridas schools have been affected by litigation at some point in time. Of those di stricts taken to court, 18 have been granted unitary status The first release from court order came in 1970 and the most recent was in 2006. Eight of the 18 releases occur during the study period. From 1999 to 2007, the percentage of schools affected by unitary status increased from 29 percent to 66 percent of all schools in the state. This means over

PAGE 21

21 onethird of all schools in Florida experienced a change in unitary status during the period of this study. I use this variation in the timing of receipt of unitary status to identify the effect unitary status has on segregation rates and the student test score distribution.6The litigation data also provide the justification for how the treatment and control groups are determined. The receipt of unitary status is conditional on having been taken to court. There may be differences between those districts that are taken to court and those that avoided litigation. Therefore, when considering which districts to include in the control group, having been taken to court is an important determinant for inclusion It is also important to consider that districts can only be treated once to the policy. Since some districts were granted unitary status prior to the start of the study period, including them in either the treatment or control group is inadvisable because they cannot be treated again. When put together, this will leave a preferred sample of 24 districts that were eligible for treatment as of 1999. Although res ults are reported for various sample configurations throughout the rest of the chapter, more emphasis should be given to the results which focus on the preferred sample since it is the one most likely to measure the treatment effect of receiving unitary st atus. Racial composition data comes from Floridas School Public Accountability Reports (SPAR) and is based on attendance in October of each school year. Figure 11 lays out the racial composition trends for the state by year. Whites are still the dominant 6 It should al so be noted that alt hough never under court order, the remaining 33 school districts are monitored by the U.S. Department of Educations Office for Civil Rights. In some ways they can be considered as also having unitary status. This needs to be consider ed when defining the relevant treatment and control groups and will be discussed later in the chapter.

PAGE 22

22 racial group, but that dominance is decreasing as the percent of students who are white decreased from 52 percent in 1999 to 45 percent in 2006.7 Monitoring desegregation and measuring its effects requires variables which express the extent to which different racial groups interact. These variables are divided into two groups: those that measure racial isolation and those that measure racial imbalance. Variables such as black white exposure rates and the percent of black students attending schools with 90 percent nonwhite enrollments measure the extent to which blacks are isol ated from white students. Dissimilarity and segregation indexes measure imbalance within a district by taking into account a given districts racial composition. They generally provide a ratio of how racial groups are distributed across schools compared to some ideal level of integration which usually corresponds to a subgroups share of the district population. The percentage of black students has decreased slightly from 25 percent to 23 percent. These decreases i n enrollment share are not due to fewer whites and blacks attending Floridas schools. Instead, these shifts are due to the increased number of Hispanics and other racial minorities. Figure 1 1 clearly shows that increases in Floridas elementary school population are mainly within these latter demographic groups. A key difference between these two types of measures is that racial imbalance variables are much better at identifying policy effects between di stricts than those measuring racial isolation because measures of racial isolation pick up differences in district level demographics absent any relevant difference in policy. This tendency can 7 In this case, 1999 and 2006 refer to the 1999 2000 and 20062007 school years respectively. This convention will be used throughout the rest of the chapter. Any deviation from this convention will be clearly stated, especially when it is necessary to def ine fiscal years later in the study.

PAGE 23

23 skew the analysis of district level desegregation efforts. A comparison of two school districts in Florida illustrates this point. Gadsden County s schools have a student population that is nearly 79 percent black while schools in Brevard County are only 14 percent black.8 The corresponding exposure rates of blac ks to other blacks are 83 percent and 24 percent respectively. A comparison across districts based on only the two exposure rates may lead some to conclude that Gadsden Countys schools are much more segregated than Brevard Countys schools. However, that is not the case. Once the racial composition differences are accounted for, the segregation indexes reveal a different story. Segregation in each is a relatively low 0.17, indicating that students are distributed fairly evenly between all schools in the district given the racial composition within the district .9 It is still necessary to calculate the exposure rate since it is a component of the segregation index. The equation for the exposure rate of blacks to whites in any given year can be stated as: Therefore, the difference in exposure rates is an artifact of the demographics rather than a result of differing policies since both counties were once under court order desegregation plans and subsequently released via the granting of unitary status. Based on this reasoning, I focus more on racial imbalance than I do racial isolation for the remainder of the study (*)ii i k i i Bw ExposureRate B ( 2 1) 8 With few exceptions, school districts in Florida are coterminous with county borders. Most of these exceptions are for experimental schools assigned their own district or juvenile delinquent programs, all of which have been excluded from this study. 9 All statistics based on calculations using SPAR data for 2004. The formulas used to calculate these statistics are discussed later in this section.

PAGE 24

24 In this case, k indexes the district, iB represents the total number of blacks in school i and iw is the percentage of the students in school i who are white. In Table 2 1, the exposure rate of blacks to whites can be interpreted as meaning that from 19992006 the average black student in Florida attended a s chool that was 55 percent white. The exposure rate is then used to calculate a segregation index of the following form: ( )/kk kkSegregationIndexwExposureRatew ( 2 2) Again, k indexes each district and kw represents the percent of students in district k who are white. This index can take on a value ranging from zero (perfect integration) to one (co mplete segregation). From 1999 to 2006, the average segregation index in Florida increased from 0.14 to 0.16. This appears to show very little change over time given the number of schools exposed to unitary status changes and is hardly near the U.S. Depa rtment of Educations acceptable upper threshold of 0.60. However, a closer look shows significant variation between district types. Column 2 of Table 2 1 shows that the average level of segregation is higher in the 34 districts that have been taken to court than in the state as a whole. Additionally, the districts that have been released from court order have an average segregation rate that is more than twice that of the districts still under court supervision. How much of this difference is due to additional districts being granted unitary status during the time period 19992006? Due to the relatively short sample period that can be constructed using publically available data from the Florida DOE, it is not possible to look at prior trends in the lev el of segregation before 1999. However, it is possible using qualitatively similar data from the U.S. DOEs Common Core of Data

PAGE 25

25 which goes back as far as 1987. Preliminary evidence is provided in Figure 2 2. From this graph, it is evident that the average level of segregation began increasing in the early 1990s. However, this increase was not uniform across different district types. The only districts not to see an increase in the level of segregation are those that remained under court supervision. Those districts released during the sample period experience first a decrease and then an increase in segregation as more districts are released from their court orders. Those districts never taken to court and those released prior to the beginning of the sample period also experience an increase in the level of segregation. These trends suggest a positive relationship between unitary status and the level of segregation. Additionally, the divergent path taken by districts released from court order suggest s that a variable mapping the receipt of unitary status over time will i dentify more than just the continuation of preexisting trends. It appears that the court orders did provide some constraint on assignment policies and that when removed, students wil l re sort into an increasingly segregated distribution. All other district level variables were obtained from the Florida School Indicator Reports (FSIR) from each district. Due to the small sample size, only a limited set of variables are used to control for various district characteristics. The fraction of students using the districts free lunch program is meant to capture the socioeconomic characteristics of the school population, while the fractions of students with limited English proficiency, disa bilities and those considered gifted are meant as proxies of the student bodys ability.10 10 There is variation across districts in what qualifies students for inclusion in the gifted program. This may limit the variables effectiveness as a proxy for ability. However, the existence of these differences highlights the need to control for this source of variation in districts. Absences, suspensions, and average teacher experience

PAGE 26

26 proxy for school quality. Total district size and per capita real income in each district account for potential differences in parental inputs across districts. The remaining variables in Table 2 1 are the dependent variables used to estimate whether the receipt of unitary status affects academic achievement. These variables measure the amount of mass in the tails of the student testing distribution. The FSIR lists the fraction of fourth graders and fifth graders who test into one of five levels on the reading and math portions of the Florida Comprehensive Assessment Test (FCAT). I combine the lowest and highest two levels in order to measure movement into and out of the tails of the distribution caused by the policy change. From 1999 2006, there was a rightward shift in the testing distribution with a lower percentage of students testing into Levels 1 or 2 and a higher percentage testing into Levels 4 or 5. This is apparent in Figures 2 3 and 2 4 It should also be noted that in all four cases, the trend for those districts currently under court order moves in tandem with the trend for those districts already released from court order. This suggests that achievement gains may have occurred for reasons other than the granting of unitary status. Description of the Empirical Models I use two main specifications in this chapter The first tests to what extent changes in the segregation index are caused by the granting of unitary status. This is a more formal statement of the rel ationship illustrated in Figure 1 2. It takes the form: 0123 ktktkt ktktSegregationIndexAYXunitarystatusu ( 2 3) This measures the segregation index in district k during year t as a function of a county level fixed effect( kA ), a year fixed effect (tY), district level controls varying by year ( Xkt), a dummy variable indicating whether district k was granted unitary status in

PAGE 27

27 year t and a random error term.11ktstatus unitary To control any withindistrict correlation of the standard errors, I cluster the standard errors at the district level. For the coefficient on to be properly identified the following condition must hold, unitary status must be received randomly. Whether it is reasonable to assume that this condition holds is discussed in the next section along with the results and specification tests. In its most robust form, the matrix of district level variables ( Xkt) contains the percent of students in district k who are Hispanic, the districts total elementary school population and a county level measure of real per capita personal income derived from the Bureau of Economic Analysis Local Annual Personal Income Data Series. The first variable is included to account for demographic variation within each county. The last two variables are meant to capture the level of sorting which takes place within county. Based on Tiebouts hypothesis, I expect larger counties to have higher segregation indexes because there will be greater opportunities for sorting between a larger number of schools. This phenomenon works through the correlation of district size, the number of schools and the presence of res idential segregation within most areas. The effect of income should also be positive as white parents, with more income and children in black majority schools, can opt for private education or can more easily move to areas of a county with schools with a racial mix they prefer. The second model looks at the distribution of academic achievement and how it is affected by the granting of unitary status. Its basic form is: 123 ktoktkt ktitYAYXunitarystatus ( 2 4 ) 11 I would prefer to use district yea r fixed effects. However, each district year cell contains only one observation, making a district year fixed effect impossible to use in this instance.

PAGE 28

28 In this specification, I use the mass in the tail s of the student testing distribution in both Math and Reading as the values for ktY. kA and tYare district and year fixed effects. ktX controls for time varying characteristics between districts. Unitary status is measured using a dummy variable which equals one if the district is deemed unitary in that year and zero otherwise. Standard errors are again clustered at the district level. Whether unitary status is received randomly is important to the identification of 3 and the applicability of this condition will be discussed in the next section. As in the previous specification, the matrix ktX accounts for district level variation in size, income and demographics. In addition to the three previous variables, parental inputs are accounted for by including the percent of students utilizing the districts Free or Reduced Price Lunch Program. Academic achiev ement is expected to fall as the percentage of poorer students in a district increases. Student inputs, and to the extent possible student ability, are accounted for by the percent gifted, percent with limited English proficiency and the level of absenteeism. More gifted students should increase achievement, while higher levels of limited English students and absenteeism should decrease it as less of the academic program can be absorbed. School inputs and the general school environment are represented by the average years of experience of the teachers and the percentage of students suspended during the year. I expect more experienced teachers to be more efficient at teaching which should increase achievement.12 12 It can be argued that higher test scores really do not represent an increase in student ability. Teacher experience exemplifies this phenomenon because it is impossible to determine if more experienced teachers are imparting more knowledge or are simply better at teaching to the test and making their students more effective test takers. More disruptive environments should reduce academic achievement.

PAGE 29

29 Results of the Empirical Models Unitary Status Effect on Segregation Up to this point, I have not addressed the issue of defining the appropriate treatment and control groups in this quasiexperimental policy analysis. The treatment group is easy to identify. It is the group of districts which experienced a change in unitary status during the eight year period starting in 1999 and continuing through 2006. This is a group of eight districts representing over onethird of the schools in Florida.13 Defining the relevant control group is more problematic since there are several possibilities. Three types of districts must be considered for inclusion in the control group; those never taken to court; those taken to court and released from court order prior to 1999; and those still under court order after 2006. To examine these differences, I use several different samples and report the results in Table 2 2. Column 1 includes all the other districts in Florida in the control group and su ggests that the granting of unitary status will have an immediate and statistically significant effect on the segregation level raising the index by 2.4 percentage points. This represents an increase of 15 percent at the sample mean of 0.15. The first alternative case to consider involves the difference between those taken to court and those that have not faced litigation. Most of the nonlitigants are located at the lower end of the population size distribution. If the effects of unitary status are so mehow correlated with unobserved characteristics associated with district size, the previously found positive and significant estimate may be inconsistent. 13 This group a nd their respective years of receiving unitary status are: Dade (2001), Duval (2001), Escambia (2004), Hillsborough (2001), Lee (1999) Pinellas (2000), Polk (2000) and Seminole (2006). See Table A1 in the Appendix for a detailed account of each districts unitary status history.

PAGE 30

30 Therefore, I exclude nonlitigants from the control group and rerun the analysis. These results ar e listed in Column 2 of Table 2 2. This restriction reduces the number of districts in the analysis to the 34 that have been taken to court. The effect of unitary status differs by only 0.1 percentage points and the parameter estimate is still significant at the five percent level. This continues to suggest that unitary status will increase the level of segregation. This is noteworthy considering that the already small sample size was reduced by half and the statistically significant result was maintained. Even though the result remained unchanged while restricting the definition of the control group to only those districts that have been taken to court, it is possible that the control group may still be improperly defined. In Column 2 of Table 2 2, the control group contained observations from 26 counties who had either already received unitary status prior to 1999 or were still awaiting adjudication after 2006. The noteworthy difference here is that the first group has already received unitary stat us and is unable to be treated again by the same policy. Therefore, it may be reasonable to exclude these counties from the control group as well. The results for this specification are listed in Column 3 and show that unitary status still has a positive and significant effect on the segregation index. When combined, the results provide robust evidence that the granting of unitary status may raise the level of segregation by roughly 15 percent within the average district. The Effect of Unitary Status over Time Prior to this point in the chapter I have assumed that the granting of unitary status causes a single shock to an affected district. This is unlikely and Pinellas Countys experience is a perfect example. The U.S. District Court granted Pinell as County unitary status in 2000. However, the agreed upon exit strategy did not alter the court -

PAGE 31

31 ordered desegregation plan until 2003. In other words, even though granted unitary status in 2000, Pinellas County continued to use the assignment plan dictated by the court ordered desegregation plan for two more years. The first year that significant amounts of re sort ing could occur due to unitary status was in 2003, not 2000. Therefore, it is misleading to assume that the treatment of receiving unitary s tatus is realized immediately. It is also unlikely that the parents who react to the decision to remove the court ordered plan do so immediately. It is likely that parents move children between schools for different reasons and at different paces. This results in the segregation index changing incrementally rather than instantaneously. This scenario is at odds with the onetime policy models estimated in the previous section. To allow the effect to vary over time, I alter the specification to include a set of dummy variables which captures the effect of unitary status over time. I continue this process as far as eight years post implementation and have a specification of the following: 8 0123 1 kttkkt ktktSegregationIndexAYXDurationu ( 2 5 ) The previous variable of inter est, ktstatus unitary, is subsumed by the duration dummies. The results shown in Column 4 of Table 2 2 reveal much more about how unitary status affects segregation over time. What becomes most apparent from this specification is the extent to whic h the effect grows over time. The immediate effect of granting unitary status is still statistically significant at the five percent level and has a magnitude of 1.7 percentage points. The coefficients on every duration variable are not only statistica lly significant, but also increase monotonically. By the eighth year

PAGE 32

32 following the policy change, the segregation index has increased cumulatively by 20.6 percentage points. Depending on the average district level mean used, this represents an increase i n segregation ranging from 71 to 137 percent. Also remarkable is the increase in the effect between years three and four. At that time, the cumulative effect more than doubles from 3.0 percentage points to nearly 6.9. This is indicative of the two scenarios discussed previously. First, it is unlikely that districts implement policy changes immediately. Secondly, it is also unlikely that parents can react simultaneously with the districts implementation of new attendance plans. Ultimately, the results in Column 4 suggest that unitary status sorts students into an increasingly less integrated educational environment and that this effect accelerates at the four year point. The Use of Weighted vs. Unweighted Observations All of the results discussed previously were observed using weighted least squares with the weighting based on total district size. This forces the regression to place extra weight on the observations from larger districts. I chose to do this due to the wide disparity in size across dis tricts in Florida. In the specification using all 67 districts, the difference between the largest and smallest is almost 175,000 students. With this wide a disparity, it appears unwise to allow the smallest district to have as much weight as the largest However, there is disagreement over the effect this procedure may have on the size of the standard errors and the statistical significance of the results. To counter this possibility, I also measured the effect of unitary status on segregation using a sample of districts with a smaller dispersion. This sample includes the ten largest districts in the state excluding Miami Dade.14 14 Miami Dade is the largest district in the state with nearly 175,000 elementary school students. The next closest district is Broward County which has over 50,000 fewer students. This shrinks the difference

PAGE 33

33 between the largest and smallest district by half. Although the size disparity is still large, it is much more likely that these districts are similar, making weighting each observation no longer necessary. The results of this specification are presented in Column 5 of Table 2 2. For each duration variable, the coefficient is similar in magnitude but measured less precisely. This may be due to the lack of weighting, but could also be due to the extremely small sample. It should be noted that the cumulative effect is still highly significant and still represents a near doubling of the level of segregation, even without the advantage of weighting individual observations. The Exogeneity of Unitary Status In order for unitary status to have a statistically identifiable effect on the level of segregation, the timing of release from court order must be exogenous. To that end, it should be noted that two actions must take place before a district can receive unitary status. The school district must decide to seek release from their court ordered desegregation plan and the court must then grant it. Any assessment of the exogeneity of unitary status must take into account both parts of the process. The first step certainly appears to be affected by district size.15 15 This phenomenon is the primary focus of Chapter 2. From 1999 to 2006, eight districts received unitary status. Of those grantees, seven are among the ten largest districts in the state while the smallest ranks thirteenth out of 67 districts. This relationship between district size and the decision to seek unitary status could confound the effect of unitary status on the segregation index if the process were solely up to the discretion of individual school districts. However, that is not entirely the case since the courts role appears to add back a significant amount of randomness into the timing of release.

PAGE 34

34 It has been noted that district courts changed their behavior towards desegregation cases in the early 1990s.16 Furthermore, if district size were causing bias in the effect of unitary status on the segregation index, changing the composition of the sampl e by using only small or large districts should change the magnitude of the coefficients on the unitary status variable. It is apparent from the results shown in Table 2 2 that this is not the case. As I decrease the sample size in C olumns 1, 2 and 3, it is predominantly smaller districts that are dropped from the sample. Even after this the coefficient on unitary status remains almost unchanged in magnitude. The same holds in C olumns 4 and 5 under a slightly different strategy. This is especially rem arkable given that column five is solely based on a sample representing 10 of the 11 largest districts. Again, it appears that the timing of release is quasirandom. Given the courts newfound desire to clear its docket of desegregation cases, it is hard to construct a scenario where a courts decision to release a district is dependent on district size. They were willing to release any district, regardless of size. Nonetheless, it is possible to check if this is the case by looking at the level of correlation between the year of release from court order and district size. For those districts released during the period of this study, the level of correlation is a relatively low 0.22. Therefore, it may be reasonable to assume that the actual timi ng of the unitary status grants is random enough to allow identification of an effect on the segregation index. Another possible issue is caused by the way the segregation index is calculated. Since it uses both the black and white student populations in its construction, the index 16 See Chemerinsky (2005) for a detailed discussion of this change.

PAGE 35

35 could be susceptible to an issue called selective migration, a phenomena where one race enters or exits a district based on the existence of, or lack of, a court ordered des egregation plan. Selective migration would be evident by divergent racial trends between those districts that have been granted unitary status and those that have not. To test whether unitary status causes selective migration, I run a set of regressions based on the following model, kt kt kt kt ktu White White status unitary size district status unitary A Population Student White or Black (2 6) This specification uses either the districts white or black student population as the dependent variable. On the right hand side, I use total district size, a dummy variable mapping the timing of unitary status and an interaction term between unitary status and a dummy variable equaling one for those observations where the dependent variable is the white student population. If the coefficient on this interaction term is significantly different f rom zero, selective migration may be an issue because unitary status affects white and black student population trends differently. The results of this regression are shown in Table 2 3. Since the coefficient on the interaction term is not statistically significant, it appears that selective migration is not an issue and the effect of unitary status on the segregation index is properly identified. The Effect of Unitary Status on Academic Achievement The Effect on the Average Student Since it appears that unitary status is affecting the way students sort into schools and increasing the level of segregation, it is important to identify any academic costs associated with these changes. To do this, I measure academic achievement as outlined in E quation 2 5. The results for Math are presented in Table 2 4 while the

PAGE 36

36 results for Reading are presented in Table 2 5. In each table, the left panel represents specifications testing the effect unitary status has on the highest two score categories of the FCAT.17 From the Math results in Table 2 4, there appears to be no significant evidence that the achievement distribution is affected by the receipt of unitary status. For each specification in the table, the coefficient measuring the effect of unitary status is too imprecisely measured to identify anything other than a zero effect. For Reading, the only result of any statistical significance is the final specification which suggests a possible decrease in the density at the low end of the achievement distribution of 1.5 percent. The right side presents the same specifications using the lower two categories. This procedure measures the density in the tails of the achievement distribution. If unitary status has a beneficial effect on achievement, I would expect to see positive coefficients in the first three columns and negative coefficients in the last three. Conversely, a shift towards the lower tail would require negative coefficients for specifications test ing the higher end of the achievement distribution and positive coefficients on those testing the lower end. The Effect by Race Based on the results of the previous section, it may seem reasonable to conclude that unitary status has no measurable effect on student achievement in the average affected district. Such a statement is certainly premature. The tests completed above only account for the net effects of all students within the average district. Since the tests above combine al l racial subgroups into one observation, it is possible that gains 17 There are five categories which students can test into on the FCAT. The dependent variable is either the sum of the percentage of students testing into the two highest categories or the two lowest categories.

PAGE 37

37 by whites are offsetting losses by blacks. Considering the original purpose of school desegregation reforms and the fears associated with the effects of resegregation, determining if this effect on achievement varies by race is possibly the most important policy outcome of the paper. Unfortunately, the data used for the achievement distribution is not broken out by race and cannot be used in that manner. However, the state does report t he percentage of students of each race that are considered proficient (at grade level) in both Math and Reading.18ktY This allows the analysis to be completed by race using equation 2 5 with now being the percentage of either black or white students who are considered proficient. The results for these tests are presented in Table 2 6. Unitary status does not appear to have a significant effect on either white or black proficiency in either subject. This extends the findings to include not only the average student, but also the average white and black student at the county level. It appears that neither is being harmed by the end of the court ordered desegregation via the granting of unitary status. Conclusions and Further Discussion The purpose of this chapter was to identify the effects that unitary status has on racial imbalance and academic outcomes. I find significant evidence that the level of segregation will rise if unitary status is received. This effect is robust to changes in the relevant control group and appears to grow over time. The majority of the effect appears to occur three to four years after unitary status is received, suggesting that it takes time not only for districts to implement new assignment plans, but for parents to 18 Unlike the achievement distribution data, the proficiency data is not collected at the elementary school level. It is a measure of all white or black students within a district r egardless of grade.

PAGE 38

38 react and move their children between schools if desired. Unitary status was also shown to have no effect on the academic achievement distribution. This finding holds for the average student, the average white student and the average black student in an affected district. These results are consistent with findings in the recent literature. Reber (2007) notes that black students in Louisiana benefitted more from the increased level of resources given to them as a result of desegregation rather than the increased interaction with white students. Rivkin (2000) also describes how black students benefitted more from increases in school quality than greater interaction with white peers. Both find that of the two changes brought by desegregation, resources appear to matter most. If their findings are correct, then resegregation caused by the end of the courts involvement should have a minimal effect on student outcomes. As this era ends, unitary status involves only one part of the process seen in both Reber and Rivkin. It allows peer groups to change while holding resources constant. I find that re sort ing does occur, as shown by increases in the level of segregation, and that the effects of that re sort ing on academic achieve ment are zero. This not only strengthens the validity of these previous studies but suggests that there are not any identifiable academic costs associated with the granting of unitary status.

PAGE 39

39 Table 21. District level variable m eans (1) (2) (3) (4) (5) Court Order Only Full Released From Currently Under Number of Districts in Sample 67 34 18 16 P < 0.05 Fraction White 0.64 0.58 0.51 0.64 Yes Fraction Black 0.19 0.24 0.29 0.29 Yes Fraction Hispanic 0.12 0.13 0.15 0.11 Yes Exposure Rate (Blacks to Whites) 0.55 0.48 0.37 0.57 Yes Segregation Index 0.15 0.19 0.28 0.13 Yes Fraction of Schools Ever Under Court Order 0.84 1.00 Fraction of Schools Granted Unitary Status 0.29 0.53 Fraction of Students with Free/Reduced Price Lunch 0.55 0.54 0.53 0.53 No Fraction of Students Considered Gifted 0.03 0.03 0.04 0.02 Yes Fraction of Students with Limited English Proficiency 0.05 0.05 0.07 0.04 Yes Teachers Average Years of Experience 13.25 13.37 13.24 13.49 No Fraction of Students Absent 21+ Days 0.08 0.07 0.07 0.07 No Fraction of Students Receiving In School Suspensions 0.03 0.03 0.02 0.04 Yes Fraction of Students Receiving Out of School Suspensions 0.03 0.03 0.03 0.03 No Fraction of Students Testing in Levels 4 and 5 (Math) 0.20 0.26 0.28 0.25 Yes Fraction of Students Testing in Levels 1 and 2 (Math) 0.56 0.49 0.48 0.51 Yes Fraction of Students Testing into Levels 4 and 5 (Reading) 0.22 0.30 0.31 0.29 No Fraction of Students Testing into Levels 1 and 2 (Reading) 0.50 0.38 0.37 0.39 No District Size 18,344 30,633 46,221 17,573 Yes Per Capita Real Income ($) 26,258 25,361 28,468 22,758 Yes Percent of Black Students Proficient in Reading 33.28 33.97 34.64 33.21 No Percent of Black Students Proficient in Math 34.49 34.71 35.70 33.6 Yes Percent of White Students Proficient in Reading 61.44 63.24 65.54 60.65 Yes Percent of White Students Proficient in Math 65.72 67.52 70.07 64.68 Yes NOTE: Column 1 shows district level means for all 67 school districts in Florida. Column 2 restricts the sample to the 34 districts ordered by federal courts to desegregate. Column 3 further restricts the sample to the 18 districts that have been released from the cour t's desegregation order (i.e. granted unitary status). Column 4 provides means for those districts ordered to desegregate and remaining under court supervision through 2007. Column 5 reflects the results of a two tailed test for equality of means at the five percent level.

PAGE 40

40 Table 22. Segregation i ndex r egressions (1) (2) (3) (4) (5) Unitary Status Granted 0.024 0.026 0.028 (2.75)** (2.65)* (2.54)* Fraction Hispanic (District) 0.081 0.189 0.097 0.304 0.661 (0.27) (0.41) (0.18) (0.62) (0.72) Total Student Population (District) 2.13*10 -6 3.01*10 -6 2.47*10 -6 3.39*10 -6 4.89*10 -6 (0.96) (1.29) (0.86) (2.03) (1.30) Per Capita Income (District) 7.43*10 -6 1.23*10 -5 1.88*10 -5 7.80*10 -6 1.92*10 -5 (2.12)* (1.84) (1.96) (1.08) (1.64) Duration 1 0.017 0.024 (2.48)* (1.47) Duration 2 0.024 0.023 (2.46)* (1.11) Duration 3 0.03 0.022 (2.27)* (0.99) Duration 4 0.069 0.07 (4.22)** (2.42)* Duration 5 0.079 0.088 (3.66)** (2.48)* Duration 6 0.096 0.096 (3.82)** (2.23) Duration 7 0.111 0.113 (2.82)** (2.15) Duration 8 0.206 0.206 (9.39)** (4.70)** Constant 0.059 0.061 0.462 0.134 0.285 (0.73) (0.41) (0.95) (0.55) (1.24) Observations 536 272 192 192 80 R squared 0.98 0.98 0.98 0.99 0.98 Weighted by District Size Yes Yes Yes Yes No NOTE: All models include districtlevel and year fixed effects. Models weighted by the total district size have errors clustered at the district-level. Absolute Value of t-statistics in parentheses represents significance at the five percent level and ** represents significance at the one percent level. Columns 1 uses the full sample of districts. Column 2 uses only those districts that were subject to court order. Columns 3 and 4 further reduce the sample by excluding those districts granted unitar y status prior to 1999. Column 5 uses only the 10 largest districts that were under court order and were eligible for treatment, excluding Miami Dade.

PAGE 41

41 Table 23. Effect of selective m igration Unitary Status Granted 1,798.77 (0.54) Unitary Status Granted*White 6,092.33 (0.82) White=1 4,107.74 (0.66) Total Student Population (District) 0.258 (4.78)** Constant 316.41 (0.10) Observations 384 R squared 0.52 NOTE: This model includes district level and year fixed effects for a sample made up of districts that were under court order and eligible for unitary status in 1999. Observations are weighted by total district size. The dependent variable is either the white or black student population in a each district. Absolute value of t statistics in parentheses, represent significance at the five percent level and ** represent significance at the one percent level.

PAGE 42

42 Table 24. Unitary status effect on math a chievement Levels 4 and 5 (High) Levels 1 and 2 (Low) (1) (2) (3) (4) (5) (6) Unitary Status Granted 0.003 0.001 0.002 0.014 0.011 0.008 (0.47) (0.15) (0.26) (1.29) (1.12) (1.03) Fraction White, District 0.044 0.156 0.202 0.034 0.002 0.325 (0.26) (0.66) (0.80) (0.19) (0.01) (1.05) Fraction Absent > 21 Days 0.064 0.049 0.194 0.007 0.014 0.132 (0.57) (0.48) (2.18)* (0.05) (0.11) (1.33) Fraction Disabled 0.019 0.131 0.016 0.094 0.066 0.11 (0.08) (0.38) (0.04) (0.41) (0.20) (0.26) Fraction w/ Free or Reduced Price Lunch 0.065 0.148 0.115 0.07 0.106 0.07 (1.12) (2.18)* (1.46) (1.41) (1.43) (0.85) Fraction Gifted 1.046 1.202 0.024 1.227 1.524 0.072 (1.82) (1.75) (0.03) (1.92) (1.97) (0.06) Fraction Limited English Proficiency 0.018 0.027 0.229 0.077 0.042 0.362 (0.10) (0.14) (1.17) (0.41) (0.24) (1.51) Fraction of Staff Dedicated to Instruction 0.064 0.140 0.263 0.058 0.179 0.292 (0.92) (1.04) (1.25) (0.94) (1.01) (1.14) Fraction Suspended In School 0.276 0.251 0.098 0.236 0.324 0.085 (1.20) (0.69) (0.23) (0.92) (0.88) (0.21) Fraction Suspended Out School 0.315 0.262 0.145 0.205 0.003 0.251 (0.98) (0.55) (0.31) (0.63) (0.01) (0.48) Fraction of Teachers w/ Advanced Degrees 0.017 0.034 0.092 0.02 0.013 0.119 (0.39) (0.71) (2.74)* (0.43) (0.24) (2.96)** Teacher Average Years of Experience 0.000 0.001 0.003 0.001 0.003 0.001 (0.24) (0.34) (1.45) (0.24) (1.01) (0.30) Real Personal Consumption (Income) <.000 <.000 <.000 <.000 <.000 <.000 (1.16) (0.82) (0.07) (1.47) (1.41) (0.50) District Size <.000 <.000 <.000 <.000 <.000 <.000 (0.38) (0.56) (1.25) (0.91) (0.85) (1.64) Constant 0.107 0.135 0.479 0.697 0.76 0.138 (0.60) (0.77) (1.74) (3.75)** (3.73)** (0.42) Observations 536 272 192 536 272 192 Number of Districts 67 34 24 67 34 24 R squared 0.9 0.91 0.91 0.92 0.93 0.93 NOTE: All specifications use either the percent of students testing into Levels 4 and 5 or Levels 1 and 2 on the Math portion of the FCAT as the dependent variable and are run with district level and year fixed effects. Columns vary by the sample size used in the regression. Absolute value of t statistic in parenthesis, represents significance at the fiv e percent level and ** represents significance at the one percent level.

PAGE 43

43 Table 25. Unitary status e ffect on r eading a chievement Levels 4 and 5 (High) Levels 1 and 2 (Low) (1) (2) (3) (4) (5) (6) Unitary Status Granted 0.004 0.005 0.002 0.024 0.022 0.028 (0.52) (0.57) (0.19) (1.16) (1.27) (1.88) Fraction White, District 0.289 0.226 0.054 0.424 0.481 0.418 (2.90)** (1.80) (0.31) (2.10)* (1.86) (0.94) Fraction Absent > 21 Days 0.177 0.194 0.156 0.278 0.312 0.261 (2.45)* (2.80)** (3.07)** (2.66)** (3.10)** (1.92) Fraction Disabled 0.024 0.037 0.104 0.251 0.335 0.424 (0.14) (0.15) (0.39) (1.00) (0.89) (1.00) Fraction w/ Free or Reduced Price Lunch 0.008 0.002 0.075 0.010 0.046 0.114 (0.21) (0.04) (1.39) (0.22) (0.69) (1.57) Fraction Gifted 0.780 1.071 0.789 1.385 1.917 1.058 (2.61)* (3.25)** (1.51) (2.92)** (3.79)** (1.17) Fraction Limited English Proficiency 0.058 0.087 0.185 0.122 0.163 0.365 (0.70) (0.83) (1.13) (0.73) (1.03) (1.45) Fraction of Staff Dedicated to Instruction 0.082 0.054 0.045 0.141 0.014 0.240 (1.50) (0.56) (0.31) (1.85) (0.09) (1.00) Fraction Suspended In School 0.150 0.283 0.054 0.005 0.020 0.116 (0.87) (0.90) (0.17) (0.03) (0.07) (0.34) Fraction Suspended Out School 0.241 0.224 0.775 0.180 0.257 1.003 (1.28) (0.72) (1.90) (0.75) (0.66) (1.86) Fraction of Teachers w/ Adv Degrees 0.014 0.028 0.058 0.094 0.107 0.041 (0.42) (0.67) (1.36) (1.27) (1.32) (0.56) Teacher Average Years of Experience 0.001 0.000 0.002 0.003 0.001 0.003 (0.95) (0.12) (1.39) (1.57) (0.68) (1.02) Real Personal Consumption (Income) 0.000 0.000 0.000 0.000 0.000 0.000 (0.93) (1.83) (1.46) (1.62) (2.77)** (1.61) District Size 0.000 0.000 0.000 0.000 0.000 0.000 (1.10) (0.66) (0.11) (0.53) (0.60) (0.43) Constant 0.055 0.062 0.116 0.904 1.216 0.557 (0.50) (0.41) (0.42) (3.99)** (4.64)** (1.08) Observations 536 272 192 536 272 192 Number of Districts 67 34 24 67 34 24 R squared 0.95 0.96 0.96 0.95 0.96 0.95 NOTE: All specifications use either the fraction of students testing into Levels 4 and 5 or Levels 1 and 2 on the Reading portion of the FCAT as the dependent variable and are run with district level and year fixed effects. Columns vary by the sample size used in the regression. Absolute value of t st atistic in parentheses represents significance at the five percent level and ** represents significance at the one percent level.

PAGE 44

44 Table 26. Unitary status e ffect on a chievement by r ace and subject Black White Math Reading Math Reading (1) (2) (3) (4) Unitary Status Granted 0.111 0.658 0.384 1.221 (0.06) (0.35) (0.24) (0.91) Fraction White, District 14.718 13.600 29.144 10.294 (0.35) (0.30) (1.25) (0.31) Fraction Absent > 21 Days 2.308 5.623 6.305 3.495 (0.20) (0.44) (0.94) (0.44) Fraction Disabled 33.869 64.937 68.295 84.735 (0.91) (1.59) (1.79) (2.05) Fraction w/ Free or Reduced Price Lunch 4.072 11.742 5.942 2.728 (0.53) (1.53) (1.26) (0.44) Fraction Gifted 128.845 131.893 52.004 109.703 (1.18) (1.11) (0.76) (1.24) Fraction Limited English Proficiency 0.330 2.510 20.605 6.246 (0.02) (0.16) (1.65) (0.52) Fraction of Staff Dedicated to Instruction 3.658 3.482 15.094 4.183 (0.19) (0.23) (1.41) (0.38) Fraction Suspended In School 8.811 27.832 40.311 10.514 (0.51) (1.17) (2.22)* (0.61) Fraction Suspended Out School 20.748 42.326 55.001 15.164 (0.37) (0.77) (1.16) (0.34) Fraction of Teachers w/ Advanced Degrees 5.235 3.398 4.534 8.226 (0.81) (0.82) (1.06) (2.14)* Teacher Average Years of Experience 0.202 0.025 0.217 0.022 (0.88) (0.12) (1.19) (0.09) Real Personal Consumption (Income) 0.001 0.001 0.001 <0.000 (1.90) (1.59) (2.46)* (0.96) District Size <0.000 <0.000 <0.000 <0.000 (0.38) (0.99) (0.60) (1.04) Constant 38.232 35.940 10.378 31.870 (1.10) (1.12) (0.57) (1.26) Observations 120 120 120 120 Number of Districts 24 24 24 24 R squared 0.96 0.95 0.98 0.97 NOTE: The dependent variable in each column is the percent of the subgroup that is proficient in a particular subject. Absolute value of t statistic in parentheses represents significance at the five percent level and ** represents significance at the one percent level.

PAGE 45

45 Figure 21. Floridas e lementary s chool e nrollment by r ace

PAGE 46

46 Figure 22. Average segregation i ndex by u nitary status, 19872006.

PAGE 47

47 Figure 23. Math achievement by u nitary status

PAGE 48

48 Figure 24. Reading a chievement by u nitary status

PAGE 49

49 CHAPTER 3 SCHOOL DISTRICT FINANCE AND UNITARY STAT US IN FLORIDA Introduction In some regards, the most active and unhindered period of court ordered desegregation was short lived. From its start with Brown v. School Board, only 14 years passed before the U.S. Supreme Court (SCOTUS) started to reverse its role in Green v. County School Board of New Kent County .1 In practice, the receipt of unitary status returns the ability to assign students to local school authorities Rather than having to meet stringent guidelines to keep school level racial populations standardized throughout the district, unitary districts can choose assignment policies regardless of the effect they may have on the distribution of students between schools. Over time, the possibility of resegregation has been explicitly condoned by SCOTUS and recent research suggests that it is not only a possi bility, but a reality as well. While the first case struck down the separate but equal standard, the second set up an exit strategy for districts placed under federal supervision. By meeting certain criteria for the assignment of students, faculty, and staff, transportation, extracu rricular activities and facilities, formerly segregated districts would be deemed to operate only one unified school system for all races rather than a separate system for blacks and another for whites. In other words, they would be deemed unitary. 2 1 347 U.S. 483 (1954) and 391 U.S. 430 (1968) respectively. Knapp (2009) shows that in Florida, districts released from court order are likely to double their level of segregation within eight years. Lutz 2 De facto segregation is allowed in Board of Education of Oklahoma v. Dowell and Freeman v. Pitts 498 U.S. 237 (1991) and 503 U.S. 467 (1992), respectively

PAGE 50

50 (2005) shows immediate and gradual increases in the level of segregation foll owing the receipt of unitary status in a nationally representative sample of districts. Since SCOTUS has already stated that resegregation via the granting of unitary status is not enough to justify sanctions on those districts that allow a return towards de facto segregation, further evidence is required to show that students are being harmed by the increased levels of segregation. On this, the literature is mixed. Knapp (2009) finds that the receipt of unitary status does not change the level of achiev ement for the average student, the average white student, nor the average black student at the district level. Lutz (2005) finds evidence that both black dropout rates and black private school attendance increase following the granting of unitary status, suggesting that the value of a public school education decreases for black students following the courts decision. Although there is a growing literature measuring the effects of unitary status on various academic outcomes, there is very little, if any, research on why school districts seek unitary status in the first place. Although SCOTUS has certainly left open the option of seeking unitary status, there is no requirement to do so. By seeking unitary status, school districts must be trying to benefi t in some manner and there could be many reasons why they may choose this route. They may be trying to satisfy parental preferences, operate their schools more efficiently or integrate their schools by choice rather than force. Regardless of the reason, the decision to seek unitary status and obtain its benefits cannot be costless. Lawyers must be hired, depositions taken, and hearings scheduled before the case even makes it to a judges bench for the final

PAGE 51

51 decision.3 To date, no one has tried to answer the basic question of why districts seek unitary status. This paper tries to bridge this gap and add to the s chool resegregation literature in several ways. First, by examining both the timing of release from court order and the size of the student population for all the districts ever placed under court order in Florida, it can easily be shown that larger distr icts receive unitary status more frequently than smaller districts. This is especially true over the last decade. This would seem to suggest that smaller districts do not find the tradeoffs to be in their favor and decide not to seek unitary status while larger districts appear to do the opposite. Secondly, it will be shown that the financial benefits, if they exist at all, are very small on a per capita basis. Smaller districts, without a large enough student population available to recapitalize the upfront costs incurred while seeking unitary status, may rationally decide to stay under the courts desegregation order in perpetuity. For larger districts that can achieve economies of scale, the proper decision may be to seek unitary status. In a classical sense, for the decis ion to seek unitary status to be rational the benefits from the outcome must outweigh this myriad of costs. Even if dist ricts do not try to take advantage of changes in district finance ex ante, they may benefit ex post via changes in the flow of revenues and expenditures. One area where this may occur is transportation. It would seem plausible that one of the most popul ar tools used to integrate schools, mandatory bussing, would be more costly 3 It might seem natural to insert some discussion of the probability of success at this point. However, that may be unnecessary for two reasons. First, according to the U.S. Civil Rights Commission, there is not a single case in Florida that has been actively pursued a nd eventually denied for a significant period of time. Secondly, some may argue that recent SCOTUS decisions have set the bar for unitary status so low that all districts have to do to achieve it is be willing to incur the costs associated with the proces s. Evidence of this can be found throughout Orfield (1996) and Chemerinsky (2009)

PAGE 52

52 to carry out than bussing students to their closer neighborhood schools. If that is the case, the receipt of unitary status may allow districts to save money on transportation and spend it on instruction in an effort to boost student achievement. Similarly, there is anecdotal evidence that desegregation cases impart nontrivial legal costs on the districts involved while they are actively in court .4 Payment of these fees diverts scarce resources away from the school districts core mission for years at a time while they actively seek removal of their court orders When districts get released from their court ordered desegregation plans, these types of costs may go away and lead t o possible changes in district level expenditures as funds can now be shifted from legal expenses to other areas .5Floridas Desegregation History and How It Relates to District Size This analysis is the first to look for these kinds of patterns. Court o rdered school desegregation came to Florida in 1956 when the National Association for the Advancement of Colored Peoples Legal Defense Fund successfully won a case against the Miami Dade County School Board. Over the next 22 years, another 33 districts w ould be placed under court ordered desegregation plans.6 4 Examples can easily be found in newspaper coverage of individual court cases. In particular, see Dries (2009), Hughes (2005), and Spinner (1999). Legal fees associated wi th court ordered desegregation have also been addressed by the U.S. Supreme Court. See Missouri v. Jenkins, 491 U.S. 274 (1989). Eventually, this list would include districts of all sizes throughout the state. Figure 1 shows the relationship between district size and desegregation status. Over this time period, two trends stand out. First, size did play a role in the probabilit y of being placed 5 This type of cost is different from what might be considered compliance costs. To some extent, all districts, regardless of litigation status, have compliance costs. Even districts that have never been taken to court must report on their level of integration to the U.S. Department of Educations Office for Civil Rights. Therefore, removal of a court order will not necessaril y decrease this type of expense because the reporting requirement does not go away. It simply shifts between government agencies. 6 For a complete list of all the districts in Florida and their court ordered desegregation status see Table A 1 in the appendix.

PAGE 53

53 under court order. Fourteen of the 15 largest districts were placed under federal supervision by 1978. At the opposite end, only three of the smallest 15 were forcefully desegregated. This means the largest districts were nearly 4.5 tim es more likely to be taken to court when compared to the smallest districts. However, it should be noted that there was some litigation within the lower end of the size distribution. Also significant is that the odds change dramatically when the distribution is trimmed. The probability of receiving a court order is nearly symmetric around the median district size of 12,196 with eight being below and nine above the median.7 Even as districts were still being placed under court order, some were beginning to be released and declared unitary. The first district to receive unitary status in 1970 did so after only seven years under court supervision. However, that is not the ordinary path taken by districts placed under court order. Of the 34 taken to court, only 19 have received unitary status after spending an average of almost 26 years under court order. For the 15 still under supervision, the average time since being found guilty is just over 40 years. When put together, there was at least some probability of prosecution for all dist ricts in Florida. For most of those still under court order, there may be no compelling reason to seek unitary status. The reason for this appears to be that size plays an even more important role in obtaining unitary status than it does in being placed under court order, especially over the last decade. Figure 2 outlines the granting of unitary status for the 34 districts placed under court order. Note first that there appear to be two groups 7 This median is based on the average number of unweighted fulltime equivalents in each district from 1997 2006. Using averages from the early part of the desegregation era does not create any substantive differences.

PAGE 54

54 within this sample, early grantees and those still under court order as late as 1997, with each being separated by a decade of time during which not a single district received unitary status. The first group is made up of eight districts that spend a relatively shorter period of time under court order. Together they average about 12.5 years under court supervision. The remaining 26 districts have spent an average of nearly 39 years under court supervision. This difference in means is significant at the 99th percentile. There is more than time that separates these two groups as the role of size appears to differ for these two groups. Among the early grantees, half are larger and half are smaller than the median district size of 34,813 for districts that have been placed under court order.8 Within the group of later grantees only one district is below the median. This highlights the increasing role that size has played in districts deciding to seek unitary status.9 Due to the time span between the two groups and data issues limiting the availability of detailed financial records earli er than 1997, the remainder of the paper will focus on the period from 1997 to 2006.10 8 This sample has fewer distric ts than the one used previously and contains districts that are larger than on average for the entire state. This is done because it may not be appropriate to consider a district that has never been taken to court as a control for a district that has been treated to unitary status. The more accurate control group is going to be those eligible for treatment and that requires a district to have been taken to court at some point in the past. It will attempt to explain why size plays 9 If the issue of what makes an appropriate control group is ignor ed for a moment and unitary status grantees are evaluated against the statewide median of 12,196 unweighted FTEs, the effect of size becomes even more striking. In this case, only two districts that receive unitary status are below while the remaining 15 are above this less restrictive, statewide median. These two districts are both in the group of early grantees. No district smaller than the statewide median has received unitary status in the last 23 years. 10 This strategy also fits t he reasoning of Or field (1996) and Chemerinsky (2009) who show that the rules used by the courts to determine whether unitary status should be granted changed in the early 1990s

PAGE 55

55 such an important role in why some districts choose to seek unitary status and others do not.11Description of the Data Litigation data for Florida was collected from the United States Commission of Civil Rights September 2007 report titled, Becoming Less Separate? School Desegregation, Justice Department Enforcement and the Pur suit of Unitary Status It contains data on every district in Florida including the year cases were initiated (if applicable), the case name, the year unitary status was received and whether districts not currently seeking unitary status pl an to do so in the future. These data were used to create the key variable of interest in this study, a dummy variable that captures the trend of districts receiving unitary status over time. Unitary Status equals zero in the years preceding receipt of unitary status and equals one in all others. This formulation of Unitary Status measures the cumulative effect of unitary status over time. Fiscal data for every district in the state is publically available from the Florida Department of Educations website and is contained in a report for each year titled, Financial Profiles of Florida School Districts .12 11 This refers to the 19971998 and 2006 2007 school years. This convention will be used throughout the rest of the chapter. Each years report contains over 300 elements detailing district level revenue and expenditure data. Due to variation in element definitions and some elements being added or dropped over time, the main focus of this paper is on aggregate data which should net out these additions, deletions and definition changes at finer levels of detail. 12These files are available at http://www.fldoe.org/fefp/profile.asp (last accessed on April 13, 2010).

PAGE 56

56 Revenue data are broken out into 18 different categories that vary by sourc e and funding objective. There are four major sources of funds for Floridas school districts; those received by districts directly from the federal government, federal funds received by districts which are first passed through the state budgeting process funds provided by the state from statelevel resources, and funds received from local sources such as property taxes. All four of these sources are managed within the Florida Education Finance Program (FEFP) to ensure an equitable distribution of resour ces across districts although s ome consideration is allowed for such things as differences in the cost of living across districts. Within each source of funds, there are several types of revenues specific to each source. These are listed in Table 3 1 along with the several statistics. Column 1 shows the mean for the entire time series from 19972006 for all 67 districts in the state Column 2 and Column 3 list sample means for the preferred control and treatment groups. When combined, t hese two columns make up a sample of districts that had previously been placed under court order but not granted unitary status prior to 1997. This excludes districts never successfully taken to court and those who received unitary status as an early grantee. Column 4 reports the result of a twotailed t test for equality of means in order to provide evidence that there is variation in the data. Each column shows revenues per student and has been converted into real 2000 dollars using the national consumer pri ce index for all urban consumers ( CPI U ) Table 3 1 appears to suggest that there are some significant differences between those districts that are granted unitary status and those that are kept under court order, absent any controls for other covariates. Eleven of the 17 categories suggest possible avenues through which

PAGE 57

57 unitary status may affect revenues. Rather strikingly, unitary status recipients appear to have $250 less per student on net when all revenue from local, state, and federal sources is co nsidered, which represents a difference of about 3.3 percent. Expenditure data are taken from the same source and listed in two ways. The first aggregates expenditures by categories such as instruction, administration, and transportation while ignoring what type of goods or services were purchased with the funds. The second method places the emphasis on the objects being purchased and tracks expenditures on salaries, benefi ts and capital outlays. Table 32 summarizes expenditures by both category and object in real 2000 dollars per STUDENT Each column follows the methodology used for Table 3 1 in creating different subsamples and testing the equality of means. Although there appears to be a significant amount of variation between the treatment and co ntrol groups by category and object, the net result appears to suggest that there is little significant difference in total expenditures between the two groups. If net revenues fall for districts receiving unitary status while expenditures are relatively constant, this may suggest that the granting of unitary status leaves districts worse off financially and begs the question of why would a rational district seek unitary status? Local school boards should be trying to ac hieve some economically rational goal in seeking unitary status and leaving their districts with fewer revenues and the same level of expenditures does not appear to meet that standard. However, it is too early to substantiate this conjecture as a more r obust model needs to be developed to account for the timeseries nature of the data and other covariates.

PAGE 58

58 Description of the Empirical Models This paper uses a timeseries model which in its simplest form can be expressed as: it i t it itStatus Unitary y ( 3 1) The dependent variable, it y will take on multiple values of per capita revenues (expenditures) in district i in year t Unitary Status measures the litigation status of district i in year t Districts still under court order in y ear t and those never taken to court receive a zero (though not all of these latter districts are included in every specification) Those that receive unitary status in year t receive a one in that and all subsequent years. The year fixed effect, t, captures all district invariant change for year t This accounts for unobserved statewide policy changes that affect districts equally. An example of this type of policy would be an administrative change to accounting rules that ca use revenues (expenditures) to be counted differently from one year to the next. Without this variable, such a change could be mistakenly attributed to a change in unitary status in districts where such changes occur simultaneously, causing the coefficient on Unitary Status to be biased. The district fixed effect, i captures all timeinvariant characteristics of district i .13 it represents a random error term which allows for district level heterogeneity. The reported results utilize a more robust version of this model that takes the form: 13 Although a district year fixed effect would help account for unobserved policy changes at the district level on a year by year basis, one cannot be used in this case due to the limited sample size.

PAGE 59

59 it i t it it itStatus Unitary y ( 3 2) It allows for more variation between districts over time by adding a vector of timevarying, district level characteristics,it. Due to the extremely small sample size, the contents of itmust be limited in number. The first element of itis the number of schools in each district. This captures the growth in the student population which occurred in some districts over the period of the study. It also captures the effect of economies of scale within the revenue and expenditure data. This is especially important for administrative costs such as school boards. The next two elements account for economic disparities between districts that may affect how districts make decisions concerning education. The first variable is the annual property tax base per STUDENT in each district and the second is the percentage of students in the district that are eligible for Free or Reduced Price Lunch.14 This chapter s identification strategy relies on the timing of the unitary status grants as its prominent source of variation. These three models will allow statistical identification of the effect of unitary status on revenues (expenditures) if unitary status occurs exogenously. In other words, unitary status must be received randomly. Anything that affects the randomness of that timing, could affect the identification strategy. The final element is the percentage of students in the district that are nonwhite. This accounts for the rapid growth in the Hispanic population across many districts in the state and any effects this may have on how districts receive and spend financing. 14 In 1997, only the percentage of FTEs eligible for free lunches is available. The percentage eligible for both was not available until 1998. The year fixed effect should account for this type of change since it occurs uniformly across districts.

PAGE 60

60 There are several issues to consider in ensuring the validity of this identification strategy. First, receiving unitary status is conditional on having been taken to court. Districts that have never been taken to court cannot receive unitary status as defined in this paper. Secondly, the study period for this paper is only ten years long, 1997 2006. Eight districts receive unitary status before then and cannot be treated by the policy a second time. Since it may be questionable to include either of these types of districts in the sample, the analysis will also be conducted with a sm aller, preferred, sample of 26 districts which includes only those districts that were eligible for treatment. F or the remaining districts, the timing of unitary status m ust be random in order for it to have a statistically identifiable effect. Here, the administrative process of dealing with the court system provides a significant amount of variation in the timing of release. This is plausible because two actions must t ake place before a district can receive unitary status. The school district must decide to seek release from their court ordered desegregation plan and the court must then grant it. Even if the first decision is affected by a nonrandom phenomenon such as size, the second part of the process, getting the case through the court system, returns randomness to the actual release date and the granting of unitary status. The administrative issues involved in going to court, having the case heard, and getting a decision made are complex and varied enough to imply that the actual year in which unitary status is received is not directly chosen by the school district and is somewhat random A simple test of this theory involves looking at the level of correlation between the year in which unitary status is received and the number of student s in the district. Figure 22 already highlights the tendency of larger districts to seek unitary status. This

PAGE 61

61 is indicative of the first part of the process. However, the relationship between the size of the district and the year in which unitary status is received is not a strong one. The correlation coefficient between the year unitary status is received and the number of students in the nine districts that received unitar y status between 1997 and 2006 is only 0.09. Those districts that are released show little sign of being released in order of their size.15 The last empirical issue concerns the weighting of observations in each regression. Districts in Florida exhibit large differences in size with the difference being nearly 350,000 students between the largest and the smallest. This size difference persists even with the preferred sample of 26 districts. As discussed in the next section, weighting does affect some of the results. This suggests that the judicial process makes the actual timing of release relatively random. 16 15 Again, it may be useful to apply the findings of Orfield (1996) and Chemerinsky (2005). They have noted that federal courts changed their behavior towards desegregation cases in the early 1990s. Given the courts newfound desire to clear its docket of desegregation cases, it is hard to construct a scenario where a courts decision to release a district is dependent on district size. They would be willing to release any district, regardless of size. Additionally, there is no evidence that a district has ever been denied unitary status during the later part of the unitary era. Therefore, it is not unreasonable to assume that small districts would receive unitary status if they were willing to seek it. The key difference between those regressions run with and without weights is in how any statistically significant results are interpreted. For t hose regressions without weights, the results represent the effect of unitary status on revenues (expenditures) in the average district. When weights are used the interpretation changes, now the results are interpreted as the effect on the average student in a district that receives unitary status. Neither specification has a clear advantage over the other as long as the distinction between the two remains clear 16 Weighted specifications utilize the number of students in the district as the weight for each district observation. This procedure is implemented using the aweights option in Stata.

PAGE 62

62 when the results are interpreted. To test the effects of weighting on the results, the prefer red sample will be reported with and without weights. In addition to reporting results for the preferred sample with and without weights, a test will be conducted on a subsample of districts where weighting should not affect the results. Weighting sh ould not be an issue when the size difference between districts is relatively small. In that case, the effect on the average district and the effect on the average student should be close to the same. To test this, Equation 3 2 will be estimated both wit h and without weights while using a sample consisting of 10 of the 11 largest districts from the preferred sample. This excludes Floridas largest district, Miami Dade and cuts the span of the size difference from nearly 350,000 to 190,000. The Effect of Unitary Status on School District Finance Unitary Status and Revenues The majority of the results are shown in two tables. Due to the large number of dependent variables, only the coefficient on Unitary Status is reported for each regression. When neces sary, further discussion of the other covariates and standard errors will occur in footnotes with additional tables provided in the appendix. Table 3 3 reports the results of using Equation 3 2 when the dependent variable,ity, takes on the value of revenues per student from various sources Table 3 4 provides results from when ity takes on the value of expenditures per student Each table uses the following pattern. First, the model is completed using all 67 districts in the state. Next, the sample is limited to the 34 districts that were taken to court. Then, the sample is reduced to the preferred sample that is considered eligible for treatment. Results for

PAGE 63

63 this sample are reported with and without w eights. Lastly, the effect of weighting is tested using a sample of ten, relatively large districts. Overall, the results from Table 3 3 suggest that Unitary Status has at best a small negative effect on revenues per student The last line reports the effect Unitary Status has on the total revenue received from all local, state, and federal (LSF) sources. The results are mixed. Five of the six specifications, weakly suggest that revenue per student falls. However, at no time, regardless of the sample or weighting used, does Unitary Status have a statistically significant impact.17 Federal funding sources are listed at the top of Table 3 3. On net, there appears to be no clear change in funding from federal sources. Any decreases appear to be offset w ithin the federal funding lines by corresponding increases. Career funding and the federal meal program show a fairly uniform sign pattern suggesting decreased levels of funding, but there is not robust statistical significance across specifications. Eve n in those specifications where there are significant results, there is little e ffect in terms of magnitude. Federal funding for other programs appears to help offset these potential losses by adding anywhere from $7 $25 per student All the remainin g sources of federal funding lack a statistically significant effect for Unitary Status This includes the total level of federal funding received which lacks any identifiable results and a clear message as to the sign of any possible effects. This is not surprising considering offsetting nature of the statistically significant results that were found for federal career, meal, and other funding. 17 Table A2 provides more detailed results from regressing total revenues on all the covariates. It should be noted that the standard errors for Unitary Status are uniformly large. What this means may depend on the results for other sources of revenue and will be discussed shortly in greater detail.

PAGE 64

64 State level funding shows weak signs of potential increases as a result of receiving unitary status. State funding for other programs is the only line within the state level section to show a consistent sign pattern and have meaningful levels of statistical significance. It suggests that districts with unitary status receive an additional $36$134 per student .18 This increase may be partially offset by a potential decrease in state categorical programs where there is some weak evidence of a decrease as much as $138 per student .19 Of the three possible sources, locally provided funding appears to be the only one to decrease with overwhelming levels of statistical significance. Here, the effect of Unitary Status on total local revenue is significant in four of the six specifications and shows that the receipt of unitary status decreases the level of resources received from all local sources by about $100 per student However, the actual effect is likely smaller due to the inconsistent sign pattern and lack of statistical significance. This offset is also apparent in the total state funding line. Results for five of the six specifications suggest that there is a total increase in state funding. In most of the cases, the increase in total state fu nding is smaller than the increase from other programs, suggesting that decreases in categorical programs provides some attenuation. Due to a lack in statistical significance, it is likely that any increases in state funding are small at best. 20 18 Table A3 shows that even with a subsample of only ten districts, the effect of Unitary Status on other state le vel funding sources is significant up to the 82nd percentile. This shows that the effect of Unitary Status is significant or close to significant in five of the six specifications. Considering the extremely small sample size and the relatively small dropoff in statistical significance for the fifth specification, there appears to be some robust evidence that an increase in this area of funding does occur as a result of having received unitary status. The majority of this decrease appears to 19 Categorical programs include such things as Florida s Teacher Lead Program, Transportation, Instructional Materials, Discretionary Lottery Funds, and the mandated Class Size Reduction Program. 20 The two remaining specifications are not statistically significant but still have negatively signed coefficients ranging from $65$86 per FTE.

PAGE 65

65 come from decreases in the amount of revenue per student sourced from local tax sources. Some additional decreases may come from the amount of other income received by unitary districts, but the magnitudes are smaller, and only one of the six specifications is statistically significant at any conventional leve l. When all sources are considered together, it is not immediately clear that any significant changes in funding occur as a result of receiving unitary status. The coefficients for the total amount of LSF funding listed at the bottom of Table 33 never r each a meaningful level of statistical significance. However, the sign and magnitude of the coefficients are in line with the results of each source when they are considered individually. The small, negatively signed effect of unitary status on total LSF funding appears to be largely driven by changes in state and local funding. Decreased local funding is only partially offset by increases in state funding, resulting in a modest decrease in overall funding. Federal levels appear to be unaffected, on net by the granting of unitary status. One possibility for the large standard errors and the lack of statistical significance for overall LSF revenues is that random noise from the federal and state categories could be masking the effect of a drop in local revenues. The results from the federal portion suggested that there was no overall effect from federal funding. The state category showed weak signs of possible increases. When these noisy signals are combined, they may drown out the statistically significant local funding results. Therefore, it may be reasonable to interpret the negative, but statistically insignificant, LSF results as supporting the notion that some decrease in local revenues does in fact occur.

PAGE 66

66 Unitary Status and Expenditures It i s important to keep in mind that revenues are only one portion of the school finance picture. It is also possible that the granting of unitary status could affect the way districts spend their finances. Therefore, it is important to also look for any pos sible changes in expenditures. Table 3 4 reports the coefficient for Unitary Status after completing regressions based on Equation 3 2 where it y takes on the values of two broad types of expenditures. The top panel lists expenditures by categories such as instruction, food service, and transportation. The bottom panel lists expenditures by objects such as salaries, benefits, and materials. The two panels end by showing the same overall result. Although there appear to be sizable decr eases in spending per student five of the six specifications return negative effects ranging $58 to $305 per student the standard errors are large. This suggests that any decrease in expenditures is, at best, very imprecisely measured. When examining the expenditure breakouts in finer detail, statistically significant results are still hard to find. Only one category, General Administration, exhibits significant changes when unitary status is received, but this occurs in only two of the six specificat ions. The results from Table 3 4 suggest that per pupil expenditures for this category increase by $7$10 per student .21 21 Table A4 presents more detailed results for the effect on General Administration expenditures. Note that the coefficient on Unitary Status is positive in all six specifications. Additionally, the standard errors are not so large as to rule out some positive effect at lower levels of statistical significance. Additionally, although there is robust evidence of statistical significance, economic significance is not very prominent. This change is understandable as exit from a court ordered desegregation plan can entail having to implement and manage complicated as signment policies. However, this increase does not appear to carry through to total expenditures and it appears that the increases in General Administration

PAGE 67

67 expenditures could be offset in several areas. Although they do not reach meaningful levels of st atistical significance, expenditures on Instruction, School Administration, Transportation, Facilities (Capitalized), and Debt Services all show a trend of decreasing following the receipt of unitary status with at least five of the six coefficients being negative.22 Of these, only Debt Service and Facilities appear to have any sizable magnitudes which could result in the total expenditure category appearing to be negative.23 22 Much like the revenue data, the aggregation of random noise from the sub level expenditure data could be why the standard errors on the total expenditure figures are so large. The dominating effect of Facilities and Debt Services is confirmed in the bottom panel of Table 3 4 where expenditures are broken out by object. Here, only two objects show any statistically significant changes. Unitary Status appears to increase expenditures on Purchased Services by just over $50 per student and decrease spending on other expenses by $12. However, these results occur in different specifications, potentially weakening their impact. This net gain washes out as total expenditures by object are predominantly negative in sign and relatively large in magnitude. Again, this offset appears to be a result of large decreases in spending on facilities and debt services. Overall, much like the results for the revenue data, there is at best only weak evidence for declines in spending as a result of receiving unitary 23 The coefficients for these two categories should be viewed with some caution. In addition to the lack of statistical sig nificance, it may be difficult to interpret what these negative coefficients actually suggest. In at least one district, construction of new schools in predominantly black neighborhoods was made a precondition of release from court order. This means expenditures on capitalized facilities were higher in the periods prior to the receipt of unitary status. Therefore, the negative coefficient could really be a signal that expenditures are returning to normal from a previously elevated level as spending on ne w construction is curtailed.

PAGE 68

68 status and that the bulk of these declines are a result of changes in facility and debt service expenditures.24The Role of Economies of Scale Overall, unitary status does not have a clear impact on school finances. Although both revenues and expenditures show some statistically significant results at certain sublevels, they do not carry over to district level bottom lines. If they did, a definitive statement could be made concerning the financial changes caused by unitary status. However, in the absence of such r esults, it is more likely that the effect of being released from a court ordered desegregation plan has modest and nuanced effects on district level finances. For revenues, there appears to be little effect on either federal or state funding per student Local funding appears to be reduced, primarily through reduction of the local tax burden, by about $100 per student However, this effect does not carry over to district level bottom lines in a statistically significant way. This could mean that unitary status has no effect on total LSF revenues or that the effect is negative, but too imprecisely measured given the current data.25 24 If the net effect on total revenues and total expenditures is considered, it appears that districts benefit from unitary status because expenditures per FTE fall by more than revenues per FTE in four of the six samples. This benefit appears to be about $100 per FTE. However, this comparison is severely weakened by the overall lack of statistical significance. Expenditures also show no clear cut results. The only statistically significant results suggest small increases for certain subcategories, but the se results also disappear when aggregated. 25 It is important to remember that state and federal funding could be adding random noise to the total LSF line. This would cause the standard errors to be large and the overall effect to be imprecisely measured. This would also obscure the effect on local tax burdens.

PAGE 69

69 Fortunately, the revenue results alone appear to be enough to explain the data. First, consider the case where the decreases in local tax revenues are offset by gains elsewhere to combine in a zero total effect scenario on LSF revenues. This would be evidence of revenue shifting, total LSF funding stays the same, but the local burden is decreased. This would be a clear benefit to a district that is able to implement such a strategy. If uni tary status provides such an avenue, the implication could be that districts seek unitary status strategically to take advantage of this revenue shifting. If districts know that they can decrease the local burden of school finance, but maintain overall rev enue levels, seeking unitary status would appear to be a worthwhile when holding all else constant. The results from the previous section also allow for another case, one which says that the true effect of unitary status on revenues is negat ive, but impre cisely measured given the current data. If this case holds, the effect on academic outcomes becomes important. If resources and outcomes decrease in tandem, unitary status may be a poor policy. However, if resources fall and outcomes remain the same or even increase, this could be a sign of increased efficiency. As outlined in Chapter 1, achievement based on standardized test scores does not change with receipt of unitary status. If the same educational outcomes are obtained at lower revenue levels, unitary status may allow districts to operate more efficiently. Therefore, a districts decision to seek unitary status may be an effort to save local tax dollars and operate more efficiently. Regardless, this case also provides a strategic reason for seeking unitary status. B oth of the previous scenarios are also consistent with the tendency of only large districts seeking unitary status. Keep in mind, t he decision to seek unitary status cannot

PAGE 70

70 be costless and it may be wise to assume that significant l egal and settlement costs may be incurred if unitary status is received.26 What does this mean going forward? The smallest district to receive unitary status among the later grantees had an average population of about 31,000 students during the study period. If we assume a balanced budget, this implies that the cost of seeking unitary status is no more than $3.1 million. Of the 15 districts remaining under court order, only three are large enough to guarantee a beneficial payback within one year This may imply that the era of unitary s tatus grants is coming to a close as fewer and fewer districts will find it in their favor to seek unitary status. This may or may not lead to efficient outcomes. If court ordered desegregation plans truly constrain the districts under such plans, then r elease can lead increased efficiency. However, if Assume for a moment that a district currently under a court ordered desegregation plan must pay a fixed cost of $1 million to receive unitary status. Table 3 3 suggests that the local burden is reduced by about $100 per student As long as the district has more than 10,000 students, the district will recoup the cost in one year. If benefit to unitary status is anything smaller than $100 per student the district must be willing to finance the cost from reserves, for e go some current expenditures or wait for the decision to pay off over time. Therefore, it is reasonable that the data suggest that small school districts do not find such an outcome desirable and decide that simply liv ing with the courts order is optimal. 26 Although good accounting data on the legal cost of seeking unitary status does not appear to be av ailable, there is some anecdotal evidence that the costs are large. In many cases, the defendant districts are required to pay the plaintiffs attorney fees. In Missouri v. Jenkins, the plaintiffs were awarded attorneys fees of over $3 million (see 131 F.3d 71 and MISSOURI V. JENKINS, 491 U. S. 274 (1989) ) In a case against Charlotte Mecklenburg Schools, the plaintiffs asked for but were denied nearly $1.5M (see 274 F.3d 814 (4th Cir. 2001)). Although the potential legal costs are not broken out, the Pinellas County, FL school board budgeted for over $4.5M in compliance costs when they received unitary status.

PAGE 71

71 districts decide to remain under court order because the upfront costs are large, scarce educational resources could be lost.

PAGE 72

72 Table 31. District level r evenues per student (1) (2) (3) (4) Treated Control Number of Districts in Sample 67 9 16 P<0.05 Direct Federal Funding 55.80 48.29 41.25 NO Federal Career Funding Through State 22.16 13.29 21.74 YES Federal Title I & V Through State 219.99 200.16 209.12 NO Federal Adult Basic Education Through State 10.02 4.06 17.36 YES Federal Meals Funding Through State 200.50 192.54 194.09 NO Federal Other Funding Through State 307.97 260.86 302.20 YES Total Federal Funding Through State 760.64 670.91 744.53 YES Total Federal Funding All Sources 816.43 719.20 785.77 YES Florida Education Finance Program 2,505.83 2,405.52 2,554.18 NO State Categorical Programs 994.46 815.71 1,011.70 YES State Racetrack Funding 37.11 3.13 42.89 YES State Other Sources 355.14 301.86 385.79 YES Total State Funding 3,892.53 3,526.22 3,994.56 YES Local Tax Revenue 2,543.65 2,710.00 2,415.33 YES Local Investment Income 92.23 99.89 89.49 NO Local Other Income 312.36 310.89 324.18 NO Total Local Revenue 2,948.25 3,120.79 2,828.99 NO Total Local, State and Federal 7,657.21 7,366.21 7,609.32 YES NOTE: Revenues are shown in real dollars (2000) deflated using the CPI U from the January of each school year starting with 19971998 and ending in 20062007. Column 1 shows district level means for all 67 school districts in Florida. Column 2 restricts the sample to the nine districts in the preferred treatment group. This includes only those districts that received unitary status between 1997 and 2006. Column 3 rest ricts the sample to the 16 that were eligible for unitary status but had not received it by 2006. These districts represent the preferred control group. Column 4 reflects the results of a twotailed test for equality of means at the five percent level.

PAGE 73

73 Table 32. District level e xpenditures per student BY CATEGORY (1) (2) (3) (4) Treated Control Number of Districts in Sample 67 9 16 P<0.05 Instruction 3,560.83 3,529.37 3,484.04 NO Pupil Personnel Services 312.93 305.75 325.49 YES Instructional Media Services 110.46 109.35 109.41 NO Instruction and Curriculum Development 194.14 167.12 199.94 YES Instructional Staff Training 67.00 68.43 67.22 NO Instructional Support Services 694.26 659.31 710.71 YES School Board 48.53 15.91 57.54 YES General Administration 78.88 47.92 82.58 YES School Administration 365.49 349.35 366.26 YES Facilities (Non capitalized) 29.03 30.09 28.16 NO Fiscal Service 60.80 34.91 68.49 YES Food Service 328.60 317.80 329.50 YES Central Services 122.04 135.82 109.15 YES Transportation 303.33 293.09 313.70 YES Operating Plant 479.07 471.27 477.32 NO Maintenance of Plant 185.11 175.10 186.37 YES Total General Support Services 2,009.62 1,878.58 2,028.97 YES Total Instruction, Instructional Support and General Support 6,264.71 6,067.26 6,223.73 YES Community Services 47.82 56.60 51.09 NO Total Current Expenditure 6,312.52 6,123.86 6,274.81 YES Facilities (Capitalized) 1,246.27 1,142.10 1,209.15 NO Debt Services 292.40 407.62 321.75 YES All Expenditures 7,851.19 7,673.59 7,805.71 NO BY OBJECT (1) (2) (3) (4) Treated Control Number of Districts in Sample 67 9 16 P<0.05 Salaries 3,865.99 3,764.15 3,851.00 YES Employee Benefits 1,096.20 1,132.67 1,093.27 YES Purchased Services 549.74 518.88 522.63 NO Energy Services 182.81 166.17 185.21 YES Materials and Supplies 394.63 365.52 398.38 YES Capital Outlays (Non Capitalized) 89.41 76.97 86.88 NO Capital Outlays (Capitalized) 1,246.27 1,142.10 1,209.15 NO Debt Services 292.40 407.62 321.75 YES Other 133.73 99.50 137.45 YES Total Expenditures 7,851.19 7,673.58 7,805.71 NO NOTE: Column 1 shows district level means for all 67 school districts in Florida. Column 2 restricts the sample to the nine districts in the preferred treatment group. This includes only those districts that received unitary status between 1997 and 2006. Column 3 restricts the sample to the 16 that were eligible for unitary status but had not received it by 2006. These districts represent the preferred control group. Column 4 reflects the results of a twotailed test for equality of means at the five percent level.

PAGE 74

74 Table 33. Effect of u nitary status on r evenues (1) (2) (3) (4) (5) (6) Direct Federal Funding 8.39 11.57 6.77 0.70 1.94 2.845 Federal Career Funding Through State 3.210* 4.442** 4.257** 0.63 1.24 0.28 Federal Title I & V Through State 2.33 3.60 2.50 6.25 2.35 2.459 Federal Adult Basic Education Through State 0.83 1.88 3.14 1.80 1.219* 1.531** Federal Meals Funding Through State 4.52 4.04 5.45 6.30 7.020** 9.034** Federal Other Funding Through State 23.55 23.50 10.12 25.36* 18.62 7.242 Total Federal Funding Through State 17.31 16.74 5.22 26.48 13.93 2.999 Total Federal Funding All Sources 8.93 5.17 11.98 27.19 15.87 0.153 Florida Education Finance Program 8.85 22.14 1.46 22.57 28.65 6.179 State Categorical Programs 44.05 59.60 138.4* 6.71 34.21 1.672 State Racetrack Funding 1.69 0.38 0.50 0.12 0.16 0.231 State Other Sources 134.1** 85.86* 99.97* 84.18** 51.19 36.14 Total State Funding 171.0* 4.51 39.39 68.44 45.46 31.4 Local Tax Revenue 41.64 71.82 58.44 77.62* 80.12 85.49* Local Investment Income 2.24 1.95 5.39 13.14 7.15 8.42 Local Other Income 25.70 16.58 47.06* 37.43 25.50 25.47 Total Local Revenue 65.11 86.45 100.1** 101.9** 98.47** 102.5** Total Local, State and Federal 114.80 76.78 151.50 6.29 37.14 71.3 Number of Districts 67 34 25 25 10 10 Weighted by Number of Students NO NO NO YES NO YES NOTE: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. Other controls include year and district fixed effects, the number of schools in the district, the size of the tax roll per capita, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, these coefficients are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted by the number of full time equivalents in the district. Columns 5 and 6 follow the same procedure but only include the ten largest districts in the state, excluding Miami Dade.

PAGE 75

75 Table 34. Effect of u nitary status on expenditures BY CATEGORY (1) (2) (3) (4) (5) (6) Instruction 33.94 35.01 44.05 48.99 17.76 7.258 Pupil Personnel Services 9.209 14.94 0.812 2.052 6.881 4.896 Instructional Media Services 2.689 1.202 1.817 3.705 2.188 1.934 Instruction and Curriculum Development 5.339 7.296 4.428 1.675 14.91 5.363 Instructional Staff Training 8.893 4.817 4.111 0.309 8.603 10.62 Instructional Support Services 4.583 16.2 0.284 1.834 0.548 10.18 School Board 0.344 1.983 4.292 1.361 2.217 0.206 General Administration 5.459 4.248 6.154 10.09*** 7.644** 4.443 School Administration 7.963 15.12 11.93 8.035 10.6 10.07 Facilities (Non capitalized) 40.19 35.95 39.27 12.69 34.74 13.79 Fiscal Service 0.943 1.455 1.032 0.322 0.11 0.662 Food Service 6.647 5.957 4.222 3.739 3.292 3.545 Central Services 2.863 0.28 2.918 7.024 3.027 9.912 Transportation 14.64 13.42 14.61 10.07 12.19 7.787 Operating Plant 10.67 9.504 2.614 6.597 7.071 5.153 Maintenance of Plant 21.53 34 32.97 0.585 29.85 15.19 Total General Support Services 23.01 27.85 55.91 21.59 52.69 25.59 Total Instruction, Instructional Support and General Support 15.51 23.35 11.58 25.57 71 8.149 Community Services 10.14 11.78 0.672 1.403 3.634 0.262 Total Current Expenditure 5.372 11.57 10.91 24.16 67.37 8.411 Facilities (Capitalized) 11.75 176.7 206.2 10.22 93.72 36.23 Debt Services 117 117.4 81.24 55.89 104.2 102.5 All Expenditures 134.1 305.6 276.6 21.5 130.6 57.87

PAGE 76

76 Table 34. Continued BY OBJECT (1) (2) (3) (4) (5) (6) Salaries 59.5 76.54 44.86 64.64 1.481 29.54 Employee Benefits 10.02 8.084 17.32 9.308 7.966 11.89 Purchased Services 3.821 35.6 37.41 32.74 52.84* 58.37 Energy Services 1.599 1.91 1.594 2.517 1.968 3.219 Materials and Supplies 8.614 5.192 9.607 0.461 10.31 7.009 Capital Outlays (Non Capitalized) 53.99 56.57 52 19.59 35.32 11.37 Capital Outlays (Capitalized) 11.75 176.7 206.2 10.22 93.72 36.32 Debt Services 117 117.4 81.24 55.89 104.2 102.5 Other 3.485 12.00* 5.116 0.483 5.964 9.666 Total Expenditures 134.1 305.6 276.6 21.5 130.6 57.87 Number of Districts 67 34 25 25 10 10 Weighted by Number of Students NO NO NO YES NO YES NOTE: Data represent the coefficient for Unitary Status obtained by regressing different types of expenditures on whether uni tary status has been received. Other controls include year and district fixed effects, the number of schools in the district, the size of the tax roll per capita, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, these coefficients are not shown. Asterisks represent increasing levels of stati stical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted by the number of unweighted full time equivalents in the dist rict. Columns 5 and 6 follow the same procedure but only include the ten largest districts in the state, excluding Miami Dade.

PAGE 77

77 Figure 31. Floridas s chool d esegregation over cases over t ime. S hows data for all 67 districts in Florida. Some displacement of points on the graph is caused by adding random variation to points which might otherwise lie on top of each other.

PAGE 78

78 Figure 32. Floridas u nitary status r eceipts over t ime. S hows data for the 34 districts that received desegregation orders. Some displacement of points on the graph is caused by adding random variation to points which might otherwise lie on top of each other.

PAGE 79

79 CHAPTER 4 A MULTI STATE ANALYSIS OF UNITARY STATUS AND SCHOOL FINANCE The results in Chapter 3 suggest that economies of scale may be a reason that only large districts in Florida seek and receive unitary status. Although Florida is representative of the entire country as a whole in some respects it is unwise to extend those results to a national level without first checking to see how school districts in other states also react to the granting of unitary status. Therefore, the analysis in Chapter 3 will now be extended to include six other states. One key difference with Chapter 3 is t hat the analysis in this chapter will use a different source of data for the fiscal data. In the previous chapter, the fiscal data was reported by the state of Florida for all of its districts. In this chapter, the data are reported by each district directly to the U.S. Department of Educations, National Center for Education Statistics (NCES). The data are qualitatively similar, but use of the NCES data will allow the inclusion of more states into the analysis. Description of the Data Litigation data for the multistate analysis also comes from the U.S. Commission on Civil Rights (USCCR) report titled Becoming Less Separate? School Desegregation, Justice Department Enforcement and the Pursuit of Unitary Status In addition to data on Florida, it also provides unitary status data on Alabama, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. When combined it provides information on over 780 districts across these seven states. Table 41 shows the breakout by litigation status for all the districts in eac h state. Over time, more than 61 percent (479 of 780) of the districts in these states were taken to court. Of those that faced court ordered desegregation, 41 percent have recei ved

PAGE 80

80 unitary status (195 of 479) .1 As in Florida, there were two distinct periods during which unitary status grants were receive d Figure 4 1 shows the relationship between the size of districts and the timing of their unitary status grants for all the districts in the sample ever taken to court It shows clearly that few districts received unitary status in the late 1980s and early 1990s. This creates two groups of unitary status grantees and acts as a natural break in the data. Figures 42 through 48 show the timing of unitary status by size for each s tate. Two things are clear from these figures. First, the bimodal nature of unitary status grants is uniform to all states. They each appear to experience a pause in the granting of unitary status around the year 1990. Secondly, it appears that Florida is an anomaly. The other six states have both large and small districts receiving unitary status. Each has a fairly even distribution of districts above and below the median for districts ever taken to court. Floridas distribution is skewed towards only large districts with only one unitary grantee being smaller than the median sized district.2 1 This data is current as of the reports 2007 publication. Since that time, another 38 districts across all seven states have received uni tary status. Since the NCES fiscal data only covers through the 20062007 school year, these recent grantees change in status can be ignored. For more details on these districts, see Holley Walker (2010). This difference across states may imply that school district finances are affected by unitary status differently as well. Therefore, further analysis of the m ultistate fiscal data is warranted. If fiscal reasons explain why districts seek unitary status, these changes are most likely to be identifiable during the latter period of unitary status grants. As shown in Table 41, more than 70 percent of all unitar y status grants occur during this latter period. Any potential fiscal effects become very important once the size of 2 This finding draws into question the economies of scale argument established in the previous chapter. If it were to hold uniformly, small districts in other states would not be receiving unitary status. Since they do, we can most likely rule out economies of scale as a nationwide reason why districts seek unitary status.

PAGE 81

81 these districts is considered. Although these 145 districts represent only 18 percent of all districts in the sample, they account for n early onethird of all s tudents in these seven states. The fiscal data for this analysis come from the NCES, Common Core of Data, Local Education Agency (School District) Finance (F33) Survey. This dataset provides broad measures of district level revenues and expenditures for all the school districts in the country. This study uses 15 years of data going back as far as the 1990 fiscal year .3 In that year, the survey was conducted with all districts in the country. It then alternated in some years betw een population and sample survey s. This study includes only years for which data is available from all district s in all seven states and ignores the years in which only a sample of districts is provided.4 3 The 1990 fiscal year corresponds to the 19891990 school year. Except for this section when fiscal years are explicitly stated, years listed in this study apply to the start of the school year. The raw finance data for the seven states contains 13,304 observations across all 15 years This total includes extraordinary education agencies such as juvenile justice programs, vocational programs, hospital/homebound programs and in North Carolina, charter schools. These types of schools account for just over 10 percent of all observations and were excluded from the analysis. An extremely small number of observations were also affected by consolidation of independent school district s. In all of these cases, none of the independent city districts had been taken to court. Since they would not be eligible for treatment to unitary status, there is no harm in excluding them from the analysis. There are also several cases in South Caroli na and Alabama where new districts were formed from school districts already under court supervision. The USCCR considered 4 Samples were drawn in the 1991, 19 93, and 1994 fiscal years.

PAGE 82

82 these districts as having never been taken to court and that notation is continued throughout this study. Upon applying these exclusions, the final data set contains 11,649 observations across fifteen years. The natural break in unitary status receipts helps define treatment and control groups based on schools being eligible for treatment in 1989. This break eliminates districts that were never taken to court and those that received unitary status prior to 1989 from the sample. This leaves 429 districts across fifteen years and totals to 7,185 district year observations in the panel. The means for all of the summary fiscal variables are provided in Table 42.5 Data are broken out by revenues and expenditures. M eans are provided in terms of per student funding levels in constant 2000 dollars.6 Total revenues and expenditures per student are nearly one quarter lower in Mississippi t han in Florida. In addition, to the differences in funding across states, there is also a tremendous size difference between states. Although not reported on this table, t he average district in Florida is nearly three times as large as the next largest state in the sample and nearly ten times the size of the smallest. Although some of the difference is eliminated by comparing median district size even that difference is striking. The median district in Florida is still twice the size of the next closes t state and five times the size of the median district in the smallest state.7 5 These summary variables are combinations of other subcategories. For brevity, only the toplevel categories are listed in Table 42. For means on all potential fiscal variables, see Table A5 in the appendix. Revenues are fur ther broken out by source and purpose. Expenditures are broken down into the objects on which the funds are spent. The differences across states make it clear that any model attempting 6 The nominal data were deflated using the CPI U for 2000 as a base year. 7 This difference may be enough to explain Florida as an anomaly.

PAGE 83

83 to estimate the effect of unitary status on fiscal variables will have to account for the large amount of variation between states or treat each state separately. Description of the Models The models used in this chapter are very similar to the models used in the previous chapters. The first specification used will take the following form: it st it it it Status Unitary y (4 1) Here, yit will take the value of various per student fiscal measures at the district level in each year itStatus Unitary measures the legal status of each district. If the district has never been taken to court or has been taken to court, but has yet to receive unitary status, the district receives a zero for that year. Once the district receives unitary status, this variable takes on a value of one in that and all subsequent years. itcaptures the timevariant characteristics for each district. These include the size of the student population, the number of teachers assigned to the district, the percent of students that are eligible to receive a free or reduced price lunch, and the percentage of students in the district that are reported as nonwhite.8 st represents a stateyear fixed effect which captures all stateyear specific variation. This strategy is worthwhile because it will pick up any variation caused by unobserved policy changes wh ich affect all the districts in one state in the same manner. The most likely example of this type of policy would be accounting changes which may affect how the data are reported from one year to the next. As long as the change is uniform across districts in a given state, 8 The number of teachers is endogenous. Alternate specifications are also run where the number of FTE instructors is omitted from matrix of time varying characteristics. This change does n ot appreciably change the results discussed in later sections.

PAGE 84

84 this state year fixed effect will account for the change. For this specification to be identified the error term, it must meet the following condition: 0 ) | ( it st it itStatus Unitary E (4 2) This condition is likely to hold since the noise in the judicial process makes the actual timing of release quasi random.9Care must be taken when interpreting the results generated by Equation 41. This specification treats all districts the same as to whether they have been taken to court. Since t he receipt of unitary status is conditional on having been tak en to court, any unobserved differences between districts never taken to court and thos e currently under court ordered desegregation plan s may bias the results. To account for this, the specification will be run using the entire sample and a sample containing only those districts that are eligible for treatment during the latter period of unitary status grants The error term is clustered at the district level to correct for heteroskedasticity. The final specification runs separate regressions for each state in the sample. In this instance a state year fixed effect cannot be implemented. D istrict year fixed effect s are also impossible due t o a cell size of one district obs ervation per year. Therefore, this specification uses year and district fixed effects. This makes the specification: it i t it it itStatus Unitary y (4 3) In Equation 43, ity, itStatus Unitary and itare the same as in Equation 41. The key difference is that the stateyear fixed has been replaced by a year fixed effect, t, and a district fixed effect, i. This specification is also likely to be identified due to the 9 Both Chapter 2 and Chapter 3 provide greater detail about the process involved in getting the unitary status grant and how the court makes the end of the process sufficiently random for ident ification.

PAGE 85

85 randomness of the judicial process The results generated by this specification will be important in determining whether there are any general trends in the fiscal data associated with unitary status. Not only should the results from Equation 41 be significa nt, but the results from Equation 43 should also show significant results across many of the states if unitary status has wide ranging effects.10The Effect of Unitary Status on Revenues The results from using Equations 41 and 43 are broken up into sever al tables Table 43 highlights the overall effect of unitary status on total revenues and expenditures. In terms of revenues, there is no clear indication of a uniform effect. Although Column 1 shows total revenue being $166 lower in districts that re ceive unitary status, the effect goes away when the control group is trimmed to include only those districts eligible for treatment as of 1990. This suggests that the result found in Column 1 is driven more by districts that have never been taken to court and/or those that received unitary status as an early grantee. It should also be noted that the results are not uniform in terms of sign and magnitude across states. This suggests that there is no clear, bottom line effect of unitary status on revenues, regardless of source. The results in Table 43 may not provide enough detail to accurately identify changes caused by unitary status. It may be possible that there are changes to subcategories of funding which offset each other in the total revenue figur es. Tables 44 through 46 present revenue data in greater detail by source. These tables show that ther e is little consistent effect on revenues from the state and federal levels. The exception may be in terms of local revenues where there is very modest evidence of 10 These statespecific regressions will also utilize the smaller sample of eligible districts due to the conditionality of unitary status on having been previously taken to court.

PAGE 86

86 some weak changes in revenues. In Table 46, local r evenues generated by District Activity Receipts show a fairly unif orm sign patter across specifications, but only reach statistically significant levels in two of the seven specifications.11 Overall, the economic significance of this result is questionable. In Alabama, average revenue per student is $6,400. This $16 change represents about a 0.2 percentage point drop in revenues. There is also a relatively consistent sign pattern in miscellaneous revenues at the local level.12The Effect of Unitary Status on Expenditures Eight of the nine specifications have a negative sign, but only two can be considered different from zero. These decreases are also minor in terms of economic significance. This lack of statistically significa nt results appears to suggest that unitary status does not appear to affect revenues in any general way The effect of unitary status on expenditures shows one interesting result that appears to be consistent through much of the state by state data. Table 43 shows a drop in Total Current Expenditures (TCE) of about $147 per student. It appears that the bulk of this drop occurs in TCE for Elementary Education which shows a drop of about $174 per st udent. Considering average total expenditure per student is about $6900 in each distr ict this represents a decrease of about 1.5 percent. This result is also significant in three of the nine specifications and has a consistent sign pattern in five of the remaining six specifications. Much like the revenue data, these categories are summations of other subcategories which also may provide relevant information. 11 It is not clear how the NCES defines District Activity Receipts The NCES Codebook does not include a definition and the district survey form only says, Gross district activity receipts for those funds under control of the custodian of district funds should be included on line 13. These two factors make it less lik ely that this is an important change in per pupil funding levels. 12 Again, it is unclear what constitutes a miscellaneous source of funding at the local level. In fiscal year 2007, the average district had $698,000 in miscellaneous funding or about $72 per student.

PAGE 87

87 Table 47 shows that these results are traceable through other parts of the dataset. TCE on Ins truction appears to show a statistically significant drop of about $101 per student. This result is not only significant for the entire sample, but is also significant for the reduced sample of unitary status eligible districts and one state. There is al so a relatively consistent trend in sign and magnitude among five of the remaining six states. This category is itself a summation with the largest component being teacher salaries .13 Table 48 shows how salaries have been affected by changes in unitary s tatus. It shows a statistically significant drop in teacher salaries in the same specifications and accounts for about 70 percent of the drop in TCE on Instruction per student .14A drop is also notable in benefits paid to instructors. This effect is shown in Table 4 9 which shows about a $20 per student drop in the benefits paid to instructors. However, some of the robustness is lost as it is statistically significant in only one specification. 15Thi s is a rather significant trend which suggests that districts and/or teachers may be reacting to unitary status in a somewhat uniform way. When combined, the drop in salaries and benefits for instruc tional staff appears to account for 90 percent of t he drop in TCE on Instruction. 16 13 Salaries accounts for nearly 72% percent of all expenditures on instruction in each district. One possible explanation for this trend in salaries and benefits is that the number of instructors in each district is 14 These results are obtained while using the number of FTE Instructors in the district as a timevarying control. Since the number of instructors is endogenous, the specifications were also run omitting this variable as a control. The results for TCE on I nstruction and Salaries for Instructors were similar in statistical significance and magnitude and are available upon request. 15 The sign and magnitude pattern is maintained across the other statelevel specifications. 16 Salaries and benefits paid for General Administration also show some downward trends as a result of unitary status. However, the magnitude is smaller and there is less of a consistent pattern across specifications.

PAGE 88

88 decreasing as a result of receiving unitary status. The current dataset provides some basic information about the number of FTEs in each district. Therefore, it is possible to regress the number of FTEs on whether unitary status has been received and other district level covariates. It is also possible to construct a crude class size measure based on the number of students in the district per FTE. Using each of thes e as a dependent variable in Equations 41 and 43 tests the extent to which the number of FTEs is affected by unitary status .17There are other possible explanations for this impact on salaries and benefits. These results are presented in Table 410 The results show that there is little consistent effect on the number of FTEs or t he number of students per FTE. This suggests that the number of FTE instructor s does not change in any uniform manner when unitary status is received. 18 First, it may be possi ble that the pool of instructors is becoming younger less experienced, or less qualified on average over time as a result of unitary status This might occur if older teachers are more likely to leave a district after it becomes unitary and are replaced by younger, less experienced recruits. Since teacher salaries are relatively formulaic, this would allow the number of FTE instructors to remain constant while the average district level profile is decreasing in age, experience, and/or quality That woul d result in lower salaries on average in unitary district s.19 17 In each of these specifications, the number of FTEs is removed from the set of time varying district level covariates contained in it. 18 Unfortunately, given the current, limited data they are untestable. 19 It might be possible that some of the potential change in the district profile is caused by the reti ring of the baby boom generation. If there is a difference in how this phenomenon affects, say, rural versus urban districts, Unitary Status could be biased since only large districts are seeking a change in their status. In this case, the coefficient on unitary status could just be picking up the contemporaneous effect of replacing the retirees with younger teachers.

PAGE 89

89 Recent work by Feng, Figlio and Sass (2010) highlights another possible contemporaneous effect. Their work suggests that teachers move in response to increased or decreased accountability measu res. Teachers facing tougher accountability measures are more likely to leave their current school while those facing lower accountability measures are more likely to stay. They also find some evidence that changes in accountability can affect the distribution of teacher quality as measured in certain valueadded measures To the extent that many unitary status grants were received during the time of increased emphasis on school accountability measures, the coefficient on itStatus Unitary may be picking up some of the effect caused by changes in accountability measures across districts. Considering that the exogenous shock the authors use to identify teacher mobility occur s at the school level, it is unlikely that a state year or combinati on of y ear level and district level fixed effect s would account for this type of variation. Therefore, the increased accountability measures could be producing bias. One thing is important to note. If the effect on per student instruction and salary exp enditures is also capturing the effect of accountability and teacher mo bility that movement must be altering the salary and benefits profile in unitary districts Although there are many ways this type of change may be occurring, there is one way that is certainly not responsible and that is withindistrict movement of teachers. Given the current level of analysis, this type of movement would result in a net effect of zero at the district level.20 20 This assumes that teachers do not accept lower salaries when they move between schools. Unfortunately, Feng et al. does not discuss the effect of accountability on the distribution of experience. Therefore, any relationship is just conjecture at this point.

PAGE 90

90 Lower salaries in unitary districts could also be a sign t hat a compensating wage differential is paid to teachers that work in districts under a court ordered desegregation plan. Since districts that are still under such plans have very high levels of integration, white teachers may need to be induced by a higher salary to teach in what are likely to be multiracial classes.21Revisiting Economies of Scale and Final Thoughts When the desegregation plan is lifted and schools become more segregated as discussed in Chapter 2, white teachers may be able to teach in classrooms more to their liking for a lower salary Martin (2010) has found evidence of compensating wage differentials in districts that have higher concentrations of minority students She also found that more highly segregated districts pay lower wages, all else equal. This studys results that sa la ries will be lower in districts treated to unitary status, are consistent with Martins findings as well. The results in this chapter make it unlikely that districts seek unitary status in order to take adv antage of any perceived fiscal benefits. There does not appear to be much of a response in terms of revenues at the local, state, or federal levels. There are some uniform results which suggest teacher salaries and benefits are affected by the receipt of unitary status. However, given the current data, it is not possible to determine exactly how and why the decrease in expenditures occurs. It is also hard to construct a st ory where districts use a prior i expectations about expenditures on salaries and benefits as a justification to seek unitary status. It is more likely that any changes to teacher salaries occur unexpectedly or as a result of a confluence of factors. 21 This would be analogous to teachers leaving teaching all together at the beginning of the desegregation era. Some of those that st ayed may have stayed only because they were paid a higher salary.

PAGE 91

91 It i s also important to note that the weak findings from Chapter 3 which suggested that the burden on local tax payers could be reduced by seeking unitary status do not carry over to data from a different source. Table 46 shows no statistically significant e ffect of unitary status on local property tax revenues. Since this was the potential source of fiscal gains for unitary districts in Florida, it makes the existence of economies of scale unlikely. This is further compounded by small districts in other st ates receiving unitary status when economies of scale would seem to suggest that they shouldnt be seeking unitary status in the first place

PAGE 92

92 Table 41. Court o rdered d esegregation status by s tate Never Taken to Court Still under Court Order Granted Unitary Status Prior to 1990 Granted Unitary Status After 1990 Total Alabama 6 53 20 52 131 Florida 33 15 8 11 67 Georgia 71 76 3 28 178 Louisiana 9 43 5 11 68 Mississippi 53 69 4 20 146 North Carolina 77 15 8 7 107 South Carolina 52 13 2 16 83 Total 301 284 50 145 780 % of Students Affected 28.9 26.6 12.8 31.7 100 Note: Districts counts based on districts in existence in 2007. With the exception of Florida, each state has a slight chang e in the number of districts over time due to the consolidation or addition of independent city school districts. Counting the number of districts on either side of 1990 is not problematic because no districts in the sample receive unitary status in 1990.

PAGE 93

93 Table 42. Summary f iscal data by s tate. AL FL GA LA MS NC SC ALL (1) (2) (3) (4) (5) (6) (7) (8) Total Federal Revenue 702 722 762 1,081 997 662 775 808 Total State Revenue 3,925 3,977 4,194 3,554 3,234 4,677 3,820 3,923 Total Local Revenue 1,784 3,000 2,515 2,378 1,600 1,895 2,768 2,193 Total Revenue 6,411 7,698 7,472 7,013 5,830 7,234 7,362 6,923 Total Current Expenditures for Elementary Education 5,703 6,186 6,463 6,157 5,230 6,316 6,316 6,020 Total Non Elementary/Secondary Expenditures 121 162 20 30 23 44 96 62 Total Capital Outlay Expenditures 576 1,178 844 528 541 711 867 729 Payments to State Governments 0 0 0 0 0 0 9 1 Payments to Local Governments 0 0 0 0 0 0 0 0 Payments to Other School Systems 5 0 33 9 0 0 35 13 Interest on Debt 90 115 79 123 84 118 180 105 Total Current Expenditures: Services 6,495 7,642 7,439 6,847 5,878 7,189 7,503 6,928 Note: Data are in constant 2000 dollars per student and were deflated using the national CPI U. Means were calculated with unweighted district level data for each state. Due to the formulaic nature of school finance, weighting does not appreciably change any of the data.

PAGE 94

94 Table 43. The effect of u nitary s tatus on t otal r evenues and t otal e xpenditures per student (1) (2) (3) (4) (5) (6) (7) (8) (9) REVENUES ALL ALL ELIGIBLE AL FL GA LA MS NC SC Total Federal Revenue 2.01 56.27 36.99* 38.64 27.99 865.20 89.82 14.49 69.29 Total State Revenue 58.06 40.80 17.85 41.90 102.00 266.80 47.44 4.33 63.69 Total Local Revenue 105.70 93.07 12.59 26.47 206.60 248.50 1.25 99.53 65.15 Total Revenue 165.70* 77.60 42.25 54.07 336.50 847.00 136.00 89.37 59.56 EXPENDITURES Total Current Expenditures for Elementary Education 173.40*** 134.60* 4.23 143.20* 180.30 114.20 11.47 97.29 108.00 Total Non Elementary/Secondary Expenditures 5.87 13.38* 4.75 25.37 1.37 5.55 3.20 9.69 6.43 Total Capital Outlay Expenditures 29.96 56.51 66.45 -66.70 250.30* 172.70 18.22 123.80 316.00 Payments to State Governments -0.21 -0.05 . -0.07 Payments to Local Governments 0.11 . -0.04 Payments to Other School Systems -0.31 1.39 -0.79 -1.90 8.13 -15.77 Interest on Debt 3.14 -2.10 3.40 3.26 6.68 -38.41 19.13 61.64 4.70 Total Current Expenditures 146.60* 92.24 69.58 181.30 424.50 22.61 22.68 273.10 203.30 Note: Data represent the coefficient for Unitary Status obtained by regressing different fiscal variables on whether unitary status has been received. C ontrols include either a state year fixed effect or fixed effects by year and district, the number of students in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches a nd the percent of the student population that is nonwhite. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts i n these seven stat es that have been subject to court order and had not received unitary status by 1989. Columns 3-9 show results for each state individually.

PAGE 95

95 Table 44. The e ffect of u nitary s tatus on f ederal r evenues per student. (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC Fed Thru State: Child Nutrition Act 4.32 2.34 4.24 4.24 1.12 9.50 12.17** 5.28 6.62 Fed Thru State: Title I 2.57 1.91 4.47 4.48 12.94 14.72 5.83 4.63 57.19** Fed Thru State: Child with Disabilities Idea 0.16 0.87 0.75 4.90 4.19 10.63 6.64 0.75 0.39 Fed Thru State: Math, Science, and Teacher 0.96 1.05 1.76 2.48 4.957** 5.42 4.32 0.56 7.38 Fed Thru State: Safe and Drub Free Schools 0.87 0.91 2.21 0.01 0.06 0.65 0.01 2.94 0.18 Fed Thru State: Title V, Part A 0.36 0.45 0.07 4.56 0.04 0.46 2.16 0.91 0.90 Fed Thru State: Vocational and Tech Education 0.39 0.84 1.29 6.18** 0.73 2.15 3.17 0.55 3.42 Fed Thru State: Other 9.02 46.26 27.40*** 16.61 3.27 1,086.00 98.78 16.15 54.34 Federal Revenue Nonspecified 2.84 3.11 7.16 18.21 23.49 Fed Direct: Impact Aid 1.25 6.95 5.06 0.06 1.69 3.87 4.01 1.12*** 0.94 Fed Direct: Bilingual Education 0.33 0.23 0.23 1.97 0.48** 0.37 0.31 0.89 0.05 Fed Direct: Indian Education 0.51 0.80 0.66 0.01 0.06 0.06 0.38 Fed Direct: Other 2.22 7.11 2.98 18.48** 13.29 73.57 11.48 32.45 0.29 Total Federal Revenue 2.01 56.27 36.99* 38.64 27.99 865.20 89.82 14.49 69.29 Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. C ontrols include either a stateyear fixed effect or fixed effects by year and district, the number s tudents in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the perc ent of the student population that is nonwhite. For brevity, the coefficients for these co ntrols are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 s hows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 39 show results for each state individually.

PAGE 96

96 Table 45. The e ffect of u nitary s tatus on state r evenues per student. (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC State Rev : General Formula Assistance 37.24 21.72 23.12 53.59 18.33 234.00 12.87 17.13 5.18 State Rev : Special Education Programs 0.79 5.66 0.84 41.14 0.00 2.78 0.03 6.65 3.48 State Rev. : Transportation Programs 9.36 0.15 13.42 6.07 0.79 2.33 State Rev : Staff Improvement Programs 3.72 0.09 0.25 0.57 8.12 8.02 State Rev : Compensatory and Basic Skills Programs 0.04 1.67 4.84 33.26 0.02 12.29 1.77 4.32 State Rev : Vocational Education Programs 1.54 4.55 1.58 4.01 0.28 3.16 5.09 19.90 State Rev : Capital Outlay And Debt Service Programs 3.62 0.27 4.13 1.54 78.71 6.17*** 17.99 82.69 State Rev : Bilingual Education Programs 4.74 4.25 0.16 15.24 0.27 State Rev : Gifted and Talented Programs 3.71 0.22 4.87 0.44 -2.67 State Rev : School Lunch Programs 0.48** 0.46 0.19 0.23 1.26 0.48** 1.98 0.39 State Rev : Other Programs 17.01 5.50 10.91 3.70 24.55 15.49 30.38* 24.05 0.83 State Rev On Behalf: Employee Benefits 1.00 0.31 3.09* 4.17 1.13 2.23 State Rev On Behalf: Not Employee Benefits 0.82 0.10 0.14 0.00 0.24 2.43 5.21 State Rev : Not Specified 6.91 8.84 31.85* . Total State Revenue 58.06 40.80 17.85 41.90 102.00 266.80 47.44 4.33 63.69 Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. C ontrols include either a state year fixed effect or fixed effects by year and district, the number s tudents in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the perc ent of the student population that is nonwhite. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 s hows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 39 show results for each state individually.

PAGE 97

97 Table 46. The e ffect of u nitary s tatus on local r evenues per student. (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC Local: Tuition Fees from Pupils and Parents 1.67 1.48 0.37 0.09 8.70 4.85 0.69 7.21 2.28 Local : Transportation Fees from Pupils and Parents 0.20 0.16 0.18 0.25 0.04 0.95 Local : Textbook Sales and Rentals 0.00 0.05 0.00 1.35 0.00 . Local : District Activity Receipts 5.60 15.76** 17.93** 34.88 5.24 3.85 8.92 Local : Other Sales and Services 0.79 3.18 0.96 10.26 6.117*** 6.84 8.72 3.24 Local : Student Fees, Nonspecified 0.32 0.20 0.02 . Local : Property Taxes 24.18 15.54 21.08 38.13 36.54 176.7* 18.26 125.80 Local : General Sales Taxes 28.97 19.70 4.40 34.52 Local: All Other Taxes 2.38 4.71 16.47 0.28 12.32 16.79 Local: From Other School Systems 1.47 0.40 2.64 0.23 0.98 0.03 18.24 0.59 1.33 Local: From Cities and Counties 21.54 16.15 67.88 56.15 4.15 16.16* 17.46 Local: School Lunch 8.81*** 2.87 2.36 6.19 2.92 3.16 1.12 3.06 0.28 Local: Interest Earnings 7.61 -5.69 1.68 -3.60 -18.72 -35.85 4.56 4.65 -0.58 Local: Miscellaneous 31.80** 39.38 60.63 54.85 93.04* 70.36 7.77 40.07 16.26 Total Local Revenue 105.70 93.07 12.59 26.47 206.60 248.50 1.25 99.53 65.15 Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. C ontrols include either a state year fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime equivale nt teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the resul ts from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 39 show results for each state individually.

PAGE 98

98 Table 47. The effect of u nitary status on t otal current e xpenditures (TCE) per student. (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC TCE: Instruction 101.20*** 84.39** 3.18 104.90** 120.30 62.71 89.16 34.08 55.54 TCE: Support Services 60.29** 43.08 2.70 37.48 2.44 30.48 77.95 126.30 49.10 TCE: Other (Elementary/Secondary) 11.61* 7.07 4.98 0.85 60.30 19.38 0.26 5.03 3.39 T CE for Elem /Secondary Education 173.40*** 134.60* 4.23 143.20* 180.30 114.20 11.47 97.29 108.00 TCE: Support Services (Pupils) 0.56 5.54 5.35 0.97 13.38 23.55 5.90 23.58 1.78 TCE: Support Services (Instructional Staff) 0.24 1.82 9.65 17.38 22.71 31.53 29.06 18.78 55.20 TCE: Support Services (General Admin) -27.21*** -11.01 1.80 5.99 15.36 29.40 -11.24 -12.90 20.04 TCE: Support Services (School Admin) 8.45 17.81** 3.74 23.26** 7.78 24.45 6.90 21.62 5.15 CE Support Services Operation and Main of Plant 14.48 13.16 14.62 0.44 2.83 100.90 77.76 24.52 24.08 CE Support Services Student Transportation 2.70 5.10 5.44 2.91 3.61 2.15 6.32 17.62 18.43 CE Support Services Nonspecified 0.44 0.27 0.77 . TCE: Support Services 60.29** 43.08 2.70 37.48 2.44 30.48 77.95 126.30 49.10 Note: Data represent the coefficient for Unitary Status obtained by regressing different sources of revenue on whether unitary status has been received. C ontrols include either a stateyear fixed effect or fixed effects by year and district, the number s tudents in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that i s nonwhite. For brevity, the coefficients for these co ntrols are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 3 9 show results for each state individually.

PAGE 99

99 Table 47. Continued TCE: Food Services 13.01** 4.71 3.10 0.85 3.78 13.85 5.24 5.03 13.61 CE -Enterprise Operations 0.79 -1.46 -8.23 -0.04 -0.30 -8.75 CE ELSEC 0.90 0.54 1.00 3.94 5.36 4.69 4.33 TCE: Other (Elementary/Secondary) 11.61* 7.07 4.98 0.85 60.30 19.38 0.26 5.03 3.39 Non Elem /Secondary Expenditures: Comm Services 0.62 1.71 4.15 11.04 1.84 1.03 1.58 9.40 12.06 Non Elem /Secondary Expenditures: Adult Education 4.44 6.05 4.74 12.24 2.81 5.95 0.65 0.31 8.94 Non Elem /Secondary Expenditures: Other 1.48 5.46 0.66 1.24 4.31 T CE: Non Elementary /Secondary 5.87 13.38* 4.75 25.37 1.37 5.55 3.20 9.69 6.43 Capital Outlay: Construction 45.10 55.24 74.06 72.28 292.30** 162.30 3.90 139.40 407.90 Capital Outlay: Land and Existing Structures 5.83 0.75 5.71 5.27 56.33* 7.71 11.57 48.10 Capital Outlay: Instructional Equipment -1.71 0.67 -2.90 2.15** -13.00 -8.71 -2.20 2.75 9.65 Capital Outlay: Other Equipment 4.27 0.85 2.92 5.55 1.51 36.98 3.76 11.07 28.31 Capital Outlay: Nonspecified Equipment 3.37 0.48 4.05 1.24 24.20* 21.13 Total Capital Outlays 29.96 56.51 66.45 66.70 250.30* 172.70 18.22 123.80 316.00

PAGE 100

100 Table 48. The e ffect of u nitary status on s alaries (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC Salaries Support Services Pupils 1.20 1.96 1.77 0.05 7.28 1.73 7.65 14.85 18.37* Salaries Support Services Instructional Staff 1.97 0.78 2.52 20.01 20.94* 18.14 3.61 15.00* 9.10 Salaries -Support Services-General Administration -14.32*** -5.53 1.68 1.90 14.72 5.15 -5.73 -8.22 6.43 Salaries Support Services School Administration 3.82 11.15* 6.22 18.94** 3.85 12.27 5.94 20.28** 1.63 Salaries Support Services Oper and Main of P lant 5.12 7.79 1.56 28.07** 11.79* 33.33** 4.37 7.60 2.82 Salaries Support Services Student Transportation 0.77 2.41 5.62 0.10 2.57 3.26 12.34 5.75 7.16 Salaries Food Services 4.99** 5.72** 0.72 3.96 2.77 14.25 6.66 0.03 4.84 Salaries: Instruction -70.33*** -72.75*** -5.56 -88.47** -82.40 -99.80* -70.84 -17.90 -57.99 Total Salaries 104.20*** 108.60** 19.49 161.60** 58.17 180.20* 98.40 45.83 88.54 Note: Data represent the coefficient for Unitary Status obtained by regressing different salary expenditures per student on whether unitary status has been received. C ontrols include either a stateyear fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (* *) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisi ana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 39 show results for each state individually.

PAGE 101

101 Table 49. The e ffect of u nitary s tatus on b enefits (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC Employee Benefits Support Services Pupils 0.37 1.20 1.02 2.17 1.33 2.44 1.40 2.94 3.83 Employee Benefits Support Services Instructional Staff 0.42 0.90 0.44 3.48 7.42** 4.67 1.14 1.96 2.75 Employee Benefits Support Services General Administration 3.24*** 1.16 1.33 4.19 2.44 0.60 1.34 2.04 2.45 Employee Benefits Support Services School Administration 1.67 3.05** 0.06 3.12 0.18 4.26 1.10 3.88* 0.23 Employee Benefits -Support Services-Op and Maint .of P ant -0.60 -1.15 -0.92 1.35 5.73** 0.97 -0.93 1.10 -0.69 Employee Benefits Support Services Student Transportation 1.19 0.73 0.06 4.92 0.07 2.63 7.33 2.35 2.56 Employee Benefits: Food Service 0.72 0.64 2.70 5.58** 0.89 2.51 2.54 3.18 0.60 Employee Benefits: Enterprise Operations 0.01 0.00 0.00 0.00 0.04 0.01 0.10 0.00 0.07 Employee Benefits: Instruction 20.05** 8.50 1.29 16.67 7.69 30.71 16.34 3.63 17.47 Total Employee Benefits 25.67** 11.89 3.73 2.76 7.74 26.14 27.83 14.45 28.35 Note: Data represent the coefficient for Unitary Status obtained by regressing different types of employee benefits per student on whether unitary status has been received. C ontrols include either a stateyear fixed effect or fixed effects by year and district, the number students in the district, the number of fulltime equivalent teachers in the district, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, the coefficients for these controls are not shown. Asterisks represent increasing levels of statistical significance start ing at the 10 (*) percent level and increasing to the five (** ) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states t hat have been subject to court order and had not received unitary status by 1989. Columns 3 9 show results for each state individually.

PAGE 102

102 Table 410. The e ffect of u nitary status on teachers (1) (2) (3) (4) (5) (6) (7) (8) (9) ALL ALL ELIGIBLE AL FL GA LA MS NC SC Number of FTEs 50.95* 61.56* 14.84 221.9 24.87 5.301 4.908 16.36 5.793 Students per FTE 0.0677 0.102 0.0429 0.239 0.263 0.0964 0.109 0.658 0.149 Note: Data represent the coefficient for Unitary Status obtained by regressing different measures of the number of instructors in each district on whether unitary status has been received. C ontrols include either a stateyear fixed effect or fixed effects by year and district, the number students in the distr ict, the percent of students eligible for free or reduced price lunches and the percent of the student population that is nonwhite. For brevity, the coeffici e nts for these controls are not shown. Asterisks represent increasing levels of statistical sign ificance starting at the 10 (*) percent level and increasing to the five (* *) and one (***) percent levels Column 1 shows the results from using all districts from Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, and South Carolina. Column 2 restricts the sample to only those districts in these seven states that have been subject to court order and had not received unitary status by 1989. Columns 3 9 show results for each state individually.

PAGE 103

103 Figure 41. Timing of u nitary status g rants for e ligible d istricts ( all s tates)

PAGE 104

104 Figure 42. Timing of unitary s tatus g rants for e ligible d istricts by s ize: A labama.

PAGE 105

105 Figure 43. Timing of u nitary status grants for e ligible d istricts by s ize: Florida

PAGE 106

106 Figure 44 Timing of unitary status g rants for e ligible d istricts by s ize: Georgia.

PAGE 107

107 Figure 45 Timing of unitary status grants for eligible d istricts by s ize: Louisiana.

PAGE 108

108 Figure 46 Timing of unitary status grants for eligible d istricts by s ize: Mississippi

PAGE 109

109 Figure 47 Timing of u nitary status grants for e ligible d istricts by s ize: North Carolina.

PAGE 110

110 Figure 48 Timing of u nitary status grants for e ligible d istricts by s ize: South Carolina.

PAGE 111

111 CHAPTER 5 CONCLUSION After several decades of active involvement, the courts that largely helped desegregate southern schools are ending their involvement in the process. The granting of unitary status would appear, at first glance, to be a watershed moment that could unleash a large amount of change on the districts that receive it. Aside from the re sort ing of students within districts, this study suggests changes in these districts are muted. As far as Florida is concerned, it is clear that court ordered desegregation plans place some constraint on district assignment of students. As shown in Chapter 2, districts that are released from a court ordered desegregation plan see their level of racial integration fall and the level of segregation rise. This change does not happen instantaneously. It takes time for districts and parents to react, but over time the level of segregation can be expected to double within eight years of receiving unitary status. Such a drastic change would be a source of concern if it were accompanied by a decrease in academic outcomes Fortunately, no back sliding in academic achievement is occurring. Unitary status appears to have no effect on academic achievement as measured by scores of the average student, the average white student, and the a verage black student. To the extent that parental preferences are being better met by the new distribution of students across schools, these results suggest that unitary status may be a pareto improvement over the status quo. Parental preferences may not be the only reason why districts seek unitary status. If court ordered desegregation plans are placing constraints on the assignment of students, it is likely that there are also other constraints. One possible area where constraints could be an issue is in school district finance. If finances are further

PAGE 112

112 constrained by court orders than they otherwise would be, seeking unitary status to relieve that constraint would be an appropriate step for local school boards. Although the logic is enticing, there i s little evidence that this is the case for most districts receiving unitary status. In Chapters 3 and 4, there is little evidence that school district finances are affected by unitary status to an extent that makes changes in finances a smoking gun which fully explains the decision by officials to seek unitary status. This is contrary to the assertions of Ryan (1999) which suggest s that school districts agree to terminate mandatory desegregation plans in exchange for large payments from state level ed ucation agencies and Moore (2002) which suggest s that court ordered desegregation creates costs which must be borne by local communities. No such effect s are noted in this multistate analysis. If such bribes occur, they are the exception rather than the rule since there is no significant change in revenues caused by the receipt of unitary status.1 Additionally, it is hard to find where costs may be avoided by seeking unitary status. With the exception of teacher salaries and benefits, f ew expenditure c ategories changed as a result of exiting a court ordered desegregation plan.2There are several other reasons why unitary status might have no effect on revenues and expenditures. First, the era of school district finance reform altered the 1 Ryan provides examples from five specific districts in five states. None of these states are included in this analysis. 2 It is possible that changes in expenditures are occurring within district and within category so that no effect is identifiable. For example, assume that a district must use bussing to meet the intent of the courts desegregation plan. This is most likely more expensive than bussing students to their closer neighborhood schools. When unitary status is received and the district is relieved of this bussing requirement, expenses on student transportation should fall. The results in Table 47 show that there is no statistical effect on student transportation. This can mean two things. First, the bussing plan was not artificially raising expenditures or the district replaced the bussing plan with an alternative that was just as costly. Both would lead to the result reported in this paper.

PAGE 113

113 way finances are done by statelevel education agencies. Most school, district and state financing decisions are now extremely rules based. Since few, if any, of these rules take into account a districts desegregation status, it would be unlikely that a change in t hat status would affect financing levels. Another more troubling possibility is that the student re sort ing that is causing the increasing levels of segregation is a within district phenomenon. This might change school level finance measures as students move between schools. The analysis completed in this study cannot rule out this type of change since such changes will net out at the district level. There is a final, straightforward reason why unitary status may have no effect on school district financ es. Simply put, court ordered desegregation worked. If the court system did its job, students should be able to re sort into increasingly segregated schools with no change occurring in the districts finances. To some extent, such an occurrence is cons istent with the results of this paper. Although this study provides evidence that court ordered segregation plans were successful and districts can be released from them without disastrous results, it does not suggest that the era of monitoring segregati on and its effects should come to an end. There are several unanswered questions within this study which provide avenues for future research on the effects of unitary status. Additional data on student level movement and test scores would provide a true measure of how black students are being affected by unitary status. Too much can be occurring at the school, classroom, and student levels to make a district level average of black student performance the final word on academic achievement. Additionally, teacher level data would help solidify the effect unitary status has on teacher mobility and salaries. This would add to

PAGE 114

114 a robust literature on teacher mobility at the beginning of the school desegregation era. Answering these questions should continue to provide insight into what causes districts to seek release from their court ordered desegregation plans and the consequences of unitary status once they receive it.

PAGE 115

115 APPENDIX A ADDITIONAL TABLES Table A 1. Litigation status by district County Year Litigation Started Year Case Closed (Unitary Status Granted) Alachua 1964 1971 Baker 1970 Pending Bay 1966 Pending Bradford 1970 Pending Brevard 1966 1978 Broward 1970 1996 Columbia 1970 1987 Dade 1956 2001 Duval 1960 2001 Escambia 1960 2004 Flagler 1970 Pending Gadsden 1970 1986 Gulf 1970 Pending Hendry 1970 Pending Hillsborough 1958 2001 Indian River 1965 Pending Jackson 1970 Pending Jefferson 1970 Pending Lafayette 1970 Pending Lee 1964 1999 Leon 1962 1974 Manatee 1965 Pending Marion 1978 Pending Orange 1970 Pending Palm Beach 1956 1979 Pasco 1970 Pending Pinellas 1964 2000 Polk 1963 2000 St. Johns 1970 Pending St. Lucie 1970 1997 Sarasota 1963 1970 Seminole 1970 2006 Volusia 1960 1970 Wakulla 1970 Pending NOTE: The following districts have never been taken to court: Calhoun, Charlotte, Citrus, Clay, Collier, De Soto, Dixie, Franklin, Gilchrist, Glades, Hamilton, Hardee, Hern ando, Highlands, Holmes, Lake, Levy, Liberty, Madison, Martin, Monroe, Nassau, Okaloosa, Okeechobee, Osceola, Putnam, Santa Rosa, Sumter, Suwannee, Taylor, Union, Walton, Washington.

PAGE 116

116 Table A 2. Unitary status e ffect on total local, state and federal r evenue per student (1) (2) (3) (4) (5) (6) Unitary Status 114.80 76.78 151.50 6.29 37.14 71.30 (99.76) (122.30) (141.60) (127.20) (115.30) (156.50) Number of Schools 0.30 5.36 1.40 0.31 0.96 1.04 (4.07) (4.35) (4.27) (2.15) (3.22) (3.15) Tax Roll per Student 0.00314*** 0.0012 0.0010 0.00189*** 0.00158** 0.00142** (0.0011) (0.0008) (0.0008) (0.0005) (0.0005) (0.0005) Fraction FRPL Eligible 1509.00 315.10 533.90 1356.00 1292.00 1832.00 (1679) (1038) (965) (1261) (1755) (1724) Fraction Non White 3171.00 8600** 8419** 8259*** 11494*** 13386*** (3304) (3438) (3521) (1822) (2890) (2592) Constant 4560*** 3803*** 3437*** 1486 1570 251.3 (942) (1103) (1087) (1394) (1130) (1139) Weighted by FTEs NO NO NO YES NO YES Observations 670 340 250 250 100 100 R squared 0.655 0.721 0.669 0.894 0.929 0.925 NOTE: Data represent the results obtained by regressing the total amount of revenue received from all local, state, and federal sources per FTE on whether unitary status has been received. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The numbers in parentheses are robust standard errors. Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted by the number of unweighted full time equivalents in the district. Columns 5 and 6 follow the same procedure but only include the ten largest districts in the state, excluding Miami Dade.

PAGE 117

117 Table A 3. Unitary status e ffect on state funding from o ther sources (1) (2) (3) (4) (5) (6) Unitary Status 134.1** 85.86* 99.97* 84.18** 51.19 36.14 (58.99) (42.23) (51.94) (33.57) (35.24) (52.78) Number of Schools 0.17 2.031** 1.74 1.437*** 0.86 1.271* (1.63) (0.86) (1.04) (0.51) (0.73) (0.65) Tax Roll per Student 0.000150* 0.0000 0.0001 0.000327** 0.000524*** 0.000589*** (0.0001) (0.0001) (0.0001) (0.0002) (0.0002) (0.0001) Fraction FRPL Eligible 971.80 442.50 921.3** 1020.00 1276.00 1615.00 (749) (435) (347) (704) (993) (1054) Fraction Non White 1135.00 191.30 170.20 1053* 1916** 2450** (1366) (922) (994) (570) (834) (975) Constant 328.7* 230.7 40.12 643.4 487.3 720.5 (197) (336) (339) (425) (451) (551) Weighted by FTEs NO NO NO YES NO YES Observations 670 340 250 250 100 100 R squared 0.298 0.507 0.498 0.63 0.658 0.656 NOTE: Data represent the results obtained by regressing the amount of revenue received from all "other" state level sources per FTE on whether unitary status has been received. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The numbers in parentheses are robust standard errors. Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only thos e districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted by the number of unweighted full time equivalents in the district. Columns 5 and 6 follow the same procedure but only include the ten largest districts in the state, excluding Miami Dade.

PAGE 118

118 Table A 4. Unitary status e ffect on general a dministration e xpenses (1) (2) (3) (4) (5) (6) Unitary Status 5.46 4.25 6.15 10.09*** 7.644** 4.44 (3.86) (3.97) (5.77) (3.49) (2.97) (3.51) Number of Schools 0.270* 0.19 0.22 0.02 0.07 0.09 (0.16) (0.18) (0.22) (0.06) (0.10) (0.10) Tax Roll per Student 1.77e 05* 0.0000 0.0000 0.00 0.00 0.00 (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) Fraction FRPL Eligible 4.58 8.16 4.16 20.91 43.60 39.50 (26) (30) (38) (47) (63) (73) Fraction Non White 47.12 55.69 47.76 20.96 107.20 230.8* (142) (136) (155) (89) (106) (111) Constant 61.11 49.08 61.86 56.81 83.65** 125.5** (40) (38) (42) (46) (37) (44) Weighted by FTEs NO NO NO YES NO YES Observations 670 340 250 250 100 100 R squared 0.863 0.885 0.88 0.725 0.787 0.751 NOTE: Data represent the results obtained by regressing the level of expenditure on General Administration per FTE on whether unitary status has been received. Asterisks represent increasing levels of statistical significance starting at the 10 (*) percent level and increasing to the five (**) and one (***) percent levels The numbers in parentheses are robust standard errors. Column 1 shows the results from a sample of all Florida districts. Column 2 restricts the sample to only those districts that have been subject to court order. Column 3 represents only those districts that were under court order and eligible for treatment. Column 4 is the same sample, but the observations have been weighted by the number of unweighted full time equivalents in the district. Columns 5 and 6 follow the same procedure but only include the ten largest districts in the state, excluding Miami Dade.

PAGE 119

119 Table A 5. Fiscal variable m eans ( a ll variables) (1) (2) (3) (4) (5) (6) (7) (8) (9) AL FL GA LA MS NC SC ALL N Total Revenue 6,411 7,698 7,472 7,013 5,830 7,234 7,362 6,923 11,649 Total Federal Revenue 702 722 762 1,081 997 662 775 808 11,649 Fed Thru State: Child Nutrition Act 206 182 241 259 303 211 231 238 11,649 Total State Revenue 3,925 3,977 4,194 3,554 3,234 4,677 3,820 3,923 11,649 Total Local Revenue 1,784 3,000 2,515 2,378 1,600 1,895 2,768 2,193 11,649 Local: Property Taxes 591 2,395 1,636 918 1,097 0 2,091 1,295 10,686 Local: General Sales Taxes 0 0 379 1,101 0 0 0 198 10,686 Local: Public Utilities 0 0 0 0 0 0 0 0 10,686 Local: Individual and Corporate Income Taxes 0 0 0 0 0 0 0 0 10,686 Local: All Other Taxes 12 0 0 0 8 0 31 7 10,686 Local: From Other School Systems 8 2 9 3 12 2 4 7 11,649 Local: From Cities and Counties 575 0 163 30 16 0 129 152 11,649 Local: School Lunch 148 119 105 61 81 169 100 114 11,649 Local: Interest Earnings 62 102 74 124 56 22 84 69 11,649 Local: Miscellaneous 198 167 98 123 214 118 138 151 11,649 NCES Local Revenue, Census Bureau Revenue 0 0 0 0 0 0 0 0 11,649 Total Current Expenditures: Services 6,495 7,642 7,439 6,847 5,878 7,189 7,503 6,928 11,649 TCE: Instruction 3,440 3,495 4,017 3,563 3,132 3,914 3,694 3,623 11,649 TCE: Instruction 3,425 3,495 4,017 3,561 3,132 3,914 3,694 3,620 11,649 TCE: Support Services 1,821 2,360 2,033 2,143 1,701 2,009 2,209 1,989 11,649 TCE: Support Services (Pupils) 246 305 284 230 200 314 395 275 11,649 TCE: Support Services (Instructional Staff) 209 347 329 301 233 234 412 286 11,649 TCE: Support Services (General Admin) 187 128 164 168 206 179 117 170 11,649 TCE: Support Services (School Admin) 344 378 394 331 279 405 376 357 11,649 Total Current Expenditures (Other Elementary Secondary) 442 332 413 451 396 393 413 408 11,649 TCE: Food Services 438 332 395 446 366 393 368 392 11,649 Total Non-Elementary/Secondary Expenditures 121 162 20 30 23 44 96 62 11,649 Total Capital Outlay Expenditures 576 1,178 844 528 541 711 867 729 11,649 Capital Outlay: Construction 444 857 610 315 341 526 605 516 11,649 Capital Outlay: Land and Existing Structures 19 71 57 58 0 46 35 37 11,649 Total Current Expenditures for Elementary Education 5,703 6,186 6,463 6,157 5,230 6,316 6,316 6,020 11,649 Payments to State Governments 0 0 0 0 0 0 9 1 11,649

PAGE 120

120 Table A5. Continued Payments to Local Governments 0 0 0 0 0 0 0 0 11,649 Payments to Other School Systems 5 0 33 9 0 0 35 13 11,649 Interest on Debt 90 115 79 123 84 118 180 105 11,649 Total Salaries 3,624 3,871 4,242 3,841 3,301 4,301 4,059 3,885 11,649 Salaries: Instruction 2,481 2,381 2,929 2,533 2,279 2,933 2,707 2,629 11,649 Long Term Debt: Outstanding at Beginning of Fiscal Year 1,621 1,978 1,547 2,315 1,728 1,798 3,558 1,946 11,649 Long Term Debt: Issued During FY 309 367 348 303 253 391 826 379 11,649 Long Term Debt: Retired During FY 121 177 186 201 166 198 359 192 11,649 Long Term Debt Outstanding at End of Fiscal Year 1,812 2,168 1,707 2,421 1,814 1,994 4,028 2,133 11,649 Short Term Debt: Outstanding at Beginning of Fiscal Year 0 0 22 0 0 0 0 5 11,649 Short Term Debt: Outstanding at End of Fiscal Year 4 0 19 0 0 0 0 5 11,649 Assets: Sinking Fund 68 102 183 318 136 0 131 141 10,686 Assets: Bond Fund 522 292 862 452 406 2 554 527 10,686 Assets: Other Funds 867 1,234 994 1,470 1,218 1 873 1,009 10,686 Fed Thru State: Title I 234 204 212 304 333 156 211 238 10,869 Fed-Thru-State: Child with Disabilities Idea 117 130 81 118 117 95 137 109 10,869 Fed Thru State: Math, Science, and Teacher 24 16 18 31 22 14 23 21 10,869 Fed Thru State: Safe and Drub Free Schools 14 7 6 8 9 7 8 9 10,869 Fed Thru State: Title V, Part A 8 11 6 13 24 6 8 11 10,869 Fed Thru State: Vocational and Tech Education 25 21 13 17 14 16 27 18 10,869 Fed Thru State: Other 55 111 124 265 139 38 97 112 10,869 Federal Revenue Nonspecified 0 0 49 0 0 99 41 29 10,869 Fed Direct: Impact Aid 8 7 10 13 5 10 4 8 10,869 Fed Direct: Bilingual Education 2 1 0 1 0 1 1 1 10,869 Fed Direct: Indian Education 2 0 0 0 0 1 0 1 10,869 Fed Direct: Other 19 43 17 84 48 20 2 30 10,869 State Revenue: General Formula Assistance 3,322 1,633 3,398 3,333 2,808 3,932 1,076 2,942 10,869 State Revenue: Special Education Programs 25 577 0 37 2 170 396 123 10,869 State Revenue: Transportation Programs 210 207 0 0 0 85 63 71 10,869 State Revenue: Staff Improvement Programs 39 22 0 0 0 16 894 106 10,869 State Revenue: Compensatory and Basic Skills Programs 54 55 0 0 196 57 199 80 10,869 State Revenue: Vocational Education Programs 12 98 0 0 69 124 300 73 10,869 State Revenue: Capital Outlay And Debt Service Programs 140 331 132 0 18 60 131 108 10,869 State Revenue: Bilingual Education Programs 1 53 0 0 0 4 0 5 10,869

PAGE 121

121 Table A5. Continued State Revenue: Gifted and Talented Programs 0 73 0 0 0 9 35 11 10,869 State Revenue: School Lunch Programs 0 8 26 0 3 7 2 8 10,869 State Revenue: Other Programs 122 885 297 210 194 184 645 314 10,869 State Revenue On Behalf: Employee Benefits 60 0 80 3 0 0 6 29 10,869 State Revenue On Behalf: Not Employee Benefits 5 0 0 0 0 66 123 23 10,869 State Revenue: Not Specified 0 0 306 0 0 0 0 70 10,869 Local Revenue: Tuition Fees from Pupils and Parents 4 1 11 9 7 6 8 7 10,869 Local Revenue: Transportation Fees from Pupils and Parents 0 2 0 0 2 0 0 1 10,869 Local Revenue: Textbook Sales and Rentals 1 1 0 0 0 0 0 0 10,869 Local Revenue: District Activity Receipts 157 174 33 0 89 0 179 84 10,869 Local Revenue: Other Sales and Services 35 24 10 10 5 30 0 16 10,869 Local Revenue: Student Fees, Nonspecified 0 12 0 0 0 0 0 1 10,869 CE Support Services Operation and Maintenance of Plant 470 681 483 638 483 520 561 525 10,869 CE Support Services Student Transportation 248 301 262 372 228 221 208 255 10,869 CE Support Services Nonspecified 13 0 0 0 0 0 0 2 10,869 CE -Enterprise Operations 0 0 9 0 1 0 22 5 10,869 CE ELSEC 4 0 8 5 29 0 13 10 10,869 Non Elementary/Secondary Expenditures: Community Services 39 46 12 9 4 45 45 26 10,869 Non-Elementary/Secondary Expenditures: Adult Education 16 102 8 21 5 1 51 22 10,869 Non Elementary/Secondary Expenditures: Other 72 1 0 0 12 0 0 14 10,869 Capital Outlay: Instructional Equipment 41 4 65 78 70 68 67 59 10,869 Capital Outlay: Other Equipment 73 250 89 73 122 75 128 107 10,869 Capital Outlay: Nonspecified Equipment 0 0 23 8 14 0 36 12 10,869 Salaries Support Services Pupils 157 211 209 173 150 233 223 191 10,869 Salaries Support Services Instructional Staff 132 235 199 218 134 157 254 181 10,869 Salaries -Support Services-General Administration 104 59 93 49 119 107 50 91 10,869 Salaries Support Services School Administration 258 281 292 251 216 319 290 271 10,869 Salaries Support Services Operation and Maintenance of Plant 137 229 157 176 115 192 175 160 10,869 Salaries Support Services Student Transportation 131 164 151 195 111 135 100 138 10,869 Salaries Food Services 157 113 146 179 112 147 126 140 10,869 Total Employee Benefits 1,030 1,142 1,188 1,154 879 971 1,100 1,057 10,869 Employee Benefits: Instruction 688 675 816 761 583 661 710 702 10,869 Employee Benefits Support Services Pupils 43 60 49 44 36 51 57 47 10,869 Employee Benefits Support Services Instructional Staff 35 62 60 54 33 35 65 47 10,869 Employee Benefits Support Services General Administration 24 22 28 23 29 25 21 25 10,869

PAGE 122

122 Table A5. Continued Employee Benefits Support Services School Administration 70 78 90 71 52 73 74 73 10,869 Employee Benefits Support Services Operation and Maintenance of Plant 51 81 41 47 39 45 55 48 10,869 Employee Benefits Support Services Student Transportation 59 63 41 65 40 22 32 44 10,869 Employee Benefits: Food Service 66 45 33 64 46 42 44 47 10,869 Employee Benefits: Enterprise Operations 0 0 0 0 0 0 1 0 10,869 Total Students 5,699 34,988 7,560 11,187 3,353 10,389 7,894 9,567 11,649 NOTE: Columns show per -capita values for different categories of expenditure and revenues in 2000 dollars. Values have been deflated using the CPI -U. Columns 1-7 show state-by -state means. Column 8 is the mean value for the entire sample. Column 9 sho ws the number of observations available for each variable. There are 11,649 district -level observations across the 15 years of data in the sample. Variables with 10,869 observations lack data from 1990. The three Asset variables are missing observations starting in 1999 which results in the lower number of observations for those three variables.

PAGE 123

123 LIST OF REFERENCES Ampadu, Osei, and Lei Zhou. 2009. Documentation for the NCES Common Core of Data School District Finance Survey (F33), School Year 2006 07 (Fiscal Year 2007) (NCES 2009339) Washington D.C.: U.S. Department of Education. Retrieved June 30, 2010 from http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2009339. Chemerinsky, Erwin. 2009. The Segregation and Resegregation of American Public Education in School Resegregation, Must the South Turn Back eds. John C. Boger and Gary Orfield Chapel Hill: The Univ ersity o f North Carolina Press. Clotfelter, Charles T. 2004. After Brown: The Rise and Retreat of School Desegregation. Princeton: Princeton University Press. Clotfelter, Charles T., Helen F. Ladd and, Jacob L. Vigdor. 2005. Federal Oversight, Local Control and the Specter of Resegregation in Southern Schools. NBER Working Paper Series W11086. Dries, Bill. County Schools Move on After Desegregation Case. The Memphis Daily News May 26, 2009. Feng, Li, David N. Figlio and Tim Sass. 2010. School Accountability and Teacher Mobility. NBER Working Paper Series, W16070. Florida State Advisory Committee to the United States Commission on Civil Rights. 2007. Desegregation of Public School Districts in Florida: 18 Public School Districts Have Unitary Status 16 Districts Remain Under Court Jurisdiction. A Report to the United States Commission on Civil Rights Washington D.C. Guryan, Jonathan. 2004. Desegregation and Black Dropout Rates. American Economic Review. 94(4) : 919943. Holley Wa lker, Danielle. 2010. After Unitary Status: Examining Voluntary Integration Strategies for Southern School Districts. North Carolina Law Review 88(3): 877910. Hughes, Bayne. The $814,187 Question. The Decatur Daily News July 18, 2005. Janssen, Julie. 2001. An Analysis of the Legal and Historical Context of the Pinellas County School Districts Separation from Court Ordered Desegregation Established in Bradley v. Board of Public Instruction. Unpublished Ph.D. Dissertation, University of South Florida, Tampa, FL. Lutz, Byron F. 2005. Post Brown vs. the Board of Education: The Effects of the End of Court Ordered Desegregation. Finance and Economics Discussion Series Federal Reserve Board.

PAGE 124

124 Martin, Stephanie M. 2010. Are Public School Teacher Salaries Paid Compensating Wage Differentials for Student Racial and Ethnic Characteristics? Education Economics. Forthcoming. Moore, Monika L. 2002. Unclear Standards Create an Unclear Future: Developing a Better Definition of Unitary Status. The Yale Law Journal 112(2): 311351. Orfield, Gary. 1996. Turning Back to Segregation. In Dismantling Desegregation: The Quiet Reversal of Brown v. Board of Education, eds Gary Orfield and Susan E. Eaton. New York: The New Press. Pinellas County School Board, Annual Budget Summary for 20002001. Last accessed at http://www.pcsb.org/budget/0001.html on May 6, 2010. Reber, Sarah J. 2007. School Desegregation and Educational Attainment for Blacks NBER Working Paper Series W13193. Rivkin, Steven G. 2000. School Desegregation, Academic Attainment, and Earnings. The Journal of Human Resources 3 5(2) : 333 346. Ryan, James E. 1999. Schools, Race and Money. The Yale Law Journal 109( 2): 249316. Schofield, Janet W. 2005 Review of Research on School Desegregations Impact on Elementary and Secondary School Students. In Handbook of Research on Multicultural Education eds. James A Banks and C herry A. McGee, San Francisco: John Wiley and Sons Spinner, Jackie. Prince Georges County to Pay Legal Costs for NAACP as Part of Settlement. The Washington Post March 9, 1999. United States Com mission on Civil Rights. 2007. Becoming Less Separate? School Desegregation, Justice Department Enforcement and the Pursuit of Unitary Status Washington D.C.

PAGE 125

125 BIOGRAPHICAL SKETCH Colin Knapp was born and raised in St. Petersburg, Florida. After receiving degrees in economics and finance from Ohio University, he served honorably for 11 years in the U nited States Air Force in various academic and financial management positions. Upon leaving the military, he returned to the University of Fl orida to complete his doctorate. He is married, has one daughter, and considers himself very lucky on both accounts.